UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Proton spin relaxation as a means of characterizing the pathology of multiple sclerosis Stewart, Wendy Anne 1989

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
831-UBC_1989_A1 S73.pdf [ 29.46MB ]
Metadata
JSON: 831-1.0059449.json
JSON-LD: 831-1.0059449-ld.json
RDF/XML (Pretty): 831-1.0059449-rdf.xml
RDF/JSON: 831-1.0059449-rdf.json
Turtle: 831-1.0059449-turtle.txt
N-Triples: 831-1.0059449-rdf-ntriples.txt
Original Record: 831-1.0059449-source.json
Full Text
831-1.0059449-fulltext.txt
Citation
831-1.0059449.ris

Full Text

P R O T O N SPIN R E L A X A T I O N AS A MEANS OF C H A R A C T E R I Z I N G T H E P A T H O L O G Y OF MULTIPLE SCLEROSIS By Wendy Anne Stewart B. Sc. (Chemistry) University of Dundee M . Sc. (Chemistry) University of British Columbia A THESIS  S U B M I T T E D IN P A R T I A L  T H E REQUIREMENTS DOCTOR  FULFILLMENT OF  FOR T H EDEGREE OF  OF  PHILOSOPHY  in THE FACULTY OF GRADUATE  STUDIES  CHEMISTRY-  We accept this thesis as conforming to the required standard  THE  UNIVERSITY  O F BRITISH  COLUMBIA  March 1989 © Wendy Anne Stewart, 1989  In  presenting  degree freely  this  at the  thesis  in  University of  available for reference  copying  of  department publication  this or of  partial  British Columbia, and study.  thesis for scholarly by  this  his  or  her  of  the  Department The University of British Columbia Vancouver, Canada  requirements  I agree  that the  I further agree  purposes  representatives.  may be It  thesis for financial gain shall not  permission.  DE-6 (2/88)  fulfilment  is  that  an  advanced  Library shall make it  permission for extensive  granted  by the  understood be  for  allowed  that without  head  of  my  copying  or  my written  Abstract  Multiple sclerosis (MS) is a demyelinating disease of the central nervous system; the cause(s) of the disease and its pathological progression are at present unknown. Magnetic resonance imaging (MRI), in particular the spin-echo radiofrequency pulse sequence, proved a powerful tool in the study of MS. The question addressed in this thesis, is whether quantitative measurement of the spin-lattice (Xi) and spin-spin (T ) relaxation 2  times can provide a means of characterizing the pathology of MS in vivo; the final goal being to use these parameters to differentiate acute inflammatory lesions from more chronic demyelinating and gliotic ones. Quantitative M R imaging methods, using twopoint determinations of Ti and T2, were first used to study post mortem and fixed brain tissue from MS patients. These data demonstrated differences in Ti and T with varying 2  degrees of demyelination and gliosis; however, chronic MS pathology is similar in all cases, and provides little information on the earlier stages of the disease process which are believed to be immune-mediated. The early immune-mediated lesion was then examined by studying experimental allergic encephalomyelitis (EAE), an animal model for MS, in primates.  The data obtained from these studies confirmed that the values of Ti and  T obtained from these two-point determinations are sensitive to variable pathology, but 2  cannot be used to distinguish acute from chronic lesions. An in-depth in vitro T2 study was then carried out on guinea pigs inoculated to produce EAE; this was followed by repeat measurements on the formalin fixed tissue. T2 measurements were also carried out on samples from the fixed MS tissue already studied using quantitative imaging methods. The relaxation data were analyzed using two different methods; the first assuming a fixed number of exponentials (minuit), and the second a continuous distribution of relaxation ii  times (NNLS).  It is clear from these studies that the models currently used to describe  the relaxation behaviour of water protons in tissue are inadequate to explain the results observed.  Both analysis methods were sensitive to the different types of pathology  present; however, assuming a continuous distribution of relaxation times not only provides a better fit to the data, but also represents a more realistic picture of the complex physical system being studied. These results show that T2 data obtained from tissue are reproducible, and can distinguish differing degrees of inflammation.  The same trends  observed in vitro were also observed in the fixed tissue; comparison of the fixed tissue data from inflammatory lesions with that from M S tissue showed that T2 can be used to differentiate acute and chronic lesions. This implies that quantitative M R measurements, together with the sophisticated methods of analysis employed in this thesis, can be used to follow the pathological progression of MS; this would provide insight into the disease process and lead eventually to an understanding of the cause(s) of the disease.  iii  Table of Contents  Abstract  ii  List of Tables  vii  List of Figures  xiv  List of Abbreviations  xxv  Acknowledgements  xxx  1  General Introduction and Thesis Goal  1  1.1  General Background  1  1.2  Basic Principles of Magnetic Resonance  9  1.3  M R Imaging and Image Enhancement  13  1.3.1  Slice Selection  15  1.3.2  Image Enhancement  17  1.4  2  Relaxation Measurements  19  1.4.1  Measurement of T i  19  1.4.2  Measurement of T2  20  1.4.3  Imaging Methods  21  M R I of Multiple Sclerosis  24  2.1  Introduction  24  2.2  Methods  30  iv  2.3  Results  3 5  2.3.1  Case 1  35  2.3.2  Case 2  42  2.3.3  Case 3  46  2.3.4  Case 4  50  2.4  Summary of Results  54  2.5  Discussion  59  3 MRI of Experimental Allergic Encephalomyelitis in Primates  64  3.1  Introduction  64  3.2  Methods  66  3.2.1  Induction of EAE  66  3.2.2  MRI Protocol .  69  3.2.3  Pathology  73  3.3  Results  7 4  3.3.1  Case 1 . . .  74  3.3.2  Case 2  80  3.3.3  Case 3  8 3  3.3.4  Case 4  91  3.3.5  Cases 6 and 7: Gd-DTPA Study  98  3.3.6  Case 8  98  3.3.7  Case 9  1 0 6  3.3.8  Case 10  1 1 3  3.4  Summary of Results  3.5  Discussion  121 • • • 124  v  4  In Vitro Quantitative Relaxation Measurements  130  4.1  Practical Aspects  130  4.2  Tissue Relaxation Models  131  4.3  Methods  138  4.4  4.5 5  4.3.1  T-i Measurements  138  4.3.2  Ti Measurements  140  4.3.3  Analysis of Relaxation Data  141  4.3.4  Spectroscopy  143  4.3.5  Protocol for Studying Tissue  143  Results  148  4.4.1  Guinea Pig Tissue  148  4.4.2  Primate Tissue  164  4.4.3  Fixed M S Tissue  169  Discussion  176  Conclusions and Suggestions for Further Work  191  5.1  Conclusions . . .  191  5.2  Suggestions for Further Work  194  A Extraction of Myelin Basic Protein from C N S Tissue  196  B Results of N N L S Analysis of Guinea Pig C N S Tissue  197  References  200  vi  List of Tables  2.1  Table showing details of scoring for pathology of sections taken from the inferior surface of each 10 mm thick brain slice. M N C - mononuclear cells, M O - macrophages  2.2  38  Pathological score for the three lesions seen in the right cerebral hemisphere of slice 3, case 1  2.3  40  Quantitative M R I data from fixed brain in case 1. T\N and T2N correspond to the average values of T\ and T2 from normal appearing white matter in the same slice  2.4  41  Correlation between lesion area as seen on post mortem and fixed brain M R I and the actual lesion area seen on the inferior surface of each 10 mm thick brain slice in case 2  2.5  42  Pathological score for section taken from the inferior surface of slice 4, case 2  2.6  44  Ti and T values from M R I number 5, which is equivalent to the inferior 2  surface of fixed brain slice 4, case 2. The area from which these values were obtained corresponds to that from where the section was taken. Cadaver T  1N  = 342± 30 ms(18), cadaver T  2N  = 217± 10 ms(6) and fixed brain T  2N  = 84± 7 ms(17), fixed brain T = 6 6 ± 7 ms(22).  lN  The numbers  in brackets give the number of areas from which the mean and standard deviation were determined  45  vii  2.7  Correlation between lesion area as seen on post mortem and fixed brain M R I and the actual lesion area seen on the inferior surface of each 10 mm thick brain slice in case 3  2.8  46  Pathological score for sections taken from fixed brain slices 5 and 6, case 3. L - left cerebral hemisphere and R - right cerebral hemisphere  2.9  49  T\ and T2 values from areas examined microscopically. Post mortem TIN = 368± 21(24), T  2N  = 91± 9(7), and fixed brain T  1N  = 218± 10(18), T  2N  = 71± 6(24). Bracketed numbers denote the number of areas from which the mean and standard deviation were determined.  P M - post mortem,  F B - fixed brain  49  2.10 Comparison of lesion area seen on the post mortem and fixed brain M R I in case 4. Slice 1 is the sagittal pilot scan and is not included  50  2.11 Pathological score for lesions taken from slices 5 and 6, case 4.  L - left  cerebral hemisphere, and R - right cerebral hemisphere  53  2.12 T i and T2 values from area examined microscopically in case 4. mortem T  IN  = 381± 12(14), T  14(20), T N = 6 5 ± 5(24). 2  2N  = 77± 9(20) and fixed brain T  1N  Post  = 207±  The numbers in brackets give the number of  areas from which the mean and standard deviation were determined. P M - post mortem, F B - fixed brain  53  2.13 Quantitative area measurements from individual lesions. P M - post mortem, F B - fixed brain, B R A I N - data from digitized brain slice  54  2.14 Quantitative M R I data from normal appearing white matter as seen on the post mortem and fixed brain SE M R I  viii  55  2.15 T i and T values from computed images, correlated with severity of de2  myelination in cases 1-4.  Lesions are denoted by the same numbers as  those used in the text. P M - post mortem, F B - fixed brain, A R E A - size and number of lesions  56  2.16 T i and T values from computed images, correlated with severity of gliosis 2  in cases 1-4.  Lesions are denoted by the same numbers as those used in  the text. P M - post mortem and F B - fixed brain  57  2.17 T i and T values from computed images, correlated with severity of in2  flammation in cases 1-4. those used in the text.  Lesions are denoted by the same numbers as P M - post mortem, F B - fixed brain, A R E A -  denotes size and number of lesions  58  2.18 Mean intensity and standard deviation from regions consisting of 18 pixels on a computed T image, from regions of normal appearing white matter. 2  P M - post mortem and F B - fixed brain  60  2.19 Post mortem Ti and T values from normal appearing white matter in case 2  4, showing the effects of multi-slice interference. The numbers in brackets give the number of R O I used to determine the mean and error 3.20 Clinical scoring for monkeys inoculated to produce E A E  63 67  3.21 T i and T values from lesions in cases 1, 2, 3, 4 and 9 obtained at the time 2  of death  123  4.22 Pathological scoring for guinea pig CNS tissue  ix  145  4.23 Results from minuit analysis of CPMG data obtained from the cervical spine of normal control guinea pigs. T (ms) values are given together 2  with the percentage contribution of each component to the signal observed. Pathological examination of this tissue confirmed that each section was normal  150  4.24 Results from minuit analysis of CPMG data obtained from the cervical spine of guinea pigs inoculated to produce E A E . T (ms) values are given 2  together with the percentage contribution of each component to the signal observed  151  4.25 Pathology from the cervical spine of guinea pigs inoculated to produce EAE.  PVI-SS - perivascular inflammation in the subarachnoid space,  PVI-VR - perivascular inflammation in the Virchow-Robin spaces, PCI - parenchymal cellular infiltration, D - demyelination and E - edema. . .  152  4.26 Results from minuit analysis of CPMG data from the upper thoracic spine of normal control guinea pigs. T (ms) values are given together with the 2  percentage contribution of each component to the total signal observed. .  152  4.27 Results from minuit analysis of CPMG data from the upper thoracic spine of guinea pigs inoculated to produce EAE. T (ms) values are given to2  gether with the percentage contribution of each component to the total signal observed  153  4.28 Pathology from the upper thoracic spine of guinea pigs inoculated to produce E A E . PVI-SS - perivascular inflammation in the subarachnoid space, PVI-VR - perivascular inflammation in the Virchow-Robin spaces and PCI - parenchymal cellular infiltration  153  x  4.29 Results from  minuit  analysis of CPMG data from the lower thoracic spine  of normal control guinea pigs. T (ms) values are given together with the 2  percentage contribution of each component to the total signal observed. . 4.30 Results from  minuit  154  analysis of CPMG data from the lower thoracic spine  of guinea pigs inoculated to produce E A E . r (ms) values are given to2  gether with the percentage contribution of each component to the total signal observed  154  4.31 Pathology from the lower thoracic spine of guinea pigs inoculated to produce E A E . PVI-SS - perivascular inflammation in the subarachnoid space, PVI-VR - perivascular inflammation in the Virchow-Robin spaces and PCI - parenchymal cellular infiltration 4.32 Results from  minuit  155  analysis of CPMG data from the lumbar spine of  normal control guinea pigs. T2 (ms) values are given together with the percentage contribution of each component to the total signal observed. . 4.33 Results from  minuit  155  analysis of CPMG data from the lumbar spine of  guinea pigs inoculated to produce E A E . T2 (ms) values are given together with the percentage contribution of each component to the total signal observed  156  4.34 Pathology from the lumbar spine of guinea pigs inoculated to produce E A E . PVI-SS - perivascular inflammation in the subarachnoid space, PVIV R - perivascular inflammation in the Virchow-Robin spaces and PCI parenchymal cellular infiltration  156  4.35 Results from minuit analysis of CPMG data from the sacral spine of normal control guinea pigs. T (ms) values are given together with the percentage 2  contribution of each component to the total signal observed  xi  157  4.36 Results from minuit analysis of CPMG data from the sacral spine of guinea pigs inoculated to produce EAE. The T (ms) values are given together 2  with the percentage contribution of each component to the total signal observed  157  4.37 Pathology from the sacral spine of guinea pigs inoculated to produce E A E . PVI-SS - perivascular inflammation in the subarachnoid space, PVIV R - perivascular inflammation in the Virchow-Robin spaces and PCI parenchymal cellular infiltration  158  4.38 Results from minuit analysis of CPMG data from the left cerebral hemisphere of guinea pigs inoculated to produce EAE. T (ms) values are given 2  together with the percentage contribution of each component to the total signal observed  158  4.39 Pathology from the left hemisphere of guinea pigs inoculated to produce EAE. PVI-SS - perivascular inflammation in the subarachnoid space, PVIV R - perivascular inflammation in the Virchow-Robin spaces and PCI parenchymal cellular infiltration  159  4.40 Results obtained from minuit analysis of CPMG data from fixed guinea pig lumbar spine. T values are given together with the percentage con2  tribution of each component to the signal observed  165  4.41 Results obtained from minuit analysis of CPMG data from fixed guinea pig lower thoracic spine. T values are given together with the percentage 2  contribution of each component to the signal observed 4.42 Results from minuit analysis of CPMG data from primate case 8.  166 T  2  (ms) values are given together with the percentage contribution of each component to the total signal observed. W M - white matter, G M - grey matter, L - lesion, L M B - left midbrain xii  166  4.43 T values obtained using minuit to analyze CPMG data from fixed MS 2  tissue. Lesions are numbered as they were in chapter 2, and W M - normal appearing white matter, G M - normal appearing grey matter and G M / W M - mixture of grey and white matter. Two values are given for case 4 W M , the data were obtained six months apart  169  4.44 Results from minuit analysis of CPMG data from the homogenized brain and spine of a normal guinea pig.  T values (ms) are given together 2  with the percentage contribution of each component to the total signal observed. B - brain tissue and C / T - cervical/thoracic spine. Two sets of data are given for C / T , determined 4 hours apart  xiii  180  List of Figures  1.1  Diagram of a nerve cell or neuron  1.2  The behaviour of spins in the presence of two orthogonal magnetic fields. A. Precession about B . Q  1  B. Behaviour of spins after the application of a  second magnetic field, B\, perpendicular to B  10  a  1.3  Diagram showing the variation of the static magnetic field, B in the presa  ence of a linear field gradient, G 1.4  13  x  Two-dimensional Fourier transform spin-echo pulse sequence for slice selective imaging  1.5  14  Diagram showing selective excitation of a slice. A. Selective rf pulse and slice select gradient. B. Magnetization in the z-direction after application of selective pulse. C. Selected slice  2.6  Diagram showing the simultaneous excitation of normal and abnormal tissue in a 10 mm  2.7  16  thick slice due to the irregular shape of the lesion.  ..  29  A. Multi-slice spin-echo pulse sequence showing variable time parameters. B. Multi-slice inversion-recovery pulse sequence, n = the number of slices in both sequences  31  xiv  2.8  A . Sagittal pilot scan obtained post mortem. Grid shows the center of each of the 12 transverse images to be obtained.  B . Sagittal pilot scan  from fixed brain. Grid shows the center of each of the 12 coronal images to be obtained.  The coronal images of the fixed brain are equivalent to  the transverse images obtained in situ. T E = 26 ms and T R = 400 ms in both images 2.9  33  Comparison of the inferior surface of a 10 mm thick brain slice in case 1 with the corresponding M R I . A . Spin echo M R I level 3, T E = 40 ms and the T R = 2 sec. B . Inferior surface of fixed brain slice at the same level.  36  2.10 Display of SE M R I level 3 together with the equivalent digitized negative of fixed brain slice on the Vax 11/750 computer. The lesions are outlined on both the image and the digitized brain slice.  Lesion area on M R I =  489 m m and actual lesion area on brain slice = 562 m m 2  2  37  2.11 Comparison of gross pathology with the spin echo M R I . A . SE M R I level 4, T E = 40 ms and T R = 2 sec. B . Inferior surface of 10 mm thick fixed brain slice at the equivalent level  39  2.12 Section taken from slice 3, case 1 stained with P T A H . The section shows lesion 8, with cystic spaces 2.13 Comparison of the SE M R I with the gross pathology.  40 A . Post mortem  SE M R I at level 4. T E = 40 ms and T R = 2 sec. B . Fixed brain SE M R I at equivalent level. Parameters as in A . C. Inferior surface of fixed brain slice at the equivalent level  43  2.14 Inferior surface of fixed brain slice 4, case 2. The box shows the position of the section taken for microscopic examination  44  2.15 Comparison of SE image with the gross pathology.  A. Post mortem SE  MRI at level 5. T E = 40 ms and TR = 2 sec. B. Fixed brain SE MRI at the equivalent level. Parameters as in A. C. Inferior surface of fixed brain slice at the equivalent level  47  2.16 Comparison of SE image with the gross pathology.  A. Fixed brain SE  MRI at level 6. Tau = 40 ms and TR = 2 sec. B. Fixed brain SE MRI at the equivalent level. Parameters as in A. C. Inferior surface of fixed brain slice at the same level  48  2.17 SE MRI from case 4. A. Post mortem MRI level 4. B. Fixed brain MRI at the same level. In both images tau = 40 ms and T R = 2 sec  51  2.18 A. Inferior surface of fixed brain slice 5, case 4. Box shows the position of the section that was removed for microscopic examination.  B. Fixed  brain slice 6, case 4, showing the two sections removed for microscopic examination  52  2.19 Computed Ti and T images from the fixed brain in case 3. A. T\ image. 2  B. T image  61  2  3.20 A. Sagittal pilot scan of monkeys brain. The grid shows the position of the 8 transverse slices to be obtained. B. Sagittal pilot scan of fixed monkey brain. The grid shows the position of the 8 coronal slices to be obtained. The coronal slices of the fixed brain are equivalent to transverse images in vivo. In both images T E = 26 ms and TR = 800 ms 3.21 Image of monkey brain showing lesions outlined using a trackball  70 71  3.22 Serial SE oo/4o MRI obtained from case 1. A. Day 15 after inoculation. 20  B. Day 18 after inoculation  75  xvi  3.23 Comparison of the gross distribution of lesions with the equivalent MRI in case 1. A. SE2000/40 image on day 23 showing lesion in the brain stem. B. The inferior surface of the 5 mm thick brain slice at the same level. .  77  3.24 Histopathology of lesion in case 1 A. H & E showing diffusion of polymorphonuclear cells, lymphocytes and macrophages into the tissue around the vessel. B. LFB- PAS-H) showing LFB-positive myelin debris at the centre of the lesion. C. Holmes' stain showing swelling, beading and destruction of axons at the centre of the lesion  78  3.25 ORO-H of frozen section showing macrophages filled with granular sudanophilic lipids in case 1  79  3.26 Plot of T\ and T2 versus time after inoculation in case 1. Ti and T2 values are given from normal appearing white matter plus or minus two standard deviations  79  3.27 Comparison of the SE MRI with the gross pathology in case 2. A.  SE2000/40  MRI on day 22 after inoculation. B. The inferior surface of 5 mm thick brain slice at the same level  81  3.28 Histopathology of lesion in case 2. Section stained with H & E showing the collar of hemorrhage around the vessel displacing the exudate peripherally. 82 3.29 Serial  SE2000/40  MRI from case 3. A. Day 11 after inoculation. B. Day  12 after inoculation  84  3.30 SE MRI from fixed brain. The lesions in both hemispheres are pale compared to the bright areas seen on the in vivo images  85  3.31 Comparison of the SE MRI with the gross pathology in case 3. A. SE ooo/6o 2  MRI slice 4. B. The inferior surface of 5 mm thick brain slice at the same level  86  xvii  3.32 Comparison of SE M R I with the gross pathology in case 3. A . SE ooo/40 2  M R I at level 3.  B . The inferior surface of 5 mm thick brain slice at the  same level  87  3.33 Comparison of the SE M R I with the gross pathology in case 3. A . SE ooo/4o 2  M R I slice 5. B . The inferior surface of 5 mm thick brain slice at the same level  88  3.34 Histopathology of lesion observed in the left cerebral hemisphere at M R I level 3. Section stained with H & E showing compact exudate similar to that seen in cases 1 and 2  89  3.35 Plot of T i and T versus time after inoculation for case 3. 2  The dashed  line denotes data from normal tissue plus or minus two standard deviations. 90 3.36 SE oo/4o obtained from case 4 on day 16 after inoculation  91  20  3.37 Serial SE oo/40 M R I from case 4. 20  A . Day 16 after inoculation.  B . Day  18 after inoculation. C. Day 19 after inoculation  92  3.38 Fixed brain SE oo/40 M R I from case 4 showing abnormal area in the left 20  cerebral hemisphere.  This gliotic lesion appears as a bright area of in-  creased T , just as was seen in the fixed brains of chronic MS patients. 2  3.39 Comparison of the SE M R I with the gross pathology in case 4. mortem SE oo/6o M R I at level 5. 20  .  93  A . Post  B . The inferior surface of 5 mm thick  brain slice at the same level  94  3.40 Histopathology of posterior and medial part of lesion in the left visual radiation of case 4. Section stained with gallo-DR  96  3.41 Histopathology of lesion in the corpus callosum of case 4. Section stained with gallo-DR  97  xviii  3.42 Plot of Ti and T versus time after inoculation for case 4. Ti and T values 2  2  from normal appearing white matter are given plus or minus two standard deviations  97  3.43 SE MRI from case 7 on day 22 after inoculation. A . Slice 5. B. Slice 6. .  99  3.44 Gd-DTPA scan from case 7 showing breakdown in the blood brain barrier where lesion is enhanced in slice 5. A. Slice 5. B. Slice 6. In both images T E = 26 ms, TR = 500 ms and number of P E increments = 256  100  3.45 SE MRI showing preclinical lesions on day 26 after inoculation in case 8. A. MRI level 4. B. MRI level 5. C. MRI level 6  101  3.46 SE MRI at level 5 from case 8, showing spread of lesions into left cerebral hemisphere  102  3.47 Plot of total lesion area versus time after inoculation in case 8  103  3.48 Plot of total slice lesion area versus time after inoculation for two SE MRI slices in case 8  104  3.49 Plot of Ti versus time after inoculation for for two areas of the lesion in slice 5, case 8  104  3.50 SE MRI from case 9. A. Slice 3 on day 26 after inoculation. B. Stice 3 on day 29 after inoculation. C. Slice 5 on day 29 after inoculation 3.51 Plot of individual lesion area versus time after inoculation in case 9. sion in left cerebral hemisphere slice 3.  107 Le-  Lesion in left cerebral hemisphere  slice 4  108  3.52 Plot of total lesion area versus time after inoculation in case 9  108  3.53 Comparison of the MRI with the gross pathology. A. SE MRI at level 4. B. The inferior surface of 5 mm thick brain slice at the equivalent level. .  109  3.54 A. Plot of Ti versus time after inoculation for lesion in slice 4. B. Plot of T versus time after inoculation for the same lesion 2  xix  Ill  3.55 A . Plot of Ti versus time after inoculation for cerebellar lesion in slice 7. B . Plot of T versus time after inoculation for the same lesion  112  2  3.56 SE M R I on day 52 after inoculation. A . Slice 4. B . Slice 5  114  3.57 A . SE M R I level 4 on day 86 after inoculation, showing new lesion in the right cerebral hemisphere. B . SE M R I level 4 on day 103 after inoculation. In both images T E = 120 ms and T R = 2 sec  115  3.58 Plot of individual lesion area versus time after inoculation in slice 4, case 10. This plot shows the appearance of the new lesion in the left cerebral hemisphere on day 86  116  3.59 Plot of total lesion area versus time after inoculation for case 10. Data is shown to day 135; little change was observed in the brain after this point. 3.60 A . Inferior surface of fixed brain slice showing area of relapse.  117  B . The  inferior surface of fixed brain slice equivalent to M R I level 5  118  3.61 A.Plot of Ti versus time after inoculation for the new lesion in the right cerebral hemisphere in M R I slice 4. B . Plot of T versus time after inoc2  ulation for the same lesion  120  4.62 Vector diagram showing the interaction of two spins 4.63 Diagram showing the possible transitions for two coupled spins.  132 0 - zero  quantum transition, 1 - single quantum transition and 2 - double quantum transition  134  4.64 C P M G decay curve for CNS tissue 4.65 F I D from lesion 1, case 4, using the modified IR pulse sequence.  139 The  pretrigger and data regions are labelled, together with the region of interest chosen. The tau value was 1 ms  140  4.66 Plot of signal amplitude versus tau for lesion 1 from case 4. . .  xx  141  4.67 A. Diagram showing division of spinal cord into five sections denoted by their approximate level in the spinal column. B. The tissue was gently rolled onto the small diameter glass rod and placed in the bottom of the 10 mm NMR tube. C - cervical, UT - upper thoracic, LT - lower thoracic, L - lumbar, S - sacral and L H - left hemisphere  144  4.68 Diagram showing the right cerebral hemisphere of the superior surface of brain slice 4 in case 1. This is the opposite face to that shown in figure 2.29, page 38. Sections were taken from areas corresponding to lesions 8 and 10. Samples of normal appearing grey and white matter were also removed  146  4.69 A. Diagram showing the superior surface of slice 7 in case 4, the opposite face to that shown in figure 2.18B, page 52. Sections were taken from the areas corresponding to lesions 1 and 3 as indicated. Samples of normal appearing grey and white matter, as indicated on the MRI, were also removed.  B. Diagram showing the superior surface of slice 6 in case 4,  the opposite face to that shown in figure 2.18A, page 52. Sections were taken from the areas corresponding to lesion 2 as indicated  147  4.70 Histologic section stained with H & E, from the sacral spine of guinea pig 14, showing a large cuff of perivascular inflammation in the subarachnoid space  149  4.71 Histologic section stained with H & E, from the upper thoracic spine of guinea pig 13, showing a perivascular cuff in a Virchow-Robin space.  . .  149  4.72 Histologic section stained with H & E, from the lower thoracic spine of guinea pig 16, showing extensive parenchymal cellular infiltration. Comparison of this section with the one shown in figure 4.72 shows the hypercellularity in the tissue around the vessel xxi  150  4.73 Summary of minuit data for spinal cord. Plot of percentage contribution to signal versus T value for guinea pig spinal cord. A . Cervical. B . Upper 2  thoracic. C. Lower thoracic. D . Lumbar. E . Sacral. controls,  tissue from normal  abnormal tissue  160  4.74 Plot of amplitude versus T value for the lower thoracic spine of guinea 2  pig 16  162  4.75 Plot of amplitude versus T value for the lumbar spines of guinea pigs 2, 2  7 and 9 (normal) and 13, 16 and 11 (abnormal)  163  4.76 Plot of amplitude versus T value (sec) for fixed guinea pig lumbar spine. 2  Guinea pig number is given in each spectrum; numbers 2, 7 and 9 are normal, and 13, 16 and 11 are abnormal. The peak around 1 sec corresponds to the T of free water in the formalin solution around the tissue. The 2  magnitude and T of this peak varied, depending on the amount of water 2  present  167  4.77 Plot of amplitude versus T value for normal appearing white matter in 2  primate case 8  168  4.78 Plot of amplitude versus T value for lesion 3 in primate case 8  168  4.79 Plot of amplitude versus T value for the left midbrain in primate case 8.  168  2  2  4.80 Plot of amplitude versus T value for normal appearing white matter in 2  case 1  170  4.81 Plot of amplitude versus T value for normal appearing grey matter in case 2  1  170  4.82 Plot of amplitude versus T value for normal appearing grey/white matter 2  mixture in case 1  171  4.83 Plot of amplitude versus T value for lesion 8 in case 1  171  4.84 Plot of amplitude versus T value for lesion 10 in case 1  172  2  2  xxii  4.85 Plot of amplitude versus T value for normal appearing white matter in 2  case 4  172  4.86 Plot of amplitude versus T value for normal appearing grey matter in case 2  4  173  4.87 Plot of amplitude versus T value for lesion 1 in case 4  173  4.88 Plot of amplitude versus T value for lesion 2 in case 4  174  4.89 Plot of amplitude versus T value for lesion 3 in case 4  174  2  2  2  4.90 Plot of amplitude versus T\ value for normal appearing white matter in case 1  175  4.91 Plot of amplitude versus T\ value for lesion 1 in case 1  175  4.92 A . Plot of amplitude versus T i value for normal appearing white matter in case 4. B . Plot of amplitude versus T\ value for lesion 1 in case 4.  . .  175  4.93 Plot of log of signal amplitude versus time after the 90° pulse in a C P M G sequence, demonstrating non- exponential behaviour of brain water proton relaxation. A . Manganese chloride solution. B . Brain tissue data fitted to one exponential. C. Brain tissue data fitted to two exponentials. D . Brain tissue data fitted to three exponentials.  177  4.94 Plot of percentage contribution to signal versus T for normal and ab2  normal guinea pigs.  predominantly perivascular inflammation in the  subarachnoid and Virchow-Robin spaces. lular infiltration.  extensive parenchymal cel-  normal tissue  180  4.95 Plots of amplitude versus T value for a smooth fit to guinea pig lumbar 2  spinal cord C P M G data. The figure on each spectrum denotes the guinea pig number; numbers 2,7 and 9 are normal and 13,11 and 16 are abnormal. 182 4.96 Plot of maximum and minimum amplitude versus guinea pig number showing differences between normal and abnormal tissue xxni  183  4.97 A . Proton spectrum from the lumbar spine of guinea pig number 13. B . Proton spectrum from the fixed lumbar spine of guinea pig 13. In both spectra tau = 400 ps  185  4.98 Plot of amplitude versus T value for fresh and fixed left hemisphere. The 2  numbers denote the guinea pig and F implies fixed tissue.  The peak  around 1 sec in the fixed tissue data corresponds to the T value for free 2  water in the formalin solution around the tissue.  The magnitude and T  2  of this peak varied, depending on the amount of water present  186  4.99 Plot of amplitude versus T value for a smooth fit to normal appearing 2  white matter in primate case 8, showing two distributions of T  2  187  4.100Plot of amplitude versus T value for normal and abnormal fixed tissue. 2  A . Normal spinal cord. B . Normal white matter from M S patient. C. Inflammatory lesion from guinea pig spine. D. Partially demyelinated and gliotic lesion from MS patient  190  XXLV  List of Abbreviations  A  angstroms, unit of measurement.  B  0  static magnetic field strength.  B  x  rotating magnetic field perpendicular to B .  BBB B  z  C CNS CP CPMG CSF  a  blood brain barrier. magnetic field in the z-direction. cervical spine. central nervous system. Carr-Purcell modified spin-echo pulse sequence. Meiboom-Gill modification of C P pulse sequence cerebrospinal fluid.  CT  X-ray computed tomography.  2D  two-dimensional.  2DFT  two-dimensional Fourier transformation.  D  demyelination.  D  self diffusion coefficient.  d DW E AE EAE  dose of myelin basic protein or steroid. dwell time. edema. energy difference between spin states. experimental allergic encephalomyelitis.  XXV  f,fb fraction of bound water. f  fraction of structured water.  /„,  fraction of free water.  st  Af  bandwidth of selective radiofrequency pulse.  FB  fixed brain.  FCA  Freunds' complete adjuvant.  FE  frequency encoding.  FID  free induction decay.  G  gliosis. magnitude of field gradient in x-, y- or z- direction.  G ,y,z x  magnitude of read gradient.  G  r  magnitude of phase encoding increment.  AG  P  gallo-DR Gd-DTPA GP H &E  gallocyanin combined with darrow red. gadolinium diethylenetriamine penta-acetic acid. guinea pig. hematoxylin and eosin.  h  Plancks constant.  I  spin quantum number.  I2r  signal intensity at time 2r after the 90° pulse in a spin-echo sequence.  I  r  signal intensity an inversion-recovery sequence with an inter-pulse delay r.  I  equilibrium value of the magnetization.  IR  inversion-recovery.  a  xxvi  k  Boltzmann constant; delay between data acquisition and reapplication of inversion-recovery pulse sequence.  L LFB-CV  lumbar spine. luxol fast blue combined with creosil violet.  LH  left cerebral hemisphere.  LT  lower thoracic spine.  MO  macrophages.  m  weight of monkey.  MBP  myelin basic protein.  MHz  megahertz.  MNC MR MRI An NMR N  a  mononuclear cells. magnetic resonance. magnetic resonance imaging. equilibrium population difference between spin states. nuclear magnetic resonance. population of spin state alpha. population of spin state beta.  v  0  ORO-H P PCI PE p.m. fi  precession frequency in hertz. oil red 0 combined with hematoxylin. perivascular inflammation. parenchymal cellular infiltration. phase encoding, post meridien. magnetic moment.  xxvii  PM PTAH PVI PVI-SS PVI-VR  post mortem. phosphotungstic acid hematoxylin. perivascular inflammation. perivascular inflammation in the subarachnoid space. perivascular inflammation in the Virchow-Robin spaces.  r  distance separating two interacting nuclei,  rf  radiofrequency.  p proton density. ROI S  region of interest. sacral spine.  cr screening constant. SE S(LD) S(t) S  0  SW  signal from 2 D F T pulse sequence in the frequency domain. signal from 2 D F T pulse sequence in the time domain. ratio obtained from two images. sweep width.  Ti Tip T  spin echo.  2  spin-lattice relaxation time. spin-lattice relaxation time in the rotating frame. spin-spin relaxation time.  Tib  spin-lattice relaxation time of bound water.  Ti  st  spin-lattice relaxation time of structured water.  w  spin-lattice relaxation time of free water.  Ti  TIN  spin-lattice relaxation time of normal appearing white matter.  T2N  spin-spin relaxation time of normal appearing white matter.  v  xxviii  Tiobs observed spin-lattice relaxation time. T  temperature in degrees kelvin.  T  a  length of the primary lobe of a sine function.  t  time after the 90° pulse in a C P M G pulse sequence. dwell time.  At t  p  pulse duration; also time of phase encoding increment in 2 D F T imaging sequence.  t  r  delay after data acquistion and reapplication of C P M G pulse sequence,  r  inter-pulse delay.  r  correlation time.  c  TE  time of the echo in a spin-echo pulse sequence.  TR  total time elapsed before reapplication of a pulse sequence; referred to as the repeat time.  8  angle between vector joining two nuclei and the static magnetic field; also flip angle of rf pulse.  UT  upper thoracic spine.  v  volume of micro-osmotic pump.  w  weight of MBP/steroid; also correction factor for 180° pulse in an IR sequence.  Ax  slice thickness.  xxix  Acknowledgements  I would like to thank my supervisor, Professor L . D . Hall, for his support throughout this thesis, and for introducing me to magnetic resonance imaging. to thank my collaborator and friend Dr.  I would also like  D . W . Paty, whose consistence support and  encouragement throughout this project have been invaluable and much appreciated. I would like to thank the members of my lab for their support and helpful discussions. Thanks also go to the members of the M S - M R I research group for their friendship and interest in my weekly updates.  I especially thank Keith Cover for his help in battling  with lATgX, and E d Grochowski for his assistance regarding the use of the programs on the Vax 11/750. The use of the whole body imaging system in the extended care unit required interaction with the M R I technicians responsible for clinical operation of the scanner. I would like to Cathy Jardine, Sharon Hall, Lesley Costley and Karen Smith for their help and cooperation throughout this project; and for giving me time on the scanner at a decent hour whenever possible! In a multi-disciplinary project such as that described in this thesis, the work could not have been completed without the assistance of the pathologists I have collaborated with. I would like to thank Dr. Ken Berry, Dr. Wayne Moore (Department of Neuropathology, V G H ) , Dr. Buster Alvord (Department of Neuropathology, University of Washington) and Dr. Kaz Watabe (UBC), for their assistance and support for the project as a whole; I especially thank Wayne for overloading the histology lab with my tissue so that I had the data for my thesis. Thanks also go to the technicians who carried out the embedding and staining of my never ending number of samples.  XXX  I must also extend thanks to Professors Myer Bloom and Alex Mackay and their graduate students for their helpful discussions and for the use of their spectrometer.  I  especially thank Julia Wallace for her friendship and patience with my questions; E d Sternin and Clare Morrison for allowing me  10,000  blocks on the microvax, (which I  constantly filled!) and for being there when the computer baffled me; Cindy Aroujo, Teresa Rind, Frank Nezil and everyone else in Hennings room 100 for adopting me as part of the lab; it made my last year a very enjoyable one. Ken Whittall wrote the NNLS programs that I used to analyze my relaxation data; I would like to thank him, not only for allowing me to use them, but also for the time he spent with me trying to optimize the analysis and for his interest in the project. Working with animals is both a difficult and time consuming operation; I extend my heartfelt thanks to Sharon Barwick, without whose technical assistance, friendship, patience and support, this thesis could not have been completed. Animals also require a lot of care, and the technicians in the animal facilities at U . B . C . have been terrific; I thank them all for being so readily available and helpful, and for sharing with me the laughter and the hard times during the project. Thanks also go to Sarka Hruby (Neuropathology, University of Washington) for teaching me how to handle primates and how to produce the animal model for MS in this species. I would also like to thank Bev Gray, Nicholas Mair and Anneke Ress for their help in the preparation of this thesis and for their encouragement throughout the project. Last but not least, I would also like to thank all my friends at Granville Chapel for their support, encouragement and prayer throughout my Ph.D.; and Anneke Rees for feeding me during the hard times and the fun times, it kept me going.  xxxi  Chapter 1  General Introduction and Thesis Goal  1.1  General Background  Multiple sclerosis (MS) is a demyelinating disease of the central nervous system (CNS) affecting 1 in 1000 people in British Columbia, most of whom are females between the ages of 20 and 45 years.  Exactly as the name suggests, demyelination results in the  destruction of the myelin sheath surrounding the nerve axons (figure 1.1).  The myelin  sheath is composed of alternate lipid and protein layers and acts as an "insulator", allowing the fast conduction of an electric potential along the axon's length. The results of myelin destruction are many and varied, and depend on the location and extent of the damaged area: any part of the CNS can be affected. M S is a multi-phasic disorder, being most commonly characterized by relapses and remissions.  The manifestation of  many different clinical symptoms may mimic a large number of monophasic disorders, dendrites  axon Figure 1.1: Diagram of a nerve cell or neuron.  1  Chapter 1. General Introduction and Thesis Goal  making MS a difficult disease to diagnose [1].  2  Other diseases also show demyelination  in the CNS, such as acute disseminated encephalomyelitis [2]. The causes of, and adequate treatments for MS have eluded researchers for decades: even with the sophisticated immunological and pathological techniques available today, the puzzle remains unsolved.  One of the main obstacles preventing any progress in  this area is the lack of knowledge about the pathological progression of the disease; this problem exists because the only pathological data available, except for a few biopsies, are post mortem, by which time the disease is usually in the chronic stage. Pathological and clinical studies are numerous [3,4,5,6], yet little is known about the intermediate stages in the disease process. In addition to clinical examination, a number of paraclinical techniques are used to support the clinical diagnosis of MS. These techniques have been in use for many years, and include: evoked potentials, which examine the ability of neurons to generate and conduct electric potentials; oligoclonal banding in the cerebrospinal fluid (CSF)(a characteristic protein banding pattern seen when the fluid is subjected to electrophoresis); and X-ray computed tomography (CT), an imaging technique which provides images dependent on the tissue density.  These techniques are useful in diagnosis, but provide  very little additional information about the nature of the disease, apart from the detection of blood brain barrier (BBB) breakdown using contrast-enhanced CT scanning [7]. In addition, these techniques examine only one variable, and serial scanning is not possible using CT, because of the problem of repeated exposure to X-rays. The introduction of magnetic resonance imaging (MRI) to clinical use in 1981 revolutionized the study of MS in vivo.  Young et al were first to publish images from  MS patients, using the inversion-recovery (IR) pulse sequence [8]. These images clearly demonstrated the anatomical detail that was available using MRI. The multi-variable nature of MRI also allowed the manipulation of the radio-frequency pulse sequences  3  Chapter 1. General Introduction and Thesis Goal  to provide different contrasting in the final image [9,10].  Subsequent studies of MS  patients showed that the spin-echo (SE) pulse sequence was much more sensitive to the number and extent of MS lesions in the brain [11,12] than the IR pulse sequence. Comparisons between MRI and other proven paraclinical techniques used to aid diagnosis, have repeatedly shown the important role this technique has in studies on MS [13,14,15,16,17,18,19,20,21,22].  MRI is already being used as a quantitative index of  disease activity in serial studies of MS patients [23,24,25,26].  Such studies provide a  comparison of disease activity with clinical activity, and have shown that there is no correlation between lesions seen on the MRI and the frequency of clinical relapse; thus, many of the new MRI lesions seen in these studies are clinically asymptomatic. MRI is also being used to monitor the effects of drug therapy on disease activity [27].  This  type of information on MS was not accessible before the advent of MRI, and it has had an enormous impact on the MS research community [28,29]. The multi-variable nature of MRI means that it can provide not only structural information, but also dynamical information. The proton signal observed in a MR experiment is dependent on the spin-lattice relaxation time (Ti), the spin-spin relaxation time (T2), the spin diffusion coefficient (D), the proton density (p), and various correlation times (t ) c  describing the interactions of the active proton spins with other spins, macromolecules and other surfaces with which they come in contact.  The proton relaxation times are  dependent on the molecular environment of the spins and are also sensitive to their restricted motion.  Thus, any changes in the molecular environment of the spins should  in turn affect the relaxation parameters. Relaxation time measurements are frequently used in chemistry to provide structural, conformational and functional information on complex molecules in solution [30,31,32] and the applications of such measurements to biological samples are already underway. In 1973, Damadian demonstrated differences in the relaxation times in normal versus  Chapter 1. General Introduction and Thesis Goal  tumour tissue [33].  4  Those early data demonstrated the future use of M R in the dif-  ferentiation of normal from abnormal tissue and subsequently many relaxation studies have been carried out on various tissues [34,35,36]. The results to date show it unlikely that simple T i and T measurements can be used to specifically characterize tissue and, 2  perhaps more importantly, pathology. Increases in the values of these relaxation times are seen in many different diseases [37,38,39,40,41,42] some of which MS can mimic, such as multiple infarcts and primary brain tumors [43,44,45]. However, this is not perhaps as disappointing as it first seems since it is unlikely that M R I alone would be used to diagnose disease regardless of its success. So, in the event of clinical diagnosis, one poses the question; can the measurement of M R parameters provide any useful information on the disease process? Thus, the specific question being addressed in this thesis is whether M R can provide information on the pathological progression of MS in vivol This requires that M R I can provide information, not only on the location and extent of MS lesions in the CNS, but also on the type of lesion visible on the M R I . The following approach has been taken to answer this question: quantitative M R I at 0.15 Tesla on a whole body imaging system has been used to study post mortem and fixed CNS tissue from MS patients. Those data are presented in chapter 2, and demonstrate that there are different intensities on the M R I , corresponding to different values for the relaxation times; microscopic examination of these areas showed them to be due to different pathology. Variations in lesion M R I characteristics have also been seen in a study on chronic, but stable, M S patients [46]. The application of this type of measurement to study early immune-mediated lesions in vivo was then tested by following the evolution of experimental allergic encephalomyelitis ( E A E ) , an animal model for M S , in primates and then comparing the quantitative M R I data with the pathology at the time of death.  Those data corroborated the evidence  provided from the studies on M S tissue. However, as will be described in detail later in  5  Chapter 1. General Introduction and Thesis Goal  this chapter, quantitative measurements on most commercial MRI instruments, including the one used in these studies, involve the calculation of Ti and T values from only two 2  points on the signal decay curve; hence, the accuracy of those T\ and T values and 2  how real the differences were for different pathology, were in question. In addition, the areas of abnormality studied in the human and primate tissues consisted of mixtures of pathology, thus preventing the isolation of the effects of each type of pathology on the quantitative M R measurements. The pathology observed in the CNS of guinea pigs after induction of E A E is less complex than that observed in primates, with some lesions being primarily inflammatory. In vitro relaxation measurements,  using the conventional spectroscopic methods described  in the next section, were therefore carried out on normal and abnormal CNS tissue from guinea pigs inoculated to produce EAE. Studies were repeated after the tissues were fixed in 10% neutral buffered formalin. Those studies showed that small differences in T are 2  reproducible from animal to animal and that differences are observed when pathology is present. Even though the samples studied were mixtures of normal and abnormal tissue, differences were seen for different degrees of inflammation involving different parts of the CNS. T values were obtained for primate CNS tissue from an animal that had also been 2  followed in vivo. As a last comparison, T\ and T values were obtained from samples 2  of the fixed MS tissue already studied using quantitative MRI. The results from these studies are presented in chapter 4. Two different methods were used to analyze the data, both of which demonstrated clearly that CNS water proton spin relaxation does not follow simple mono-exponential behaviour. The data also show that relaxation measurements on tissue can be reproduced, with errors around 5%. Spectroscopic studies carried out simultaneously, showed no contribution to the MR signal from lipid protons. The fixed tissue showed the same trends observed in vitro: this is important since most of the MS tissue available is already fixed, and these results show that useful information can  Chapter 1.  General Introduction and Thesis Goal  be obtained from the study of fixed tissue.  6  The studies on fixed M S tissue showed  differences between the areas that were completely demyelinated and gliotic, and those only partially demyelinated and gliotic.  These data also differed from the primarily  inflammatory lesions studied i n the guinea pigs. The multi-disciplinary nature of this thesis implies a large plurality of readers; therefore, although this is an integrated thesis, the author has chosen to subdivide the data into separate chapters, thereby allowing the specialist reader to access easily the information of interest to him. The multi-disciplinary nature of the thesis also means it includes many medical terms that may be unfamiliar to some readers; the large number of terms precludes the inclusion of a glossary, but they are listed below and the reader is referred to Dorlands Medical Dictionary, which is available in the University of British Columbia bookstore and in the Woodward Biomedical library. In addition, where possible, the terms are defined briefly as they appear in the text of this thesis.  7  Chapter 1. General Introduction and Thesis Goal  List of Medical Terms acute  demyelination  immunology  akinesia  dendrites  incontinence  allergic  dexamethasone phosphate  infarct  anaesthesia  disseminated  inflammation  anorexia  dorsum septum  inoculation  anterior fissure  dura-mater  in situ  arachnoid membrane  edema  internal capsule  astrocytes(fibrous, protoplasmic)  encephalomyelitis  intracardially  ataxia  euthanyl  in vitro  atropine  exudate  in vivo  axon  fibrin  ketamine  blood brain barrier  formalin  lamellae  brainstem  Freunds' complete adjuvant  lesion  central nervous system  gitter cells  leucocytes  cerebellum  glia/glial  leucoencephalopathy  cerebral cortex  gliosis  lymphocytes  cerebral hemispheres  gross pathology  lysosome'  cerebrospinal fluid  hemiplegia  macrophage  chronic  hemorrhage  microglia  chronic-progressive  histiocytes  microscopic pathology  corpus callosum  histology  midbrain  corticomedullary  hyperacute  mononuclear cells  corticosteroids  hypercellularity  multiple sclerosis  decerebration  hyperplasia  myelin  decortication  immunization  myelin basic protein  Chapter 1.  General Introduction and Thesis Goal  necrotic  periventricular  rompun  neuroglia  perivascular cuffing  strabismus  neurological  phagocytes  subarachnoid space  nystagmus  pia-mater  subclinical  oligoclonal  plaque  subcortical  oligodendrocyte  p l a s m a cells  subpial  optic c h i a s m  p o l y m o r p h o n u c l e a r cells  sudanophilic lipids  paraclinical  post m o r t e m  synapse  paraplegia  presynaptic  thalamus  parenchyma  ptosis  tremor  paresis  quadriplegia  V i r c h o w - R o b i n spaces  pathogenesis  remitting-relapsing  visual r a d i a t i o n  perivenous  remyelination  W a l l e r i a n degeneration  8  Chapter 1.  9  General Introduction and Thesis Goal  Before presenting the results of the studies briefly described above, the reader is now introduced to the basic concepts of the pulse Fourier transform M R experiment [47]. Detailed M R theory is not presented, since many excellent references already exist on the subject [48,49]; however, the information given should enable the reader to understand the description of M R I which follows in the next section.  1.2  Basic Principles of Magnetic Resonance  When a spin system is placed in a static magnetic field, the system becomes ordered and the spins may take up one of 21 + 1 orientations in the magnetic field (where J is the spin quantum number). For protons, J = 1/2; thus there are two energy levels, and the energy difference between them is given by equation 1.1;  where 7 is the gyromagnetic ratio and h is Planck's constant.  B y convention, the  direction of the static magnetic field strength, B , is assigned to be along the z-axis. The Q  spins precess about the direction of B (figure 1.2), and since AE = hv, the frequency 0  of the precession is given by equation 1.2 [47]; v =  lB  (1.2)  0  2TT  A t thermal equilibrium, the populations of these two energy levels are given by the Boltzman distribution (equation 1.3). Radiation corresponding to the energy difference between these two states will induce transitions in both directions.  w = "d-tr*  ( L 3 )  a  where k = Boltzman constant, Np = population of upper state and N  a  lower state.  = population of  However, since there is initially a greater population in the lower energy  Chapter 1.  General Introduction and Thesis Goal  A  10  B  Figure 1.2: The behaviour of spins in the presence of two orthogonal magnetic fields. A . Precession about B . B . Behaviour of spins after the application of a second magnetic field, Bi, perpendicular to B . Q  Q  state, there is a net absorption of energy.  If there were no relaxation processes, the  populations of the two states would soon become equal, a process known as saturation. Although this can occur (and is often used in experiments [47,50]), it is usually possible to observe an absorption continuously when the appropriate resonance frequency is applied. This indicates that non-radiative relaxation processes are occurring that prevent saturation.  These processes are characterized by the spin-lattice relaxation time, T\,  and the spin-spin relaxation time, T , the two time constants associated with the spin 2  system's return to its equilibrium state. Relaxation is considered further in chapter 4. Transitions are induced by an oscillating radiofrequency (rf) pulse,  by applying a  current in a coil wound perpendicular to the direction of B„. This transmitter generates a linear field oscillation along the x-axis which is equivalent to two counter-rotating field vectors in the xy plane, but interaction only occurs between the precessing nuclear magnetization and the vector rotating in the same direction [50]. When the frequency of  Chapter 1.  11  General Introduction and Thesis Goal  B\ is equal to the precession frequency of the nuclear spins, this is known as the resonance condition and maximum signal is observed. Application of B\ produces a torque which moves the magnetization towards the xy plane. The final position of the magnetization will depend upon the length of time for which the rf is applied. The B\ field is applied as a pulse of duration t , which usually lasts a few microseconds. The angle, 0, (usually p  referred to as the flip angle or tip angle) through which the magnetization is tipped from the z-axis is easily calculated from equation 1.4, 9 = iB t x  (1.4)  p  The magnitude of B\ is chosen to be large enough to tip the magnetization from the z-axis into the xy plane in a few microseconds (figure 1.2).  It usual to refer to the  pulse magnitude in terms of the flip angle rather than the pulse duration, since the former is a reproducible parameter. After the application of a 90° pulse, the precessing magnetization induces in the receiver coil an oscillating current which can be detected. This signal is called the free induction decay (FID): free of the influence of the rf field, induced in the coil, and decaying back to equilibrium. Since the signal decays away after excitation, it is referred to as a transient The N M R signal observed must be digitized in order to be processed by the computer: this is referred to as analogue- to-digital conversion. This process involves the conversion of a voltage at a particular moment, to a binary number; this number is then stored as a 'data point' and represents the signal amplitude at that point in time. The signal is sampled at regular intervals, defined by the dwell time, DW (equation 1.5);  D  W  = - k  (  L  5  »  where SW is the spectral width. The data points must be collected at a sampling rate greater than or equal to twice the signal frequency, if the frequency of a signal is to be  Chapter 1.  12  General Introduction and Thesis Goal  represented correctly. The data collected in this manner, represent the variation of the signal over time; for useful information to be obtained this signal must be converted to the frequency domain. This is achieved using the mathematical process of Fourier transformation [51], which converts a function of time f(t) into the corresponding function of frequency, F(u>), as shown in equation 1.6. F(u) = [ °° f(t)e~ dt +  J  iut  (1.6)  —oo  The inverse relationship also holds; F(u)e' du ut  oo  (1.7)  Due to the effects of a phenomenon referred to as chemical shift, the field experienced by a nucleus, i, is not exactly B  0  but; Bi = B {l-<r) 0  (1.8)  where o is the screening constant, and characterizes the effects of the spatial electronic distribution around nucleus i. The resonance frequency of the nucleus is then given by;  Ut =  i  {  1  ~  a  )  B  °  -  (1  9)  The range of frequencies of interest is usually known for the nucleus being studied; thus, the spectral width can be defined by setting the dwell time to an appropriate value (equation 1.5). For protons, the majority of signals occur over a narrow range of frequencies; these signals are excited simultaneously using a broadband rf pulse. Subsequent Fourier transformation of multiple resonances then gives rise to the familiar N M R spectra; the frequency differences between the resonances being determined by cr for each nucleus. A t low magnetic field strength, the effects of chemical shift are negligible; however, as the field increases, so too the effects of chemical shift. For the CNS studies carried out  Chapter 1. General Introduction and Thesis Goal  13  Figure 1.3: Diagram showing the variation of the static magnetic field, B in the presence of a linear field gradient, G . Q  x  in this thesis, chemical shift was not a problem; this is due to the fact that the images obtained were of tissue water. There are protons in other electronic environments (such as lipid proton); however, these nuclei are bound up in membranes and on the time scale of the imaging experiment, are not observed due to their short T -values. 2  1.3  M R Imaging and Image Enhancement  The basic principles of M R I are the same as those for pulse F T spectroscopy described above. The difference is the addition of linear field gradients to spatially define the object under investigation. In the presence of such a magnetic gradient, G , the magnetization x  along the z-axis is given by equation 1.10: B (x) = B + G .x z  Q  x  (1.10)  This is demonstrated in figure 1.3. The resonance frequency of the system also becomes spatially dependent, as shown in equation 1.11; nfB  0  2ir  7<7x.x  + 2IC  (1.11)  where v is the resonance frequency of nuclei at position x during the application of x  gradient G . x  Conventional N M R spectroscopy examines the whole sample, while imaging  Chapter 1.  14  General Introduction and Thesis Goal  selective 90° RF  A  180°  echo  G:  G> Gx-  Figure 1.4: Two-dimensional Fourier transform spin-echo pulse sequence for slice selective imaging. methods examine a section, or sections, of a sample [52,53,54].  Selection of a slice is  achieved by using three orthogonal linear field gradients and a selective rf pulse.  The  combination of one gradient with the selective pulse provides a slice of known thickness with an in-plane resolution that is defined by second and third gradients.  The use of  the F T method in imaging was first demonstrated by Kumar, Welti and Ernst [55], and is an extension of the two-dimensional F T (2DFT) spectroscopic methods [56] used in the elucidation of complex molecular structures [57]. A typical 2 D F T pulse sequence for imaging is shown in figure 1.4. The x- and y-gradients in figure 1.4 spatially define the plane of the slice. One is referred to as the read gradient, which is a constant value, and the other as the phase-encoding gradient, which is incremented in amplitude for each successive excitation. Together they produce a 2D matrix whose size and resolution are  Chapter 1.  15  General Introduction and Thesis Goal  determined by the number of phase encoding increments chosen. The signal observed, S(t) can be written as S(t) = J J p(r)s(r,t)dxdy  (1.12)  where s(r,t) is the contribution from the area dxdy, volume averaged over the slice thickness at position r, and p{r) is the spin density. 2 D F T of S(i) gives 5(u) = 5 K , w „ )  (1.13)  such that S(u) = J J S{t)exp{iut)dt dt x  y  (1.14)  The field of view (FOV) from which the image is obtained, is defined by equations 1.15 and 1.16, which describe the dependency of the F O V on the read or frequency encoding (FE) and phase encoding (PE) gradients [58].  where G is the magnitude of the F E gradient and At is the duration of the dwell time. r  PE=  , A 2 ?V  (1.16)  where AG is the magnitude of the P E increment and t is the duration of the increment. P  1.3.1  p  Slice Selection  In all the experiments carried out in this thesis, slice selection was achieved by application of a sinc-shaped frequency-selective (narrow bandwidth) rf pulse in the presence of a linear field gradient (figure 1.5) [53]. As the diagram shows, only those spins that are precessing with a frequency contained in the narrow band of rf used to excite the slice will be detected.  The other spins remain unperturbed and do not contribute to the  Chapter 1. General Introduction and Thesis Goal  16  Figure 1.5: Diagram showing selective excitation of a slice. A . Selective rf pulse and slice select gradient. B . Magnetization in the z-direction after application of selective pulse. C . Selected slice.  Chapter 1.  General Introduction and Thesis Goal  signal observed.  17  The position of the selected slice is dependent on the direction of the  applied gradient and the frequency band excited by the selective rf pulse. Multiple slices can be obtained simply by off-setting the frequency of the rf pulse to excite a different frequency range. The slice thickness, A x , is defined by equation 1.17. Ax = ^  (1.17)  lG  z  where A / is the bandwidth of the selective pulse which is given by A/ =2 x i  (1.18)  where T is the length of the primary lobe in the sine function. 0  The F O V and slice thickness are chosen to optimize the signal-to-noise ratio and resolution in the final image of the sample under study.  1.3.2  Image Enhancement  The contrast observed in the final image is dependent on the variables in the rf pulse sequence used, and on the relaxation parameters in the different areas of the sample. If we consider a SE pulse sequence (1.19) [59], for example, then the signal intensity observed is dependent on 2r and T2, according to equation 1.20. (90° — r — 180° — r — Acquisition of echo)  I  2r  = I exp(-2T/T ) 0  2  (1.19)  (1.20)  where 7 is the signal intensity at time 2r and I is the equilibrium value of the mag2t  netization.  Q  Similarly, for an IR pulse sequence (1.21), the signal intensity observed is  given by equation 1.22. (180° — T — 90° — Data Acquisition - Delay)  (1.21)  18  Chapter 1. General Introduction and Thesis Goal  I = I [l - 2exp(-r/T )} T  0  1  (1.22)  where I is the intensity of the signal observed after the 90° pulse when the inter-pulse T  delay is equal to r .  Variation of the parameters in these pulse sequences will provide  different contrast in the final image. The delay time before reapplication of the pulse sequence should be at least bT\ if the system is to return to its equilibrium state. If the delay time is shorter than this, then the signal magnitude is also dependent on Ti according to equation 1.23; / = J [ l - 2exp(-T/T ) 0  l  + expiTR/Tr)}  (1.23)  where TR is the total time elapsed before reapplication of the pulse sequence. For many years, scientists utilized the ability of paramagnetic metal ions to alter the relaxation times of various nuclei [60] and in the past few years attention has focused on the possibility of using such ions to enhance M R I contrast in vivo. The ion most commonly used is gadolinium ( G d ) ; it is administered in the form of a complex which the 3+  body can excrete, and which is not harmful to the vital organs. The preferred complex is gadolinium diethylenetriaminepenta-acetic acid (Gd-DTPA) [61,62,63,64]. Enhancement occurs due to a reduction i n the Ti and T values of the water protons in the 2  presence of the dominating dipolar interaction with the unpaired electrons on the paramagnetic ion.  This is due to the large local fields produced by the unpaired electrons,  which have magnetic moments that are much greater than those of the water protons. The rf pulse sequence is chosen to utilize the decrease in Ti, and is usually a SE sequence with a small r-value and a short T R [65]. The signal intensity dependence on the T R is given i n equation 1.23 above. Areas penetrated by the paramagnetic complex will then be brighter i n the image than normal tissue, which, i n most tissues, is impermeable to the complex. In normal human brain tissue, there is no enhancement after intravenous  Chapter 1. General Introduction and Thesis Goal  19  administration of G d - D T P A because the blood brain barrier ( B B B ) prevents the contrast agent reaching the cerebral extravascular space [66]. The pituitary gland and stalk and intracavernous portions of cranial nerves III-VI, which have an incomplete B B B , do enhance with G d - D T P A .  In many diseases of the CNS, the B B B is damaged in local  areas; these areas are then permeable to the G d - D T P A complex and will be enhanced on the M R I [67]. Image enhancement studies on M S patients [68,69,70] have shown that lesions do have different characteristics; some enhance very quickly while others form an outer ring of enhancement which eventually fills the lesion.  It is reasonable to assume  that the pathology of these two sets of lesions is likely to be different; areas which enhance rapidly are probably inflammatory lesions, in which it is believed the B B B is most permeable. The other areas are probably demyelinating and gliotic, where it is common to see inflammation on the periphery; this would explain the ring of enhancement rapidly observed around the lesion before all of it is enhanced.  Contrast enhancement studies JK;  were carried out on three of the primate cases presented in chapter 3.  1.4 1.4.1  Relaxation Measurements Measurement of T i  There are a number of methods available for measuring the T i of magnetic nuclei [71, 72,73,74]. The one most commonly used is the IR method; this method employs the IR pulse sequence 1.21; as described above the magnitude of the signal obtained after a 90° pulse is given by equation 1.22 (page 16). As previously mentioned, a delay time after data acquisition should be at least 5Ti to allow the spin system to return to equilibrium. The r-value is varied, and T i can then be determined from a three-parameter exponential fit to the data (equation 1.24) [74]. I = / „ ( ! - [1 + w(l - expi-k/TiWexpi-T/Ti)) r  (1-24)  Chapter 1.  20  General Introduction and Thesis Goal  where I is the signal intensity at T ^> T i , k is the waiting period between the data 0  acquisition and reapplication of the 180° pulse and w is the correction factor for the 180° pulse.  Detailed error analysis in the use of this method has been carried out [75], and  shows it to be a reliable method when conditions are optimized.  1.4.2  Measurement of T  2  The first measurements of T were carried out by Hahn [59] i n 1950; he demonstrated 2  that the intensity of the signal obtained from a sample after a spin echo pulse sequence (1.19) had been applied is dependent on T , as given by equation 1.20 (page 16). Carr 2  and Purcell then showed that in the presence of a field gradient, the signal intensity is also dependent on the diffusion coefficient, as shown in equation 1.25 [76]. I  2r  = I exp(-2r/T )exp(-l<y G DT ) 2  0  2  3  2  (1.25)  where G is the magnitude of the field gradient and D is the diffusion coefficient. For larger values of T dependence.  2  the effects of diffusion are much more pronounced, due to the r  3  The effects of diffusion can be reduced by using the Carr-Purcell (CP)  pulse sequence (1.26), which is a modification of the Hahn spin-echo method. (90° - r - 180° - r - echo - r - 180° - r - echo - ...)  (1.26)  Further modifications to this sequence have been made by Meiboom and G i l l ( C P M G ) [77], as shown in 1.27; that method overcomes the accumulation of errors due to pulse imperfections, which occur when the C P method is used. (90° - r - 180° - r - echo - T - 180° - r - echo - ...) y  y  (1.27)  If a short value of r is used i n this sequence, problems arise due to effective spin-locking of the magnetization about the y-axis. If spin locking occurs, the measurement of T is 2  Chapter 1.  General Introduction  and Thesis Goal  21  then equivalent to T i , the T i in the rotating frame [47]. This led Freeman and Hill to p  a further modification of the method, given in 1.28 [78]. (90° — T — 18(£ — T — echo - r - 1 8 0 1 , - r - e c h o - r - 18(£ - ...)  (1.28)  In the limit of short interpulse spacing, the Freeman-Hill modification does not lead to spin locking. In all of the above methods, T can be determined from an exponential fit to equation 2  1.20 (page 16), assuming monoexponential relaxation behaviour and neglecting the effects of diffusion.  1.4.3  Imaging M e t h o d s  As described briefly in the general background, commercial imaging methods for determining Ti and T generally involve obtaining at least two images of the same slice using 2  two different pulse sequences.  Each pulse sequence has two variables, thereby allowing  Ti and T values to be determined by taking the ratio of the signal intensity in each 2  pixel from both images [79,80]. Others methods employ a single sequence that provides data to determine Ti and T simultaneously [81,82,83]. The relaxation times are usually 2  obtained for the whole slice, pixel by pixel, or from a chosen region of interest (ROI). The whole body M R I system used for many of the experiments described in this thesis, is a Picker International N M R Cryogenic '2000' system operating at 0.15 Tesla. Computations to produce T i and T images were performed as follows, using the manufacturers 2  software: Ti computation: one IR and one SE sequence are required to compute T i ; for optimal results, T R and T E should be equal in both images. The intensities observed in the IR and S E images are given in equations (1.29) and (1.30). The ratio of these images is  Chapter 1.  22  General Introduction and Thesis Goal  then taken (equation 1.31), which eliminates the dependence on the equilibrium magnetization, I , and T?,. T i is then obtained from a look-up table, which is a plot of S versus 0  0  T i . Studies of water doped with either copper (II) sulphate, or manganese (II) chloride, demonstrated that the computed Ti values are reproducible and comparable to the value obtained using conventional spectroscopic methods [84]. As previously mentioned, the disadvantage of this two-point computational method is that it is not possible to make error estimates on the values obtained; only the reproducibility can be determined. However, the values obtained on the same instrument using the inversion-recovery method [74] show the errors to be no greater than 10-12%.  I  = I exp{-2t/T )[l  IR  0  2  - (2 - 2exp(-k/T ) 1  + exp({-k - r ) / T i ) ] e x p ( - i / T ) 1  (1.29)  where I is the equilibrium magnetization, 2t is the echo delay (TE) for the spin-echo 0  readout after the 90° pulse in the IR pulse sequence, k is the time between data acquisition and reapplication of the pulse sequence and r is the time between the 180° pulse and the 90° pulse in the IR pulse sequence.  I  = I„exp(-2r/T )[l  SE  s  =  °  Im I  2  =  - 2exp(-k/T ) 1  1 - (2 - 2exp{-k/T )  + exp(-r -  i  (l-2exp(-k/T )  SE  + exp{-TR/T{)]  k/T^expj-tm  + exp(-TR/T )  1  (1.30)  1  l  '  '  T computation: two SE images with equal T R and different T E are required to com2  pute T . 2  Once again, the ratio of these two images is taken (equation 2.26) eliminating  the dependence on I and T\. Just as for T i , T is obtained from a look-up table which 0  2  Chapter 1.  23  General Introduction and Thesis Goal  is a plot of S versus T . 0  2  Phantom studies were also carried out to check the T values 2  obtained using this two-point computational method [84]. The computed T values have 2  been shown to be reproducible and comparable to values obtained using the spin-echo method [59]. Just as for T\, it is not possible to make error estimates on these computed T values; however, values obtained using the SE method show the error to be less than 2  12- 15% for the range typical of CNS tissue T values. 2  S  [1 - 2exp(-k /T )  I_SE  =  °  1  L=  IE S  2  l  [l-2exp(-k /T ) 2  1  +  exp(-TR/T )]exp(2r /T ) 1  1  + exp(-TR/T )]exp(2T /T ) 1  2  2  2  K  '  )  These two-point computational methods for determining T\ and T do not provide any 2  information about the relaxation behaviour of the water protons; that is, whether it is mono-exponential or multi-exponential, and provide no indication as to the environment of the protons giving rise to the signal seen in the M R I . Proton spin relaxation behaviour in tissue is the subject of chapter 4, where the results of multi-exponential relaxation studies on CNS tissue are presented.  Chapter 2  M R I of Multiple Sclerosis  2.1  Introduction  The data presented in this chapter were obtained from immediate post mortem and fixed brain tissue from M S patients, and address two questions: 1. How accurately does M R I reflect the extent of disease in the CNS? and 2. What do variations in signal intensity on the M R I correspond to pathologically? There are essentially two classes of cells in the CNS: the nerve cells (or neurons) and the neuroglial cells (or glia).  As shown in the general introduction, the typical neuron  has four morphological regions; the cell body, dendrites, axon and presynaptic terminals of the axon.  Neurons interconnect with one another via chemical or electrical means,  and the point of contact is known as the synapse.  The neurons are responsible for  sending motor commands to the muscles and for the brain receiving sensory information from the different parts of the body.  Glial cells, however, have a supportive role and  also segregate groups of neurons from one another. nutritive functions.  They are also believed to have  In addition, there are certain glial cells that make myelin, the  fatty insulating sheath around large axons.  The features of the demyelinating disease  known as M S can be related to the destruction and loss of a single type of glial cell, the myelin-producing oligodendrocyte [85]. The areas depleted of these cells are usually large enough to be seen with the naked eye as discrete plaques; they tend to be multiple, and to occur, often symmetrically, in particular locations which include the optic nerves  24  Chapter 2.  MRI of Multiple Sclerosis  25  and chiasm, periventricular white matter, the corticomedullary junction, subpial regions of the cerebral cortex and brain stem, and certain parts of the spinal cord, such as the dorsum septum, subpial regions and on either side of the anterior fissure.  For  anatomical information, the reader is referred to references [86,87]. In all regions, lesions are almost invariably perivenous in location. Microscopically, the lesions show a loss of oligodendrocytes and myelin, other structures including axons, nerve cells and blood vessels tending to be preserved.  Astrocytes, another type of glial cell, can be divided  into two subclasses; protoplasmic and fibrous, each being associated with different parts of the neuron.  Fibrous astrocytes contain many filaments and are found extensively in  areas of the CNS containing myelinated axons (these areas are known as white matter because of the white appearance of the myelinated axons in unstained, freshly cut brain sections).  The protoplasmic astrocytes contain few filaments and are associated with  the cell bodies, dendrites and synapses of neurons, which are located in the grey matter (so called because of the grey appearance in brain sections).  Both types of astrocytes  contact blood capillaries and are therefore thought to have a nutritive function; in M S , these cells exhibit non-specific reactive changes.  Perivascular cuffs of lymphocytes and  plasma cells are common in subacute lesions. A third class of glial cell, the microglia, are small, ovoid cells that are capable of acting as phagocytes; that is they remove the myelin and axonal debris that arises from demyelination and other destructive processes in the CNS. There now follows an in-depth description of the complex pathology of MS which will provide the reader with some insight into the complexity of solving the MS puzzle and to make them aware of the conflicting theories that exist about the pathogenesis of the disease.  The effects of different pathology on the M R signal observed are then  discussed. The pathological features of MS resemble other disorders such as post-infectious and sporadic perivenous encephalomyelitis and the human [88] and experimental forms of  Chapter 2.  26  MRI of Multiple Sclerosis  allergic encephalomyelitis [89]. Theories about the pathogenesis of M S are usually concerned with the selective destruction of myelin and oligodendrocytes in affected white matter in this disease.  There is no general agreement about the nature of the myelin  destruction or how it occurs. Many authors have examined M S tissue at various stages of the disease process; some describe myelin breakdown occurring in the absence of an increase in the number of cells, including microglia [90,91,92,93]. Others have described hypercellularity of the tissue as an accompanying feature of commencing myelin breakdown [94,95]. A third pattern was recognized on examination of plaques with bordering hypercellular white matter [96,97]; these cells were identified as oligodendrocytes, based on their nuclear morphology and oxidative enzyme activity. These authors concluded that oligodendrocytes may be the source of increased protease activity present at plaque margins and that microglial cell infiltration is a secondary response to myelin breakdown. Electron microscope studies have failed to confirm the existence of a type of myelin breakdown that occurs in the absence of infiltrating macrophages.  It is now known  for certain that oligodendrocyte hyperplasia does occur in areas of active myelin breakdown but this is believed to be due to co-existing remyelination.  The first electron  microscope studies of periplaque white matter revealed changes in myelin and oligodendrocytes in the absence of infiltrating cells. The changes were interpreted as evidence of a toxic or degenerative process primarily affecting oligodendrocytes, with myelin breakdown occurring as a secondary event [98,99,100,101,102,103].  Two different types of  myelin breakdown were observed frequently in the above studies: one involving granular vacuolar degeneration, consisting of focal splits containing granular osmiophilic debris, myelin fragments and 600 A diameter particles in sheathes of normal thickness; this change is now known to occur in normal white matter.  The second involved compact  myelin internodes and paranodes that were abnormally thin and/or exhibited premature termination of superficial or deep lamellae. Granular vacuolar degeneration as well as  Chapter 2.  MRI of Multiple Sclerosis  27  Wallerian degeneration was also observed in white matter remote from plaques, together with conspicuous secondary lysosome formation in oligodendrocytes, astrocytes and microglia, which was also interpreted as possibly indicating a primary glial cell alteration causing secondary myelin breakdown [104]. The lesions examined in the above studies were mostly quiescent plaques with little evidence of recent myelin breakdown, suggesting that the changes observed may have been reactive or regenerative in nature. In 1975 Prineas [105] reported an electron microscope study of 16 'active' plaques from cases where fixation was achieved by in situ perfusion within a few minutes of death. It was found that hypercellularity was mainly due to the presence of infiltrating microglia. Most myelin sheaths in the infiltrated zone, including many directly contacting microglial cells, appeared normal. Some however, were abnormally thin, with superficial or deep lamellae terminating prematurely along the internode. Lymphocytes were present in the Virchow-Robin spaces but none were observed in the parenchyma. It was not uncommon however, to find plasma cells in the parenchyma. It was concluded that myelin thinning in the presence of infiltrating microglia of uncertain origin was a major pattern of myelin destruction in M S . More recent studies support this picture of myelin breakdown in the presence of microglia, and have also provided further evidence for localized areas of remyelination [106,107,108]. The water in tissues may be in many different environments. For the most part, however, the exchange between these environments will be rapid compared to the timescale of the M R experiment.  The result of this is the observation of a single relaxation time  which is a weighted average of the values from all the different environments.  Changes  in the number of cells present in an area of the CNS may therefore lead to changes in the ratio of water with restricted motion (that is, associated with macromolecules, organelles or other surfaces) and water in the cytoplasm, which has relaxation characteristics more  Chapter 2. MRI of Multiple Sclerosis  28  like that of bulk water. The relaxation rate observed is given in equation 2.33 [109]; — = J- + I z l  T  x  T»  T  (2.33)  K  lf  '  where / is the fraction of "bound" or restricted water, Tu, is the T\ of this water and Tij is the T i of the free or unrestricted water.  If the tissue exhibits multiexponential  relaxation behaviour, hypercellularity could affect both the relaxation rates of the different components and the relative contribution to the M R signal from each component. This is demonstrated clearly in the data presented in chapter 4. These changes will be affected by the location and numbers of cells involved. Myelin breakdown products will be present in the extracellular space for a short time before they are consumed by macrophages.  These breakdown products comprise the  proteins and lipids that combine to form the myelin sheath.  Their presence in the  extracellular space could affect free and bound water ratios, and could also give rise to a M R signal from the lipid protons, which would have different relaxations characteristics from those of water.  In normal CNS tissue, lipids are bound up in membranes; as a  result, the lipid protons behave like a solid, having a very short T . 2  On the timescale  of the M R I experiment, the signal from these lipids has already decayed to zero and is not observed.  The destruction of the myelin sheath gives rise to free lipid that has a  longer T and can be observed. Spectroscopic studies on primates with E A E have shown 2  a large lipid signal in the lesions resulting from the disease process [110], and this could also be the case with M S . It is clear that a technique such as M R I is not sensitive enough to detect the different mechanisms that may be causing myelin breakdown. However, these plaques can be seen grossly, and the changes are occurring at a molecular level. Thus, it is not inconceivable that M R I could detect and differentiate between an inflammatory response, involving perivascular cuffing and possible infiltration of inflammatory cells into the surrounding  Chapter 2. MRI of Multiple Sclerosis  29  {10mm  Figure 2.6: Diagram showing the simultaneous excitation of normal and abnormal tissue in a 10 mm thick slice due to the irregular shape of the lesion. parenchyma, and more chronic demyelinating and gliotic plaques. be possible to differentiate degrees of demyelination and gliosis.  In addition, it may If this were the case,  the pathogenesis of the disease could be followed over time, providing insight into the different mechanisms that are involved in each stage of the disease process. This chapter describes the initial quantitative M R I studies on M S tissue that prompted further study in this area, and provided the first evidence that M R I may be sensitive to different types of MS pathology. MRI,  The approach taken was as follows: abnormal areas on the  as indicated by increased signal intensity on the SE image [11], were compared  with the gross pathology seen in the fixed brain. Qualitative visual comparisons of all the data were made; in addition, where appropriate, the areas of abnormality on the MRI  and the actual pathology seen in the fixed brain were quantified and compared.  This is not a trivial comparison, since the MRI are volume-aver aged over a 10 mm slice thickness.  This means that the resultant image is an average over this thickness, and  the appearance of a particular area may be dependent on more than one tissue type. The  lesion borders are therefore sometimes difficult to define because the lesion may  be irregularly shaped through the slice (figure 2.6). The average T i and T values for 2  regions of interest (ROI) were obtained using quantitative imaging methods.  Sections  Chapter 2. MRI of Multiple Sclerosis  30  were then taken for microscopic examination, and the data compared with the T i and T  2  data obtained from the equivalent area on the M R I . Volume-averaging also affects these comparisons, since the area seen on the M R I may consist of normal and abnormal tissue, especially if the lesion is seen in only one 10 mm thick slice. The data shows M R I to be an accurate indicator of diseases extent in the brain.  The quantitative M R I did show  a correlation between the degree of gliosis and the Ti-values. The problems associated with such studies and their implications are identified and discussed.  2.2  Methods  M R I data were collected on a Picker International N M R Cryogenic '2000' system, operating at a field strength of 0.15 Tesla, corresponding to a proton resonance frequency of 6.4 M H z .  A receiver coil with an aperture of 30 cm was used to acquire the images.  Multi-slice S E pulse sequences were employed, together with single and multi-slice IR pulse sequences (figure 2.7).  The multi-slice sequences provided 12 contiguous 10 mm  thick slices. The single slices were also 10 mm thick. Echo-delay times (TE) of either 40 and 60 ms or 40 and 80 ms and an inter-pulse delay (TI) of 400 ms were used in the SE and IR sequences respectively. In all sequences, the repeat time (TR) was 2 sec. These sequences were chosen to allow comparison with in vivo MS M R I data and to allow the computation of spin-lattice (Ti) and spin-spin (T ) relaxation time images. 2  data were transferred from magnetic tape to a Vax-11/750 computer.  The M R I  A n interactive  computer program is available for outlining R O I using a trackball and cursor, and calculating their area. The areas of increased signal intensity on the S E images were outlined using this program and their areas determined.  The R O I were then transferred to the  computed Ti and T images; the intensity obtained from these regions corresponds to the 2  average T i and T value from that area. In addition to this, R O I were chosen from the 2  Chapter 2. MRI of Multiple Sclerosis  A.  31  selective 180° selective 90°  A  selective 90°  e c h o  — J \ / —  B. selective 180°  selective 90°  tau  selective 180°  selective 180°  •  A  A  echo  — TE TR  J  n  Figure 2.7: A . Multi-slice spin-echo pulse sequence showing variable time parameters. B . Multi-slice inversion-recovery pulse sequence, n = the number of slices in both sequences.  Chapter 2.  32  MRI of Multiple Sclerosis  centre of the lesions corresponding to those examined microscopically, and transferred to the computed images.  In an attempt to minimize the effects of volume averaging, the  Ti and T data given in the results section are from these central regions. 2  of the lesions is also more likely to be pathologically homogeneous.  The centre  Direct comparison  of the Ti and T values between patients was questionable since the elapsed time after 2  death before scanning varied [111,112,113].  Although literature studies show that the  Ti and T values remain stable for up to 24 hours, changes in temperature could lead 2  to differences in relaxation time values, particularly Ti [114].  The length of time for  fixation also varied [112,115]. The effects of fixation have only been studied over a few days, and the data on brain tissue show a tendency for Tj and T to decrease over time. 2  In the studies described here, tissue had been fixed in 10% neutral buffered formalin for up to 3 years.  It seems likely that T i and T would eventually stabilize. However, to 2  overcome this potential problem, values of T i and T for normal appearing white matter 2  were obtained from each slice.  A t least 10 R O I , each with an area of 14 pixels, were  chosen on the S E images and transferred to the computed images. The ratios of lesion Ti and T to normal white matter T i and T were then calculated. 2  2  This method, of  course, makes the assumption that the effects of elapsed time and temperature changes after death and time of fixation affect normal and abnormal tissue to the same extent; this is not necessarily the case. Three chronic cases of MS were scanned as soon after death as possible. A sagittal pilot scan was obtained before using the sequences described above, to allow the positioning of the first transverse slice on the superior surface of the brain (figure 2.8A). This is important for later comparison with pathology.  The brains were then removed and  placed in 10% neutral buffered formalin. These fixed brains together with one other that was already fixed, were then scanned. The fixed brain was placed on its superior surface and a sagittal pilot scan obtained. This allowed comparison of the positioning with that  Chapter 2. MRI of Multiple Sclerosis  33  Figure 2.8: A. Sagittal pilot scan obtained post mortem. Grid shows the center of each of the 12 transverse images to be obtained. B. Sagittal pilot scan from fixed brain. Grid shows the center of each of the 12 coronal images to be obtained. The coronal images of the fixed brain are equivalent to the transverse images obtained in situ. TE = 26 ms and TR = 400 ms in both images.  Chapter 2. MRI of Multiple Sclerosis  34  in situ. The first coronal slice was then positioned on the superior surface of the brain and the data acquired(figure 2.8B). The coronal scans of the fixed brain corresponded to the transverse slices in the post mortem scans. Using the sagittal pilot scans to visually orient the brain, horizontal, 10 mm thick brain slices were cut using a standard brain knife guided by 10 mm outside diameter rods; the inferior surface of each brain slice then corresponded, as closely as possible, to the centre of the equivalent M R I . Black and white photographs of the inferior surface of each 10 mm thick brain slice were obtained and digitized so that they could be displayed on the Vax. Focussing was carried out on the largest brain slice, and thereafter kept constant. The brain size was scaled on the computer using a grid that was photographed at the same time with the same focus setting as that used for the brain. Lesions were identified on the actual brain slices, and if they were visible on the digitized image of the slice, they were outlined using the program described above, and the lesion areas determined. After comparison of the gross lesions with the M R I , appropriate blocks were taken for microscopic examination. They were embedded in paraffin and stained with Haematoxylin and Eosin (H k E ) , Phosphotungstic Acid Haematoxylin ( P T A H ) and Luxol Fast Blue- Cresyl Violet ( L F B - C V ) . Occasional sections were also stained with Bodian's stain for axons and Marchi's stain for neutral fat [116]. The following criteria were used to differentiate the pathology present in each lesion: 1. Inflammation: sections stained with H & E were examined for the presence of perivascular inflammation.  Lesions were then graded according to the number of  perivascular cuffs present. 2. Demyelination: L F B was absent in areas of total demyelination and pale in areas of partial demyelination.  Chapter 2.  35  MRI of Multiple Sclerosis  3. Gliosis: using sections stained with P T A H , lesions were examined for fibrillary gliosis. The more fibrocytes present, the denser the gliosis was considered to be. The detailed method of evaluation is given in table 2.1. The pathology of each section was then compared with the quantitative M R I data obtained from the same area.  2.3 2.3.1  Results Case 1  A post mortem scan was not obtained from this patient.  Comparison of the inferior  surface of the 10 mm thick brain slice with the equivalent M R I (figure 2.9) showed a remarkable similarity between the two.  Quantification of these areas showed the  accuracy with which M R I could detect the extent of disease in this case (figure 2.10). Comparison of a slice at a lower level with the equivalent M R I again corresponded well with one another (figure 2.11).  A section was removed from the three lesions seen in  the right cerebral hemisphere, just above the level of the lateral ventricles. Microscopic examination of these lesions showed them to be completely demyelinated and gliotic. The degree of gliosis varied in each lesion, and in lesion 8 cystic spaces were also present (figure 2.12). The detailed histopathology of each lesion is given in table 2.2. On the M R I , the lesions did not all have the same intensity. The T i and T values reflected this 2  observation (table 2.3).  Chapter 2. MRI of Multiple Sclerosis  36  B.  Figure 2.9: Comparison of the inferior surface of a 10 mm thick brain slice in case 1 with the corresponding MRI. A. Spin echo MRI level 3, TE = 40 ms and the TR = 2 sec. B. Inferior surface of fixed brain slice at the same level.  Chapter 2. MRI of Multiple Sclerosis  37  Figure 2.10: Display of SE M R I level 3 together with the equivalent digitized negative of fixed brain slice on the Vax 11/750 computer. The lesions are outlined on both the image and the digitized brain slice. Lesion area on M R I = 489 m m and actual lesion area on brain slice = 562 m m . 2  2  38  Chapter 2. MRI of Multiple Sclerosis  I. Demyelination (D)  Number and Size (0-4) 0. none 1. single small foci 2. several small foci 3. one large confluent area 4. several large confluent areas.  Severity a. none b. identified only microscopically c. macroscopically identified but mild. d. marked but partially preserved myelin. e. totally demyelinated (no stained myelin).  II. Perivascular Inflammation (P)  Number (0-4) 0. none 1. 1-3 vessels 2. 4-6 vessels 3. 6+ vessels 4. almost all vessels  Severity a. none b. only a few MNC c. 1-2 layers of MNC d. 3-5 layers of MNC e. 5+ layers of MNC  III. Gliosis Astrocytosis (G)  0. none 1. slight, marginal 2. slight, diffuse 3. moderate, diffuse 4. marked (densely gliosed)  f.fibrillarydominant p. protoplasmic dominant  IV. Others  1. MO : presence of MO  2. Ax : axonal degeneration  Table 2.1: Table showing details of scoring for pathology of sections taken from the inferior surface of each 10 mm thick brain slice. MNC - mononuclear cells, MO - macrophages.  Figure 2.11: Comparison of gross pathology with the spin echo MRI. A. SE MRI level 4, TE = 40 ms and TR = 2 sec. B. Inferior surface of 10 mm thick fixed brain slice at the equivalent level.  LESION  DEMYELINATION  INFLAMMATION  GLIOSIS  8 9 10  4e 4e 4e  lb lb 4d,M0  2f 3p 3-4p  Table 2.2: Pathological score for the three lesions seen in the right cerebral hemisphere of slice 3, case 1.  Chapter 2.  41  MRI of Multiple Sclerosis  Lesion  Zi(ms)  XiTv(ms)  8 9 10  460 690 430  186± 17 186± 17 186± 17  T (ms)  T r(ms)  110 120 100  54± 6 54± 6 54± 6  2  2.25 3.63 2.25  2A  T2/T2N 2.06 2.26 1.81  Table 2.3: Quantitative M R I data from fixed brain in case 1. TIN and T AT correspond to the average values of T\ and T from normal appearing white matter in the same slice. 2  2  Chapter 2.  2.3.2  42  MRI of Multiple Sclerosis  Case 2  The gross pathology seen on the inferior surface of the 10 mm thick brain slices again corresponded well with the areas of increased signal intensity on the S E M R I (figure 2.13).  The total lesion area seen on the fixed brain was quantified and compared with  that seen on the SE M R I (table 2.4).  A section taken from brain slice 4 in this case  (figure 2.14) showed lesions similar to that seen in case 1. The detailed histopathology  POST M O R T E M  FIXED BRAIN  DIGITIZED B R A I N  Sli ce  Area(mm )  Slice  Area(mm )  Slice  Area(mm )  2 3 4 5 6 7 8 9 10 11 12  0 11 90 560 415 575 587 371 209 9 119  2 3 4 5 6 7 8 9 10 11 12  0 19 141 632 799 465 501 671 93 35 37  1 2 3 4 5 6 7 8 9 10 11  39 115 342 748 709 607 481 889 296 108 24  2  2  2  Table 2.4: Correlation between lesion area as seen on post mortem and fixed brain M R I and the actual lesion area seen on the inferior surface of each 10 mm thick brain slice in case 2 is given in table 2.5, and the quantitative data for this lesion are given in table 2.6.  Chapter 2. MRI of Multiple Sclerosis  43  Figure 2.13: Comparison of the SE M R I with the gross pathology. A . Post mortem SE M R I at level 4. T E = 40 ms and T R = 2 sec. B . Fixed brain S E M R I at equivalent level. Parameters as in A . C. Inferior surface of fixed brain slice at the equivalent level.  Chapter 2. MRI of Multiple Sclerosis  Chapter 2. MRI of Multiple Sclerosis  44  Figure 2.14: Inferior surface of fixed brain slice 4, case 2. of the section taken for microscopic examination.  The box shows the position  LESION  DEMYELINATION  INFLAMMATION  GLIOSIS  7  4e  lb  2f  Table 2.5: Pathological score for section taken from the inferior surface of slice 4, case 2.  Chapter 2. MRI of Multiple Sclerosis  45  SLICE  LESION  2i(ms)  r (ms)  5,PM FB  7 7  520 460  100 140  2  T /T x  lN  1.41 2.14  T2/T2N  1.22 2.13  Table 2.6: Ti and T values from M R I number 5, which is equivalent to the inferior surface of fixed brain slice 4, case 2. The area from which these values were obtained corresponds to that from where the section was taken. Cadaver TIN = 342± 30 ms(18), cadaver T N = 8 4 ± 7 ms(17), fixed brain TJAT = 217± 10 ms(6) and fixed brain T N = 66± 7 ms(22). The numbers in brackets give the number of areas from which the mean and standard deviation were determined. 2  2  2  Chapter 2.  2.3.3  46  MRI of Multiple Sclerosis  Case 3  The areas of increased signal intensity on the SE images corresponded well with the gross pathology see on the inferior surface of each brain slice (figure 2.15 and 2.16).  The  areas of abnormality on the SE images were quantified and compared with the actual pathology seen on the fixed brain slices, and the data are given in table 2.7.  POST M O R T E M Slice Area(mm ) 2  4 5 6 7 8 9 10 11 12  FIXED BRAIN Slice Area(mm )  41 79 515 877 575 448 245 138 0  2  0 47 242 943 555 254 148 201 0  3 4 5 6 7 8 9 10 12  Sections  DIGITIZED B R A I N Area(mm ) Slice 2  1 2 3 4 5 6 7 8 9  9 73 213 1003 494 430 130 69 32  Table 2.7: Correlation between lesion area as seen on post mortem and fixed brain M R I and the actual lesion area seen on the inferior surface of each 10 mm thick brain slice in case 3. were taken from brain slices 5 and 6 in this case; sections were removed from the white matter around the posterior horns of the both lateral ventricles in slice five, and from the right posterior horn in slice 6.  Microscopic examination of these sections showed  demyelinated and gliotic lesions similar to those in cases 1 and 2. The histopathology of each lesion is given in table 2.8. The lesions were clearly seen on the post mortem and fixed images from this patient. The quantitative M R I data from the three areas studied microscopically are given in table 2.9.  Chapter 2. MRI of Multiple Sclerosis  47  Figure 2.15: Comparison of SE image with the gross pathology. A . Post mortem SE M R I at level 5. T E = 40 ms and T R = 2 sec. B . Fixed brain SE M R I at the equivalent level. Parameters as in A . C. Inferior surface of fixed brain slice at the equivalent level.  Chapter 2. MRI of Multiple Sclerosis  48  Figure 2.16: Comparison of SE image with the gross pathology. A . Fixed brain SE M R I at level 6. Tau = 40 ms and T R = 2 sec. B . Fixed brain SE M R I at the equivalent level. Parameters as in A . C. Inferior surface of fixed brain slice at the same level.  Chapter 2. MRI of Multiple Sclerosis  Chapter 2.  49  MRI of Multiple Sclerosis  SLICE  LESION  DEMYELINATION  INFLAMMATION  GLIOSIS  5,L 5,R 6,R  4 5 6  3e 3e 3e  3c 4a,M0 3b  2f 3p 2f  Table 2.8: Pathological score for sections taken from fixed brain slices 5 and 6, case 3. L - left cerebral hemisphere and R - right cerebral hemisphere.  SLICE  LESION  r,(ms)  T (ms)  7,PM 6,FB  4 4  666 460  300 180  1.83 2.07  3.18 2.47  7,PM 6,FB  5 5  600 450  320 180  1.66 2.01  3.39 2.46  8,PM 7,FB  6 6  540 300  120 100  1.45 1.43  1.41 1.39  2  Tr/T  lN  T2/T2N  Table 2.9: T\ and T values from areas examined microscopically. Post mortem TIN = 368± 21(24), T = 9 1 ± 9(7), and fixed brain T = 218± 10(18), T = 71± 6(24). Bracketed numbers denote the number of areas from which the mean and standard deviation were determined. P M - post mortem, F B - fixed brain. 2  2N  1N  2N  Chapter 2. MRI of Multiple Sclerosis  2.3.4  50  Case 4  The MR images in this case showed massive periventricular lesions throughout the cerebrum (figure 2.17). When the brain was cut, these lesions could not be seen grossly (figure 2.18).  The areas of increased signal intensity on the post mortem and fixed  brain spin echo images were quantified and compared (table 2.10). Sections were taken  POST MORTEM Slice Area(mm ) 2  2 3 4 5 6 7 8 9 10 11 12 13  0 0 264 827 1,218 1,064 924 916 363 159 0 0  FIXED BRAIN Slice Area(mm ) 2  2 3 4 5 6 7 8 9 10 11 12 13  0 0 157 646 1,743 629 721 881 228 0 0 0  Table 2.10: Comparison of lesion area seen on the post mortem andfixedbrain MRI in case 4. Slice 1 is the sagittal pilot scan and is not included. from slices 5 and 6 (figure 2.18). Histologic examination showed the areas of increased signal on the SE images to be large areas of partial demyelination with gliosis present. The detailed histopathology is given in table 2.11. The quantitative MRI data showed variations where the histopathology varied (table 2.12) and were different from the values obtained from the totally demyelinated and gliotic lesions seen in the previous three cases.  Chapter 2. MRI of Multiple Sclerosis  51  Figure 2.17: SE MRI from case 4. A. Post mortem MRI level 4. B. Fixed brain MRI at the same level. In both images tau = 40 ms and TR = 2 sec.  Figure 2.18: A . Inferior surface of fixed brain slice 5, case 4. Box shows the position of the section that was removed for microscopic examination. B . Fixed brain slice 6, case 4, showing the two sections removed for microscopic examination.  Chapter 2.  53  MRI of Multiple Sclerosis  SLICE  LESION  DEMYELINATION  INFLAMMATION  GLIOSIS  6,R 5,R 6,L  1 2 3  3d 3d 3e  Oa 2b lb  2 3p 3p P  Table 2.11: Pathological score for lesions taken from slices 5 and 6, case 4. cerebral hemisphere, and R - right cerebral hemisphere.  LESION  Ti(ms)  T (ms)  7,PM FB  1 1  510 340  90 150  1.36 1.63  1.31 2.42  6,PM FB  2 2  830 540  180 200  2.10 2.66  2.14 3.03  7,PM FB  3 3  640 340  130 190  1.71 1.60  1.87 2.94  SLICE  2  Tr/T  1N  L - left  T2/T2N  Table 2.12: T\ and T values from area examined microscopically in case 4. Post mortem T = 381± 12(14), T = 77± 9(20) and fixed brain T = 207± 14(20), T = 65± 5(24). The numbers in brackets give the number of areas from which the mean and standard deviation were determined. P M - post mortem, F B - fixed brain. 2  lN  2N  1N  2N  Chapter 2.  2.4  54  MRI of Multiple Sclerosis  S u m m a r y of R e s u l t s  A summary of the quantitative results from all 4 cases is given in the following tables. Examples of individual lesion areas quantified from the SE M R I and the actual fixed brain slices are given in table 2.13. Quantitative M R I data from normal appearing white matter are given in table 2.14; the data from pathological areas are then presented in order of increasing degrees of demyelination, gliosis and inflammation (tables 2.15, 2.16 and 2.17).  PM MRI 361 206 184 56 63 132 132 184  FB MRI 209 266 139 149 149 149 105  BRAIN 396 397 15 11 203 203 156 254  Table 2.13: Quantitative area measurements from individual lesions. P M - post mortem, F B - fixed brain, B R A I N - data from digitized brain slice.  55  Chapter 2. MRI of Multiple Sclerosis  CASE  . PM  1 2 3 4  I \ (ms) FB  342±30 368±21 381±12  186±17 217±10 218±10 207±14  T ( ms) FB PM 2  84±7 91±9 77±9  54±6 66±7 71±6 65±5  (  Table 2.14: Quantitative M R I data from normal appearing white matter as seen on the post mortem and fixed brain SE M R I .  56  Chapter 2. MRI of Multiple Sclerosis  LESION  T^ms)  T (ms)  AREA  SEVERITY  1,PM FB  3 3  d d  510 340  90 160  2,PM FB  3 3  d d  830 540  180 220  3,PM FB  3 3  e e  640 340  130 190  4,PM FB  3 3  e e  670 460  300 180  5,PM FB  3 3  e e  600 450  320 180  6,PM FB  3 3  e e  540 300  120 100  7,PM FB  4 4  e e  520 460  100 140  8,FB  4  e  460  130  9,FB  4  e  690  130  10,FB  4  e  430  100  2  Table 2.15: T\ and T values from computed images, correlated with severity of demyelination in cases 1-4. Lesions are denoted by the same numbers as those used in the text. P M - post mortem, F B - fixed brain, A R E A - size and number of lesions. 2  57  Chapter 2. MRI ol Multiple Sclerosis  LESION  SEVERITY  TYPE  TaCms)  T (ms)  670 460  300 180  540 300  120 100 100 140  2  4,PM FB  2 2  6,PM FB  2 2  f  7,PM FB  2 2  f f  520 460  8,FB  2  f  460  130  1,PM FB  2 2  P P  510 340  90 160  3,PM FB  3 3  P P  640 340  130 190  2,PM FB  3 3  P P  830 540  180 220  5,PM FB  3 3  P P  600 450  320 180  9,FB  3  P  690  130  10,FB  3-4  P  430  100  f  f  f  Table 2.16: T i and T values from computed images, correlated with severity of gliosis in cases 1-4. Lesions are denoted by the same numbers as those used in the text. P M post mortem and F B - fixed brain. 2  58  Chapter 2. MRI of Multiple Sclerosis  AREA  SEVERITY  r,(ms)  T (ms)  1,PM FB  0 0  a a  510 340  90 160  3,PM FB  1  b b  640 340  130 190  b b  520 460  100 140  LESION  1  2  7,PM FB  1  8,FB  1  b  460  130  9,FB  1  b  690  130  2,PM FB  2 2  b b  830 540  180 220  6,PM FB  3 3  b b  540 300  120 100  4,PM FB  3 3  c c  670 460  300 180  5,PM FB  4 4  a a  600 450  320 180  10,FB  4  d  430  100  1  Table 2.17: T\ and T values from computed images, correlated with severity of inflammation in cases 1-4. Lesions are denoted by the same numbers as those used in the text. P M - post mortem, F B - fixed brain, A R E A - denotes size and number of lesions. 2  Chapter 2.  2.5  59  MRI of Multiple Sclerosis  Discussion  Two questions were asked at the beginning of this chapter: (1) How accurately does M R I reflect the extent of disease in the CNS? and (2) What do variations in signal intensity on the M R I correspond to pathologically? The problem of volume averaging and the shape of the lesion within the slice suggest that comparisons between the M R I and the actual pathology would not be very impressive. The data showed clearly that this is not the case.  The qualitative visual comparisons showed an excellent correlation between the  gross pathology and the areas of increased signal intensity seen on the SE M R I . When these areas were quantified, the difference observed between the abnormal areas on the SE image and the actual pathology seen in the brain was 5-60% when individual lesions were considered. When total brain lesion areas were compared, the range difference was considerably lower, 20-30%. There are a number of reasons for these discrepancies. It is extremely difficult to define the margins of lesions in the subcortical white matter of both fresh and fixed tissue, since the grey matter has a T value of the same order as 2  many of the abnormal areas. In the fixed tissue, the lateral ventricles often appear white, making periventricular lesions margins difficult to define.  It is also difficult to define  small lesions in these areas, giving rise to large errors, and the presence of many small plaques could lead to large cumulative errors. If these small plaques do not fill the slice, the signal intensity will be much lower than it should be, due to the excitation of both normal and abnormal tissue within the 10 mm slice thickness. These studies show that M R I does reflect the extent of M S in the brain, with problems arising when there are small lesions and/or subcortical lesions whose margins cannot be defined.  Some small  lesions may be missed completely using the 10 mm slice thickness. However, unless an MS case consisted only of such lesions, the resulting errors would be small compared to the errors involved in quantifying areas over time in a drug efficacy trial, for example.  Chapter 2. MRI of Multiple Sclerosis  60  Question two is more difficult to answer.  The first point to note is that the effects  of fixation do seem to stabilize over time, as shown in table 2.14.  The values for  normal appearing white matter overlap for the cases studied; this suggests that direct comparisons can be made between the quantitative data from the lesions in each case, although this assumes that the effects of fixation on different types of pathology are the same; for the present time this is assumed to be the case.  Literature data have shown  that fixation decreases the T\ and T values obtained from tissue; this was the true for 2  all cases studied here except case 4; the T values from the fixed brain M R I data were 2  higher than those obtained post mortem. observation.  No explanation has yet been found for this  When the summarized data in tables 2.15, 2.16 and 2.17 are examined,  the T values do not seem to follow any trend. 2  to the computation used to determine T .  This is thought to be due in part  The computed images produced are often  2  extremely noisy (figure 2.19), and when average values are obtained over a R O I , the standard deviation within these regions can vary substantially (22-33%), as shown in table 2.18.  The effects of demyelination and inflammation on the values of T i do not  TISSUE PM PM FB FB  T  2  (ms) 83 81 84 91  Std. Dev. 17.9 26.9 18.7 23.4  Table 2.18: Mean intensity and standard deviation from regions consisting of 18 pixels on a computed T image, from regions of normal appearing white matter. P M - post mortem and F B - fixed brain. 2  seem to follow any particular trend.  In these chronic inactive lesions, however, the  Chapter 2. MRI of Multiple Sclerosis  Figure 2.19: Computed Ti and T images from the fixed brain in case 3. B. T image. 2  2  61  A. Ti image.  Chapter 2.  62  MRI of Multiple Sclerosis  extent of inflammation is minimal and would not be expected to have major effects on the relaxation times.  It is not certain what the effects of demyelination are, since for  the cases studied here, it cannot be separated from the effects of gliosis. How real the differences are between the partially demyelinated and completely demyelinated lesions is questionable, since most of the completely demyelinated lesions were only seen in one slice.  The values for the relaxation times from these areas are likely to be a mixture  of normal and abnormal tissue, giving rise to lower values for the relaxation times than expected. The T\ and T values for case 4 are probably the most accurate, since these 2  lesions extended throughout both hemispheres and the data were obtained from slices where the lesions could be seen in the two adjacent slices. The data on gliosis indicate that the more gliotic a lesion is the higher its T\ value.  Although the T\ values from  lesions 9 and 10 are not as high as that from lesion 2, lesions 9 and 10 are small and could have lower values than expected due to volume averaging as described above. A n increase in T\ with increasingly severe gliosis makes sense, since it is known that the water content of tissue increases as the tissue becomes more gliotic [118].  It was also  noted i n these studies that the effects of multi-slice interference contribute to the observed values of the relaxation times. In the multi-slice pulse sequences used, excitation of the different slices is interleaved to minimize the effects of slice interference; that is, the order of excitation is slice 1,3,5,7,9,11,2,4,6,8,10,12. However, i n case 4, for example, data from normal appearing white matter showed differences as high as 15% (table 2.19).  Fixed  tissue was less affected due to the shorter values for the relaxation times, which means a more rapid return to equilibrium after excitation and data acquisition. It is clear that multi-slice interference could have a significant effect on the lesion T\ and T since they 2  have even longer T\ values than normal white matter.  The lesions i n odd numbered  slices would tend to show lower values than expected. The values for lesions 5, 9 and 10, which are all present in odd numbered slices, would then have longer T\ and T values 2  Chapter 2. MRI of Multiple Sclerosis  63  Slices  Average TIN  Average T N  4,6 5,7  389±12(17) 367±12(18)  84±5(20) 71±8(21)  2  Table 2.19: Post mortem Tj and T values from normal appearing white matter in case 4, showing the effects of multi-slice interference. The numbers in brackets give the number of R O I used to determine the mean and error. 2  than those determined. In conclusion, there is no doubt that M R I is an accurate indicator of the extent of MS in the CNS. However, there are limits to the useful quantitative information that can be obtained using quantitative M R I methods to determine T\ and T . 2  The reasons for  this are threefold: firstly, the effects of different pathology cannot be separated from one another. In addition, the pathology in chronic MS is similar in each case and provides no information on the early lesions, which are believed to be immune-mediated; secondly, volume averaging and multi-slice interference give rise to lower relaxation time values than expected and lastly, the observed T\ and T values obtained using the two-point 2  computational method provide no indication as to the source of the signal in the tissue. This means that it is not possible to determine whether the signal follows a mono- or multi-exponential decay.  In the next chapter, the first of these limits is addressed by  studying an animal model for M S in primates.  The disease was followed in vivo using  M R I with a slice thickness of 5 mm, and the histology compared with the quantitative M R I data at the time of death.  Chapter 3  M R I of Experimental Allergic Encephalomyelitis in Primates  3.1  Introduction  Experimental allergic encephalomyelitis ( E A E ) is an animal model for M S in which attempts have been made to understand the mechanisms of development of early, as well as late, immunologically-rnediated, inflammatory and demyelinating lesions in the CNS. E A E can be produced in susceptible species and strains by sensitization to white matter or to myelin basic protein ( M B P ) in Freund's complete adjuvant ( F C A ) [117]. M B P makes up 30-35% of the proteins in normal CNS myelin.  Manipulation of the exper-  imental conditions for induction of E A E allows the production of not only hyperacute and acute forms but also chronic and remitting-relapsing courses of the disease [119]. As previously described in chapters 1 and 2, M S can run several clinical courses: acute, chronic progressive, chronic relapsing-remitting, or various combinations thereof. Histologically, the disease is characterized by relatively sharply defined demyelinative lesions with preservation of axons. E A E can also present with a variety of clinical courses. The classical form is usually an acute disorder with inflammatory and/or demyelinatingnecrotic processes depending on the species studied: in guinea pigs and rats it is almost purely inflammatory, whereas in rabbits and monkeys it is both inflammatory and demyelinating, and in monkeys even necrotic.  The early lesions are relatively small and  perivascular but they may coalesce into large, irregularly shaped ones. Monkeys sensitized with M B P in F C A may show all three forms: hyperacute, acute  64  Chapter 3. MRI of Experimental Allergic EncephalomyeUtis in Primates  65  or chronic, the last being the least common unless treatment is given to prolong the course. Thus, most monkeys sensitized to MBP in FCA die with either hyperacute or acute EAE. Clinically, the acute form is characterized by rapid onset of abnormalities in vision and paralysis, leading to death within a few days. Pathologically, hyperacute E A E in monkeys is marked by severe inflammation (with neutrophils, hemorrhage, edema and fibrin), demyelination and varying degrees of necrosis. Although many of these changes resemble acute necrotizing hemorrhagic leucoencephalopathy in humans, some lesions resemble acute MS. Acute primate E A E was studied in order to observe the evolution of acute, immunologicalry-mediated, MRI-visible, CNS lesions. Alvord et al manipulated the acute model to produce remissions by treatment with myelin basic protein (MBP) and corticosteroids, followed by relapse on withdrawal of therapy [120]. At post mortem, these relapsing primates frequently show demyelinating lesions with preservation of axons, similar to demyelinating plaques seen in MS. Four animals were treated in this manner, and although changes were observed both clinically and on the MRI, the monkeys did not go on to fully recover and relapse.  Prior to this time, it has not been possible  in primate EAE, to mimic the spontaneous relapsing-remitting course seen in MS (both clinically and on the MRI). However, the results from case 10 describe the development of subclinical E A E with MRI lesions showing both resolution and a relapse, demonstrating that spontaneous dissemination of lesions in time and space in E A E is possible following a single immunization and without subsequent treatment.  This chapter describes the  use of both qualitative and quantitative MRI to study acute and chronic primate EAE.  Chapter 3. MRI of Experimental Allergic Encephalomyelitis in Primates  3.2 3.2.1  66  Methods Induction of E A E  Acute E A E Permission for carrying out these studies was granted by the Animal Care Committee at the University of British Columbia. Four male Macaco, fascicularis monkeys, weighing 3-5 kg, were fed a regular diet of monkey chow and fresh fruit before and after induction of E A E . E A E was induced by intradermal injection of 0.15 ml of a water-in-oil emulsion containing 0.5 mg of Mycobacterium tuberculosis and 15 mg of M B P , divided among three sites in the hind foot pads [120]. The M B P was extracted using the batch method of Diebler et al.  [121](see appendix A for details of extraction process).  Clinical  observations of the monkeys were made 3 times daily beginning 7-9 days after inoculation (Table 3.20) [122].  Production of Chronic E A E As mentioned in the introduction, acute E A E can be treated to produce remission followed by a relapse on withdrawal of therapy [120].  Four animals were treated at the  appearance of definite neurological signs with M B P (2 mg/kg/day) and dexamethasone phosphate (0.5 mg/kg/day).  The M B P and steroid were dissolved in 10% D M S O in  saline to facilitate absorption, and placed in Alzet mini-osmotic pumps (model 2001, Alza, Palo Alto, Ca. 94304) with known volumes and flow rates.  The calculation of  amounts of M B P and steroid required is shown in equation 3.34. w  dmv  /  (3.34)  where w = weight of M B P or steroid required, d = dose of steroid or M B P (mg/kg/day), m = weight of monkey (kg), v = volume of micro-osmotic pump ((A) and / = flow rate  Chapter 3. MRI of Experimental Allergic EncephalomyeHtis in Primates  1. ?: Prodromal Signs Anorexia (food pellets scattered on the floor of the cage) Weight loss Inactivity Slow response to startle Irritability Yawning  2. ± : Mild Neurological signs Drooling Nystagmus Transient tremor Clumsiness Head tilt Apathy (indifference) "Headache" (acute distress)  3 . +: Moderate Neurological Signs Pupillary abnormalities (unequal, dilated, poorly reactive to light) Ptosis Strabismus Tremor Seizure Paresis Incontinence Body twisting Blindness Ataxia Akinesia  4. -f+: Severe Neurological Signs Paraplegia Hemiplegia Quadriplegia Lethargy or somnolence  5. ++: Moribund Semicoma Coma Decerebration Decortication  6. + + + + : Death  Table 3.20: Clinical scoring for monkeys inoculated to produce E A E .  67  68  Chapter 3. MRI of Experimental Allergic EncephalomyeHtis in Primates  of pump (/zl/day). The pumps were then implanted subcutaneously in the backs of the animals between the shoulder blades.  Placement of the pumps in this area prevented  the animals tampering with the stitches, which would inhibit healing and could cause infection. After the first four cases were completed, Alvord and co-workers altered the experimental protocol to use intradermal injections i n the hind legs rather than the footpads. This modification was shown to produce a milder form of E A E with later onset. protocol was adopted for subsequent studies on the primates.  This  Case 5 was inoculated  using the new protocol with the same dose of M B P and F C A as previous cases. However, the disease still progressed quite rapidly and the pathology observed had a hemorrhagic component, which is not analogous to M S . Treatment to produce remission was not successful. The data from case 5 were similar to that obtained from cases 1-4, therefore the data are not included in the results section. In an attempt to produce both a milder form of the disease, which would increase the chances of successful treatment to produce remission, and to alter the pathology observed, cases 6 and 7 were inoculated with two thirds of the M B P dose given to previous animals (total M B P dose of 10 mg).  These  two cases again progressed quite rapidly, and although they did respond to treatment, they did not fully recover. The M B P dose was reduced still further to a total dose of 5 mg; cases 8, 9 and 10 were inoculated using this dose.  On the appearance of clear  neurological signs, case 8 was nursed to see if she would recover, but survived only 10 days from the onset of neurological signs. Case 9 was treated with steroid and M B P as in the previous cases, but again failed to recover completely. Case 10 developed subclinical E A E without treatment, and went on to develop neurological signs at a later date. Cases 6-10 were monitored daily from day 8 after inoculation for the onset of clinical signs.  Chapter 3. MRI of Experimental Allergic EncephalomyeHtis in Primates  3.2.2  69  M R I Protocol  M R I data were collected on the Picker International imaging system described previously; a receiver coil, (saddle design, built by Picker International) with an aperture of 15 cm was used to obtain the images.  Multi-slice SE and IR pulse sequences were employed  (figure 2.7, page 31), both of which provided 8 contiguous, 5 mm thick slices.  Echo-  delay times (TE) of either 40 and 60 ms or 40 and 80 ms and an inter-pulse delay (TI) of 400 ms were used; in all sequences the repeat time (TR) was 2 sec. Cases 6-10 were scanned using multi-slice multi-echo pulse sequences; T E values of 40/80 ms or 40/120 ms were employed. Spin-echo sequences are referred to as SETR/TE  m  a  U figures. The  monkeys were positioned in the scanner using the laser available for patient alignment, such that the inner canthus of both eyes were aligned perpendicular to the direction of the static magnetic field; the appropriate flexion of the neck was determined visually. For comparisons to be made, repositioning was critical i n the serial scans: a sagittal pilot scan was obtained in 0.8 min, allowing the position of the monkey to be checked and positioning altered if necessary.  The pilot scan also allowed the centre of the first  transverse slice to be positioned on the superior surface of the brain (figure 3.20A); this was important for subsequent comparison with the gross and microscopic lesions seen pathologically. The pulse sequences were chosen to allow direct comparison with human MS data, and also to allow the computation of the spin-lattice relaxation time (T\) and the spin-spin relaxation time (T2); the latter computations were performed using the manufacturers software as described in chapter 1, page 21. The quantitative M R I data were transferred to a Vax 11/750 and using the program described in chapter 2, T i and T values were obtained from the computed images by 2  outlining an area of 14 pixels at the centre of the lesions or from an area of normal appearing white matter.  The same lesions were followed serially and, where possible,  Chapter 3. MRI of Experimental Allergic Encephalomyehtis in Primates  70  Figure 3.20: A . Sagittal pilot scan of monkeys brain. The grid shows the position of the 8 transverse slices to be obtained. B . Sagittal pilot scan of fixed monkey brain. The grid shows the position of the 8 coronal slices to be obtained. The coronal slices of the fixed brain are equivalent to transverse images in vivo. In both images T E = 26 ms and T R = 800 ms.  Chapter 3.  MRI of Experimental Allergic EncephalomyeHtis in Primates  71  Figure 3.21: Image of monkey brain showing lesions outlined using a trackball. lesions that extended through three slices were chosen to minimize the effects of volumeaveraging.  In addition to obtaining quantitative M R I data, for cases 8-10 total lesion  area, slice lesion area and individual lesion area were followed over time on the SE images (figure 3.21). The monkeys were anaesthetized for scanning using a 12:1 combination of ketamine hydrochloride (Roger/STB Inc. Chemical name ±-2-o-chlorophenyl-2-methylaminocyclohexanone hydrochloride, 100 mg/ml) and rompun (Haver-Lockhart. Chemical name 2(2,6-di-methylphenylamino)-4H-5,6-dihydro-l,3-thiazine, 20 mg/ml) (0.3 mg/kg). Since the monkeys were supine while being scanned, atropine (Clark Rogers L t d . 0.5 mg/ml) was also administered at the time of anaesthesia (0.2-0.4 ml) to prevent oral fluid accumulation and aspiration.  Chapter 3. MRI of Experimental Allergic EncephalomyeHtis in Primates  72  Serial M R I of Acute E A E The monkeys were scanned before inoculation and then daily from day 9 after inoculation.  Before each set of scans was obtained, the monkeys were checked for evidence of  neurological impairment. Following the earliest detection of changes on the M R I (as indicated by increased signal intensity on the S E images) [13], images were obtained 3 times daily until death.  T\ and T computations were carried out for each set of M R images 2  and the values recorded for the abnormal areas. The ranges of T\ and T corresponding 2  to normal-appearing white matter or brainstem (case 1) were obtained from statistical analyzes on 8-24 areas from the equivalent slice in 3 serial scans; both are comparable to those seen in humans [123]. When each monkey became incapacitated, being unable to feed or drink by itself, it was sacrificed using 3 ml of euthanyl ( M T C Pharmaceuticals) intracardially while under anaesthetic. When possible, immediate pre-mortem and postmortem scans were obtained. buffered formalin.  The brain was then removed and fixed in 10% neutral  After a period of at least 1 week a set of M R images was obtained  from the fixed brain.  The brain was positioned on its superior surface and a sagittal  pilot used to check its positioning. The centre of the first horizontal slice was positioned on the superior surface of the brain (figure 3.20B), thereby allowing direct comparison between the images of the fixed brain with both the in vivo images and the gross and microscopic pathology.  Serial M R I of Chronic E A E A baseline scan was obtained at the time of inoculation, and serial M R I data were obtained every few days beginning any time between day 9-15 after inoculation. Following the earliest detection of changes on the M R I , the monkeys were scanned daily. Contrast enhanced scans were obtained from cases 6, 7 and 10 after intra-venous injection of Gd-  Chapter 3. MRI of Experimental Allergic Encephalomyehtis in Primates  73  D T P A [61,62,63,64]. Multi-slice SE pulse sequences were used to obtain these contrast enhanced scans, with T E = 26 ms and T R = 500 ms. This sequence provided 4-6 slices, each 5 mm thick. When clear neurological signs were observed, mini-osmotic pumps were implanted as described above, and daily scanning continued until death. Pre-mortem or immediate post-mortem scans were obtained from each animal. As for the acute cases, the brains were removed and fixed in 10% neutral buffered formalin, and a set of M R I obtained from the fixed brain.  3.2.3  Pathology  Using the sagittal pilot scan to visually orient the brain, horizontal brain slices, each 5 mm thick, were cut using a standard brain knife guided by 5 mm outside diameter glass rods; the inferior surface of each slice then corresponded as closely as possible to the centre of the equivalent M R I . After comparison of the gross lesions with the M R I data, appropriate blocks were taken for microscopic evaluation, embedded in paraffin and stained with Hematoxylin and Eosin (H & E ) , Luxol fast blue combined with periodic acid-Schiff and hematoxylin (LFB-PAS-H), with Holmes' axon stain (LFB-Holmes) or with Nissl's stain (LFB-Nissl) and Gallocyanin for myelin combined with Darrow red (gallo-DR). In addition, frozen sections were stained with oil red 0 and hematoxylin (ORO-H) [116].  Chapter 3. MRI of Experimental Allergic Encephalomyelitis in Primates  3.3  74  Results  3.3.1  Case 1  M R I and Clinical Observations A n area of increased signal intensity on the SE image was observed in the brainstem (midbrain) 15 days after inoculation. lethargy and bilateral ptosis.  This change was associated with the onset of  Over the course of the next few days there was little  change in the M R I (figure 3.22), and the monkey remained clinically stable; hence the monkey was only scanned once a day on days 17 through 21. On day 21 after inoculation, the MRI-visible lesion spread rostrally and caudally. This change was associated with worsening of the ptosis in the left eye. On day 22, the M R I remained positive in the same three slices and the ptosis in the left eye was worse.  B y late evening of the same day, the monkey was off balance and  the right arm appeared spastic. On day 23 the monkey lay prostrate in the cage.  He was nursed all day to see if  he would recover but had to be sacrificed at 7.35 p.m. A n immediate post mortem set of scans was obtained. The brain was removed and immediately placed in 10% neutral buffered formalin. The lesion, previously easily seen on the M R I (both S E and IR), could not be seen in the images of the fixed brain. This is in contrast to the studies on brains from humans with chronic MS as described in chapter 2, in which the demyelinated and gliotic lesions were easily and reproducibly seen on both immediate post mortem and fixed brain scans.  Figure 3.22: Serial SE2000/40 M R I obtained from case 1. B . Day 18 after inoculation.  A . Day 15 after inoculation.  Chapter 3.  MRI of Experimental Allergic Encephalomyelitis in Primates  76  Comparison of the M R I with the Gross Distribution of Lesions Photographs of the 5 mm thick brain sections and the histologic preparations were compared with the equivalent M R images, as shown in figure 3.23.  A 10 x 10 mm lesion,  mainly in the midbrain and extending into the rostral pons, could be seen on the gross sections. As was observed on the M R I , the lesion was present in three 5 mm thick brain sections.  Another small lesion consisting of inflammation surrounding about 1 dozen  blood vessels over an area of 5 x 2 mm could be seen in the left occipital subcortical white matter; this small lesion was not detected on the M R I .  Histology The initial area of increased signal observed on the SE image was found to correspond to a confluent, perivascular inflammatory lesion, in which the vessel walls were infiltrated by white blood cells diffusing out into the tissue (figure 3.24A). Most of the cells were polymorphonuclear leucocytes with a small number of lymphocytes and a large number of macrophages; most of the macrophages were large round cells (gitter cells) filled with granular sudanophilic lipids (figure 3.25) and LFB-positive myelin debris (figure 3.24B). There was no hemorrhage in this lesion but there was evidence of subtotal necrosis with swelling, beading and destruction of axons. A t the edges of the lesion, the myelin sheaths were remarkably swollen (figure 3.24C).  Quantitative M R I The Ti and T values from the lesion were measured as a function of time after inoculation 2  (figure 3.26) and were shown initially to increase, then stabilize while the monkey was clinically stable.  Figure 3.23: Comparison of the gross distribution of lesions with the equivalent M R I in case 1. A . SE ooo/4o image on day 23 showing lesion in the brain stem. B . The inferior surface of the 5 mm thick brain slice at the same level. 2  Chapter 3. MRI of Experimental Allergic Encephalomyelitis in Primates  A.  B.  Figure 3.24: Histopathology of lesion in case 1 A . H & E showing diffusion of polymorphonuclear cells, lymphocytes and macrophages into the tissue around the vessel. B . L F B - PAS-H) showing LFB-positive myelin debris at the centre of the lesion. C. Holmes' stain showing swelling, beading and destruction of axons at the centre of the lesion.  78  Chapter 3. MRI of Experimental Allergic EncephalomyeHtis in Primates  "IH 15  ' ' ' ' • ' ' 16  17  18  19  20  21  22  Time after inoculation (days)  79  r23  Figure 3.26: Plot of Ti and T versus time after inoculation in case 1. T i and T values are given from normal appearing white matter plus or minus two standard deviations. 2  2  Chapter 3. MRI of Experimental Allergic Encephalomyelitis in Primates  3.3.2  80  Case 2  MRI and Clinical Observations A n area of increased signal on the SE image was observed in the left cerebral hemisphere on day 22 after inoculation (figure 3.27A). A non-contrast C T scan obtained the same day did not show any evidence of hemorrhage or low-density. The monkey was sacrificed on that day and an immediate post mortem scan obtained.  The appearance of this  abnormal area was not associated with any neurological signs apart from a small loss in weight. The lesion was not visible on the images of the fixed brain.  Comparison of the MRI with the Gross Distribution of Lesions Photographs of the 5 mm thick brain section and the histologic preparation were compared with the equivalent M R I . The left cerebral hemispheric lesion present on the M R I was seen clearly on the gross section (figure 3.27B), as mildly hemorrhagic.  On the  histologic preparation the lesion was 1 0 x 8 mm, centred around the genu of the internal capsule, extending slightly into the striatum (caudate and putamen) and into the nucleus ventralis anterior thalami.  This lesion consisted of small areas of inflammation which  did not coalesce.  Histology The lesion had many of the same characteristics as the lesion in case 1, but it was less confluent and was hemorrhagic. The medial half of the lesion consisted of smaller perivascular lesions, while the lateral half consisted of larger haemorragic lesions. Within this lesion there were collars of hemorrhage around the blood vessels, displacing the exudate peripherally (figure 3.28). The exudate consisted mostly of polymorphonuclear cells. A small amount of hemorrhage was seen in the anterior portion of the lesion, but  Figure 3.27: Comparison of the SE M R I with the gross pathology in case 2. A . SE oo/40 M R I on day 22 after inoculation. B . The inferior surface of 5 mm thick brain slice at the same level. 20  Chapter 3.  MRI of Experimental Allergic Encephalomyelitis in Primates  82  Figure 3.28: Histopathology of lesion in case 2. Section stained with H & E showing the collar of hemorrhage around the vessel displacing the exudate peripherally. the major component was posterior and lateral, and these two areas coalesced. A t the site of thrombosed venules there were many polymorphonuclear cells.  Some of the smaller  lesions that comprised the large lesion seen grossly, had as many polymorphonuclear leucocytes but no hemorrhage; others were composed of predominantly mononuclear cells (lymphocytes and a few plasma cells) distending the Virchow-Robin spaces and infiltrating diffusely into the parenchyma.  Chapter 3.  3.3.3  MRI of Experimental Allergic Encephalomyehtis in Primates  83  Case 3  M R I and Clinical Observations Areas of increased signal intensity on the S E images were seen in both cerebral hemispheres on day 11 after inoculation. There were no neurological signs observed at that time. Twelve hours later, the lesions in both hemispheres had enlarged (figure 3.29A, B). The monkey's movements were slow and he was urinating frequently.  A further twelve  hours later, the lesions had spread dramatically in both hemispheres.  The monkey was  clumsy and his peripheral vision was diminished. Seven hours later, he was found dead in his cage. A set of scans was obtained post mortem, and the brain removed and placed in 10% neutral buffered formalin. After 10 days of fixation, a set of scans was obtained from the fixed brain; the lesions were barely visible on the images of the fixed brain (figure 3.30). Comparison of the M R I with the Gross Distribution of Lesions Comparison of the photographs of the 5 mm thick brain sections and the histologic preparations with the equivalent M R I showed that the lesions in both hemispheres corresponded well with the M R I appearance (figure 3.31A, B ) . A t M R I level 3 (figure 3.32A, B), two lesions could be seen.  The left cerebral hemispheric lesion was present in the  white and grey matter of the occipital lobe, and measured 12 x 15 mm. The right-sided lesion was more anterior, and present predominantly in the white matter; it measured 12 x 11 mm. A t a lower level ( M R I number 5) the lesions were larger (figure 3.33A, B ) . The left cerebral lesion involved the temporal cortex and white matter, and measured 17 x 10 mm. The lesion included the choroid plexus of the trigone and extended to the posterior edge of the putamen.  Isolated perivascular lesions could also be seen at the level of the  Figure 3.29: Serial SE oo/4o M R I from case 3. A . Day 11 after inoculation. after inoculation. 20  B . Day 12  Chapter 3. MRI of Experimental Allergic Encephalomyehtis in Primates  85  Figure 3.30: SE M R I from fixed brain. The lesions in both hemispheres are pale compared to the bright areas seen on the in vivo images. anterior commissure and beneath the ependyma overlying the head of the caudate. The right cerebral lesion measured 18 x 12 mm and involved the lateral edge of the thalamus (lateral geniculate and pulvinar), the posterior edge of the globus pallidus and the posterior half to two thirds of the putamen; it also included the external and extreme capsules and the anterior portion of the visual radiation.  Histology In M R I number 3, corresponding to brain slice 2 (figure 3.32A, B) there were two lesions, one in each hemisphere.  The left lesion consisted of two areas: posteriorly the lesion  was more hemorrhagic, being diffusely so in the white matter; the hemorrhage occurred around small vessels that were only slightly inflamed. The lesions in the white matter were large compared to those in the cortex, which were not hemorrhagic. The exudate was predominantly polymorphonuclear cells.  The anterior portion of the lesion was  mostly present in the white matter and it was similar to the posterior portion with a  Chapter 3. MRI of Experimental Allergic Encephalomyehtis in Primates  Figure 3.31: Comparison of the SE M R I with the gross pathology in case 3. A . S E M R I slice 4. B . The inferior surface of 5 mm thick brain slice at the same level.  Chapter 3. MRI of Experimental Allergic EncephalomyeUtis in Primates  87  Figure 3.32: Comparison of SE MRI with the gross pathology in case 3. A. SE oo/4o MRI at level 3. B. The inferior surface of 5 mm thick brain slice at the same level. 20  Figure 3.33: Comparison of the SE M R I with the gross pathology in case 3. A . SE M R I slice 5. B . The inferior surface of 5 mm thick brain slice at the same level.  Chapter 3.  MRI of Experimental Allergic Encephalomyehtis in Primates  89  Figure 3.34: Histopathology of lesion observed in the left cerebral hemisphere at M R I level 3. Section stained with H & E showing compact exudate similar to that seen in cases 1 and 2. large exudate mainly containing polymorphonuclear cells. The lesion in the left cerebral hemisphere was a mixture of hemorrhagic and nonhemorrhagic areas. The hemorrhagic areas were found on the periphery of the lesion with very little inflammation; in the centre of the lesion there were areas of perivascular inflammation with exudate similar to that in cases 1 and 2 (figure 3.34). The right cerebral lesion in M R I number 3, corresponding to brain slice 2 contained many axons which showed swelling and fragmentation.  A large number of axons re-  mained intact in spite of the large size and compactness of the exudate. lesions within this area the axons were almost completely destroyed.  In the larger  There was no su-  danophilic debris observed in this lesion. The hemorrhagic areas were only present in the white matter. In M R I number 5, corresponding to brain slice 4 (figure 3.33A, B ) , lesions were again seen in both cerebral hemispheres.  The left-sided lesion was quite hemorrhagic, being  Chapter 3.  MRI of Experimental Allergic EncephalomyeUtis in Primates  90  Time after inoculation (days)  Figure 3.35: Plot of T\ and T versus time after inoculation for case 3. The dashed line denotes data from normal tissue plus or minus two standard deviations. 2  more so laterally, and the most anterior portion of the lesion was not hemorrhagic. The exudate was, again, predominantly polymorphonuclear cells. The right-sided lesion was hemorrhagic laterally, with no hemorrhage medially. The exudate was similar to that in the left hemisphere, being predominantly polymorphonuclear cells. The lesions in case 3 were more hemorrhagic and necrotic than the lesions in cases 1 and 2. In addition, the areas of inflammation and hemorrhage were more confluent than in cases 1 and 2. Quantitative M R I The T. and T values of the left hemisphere lesion in M R I slice 4 were measured as a 2  function of time after inoculation and were shown to increase from the time of detection until death. The increase was not dramatic (figure 3.35).  Chapter 3. MRI of Experimental Allergic EncephalomyeHtis in Primates  Figure 3.36: 3.3.4  SE oo/40 2 O  91  obtained from case 4 on day 16 after inoculation.  Case 4  M R I and C l i n i c a l Observations A n area of increased signal intensity on the SE image was observed in the left cerebral hemisphere on day 16 [124] after inoculation (figure 3.36). This lesion was not associated with any neurological signs, apart from a small loss in weight. The lesion spread laterally, medially, superiorly and inferiorly over the course of the next 3 days (figure 3.37 A , B , C). On day 17 after inoculation, the monkey appeared spastic in his right arm which became progressively more severe. Later the same day he seemed to neglect his right visual field, suggesting a right hemianopia; on day 18 his left leg was affected.  Twentyfour hours  later he was prostrate in his cage and his whole right side appeared spastic.  He was  sacrificed at 2 a.m. on day 20 after inoculation. A n immediate post mortem scan was obtained, and the brain was removed and placed in 10% neutral buffered formalin. Only parts of the lesion could be seen on the M R I of the fixed brain (figure 3.38).  Figure 3.37: Serial SE oo/4o M R I from case 4. A . Day 16 after inoculation. B . Day 18 after inoculation. 20  Chapter 3. MRI of Experimental Allergic EncephalomyeHtis in Primates  93  Figure 3.38: Fixed brain SE oo/40 MRI from case 4 showing abnormal area in the left cerebral hemisphere. This gliotic lesion appears as a bright area of increased T , just as was seen in the fixed brains of chronic MS patients. 20  2  Comparison of the MRI with the Gross Distribution of Lesions Comparison of photographs of the 5 mm thick brain sections and the histologic preparations with the equivalent MRI, showed a large hemorrhagic lesion in the left cerebral hemisphere (figure 3.39 A, B). The lesion measured 21 x 18 mm and involved the genu and posterior limb of the internal capsule, the posterior two-thirds of the putamen, the external and extreme capsules and the insular cortex.  It also involved the temporal  cortex and white matter anteriorly and extended back to the anterior portion of the visual radiation, the tail of the caudate and the lateral thalamus (lamina medullaris or internus thalami). Smaller inflammatory lesions were also seen in the splenium of the corpus callosum, the columns of the fornix and in the right anterior thalamic nucleus. Smaller sub-ependymal lesions were also present at the head of the caudate on the left side.  At lower levels the lesion also involved the lateral geniculate nucleus, the optic  tract and the lateral half of the amygdaloid nucleus. A smaller lesion, measuring 6x3  Figure 3.39: Comparison of the SE M R I with the gross pathology in case 4. A . Post mortem SE ooo/60 M R I at level 5. B. The inferior surface of 5 mm thick brain slice at the same level. 2  Chapter 3.  MRI of Experimental Allergic EncephalomyeHtis in Primates  95  mm, was present in the right cerebral hemisphere; it involved the retrolenticular part of the internal capsule and was visible on the M R I . Histology The larger lesion in the left cerebral hemisphere was quite edematous, necrotic and hemorrhagic laterally and anteriorly with extensive destruction of axons.  More posteriorly  and medially, involving the thalamus, it was only slightly hemorrhagic. The most posterior part of the lesion involving the visual radiation (figure 3.40) was more demyelinated than necrotic, with abundant myelin debris, many polymorphonuclear cells and lymphocytes. There was almost complete preservation of axons with only a few axon swellings. A n old infarct could also be seen in the anterior temporal lobe and cortex; it showed gliosis with Wallerian degeneration of the underlying white matter, and only slight inflammation.  This old infarct was the region clearly seen on the images of the fixed brain  (figure 3.38).  In the corpus callosum (figure 3.41), the exudate displaced most of the  axons to the periphery, but a few myelinated axons coursed through the exudate.  The  smaller lesion in the right hemisphere showed moderate destruction and displacement of axons, but many myelinated and demyelinated axons coursed through the exudate. The exudate was made up of predominantly polymorphonuclear cells with a large number of lymphocytes and histiocytes.  Quantitative M R I The Ti and T values of the lesion first observed on day 16 were measured as a function 2  of time after inoculation (figure 3.42); both sets of values increased from the day of detection until death.  Chapter 3. MRI of Experimental Allergic EncephalomyeHtis in Primates  96  Figure 3.40: Histopathology of posterior and medial part of lesion i n the left visual radiation of case 4. Section stained with gallo-DR.  Chapter 3. MRI of Experimental Allergic Encephalomyehtis in Primates  97  Figure 3.41: Histopathology of lesion in the corpus callosum of case 4. Section stained with gallo-DR. 700 600.  Ij LESION  £ «00  4VA  Tj mmi  16  17  11  W  20  Time after inoculation (days) Figure 3.42: Plot of T i and T versus time after inoculation for case 4. T i and T values from normal appearing white matter are given plus or minus two standard deviations. 2  2  Chapter 3.  3.3.5  MRI ol Experimental Allergic EncephalomyeUtis in Primates  98  Cases 6 and 7: G d - D T P A Study  M R I Data Cases 6 and 7 were inoculated together and developed identical areas of increased signal intensity on day 22 after inoculation, in slices 4, 5 and 6 (figure 3.43). Later the same day, the lesions had spread rostrally and caudally into slices 3 and 7. Without removing the animals from the coil during their respective scans, G d - D T P A was administered I.V. (0.16 mmole/kg to case 6 and 0.3 mmole/kg to case 7) and serial images obtained over the next 45 minutes.  In both cases, this data showed breakdown in the blood brain  barrier (figure 3.44). Two mini osmotic pumps, containing 50 mg M B P (54 mg) and 15 mg (14 mg) steroid respectively for case 6 (case 7), were implanted in each animal after this scan, together with an intra-peritoneal (LP.) injection of 1 mg of steroid dissolved in saline. Eight hours later, the monkeys were rescanned; case 6 died during the procedure; case 7 developed pulmonary edema during the procedure.  She seemed to recover, but  then developed heart failure and could not be revived; a post mortem scan was obtained from each animal. It was not possible to obtain pathology from these two brains, due to an error in the fixation process.  3.3.6  Case 8  M R I and Clinical Observations Bilateral areas of increased signal intensity were first observed on day 23 after inoculation, in slices 4,5 and 6 (figure 3.45). These lesions were not associated with any neurological signs. On day 24, the lesions had spread and were now seen in slices 3,4,5 and 6 in the right cerebral hemisphere, and slices 4 and 5 in the left cerebral hemisphere (figure 3.46). No clinical signs were apparent at this time.  On day 26 after inoculation, the lesions  had enlarged further (figure 3.47), but only prodromal signs were present clinically. She  Figure 3.43: SE M R I from case 7 on day 22 after inoculation. A . Slice 5. B . Slice 6.  Chapter 3. MRI of Experimental Allergic Encephalomyehtis in Primates  100  Figure 3.44: G d - D T P A scan from case 7 showing breakdown in the blood brain barrier where lesion is enhanced in slice 5. A . Slice 5. B . Slice 6. In both images T E = 26 ms, T R = 500 ms and number of P E increments = 256.  Chapter 3. MRI of Experimental Allergic Encephalomyelitis in Primates  A.  B.  Figure 3.45: S E M R I showing preclinical lesions on day 26 after inoculation A . M R I level 4. B . M R I level 5. C . M R I level 6.  Chapter 3. MRI of Experimental Allergic Encephalomyehtis in Primates  102  Figure 3.46: SE M R I at level 5 from case 8, showing spread of lesions into left cerebral hemisphere.  Chapter 3. MRI of Experimental Allergic EncephalomyeHtis in Primates  103  Figure 3.47: Plot of total lesion area versus time after inoculation in case 8. was watched for the next few days, and when scanned on day 29, some of the lesions had receded while others had increased in size (figure 3.48). On day 32 after inoculation, an occipital lesion was no longer visible; others had increased in size in parts and receded in others. There were also changes in signal intensity.  Clinically, her pupils were dilated and poorly reactive to light; she was apathetic  and her movements were slow. would not eat by herself.  Later the same day, her left leg was affected and she  She was nursed over the next few days to see if she would  recover, but she continued to deteriorate chnically. On day 34, some areas were brighter on the SE images, but had decreased in size. Dramatic increases were seen in T i on slice 5 (figure 3.49), and the lateral ventricles appeared deformed on slice 4. The lesion size and intensity continued to change over the next 9 days (figure 3.47). Although nursing was continued throughout this time, she continued to deteriorate chnically and died on day 43 after inoculation. A post mortem scan was obtained. The brain was removed, and sections taken for in vitro study based on the M R I as follows:  Chapter 3. MRI of Experimental Allergic Encephalomyehtis in Primates  104  A r e a (sq.mm)  24  26  29  32 Time  34  35  37  39  43  (days)  Figure 3.48: Plot of total slice lesion area versus time after inoculation for two SE M R I slices in case 8.  25  30 Time  35  40  45  (days)  Figure 3.49: Plot of Ti versus time after inoculation for for two areas of the lesion in slice 5, case 8.  Chapter 3. MRI of Experimental Allergic Encephalomyelitis in Primates  105  1. Two areas of normal appearing white matter. 2. Two areas of normal appearing grey matter. 3. Section taken from the brain at approximately M R I level 5 , close to the corpus callosum, and labelled lesion 1. 4 . Section taken from the white matter at approximately M R I level 3 / 4 , adjacent to the right lateral ventricle, and labelled lesion 2. 5. Section taken from the right corpus callosum area at approximately M R I level 6, and labelled lesion 3 . 6. Section taken from the left midbrain. 7. Section taken from the corpus callosum area containing both grey and white matter. The results of this in vitro study are presented in chapter 4 .  Chapter 3.  3.3.7  MRI of Experimental Allergic Encephalomyehtis in Primates  106  Case 9  M R I and Clinical Observations A n area of increased signal was seen on the S E M R I on day 26 after inoculation (figure 3.50A); no clinical signs were apparent at this time.  Two days later, the disease had  progressed dramatically, and areas of increased signal intensity were now seen in slices 2,3,4,5,6,7 and 8 (figure 3.50B). Clinically, her pupils were dilated and she had developed nystagmus, tremors and appeared to be completely disoriented, having no concept of her position in relationship to her surroundings; her hearing may also have been affected. Later the same day, two mini-osmotic pumps, containing 43 mg of M B P and 11 mg steroid respectively, were implanted subcutaneously and 1 mg steroid dissolved in saline was administered L P . On day 30 after inoculation, she was completely disoriented and had developed ataxia; her pupils were still dilated. Later the same day, she improved clinically, being able to sit up and reach to grasp food in her hand. B y day 31 after inoculation, she had again deteriorated, being completely disoriented and ataxic; she could also be lethargic. Six hours later, she was found dead in her cage; a post mortem scan showed large lesions in both hemispheres and in the cerebellum and brainstem.  Individual lesions showed  changes in size as shown in figure 3.51, and the total lesion area varied as shown in figure 3.52. Comparison of the M R I with the Gross Distribution of Lesions The areas of increased signal intensity in the SE images corresponded well with the lesions seen grossly on the fixed brain slices (figure 3.53). A large lesion in the left hemisphere extended from slice 2 to slice 6; lesions were also seen in the cerebellum and brainstem, just as was present in the M R I .  Figure 3.50: SE MRI from case 9. A. Slice 3 on day 26 after inoculation, B. Slice 3 on day 29 after inoculation. C. Slice 5 on day 29 after inoculation.  Chapter 3. MRI of Experimental Allergic Encephalomyehtis in Primates  108  400  Area (sg. mm)  26  29  30 Time  31  (days)  Figure 3.51: Plot of individual lesion area versus time after inoculation in case 9. in left cerebral hemisphere slice 3. Lesion i n left cerebral hemisphere slice 4.  Area  2000 1800 1600 1400 1200 (sq.mm) 1000  o  14  26  29 Time  30  31  (days)  Figure 3.52: Plot of total lesion area versus time after inoculation i n case 9.  Lesion  Figure 3.53: Comparison of the M R I with the gross pathology. A . SE M R I at level 4. B . The inferior surface of 5 mm thick brain slice at the equivalent level.  Chapter 3. MRI of Experimental Allergic Encephalomyelitis in Primates  110  Histology  Sections were taken from the centre of the large lesion in the left cerebral hemisphere and from the lesion in the cerebellum. Histologic examination of these areas showed them to be mainly inflammatory, with many perivascular cuffs containing predominantly polymorphonuclear cells; lymphocytes and mononuclear cells were seen in smaller numbers.  Some areas contained dense collections of polymorphonuclear cells wiping out  the parenchyma, with only fragments of axons present.  The primary process in both  lesions was necrosis; this resembles the pathological features of acute hemorrhagic leukoencephalitis. A few perivascular cuffs were predominantly mononuclear cells, which is consistent with the early autoimmune disease process. Quantitative M R I  Changes in T\ and T over time for the lesions studied histologically are shown in figures 2  3.54  and 3.55.  Chapter 3. MRI of Experimental Allergic Encephalomyelitis in Primates  111  Figure 3.54: A . Plot of T versus time after inoculation for lesion i n slice 4. B . Plot of T versus time after inoculation for the same lesion. x  2  Chapter 3. MRI of Experimental  Allergic Encephalomyehtis  in  Primates  112  A.  380 -I 26  1  1  1  1  1  1  1  1  26.5  27  27.5  28  28.5  29  29.5  30  Tine  .  30.5  31  (days)  B.  T2  (ms)  26  26.5  27  27.5  28  28.5  Time  29  29.5  30  30.5  31  (days)  Figure 3.55: A . Plot of T i versus time after inoculation for cerebellar lesion in slice 7. B . Plot of T versus time after inoculation for the same lesion. 2  Chapter 3. MRI of Experimental Allergic Encephalomyelitis in Primates  3.3.8  113  Case 10  MRI and Clinical Observations Serial M R I data were obtained every few days from day 11 to day 26 after inoculation. No lesions were visible on the M R I until this point, and the monkey was chnically asymptomatic. Daily clinical monitoring was continued until day 52 after inoculation, at which time she was scanned at 0.15 T using the multi-slice sequences.  Bilateral hemispheric  lesions were identified on these scans. The SE images showed the right hemisphere lesion to extend from slice 3 to slice 7, with the largest area occurring in slice 5 (155 mm ). 2  The lesion was irregularly shaped and had variable intensity throughout the hemisphere. The extent of the lesion could also be seen clearly on the IR images, on which, lesions appear as dark areas of increased T i .  Areas of increased signal intensity could also be  seen in the left cerebral hemisphere in slices 4 and 5 (figure 3.56). These areas were seen to increase and decrease in size over time. On day 86 after inoculation, a new lesion appeared in the left cerebral hemisphere in an area previously unaffected (figure 3.57A). This lesion could be identified on both the SE and the IR images. Over a period of 8 days the lesion changed in size and intensity and after 23 days was barely discernable on either SE or IR images (figure 3.57B, 3.58). By day 109 post inoculation, all areas of increased signal intensity had almost completely disappeared, apart from pale areas still discernible on slices 4 and 5 in the right cerebral hemisphere. Until this point, no clinical signs were apparent in this animal; periodically however, she demonstrated possible left arm weakness; this was not marked and did not affect her agility. Subtle signs of disease were food scattering, irritability, weight loss and a period of anorexia from day 54 to day 58; these signs are considered prodromal ones, and are not specific for E A E [123]. This animal maintained the use of all four limbs, and vision did not deteriorate. Daily clinical monitoring and serial scanning were continued.  Figure 3.56: SE M R I on day 52 after inoculation. A . Slice 4. B . Slice 5.  Figure 3.57: A . SE M R I level 4 on day 86 after inoculation, showing new lesion in the right cerebral hemisphere. B. SE M R I level 4 on day 103 after inoculation. In both images T E = 120 ms and T R = 2 sec.  Chapter 3.  MRI of Experimental Allergic Encephalomyehtis in Primates  116  160  Time (days)  Figure 3.58: Plot of individual lesion area versus time after inoculation in slice 4, case 10. This plot shows the appearance of the new lesion in the left cerebral hemisphere on day 86. On day 167 after inoculation, she demonstrated obvious clinical signs, holding her left arm to her chest; she did not use her left hand to grasp or eat. This clinical development was not associated with any obvious change in the M R I . She subsequently recovered, then on day 197 after inoculation, a new lesion or an exacerbation of an existing lesion was seen in the right temporal lobe in slice 7; this was associated with the onset of /  apathy and disuse of her left arm.  She once again recovered, and was clinically stable  until sacrificed on day 352 after inoculation. Intra-venous injection of Gd- D T P A during the pre-mortem scan caused no enhancement in the lesion areas. The variation in total lesion area versus time after inoculation is shown in figure 3.59. Comparison of the M R I with the Gross Distribution of Lesions The brain was cut into 5 mm thick slices as described in the methods section. Grossly, there was little pathology to be seen. Figure 3.60A shows the slice in which the new lesion  Chapter 3.  MRI of Experimental Allergic EncephalomyeHtis in Primates  117  Figure 3.59: Plot of total lesion area versus time after inoculation for case 10. Data is shown to day 135; little change was observed in the brain after this point. was observed on day 86 after inoculation; a small abnormal area can be seen grossly. The lesion in the left cerebral hemisphere was not clearly discernible (figure 3.60B), however other areas, which had previously been abnormal on the SE M R I , appeared to be gliotic, and sections were taken for histologic examination from these areas, together with the other areas of interest already mentioned.  Histology The left hemisphere lesion corresponding to the new lesion seen on day 86 after inoculation, consisted of minimal perivascular inflammation; the area was mildly gliotic and hypercellular.  In one area, the number of oligodendrocytes was decreased and reactive  astrocytic cytoplasm could be seen; the lesion was obviously old and inactive. The lesion still visible in the left hemisphere in M R I slice 5, showed one perivascular  Figure 3.60: A . Inferior surface of fixed brain slice showing area of relapse. inferior surface of fixed brain slice equivalent to M R I level 5.  B . The  Chapter 3. MRI of Experimental Allergic Encephalomyelitis in Primates  119  inflammatory cuff; reactive astrocytes were seen and one area was intensely gliotic. The lesion was located close to the lateral geniculate body, as was seen in the M R I , and extended into the cortex. The other areas, which had previously been abnormal on the S E M R I , showed areas of hypercellularity and mild gliosis.  These areas were located close to the posterior  horn of the left lateral ventricle and in the right occipital lobe.  Just as for the lesions  described above, these areas were clearly old and inactive, with no evidence of recent inflammation. Quantitative M R I The changes in T\ and T for the new lesion seen in the right cerebral hemisphere on day 2  86 were plotted, as shown in figure 3.61.  Chapter 3. MRI of Experimental Allergic EncephalomyeHtis in Primates  120  A.  Tl  (ms)  400 395 390 385 380 375 370 365 360  T  70  B.  A.  wv 95  T2  (ms)  90 85 80 •7 S  70  80  90  100 Time  110  120  130  140  (days)  Figure 3.61: A.Plot of Ti versus time after inoculation for the new lesion in the right cerebral hemisphere in M R I slice 4. B . Plot of T versus time after inoculation for the same lesion. 2  Chapter 3. MRI of Experimental Allergic EncephalomyeHtis in Primates  3.4  121  Summary of Results  1. M R I detected acute E A E lesions in primates when only prodromal signs were present chnically.  This M R I change was due to an increase in the T i and T  values in the lesion area.  2  Preclinical lesions have also been seen in the CNS of  guinea pigs with E A E [125]. 2. M R I was an accurate indicator of the gross extent of the lesions of acute E A E [126]. 3. The oldest lesions were the most hemorrhagic, but the hemorrhage occurred in two patterns, one near vessels displacing the exudate peripherally and one at a slight distance from a much less inflamed vessel. Only the former pattern is consistent with the classical belief that hemorrhage is a secondary event in E A E . 4. In lesions with varied histologic components the M R characteristics of the lesions also varied (table 3.21). Conversely, microscopically similar lesions had the same M R characteristics. The order of T\ and T values given in table 3.21, corresponds 2  to increasing degrees of inflammation, demyelination and hemorrhagic necrosis. 5. As the disease progressed, the M R I characteristics also changed. Since the T\ and T values are sensitive to the molecular environment around the protons, these M R I 2  changes are indicative of progressive lesion evolution at a molecular level. 6. Acute inflammatory lesions were not clearly defined on the M R I of the fixed brain. Formalin fixation of these lesions must give rise to a molecular environment that is indistinguishable from normal tissue using M R I . However, just as seen in MS tissue, the small gliotic lesion in case 4 could be clearly seen in the images of the fixed brain.  Chapter 3. MRI of Experimental Allergic Encephalomyehtis in Primates  122  7. The results show, for the first time, that it is possible to produce spontaneous relapsing-remitting E A E in a primate following a single immunization and without treatment. This mimics the M R I and clinical course observed i n the serial studies of relapsing M S patients. 8. Just as for acute E A E , lesions were detected before the onset of clinical signs. The lesions i n case 10 were notably less intense that those seen in the acute cases; this corresponds to a lower T value. A t the time of death, these areas corresponded 2  to mildly gliotic lesions with little evidence of recent inflammatory activity. 9. Treatment with dexamethasone phosphate causes measurable changes in the size and intensity of lesions, but for the cases presented here, was unsuccessful in producing complete clinical remission followed by relapse upon withdrawal of therapy. 10. G d - D T P A studies showed breakdown in the B B B within a few hours of disease onset, and before clinical signs were apparent.  No enhancement was seen in the  lesions i n case 10, which were mildly gliotic. 11. The location of lesions in case 9 could be correlated with the clinical symptoms observed; this is not always possible.  Chapter 3. MRI of Experimental Allergic Encephalomyelitis in Primates  CASE  Ti(ms)  T (ms)  1 9 9 2 3 4  500 500 520 460 540 600  120 120 160 110 260 270  123  2  Table 3.21: T i and T values from lesions in cases 1, 2, 3, 4 and 9 obtained at the time of death. 2  Chapter 3.  MRI of Experimental Allergic EncephalomyeHtis in Primates  124  3.5 Discussion The studies show that MRI can detect EAE in primates before the onset of neurological signs; similarly, in MS, MRI-visible lesions may appear in the CNS unaccompanied by any new clinical symptoms. It remains to be proved whether the first lesion in EAE is pathologically similar to the chnically asymptomatic lesions in MS. It is recognized that hemorrhage is not frequently seen in the CNS of MS patients; however, it may be that the perivascular inflammatory lesions, which may or may not be confluent, could be similar in both acute EAE and MS. As mentioned previously, serial MRI studies carried out on MS patients at U.B.C. have shown asymptomatic MRI detected lesions to come and go with a frequency that is much higher than the development of clinical signs; most of these lesions become smaller and some disappear over time. The time relations make it unlikely that these new and disappearing lesions are demyelinating; it is likely that they are primarily inflammatory. One must be wary of assuming a complete recovery in these areas of the brain, since the disappearance of a lesion on the MRI does not necessarily indicate that the tissue has returned to normal, as was clearly demonstrated in case 10. In cases 1-4 (acute EAE), the monkeys developed clinical signs within 12-24 hours after the areas of increased signal intensity were seen on the SE images. These earlier cases were also characterized by very bright areas of increased signal intensity on SE images and correspondingly dark areas on the IR images due to the significant increases in T and T i .  2  The lesions in case 10 were notably less intense. This is an important point,  since this animal did not go on to develop clinical signs until much later in the course of the disease, even though the lesion size was significant, and comparable to the lesion size in all other animals. The pathology of the lesions in this case was different from previous cases, being mildly gliotic; evidence of the earlier inflammation was observed in the lesion still visible on the MRI at the time of death.  Chapter 3.  MRI of Experimental Allergic EncephalomyeUtis in Primates  125  It has been interesting to note the disappearance of the inflammatory lesions on the MRI of fixed tissue. It appears that as the tissue becomes more severely affected by the disease process, then the lesions become more visible on thefixedbrain SE MRI. The reaction believed to be of major importance in the formaldehydefixationof tissue, is the formation of methylene bridges between the e-amino group of lysine and available amide groups in peptide linkages of other proteins [116]. The proteins 0=C HC—(CH ) 2  4  N H + HOCHgOH + H ~ N 2  I  NH  (3.35)  present in the exudate of the inflammatory lesions will be cross- linked onfixation,which must give rise to a relationship between bound and free water similar to that seen in normalfixedtissue. This is in contrast tofixeddemyelinated and gliotic lesions, which are both MRI visible.  In vitro quantitative  MR studies on fresh andfixedtissue are  described in chapter 4 [127]. These studies provide more detailed information regarding the pathological changes which give rise to the changes in Ti and T . 2  The role that demyelination plays in the relaxation mechanisms of the tissue water protons is still not clear. The studies on human MS tissue presented in chapter 2, show that demyelinating lesions cause an increase in Ti and T , but how this change compares 2  with the effects of inflammation is not yet established. The areas of abnormality seen grossly on thefixedbrain slices in EAE correspond  Chapter 3. MRI of Experimental Allergic EncephalomyeHtis in Primates  well with the areas of increased signal intensity seen on the S E images.  126  This has im-  portant implications when studying the evolution of the pathological process over time in both E A E and M S . The technique used to measure the extent of disease must be both accurate and reproducible and if disease is being followed over time using M R I , repositioning of the subject i n follow-up studies is critical.  Studies i n M S patients at  U . B . C . have shown that a 5 mm deliberate offset in the slice position causes an overall lesion area change of 11%; changes i n individual slice lesion area can be as high as 50%, particularly when small lesions are present [23]. Repositioning is also critical for the quantitative measurement of M R I parameters. obtained by selectively exciting a slice 5 mm thick.  The images of the monkeys were Within this thickness there may  be water protons i n different compartments or i n different types of tissue which are not rapidly exchanging; as previously mentioned, this may give rise to multi-exponential relaxation behaviour [128,129]. The signal contribution from each proton source depends on its volume and on the pulse sequence used to excite the slice, and the resulting image intensity in each pixel is the sum of the signal contributions through the slice. For a SE image the observed intensity would be given by equation 3.36; hr = J2l exp(-2T/T )[l i=i 0  2i  - exp(TR/Ti)]  (3.36)  where n is the number of exponentials required for an adequate description of the data. This means that the computed images can only provide observed Ti and T values, giving 2  no indication as to their source. Accurate repositioning is essential, since variations in Ti and T may be due to variations in the contributions to the signal from different 2  compartments within the slice.  In addition, the lesion may increase i n size, filling the  slice: this gives rise to an "observed" increase in the relaxation times because of the increased contribution to the signal from the abnormal area. The results of the quantitative M R I studies show that the T i and T values increase 2  Chapter 3. MRI of Experimental Allergic Encephalomyelitis in Primates  127  over time; when clinical signs were stable, as in case 1, the T\ and T values also stabilized. 2  This finding seems to indicate that the pathological progression of the disease also halted, since T\ and T are sensitive to changes in the molecular environment. It must also be 2  noted, however, that small changes i n T and T (< 10%) will not be observed, and x  2  some molecular changes may cancel each other out i n terms of these observed two-point determinations of T\ and T . 2  When studying the development of E A E lesions in case 4,  it was noted that T\ began to increase before the lesion became visible on the SE image. This was also seen i n the other cases of E A E in primates; i n particular, the new lesion seen on day 86 in case 10, and indicates that the initial pathological changes affect T\ rather than T . 2  A t different points in the study of case 10, the changes in T i were more  dramatic than T , showing small dark areas of increased T i on the IR images.  As a  results of this, parts of the lesions were more easily seen on IR than S E images.  The  2  initial increase in T\ is small, and falls within the error boundaries; therefore, although it is reproducible, it is questionable whether it could be used to predict the appearance of a lesion using these simple quantitative M R I methods. In order to use T\ and T to distinguish between different types of lesions, the changes 2  in the measurements must be large compared to the errors involved. Of the nine E A E cases presented here, the most dramatic changes in T\ and T were observed in case 4, 2  corresponding with lesions that were more demyelinated and hemorrhagic than those in the other cases.  The difference in T\ between a lesion with no hemorrhage and little  demyelination (case 1) and the lesions of case 4 is 100 ms. This difference is large enough to be considered significant. If we look at table 3.21, we see that the relaxation times for case 2 are lower than might be expected, based on the type of pathology that was present; these low values can be explained in terms of volume averaging. The lesion i n this case, consisted of small inflammatory areas which did not coalesce; the area of increased signal intensity on the S E image is therefore not homogeneous, and any area chosen to obtain  Chapter 3.  MRI of Experimental Allergic Encephalomyehtis in Primates  128  values for the relaxation times will consist of normal and abnormal tissue, giving rise to the lower T\ and T2 values. A t the present time, the pathological process primarily responsible for the difference in M R I parameters in the lesions seen in these four cases cannot be identified. Further studies on demyelinating lesions will be necessary before we can state which pathological process gives rise to the different changes in T i and T . 2  From the post mortem M S studies, the data indicates that the most gliotic lesions have the highest T i values. Where inflammation fits into this picture is not yet understood. Both inflammatory and gliotic lesions are associated with an increase in water content [118], which in turn affects the relaxation parameters.  It may be that inflammation and  gliosis will have similar M R I characteristics; they could, however, be distinguished using the length of time they have been present in the brain. In addition, case 10 shows that old inactive areas do not show enhancement with G d - D T P A ; this could be due to the lack of severity of these areas but does suggest that gliotic lesions will behave differently from inflammatory ones. Treatment of E A E with steroid and M B P shows significant changes in the size and intensity of the lesions, as seen in case 9.  Case 10 has shown, for the first time, that  spontaneous relapsing and clinically asymptomatic M R I lesions are possible following a single immunization to produce E A E in a primate. The observations made in this case were comparable to the M R I changes seen in serial studies of MS patients. In conclusion, it is clear that there are measurable differences in the relaxation times when different pathology is present. However, just as for the studies on M S tissue, problems arise due to limitations in the slice thickness possible (causing volume-averaging), and in the inability to isolate the effects of inflammation, demyelination and hemorragic necrosis on the relaxation times. The limits of the quantitative M R I methods used on this system have already been alluded too.  The next chapter describes quantitative  in vitro T\ and T measurements on CNS tissue; these studies demonstrate the lack of 2  Chapter 3. MRI of Experimental Allergic EncephalomyeUtis in Primates  129  information available using these two-point computations to determine the relaxation times.  Chapter 4  In Vitro Quantitative Relaxation Measurements  4.1  Practical Aspects  Measurement of relaxation times in tissue presents many problems; not least of these is the complexity of the tissues themselves.  In addition, measurements on tissue in vitro  may be affected by sample preparation; which includes the effects of blotting, squashing, freezing and thawing, elapsed time after death and excision, and fixation; ideally, the tissue should also be studied at physiological temperature (37°C), but this is not always practical.  Many of these factors have been studied, and the data can be summarized  as follows: Beall et al carried out extensive studies on many different tissues, and have made numerous suggestions for successful relaxation measurements to be obtained from tissue samples [131].  Any action which causes a change in the water content of the  sample should be avoided since this could potentially alter the values of T i and T . 2  Samples should therefore be kept in tightly sealed containers and should not be rinsed with saline or blotted with any absorbant substance, since this could also affect water content. When removing tissue, it should be cut firmly and quickly with a sharp scalpel; cutting with scissors is not recommended as it forces fluid out of the tissues.  Likewise,  soft tissue should not be forced into the N M R tube, since this will also result in fluid loss.  It is important that the same piece of tissue studied using N M R should be sent  for pathological examination, since tissue heterogeneity may lead to erroneous results when compared with the pathology. A group of samples will display a certain variation,  130  131  Chapter 4. In Vitro Quantitative Relaxation Measurements  therefore as much data as possible should be collected; Beall also recommends that extensive reproducibility studies be carried out before acquisition of final data.  This  is not always practical, since samples may be obtained at short notice and although other authors have shown that the effects of elapsed time on T\ and T after excision 2  are minimal up to 24 hours [111,112,113], tissue deterioration begins immediately upon excision and it is important that measurements are carried out as soon after death as possible if comparisons are to be made with data obtained in vivo. Studies on the effects of temperature show a linear dependence of T\ from 20 — 50°C; T showed very little 2  dependence on temperature in this range [114]. The many different methods available for determining T j and T in vivo and the many different magnetic field strengths being 2  employed, result in large variations in the values obtained from the same tissue type. The lack of any standard measurement and the field dependence of T i thus make it difficult to evaluate the true potential and usefulness of making such measurements; the large variation in values of the relaxation times also complicates the interpretation of the data.  4.2  Tissue Relaxation Models  Spin relaxation occurs due to fluctuating magnetic fields at the site of the nucleus. For relaxation in tissue, where for the most part we are looking at tissue water, the most important mechanisms for this relaxation are dipole-dipole interactions and rotational and translational motions.  Correlation times for these interactions must therefore be  considered i n mechanistic models for relaxation in tissue. This includes exchange rates for water molecules which are bound (hydrogen bonded to hydrophilic regions) to macromolecules with those that are in the bulk water. When placed in a static magnetic field, B , all particles with spin (that is, species 0  Chapter 4. In Vitro Quantitative Relaxation Measurements  132  Figure 4.62: Vector diagram showing the interaction of two spins. containing nuclei having spin or unpaired electrons) have magnetic moments, and will therefore produce local magnetic fields. In the M R experiment, the spins are perturbed from their equilibrium state using a rotating field, B\, perpendicular to B . 0  After  perturbation, due to the motion of these particles, a particular nucleus will experience a fluctuating field from neighbouring spins, due to changes in r and 8 (figure 4.62). When these fluctuations occur at the resonance frequency corresponding to a spin transition, energy transfer will occur: the spin relaxing to the lower state and the energy quantum being dissipated as thermal energy in the bulk of the sample. If A n is the disturbance from the equilibrium population difference at a time t, and An  0  is the value of A n at  time r , then: 0  A n = An e^T 0  (4.37)  where T i is the spin-lattice relaxation time, and the magnetization decay is monoexponential.  The rate, or efficiency of dipole-dipole relaxation depends on both the  strength and frequency of the fluctuating fields. These i n turn depend on three factors,  133  Chapter 4. In Vitro Quantitative Relaxation Measurements  namely: a) the distance between the nuclei involved and the angle between the vector that joins the nuclei and B , b) the effective correlation time, T , of the vector that joins 0  c  the nuclei and c) the nature of the nuclei themselves. For solutions, we make the assumption that molecules tumble isotropically, that is, tumbling occurs equally easily in all directions. For large molecules and macromolecules, this may not be possible, in which case the tumbling is said to be anisotropic [132]. This is important in tissue, where the water molecules are, at least part of the time in association with membranes and macromolecules in the cytoplasm, and this in turn affects the relaxation rates. For two interacting spins undergoing isotropic reorientation with a single correlation time and exponential correlation function, the dipole-dipole relaxation rates are given by equations 4.38 and 4.39 [133,134]. -K[—^—^r-^—)  1  T  l  1DD  l+w2r 2 c  K p T c + r P ^  1  T D  1  l + 4u; r 2  +  2D  (4.38)  2  (4-39)  T r ^ }  l+^T,?  where K = 3/x /i 7 /1607r r , and is a constant for a particular nucleus; u = 2irv is the 2  2  4  2  6  0  0  resonance frequency and T is the correlation time, both these parameters are variables. c  At 100 MHz (which is close to the resonance frequency at which the measurements described in this chapter were carried out), u> = 6.4 x 10 5ec ; since r for free water is 8  -1  c  approximately 1 0 (T U> ) is 2  2  - 1 2  sec, the product UT = 6.4 x 1 0 , which is much less than 1 and - 4  C  even smaller. Thus the above equations reduce to; = K[T -r4T ]  ±  c  c  = 5KT  ±? = K[3T + 5r + 2r ] = C  c  (4.40)  c  = 5Kr  c  c  (4.41)  *  ±2  giving 1/Ti = I/T2 = 5KT for noviscous liquids. C  fluctuations at the Larmor frequency, u . 0  Relaxation requires magnetic field  Since these fluctuations are produced by  Chapter 4.  134  In Vitro Quantitative Relaxation Measurements  2  aa  Figure 4.63: Diagram showing the possible transitions for two coupled spins. 0 - zero quantum transition, 1 - single quantum transition and 2 - double quantum transition. proton tumbling at a rate ( T ) , relaxation involving single quantum transitions (figure - 1  c  4.63) is most efficient when  U> T 0  C  ~ 1. Not all molecules tumble at the same rate, and  they are continually colliding thereby changing their rate and direction of tumbling. A t temperatures below room temperature, molecules will undergo slow reorientation, giving rise to a long r ; conversely, higher temperatures give rise to short r . c  c  The relaxation  time increases with increasing temperature, and decreases with increasing correlation time; this increase with temperature is due to the decrease in r at higher temperatures. c  The number of protons rotating at each frequency across a range of frequencies is defined as the spectral distribution, J (equation 4.42); w  /•+00  Ju =  G(T)exp(iuT)dr  (4.42)  J—oo  where G(T) is the correlation function. For large molecules or viscous solvents, where  UT 0  C  is greater than 1, slow tumbling  cannot give rise to fields at the Larmor frequency. As a result, single quantum relaxation processes cease to be efficient.  When  UT O  0  is close to 1, the spectral density, or the  intensity of appropriate field fluctuations, is at a maximum, and so relaxation is also most efficient. Since u> depends on the field strength, the relaxation rate of a particular 0  135  Chapter 4. In Vitro Quantitative Relaxation Measurements  spin is field dependent in molecules, or parts of molecules, where U> T ~ 1. Zero quantum Q  C  transitions (figure 4.63) involve a mutual spin-flip between two nuclei that have similar energy levels.  Such transitions do not give rise to spin-lattice relaxation, but involve  the excess energy being transferred from one spin to another.  When u> t 0  c  1, zero  quantum transitions are efficient. Thus, spin-lattice relaxation is slow, but the life-time of an individual spin state is short because the excitation passes via mutual spin-flips to other spins. As a result, Ti is long but T is short, giving rise to the broad N M R signals 2  seen from solids; the T for solids is therefore field independent for r X 1 0 . - 5  2  c  As the internuclear distance between two particles increases, the intensity of the field produced by their interaction decreases, due to the r dependence, and so the efficiency of 6  relaxation also decreases. As was described in section 1.3.2, the presence of paramagnetic species also affects relaxation, since  ^electron  is about 1000 times bigger than 71//. The  most common paramagnetic species is dissolved oxygen and in its presence the nuclear relaxation rate will increase; in tissue, changes in amounts of dissolved oxygen may affect relaxation rates in tumors for example [135]. The application of the relaxation theory described above to tissue, is complicated by the heterogeneity in the sample; protons may be in different compartments or in different chemical environments within a compartment, with exchange processes between these different types. The observed Ti and T values for tissue could not be explained i n terms 2  of a simple solution to the Bloembergen, Purcell and Pound theory; however, the data could be explained if a small fraction of cell water was assumed to be immobilized on the surface of macromolecules, thus giving it a correlation time of the order of ice. Fast exchange between the free and bound water then gives rise to an observed relaxation time dependent on the fraction of bound water;  Chapter 4. In Vitro Quantitative Relaxation Measurements  136  This is referred to as the fast exchange two-state (FETS) model of Zimmerman and Brittin [109].  Other models invoke three or more separate water states with variable  rates of exchange between pairs of states [147,148]. A n example of such a model will now be considered; from relaxation measurements on protein solutions at different frequencies, Grosch and Noack [137] believe a three compartment model is sufficient to fit their data; this is supported by other literature data [138,139].  This three-compartment model  consists of the following; 1. Bound water - water directly hydrogen bonded to a fixed polar or ion site on the macromolecule. 2. Structured water - water motionally perturbed by being in the vicinity of a macromolecule, but not bonded to it. 3. Bulk water - water in the bulk phase where molecular motion is determined solely by the interaction characteristics of the water molecule. The water spin-lattice relaxation rate is then given by; m  = ^-r^J-lw  + ^-  (4.44)  -list *lb  where /„,, f and fb are the fractions of bulk, structured and bound water respectively, st  and there is fast exchange between the three fractions. Studies on lysozyme have shown that the bound water fraction, /„,, determined by an N M R titration method[140], consists of water molecules directly hydrogen bonded to the macromolecule with typically two or more ligands.  These bound water molecules are in intimate contact with the  macromolecule, being spatially close to it.  The mean separation of bound water from  polar portions of a polypeptide is only 2.9 A, while water molecules next to nonpolar surfaces are at a distance of 3.7 A [1.41]. As described above, the dipole-dipole interactions  137  Chapter 4. In Vitro Quantitative Relaxation Measurements  vary as 1/r ; therefore the influence of macromolecular motion by cross-relaxation will be 6  visible in 1/Tib but not 1/Ti, [142]. The correlation times determined using the N M R t  titration method are comparable to those obtained by dielectric study [136]. Just as for the F E T S model, the observed relaxation time in this three-compartmentment model is dominated by the fraction of bound water; it is also dependent on the molecular weight of the protein, which determines the rate at which the molecule tumbles [143]. Other investigators propose models assuming a continuous distribution of water proton correlation times [144,145] to simultaneously explain T i and T observations. More 2  recently, relaxation data has been analyzed in terms of a continuous distribution of relaxation times [146]; this approach is the most valid when the complex nature of the tissues is considered. The water molecules will sample many different environments during the time course of the M R experiment; depending on the correlation times for the interactions involved, a single or multi-exponential decay will be observed. The data presented here support the recent literature data; simple models involving compartmentalization of water in tissue are inadequate to explain the observed results. Many literature studies have shown that a single exponential fit to relaxation data is insufficient to describe the decay observed; a number of authors report three-exponential fits for some tissue data, ascribing the values to bound (~ 10 ms), intracellular (~ 70 ms) and extracellular (~ 200 ms) water [147,148]. Such compartmentalization is unlikely, due to the effects of diffusion, since water can diffuse readily through the cell membrane. Recent observations on model systems for edema have shown that cell membranes alone are insufficient to produce multi-exponential relaxation behaviour [149].  These studies showed that  for suspensions of red blood cell ghosts, T i and T had mono-exponential decays; multi2  exponential decay was only observed in samples with the cells packed in layers, and was only observed for T . For normal CNS tissue, there is no extracellular space, suggesting 2  138  Chapter 4. In Vitro Quantitative Relaxation Measurements  that a maximum of two exponentials will be observed; the data presented here demonstrate that such a simple interpretation does not explain the observations.  The data  presented here are analyzed in terms of a fixed number of exponentials and i n terms of a continuous distribution of relaxation times; the advantages and disadvantages of both models and possible explanations for the observed data are addressed in the discussion.  4.3 4.3.1  Methods T Measurements 2  A l l measurements were carried out on a modified Bruker S X P 4-100 N M R spectrometer operating at 2.1 Tesla (90 M H z for protons).  The signal was acquired using a Nicolet  2090AR dual channel digital oscilloscope and successive scans accumulated by a National 32016 microcomputer serving a microVax II computer. The 90° pulse varied from 3—4 pis, and the 180° pulse was equal to twice the 90° pulse length. Before measurements were made on tissue, the spectrometer was set up using water doped with copper II sulphate. The phase, amplitude and pulse lengths were optimized using the rf pulse sequences shown i n 4.45, 4.46 and 4.47. (90° — data acquisition — t —)„  (4.45)  (90° — data acquisition — t —)„  (4.46)  (90°_ — data acquisition — t —)„  (4-47)  T  r  x  r  This was then repeated for the tissue samples before the experimental run.  T mea2  surements were made using the C P M G pulse sequence 4.48; the phase of the 90° pulse was alternated 180° for each excitation. Four points were collected for each of the first 224 echoes, followed by every eighth echo up to 512, giving a total of 736 echoes. The  139  Chapter 4. In Vitro Quantitative Relaxation Measurements  Figure 4.64: C P M G decay curve for CNS tissue interpulse delay, r = 400 /xs, acquisition time was 1.7 sec and a delay, t , of r  (90:(90°_ ) - - - 180°. - (r - 180°.-)„ - r - 180p T  x  (4.48)  10 sec was left to allow the system to return to equilibrium before reapplication of the pulse sequence. A number of samples were also studied using a r-value of 800 [is to determine whether the measurements were tau- dependent. Fifty transients were collected for each sample, each odd numbered transient being added and each even numbered transient substracted to give the final data. The baseline was determined from the raw C P M G data and the data outputted as follows; 35 points at 1 point intervals, 50 points at 2 point intervals and every 5th point to the end of the data.  Each data set then  consisted of 110-130 points to be analyzed using the programs described in section 4.3.3. A typical decay curve is shown in figure 4.64. Measurements were carried out at room temperature (24°C).  Chapter 4.  140  In Vitro Quantitative Relaxation Measurements 3000  data  —I I I i I I ' ,'  2500 -  c  •?  ROI  2000 -  1500  -  1000  -  500  -  (0  w c  pretrigger  -500  - i — i — | — i — i — i — i — j — i — i — r — i — | — i — i — i — i — ] — i — i — i — • — i — i i50  100  150  Time (ps)  200  250  300  Figure 4.65: F I D from lesion 1, case 4, using the modified IR pulse sequence. The pretrigger and data regions are labelled, together with the region of interest chosen. The tau value was 1 ms. 4.3.2  For  T i Measurements  these measurements, a modified IR sequence was used (4.49). The signal obtained  after the first (second) 90° pulse was added (subtracted) to (from) the accumulated signal in memory. (~t - 90° -t r  The  delay t was 10 sec, the 90° and 180° r  r  - 18C£  - r - 90°-)„  (4.49)  pulse lengths were 3.1 ps and 6.2 ps respectively  and 30 values of tau were used, varying from 1 ms to 3 sec. Data for each tau were collected in a random order to average out any systematic errors in the data acquistion. The  number of transients, n = 16, the block size was 512 and the dwell time 0.5ps.  The  F I D from one sample is shown i n figure 4.65. The pretrigger region was used to  determine the baseline noise, and a 10 point region of interest chosen in that part of the decay curve corresponding to mobile protons, from which the average signal intensity was determined.  A plot of signal intensity versus tau then gives the result shown in figure  141  Chapter 4. In Vitro Quantitative Relaxation Measurements  1. D  Tau  1.5  value  (sec)  Figure 4.66: Plot of signal amplitude versus tau for lesion 1 from case 4. 4.66.  This modified IR sequence was used since it allowed the Tj data to be analyzed  using the same programs as those used for analysing T data; it is also less susceptible 2  to systematic error than the conventional IR method. 4.3.3  Analysis of Relaxation D a t a  Minuit  The  C P M G data were analyzed using minuit, a non-linear functional minimization  program [150].  The program allows the number of exponentials to be specified and  the optimum number for each sample was chosen to be the minimum number that adequately described the data.  The values of T obtained must also be separated by at 2  least a factor of three. For three exponentials, the data are fitted to equation 4.50; I = Aexp{—^-) + Bexp(-^r)  +  t  ±2a  J-2b  Cexp(-^r-)  (4.50)  J-2c  where I is the signal intensity for a given time t after the 90° pulse, and A , B and C are t  the respective contributions to the signal for each T value denoted by a, b and c. 2  Chapter 4. In Vitro Quantitative Relaxation  Measurements  142  Non-Negative Least Squares (NNLS)  In contrast to minuit, this method analyzes the data in terms of a continuous distribution of relaxation times. This has already been done as a linear problem by Provencher [151] and first applied to M R by Kroeker and Henkelman [146]. For the data presented here, least squares and linear programming algorithms were used to interpret relaxation in terms of a spectrum of relaxation times [152]. These stable methods are non-iterative and avoid all of the difficulties associated with nonlinear optimization schemes.  In  addition, they are not restricted by any assumptions which limit the number of possible relaxation times. The general integral equation describing multi-exponential relaxation is; y(U)=  f s(T)e^ dT a  T  (4.51)  Jb  where i = 1,2, ...iV, the N data yi are measured at times  and s(T) is the unknown  amplitude of the spectral component at relaxation time T.  The limits on the integral  are restricted to the finite values a and 6, which are chosen to contain the values of T\ and T expected for the physical system being analyzed. 2  The approach taken is to assume a large number of known times Tj, and then solve for the corresponding amplitudes, s,, some of which may be zero. For computer implementation, equation 4.51 must be discretized. The general form of the discrete version of equation 4.51 is given in equation 4.52; Af  y. = E ^ i where i = 1,2, ...N.  5  i  (4-52)  The matrix system defined by equation 4.52 can then be solved  using the NNLS algorithm of Lawson and Hanson [153].  Chapter 4. In Vitro Quantitative Relaxation Measurements  4.3.4  143  Spectroscopy  As mentioned previously, breakdown products of myelin and other CNS components may giveriseto a signal from mobile lipid components. In order to determine whether the signal observed was only from water, or from lipid and water, spectra were collected at points along the CPMG pulse train; this was achieved using the sequence shown in 4.53. (90° - I _ 180°. - (r - 180°,-) n  - - data acquisition)  T  (4.53)  In the sequence, n was varied to allow echoes to be collected 0.4-1000 ms after the 90° pulse. The data were then Fourier transformed and examined. 4.3.5  Protocol for Studying Tissue  EAE was produced in 14 Hartley guinea pigs by intradermal injection of a water-in-oil emulsion containing homogenized whole CNS tissue and FCA [117]; five controls were inoculated with FCA only. The animals were monitored daily for the onset of clinical signs; typical signs are weight loss, a slowrightingreflex, and hind leg weakness which can progress to complete paralysis of the hind legs. Clinical signs may appear at any time after day 7 post inoculation. Once signs were detected, the animals were sacrificed at different stages of the disease; that is, some with hind leg weakness, some with a slow righting reflex and others that were completely paralyzed. The CNS was then removed intact. The spinal cord was divided into 5 sections, using a sharp scalpel, denoted by their approximate level in the spinal column: cervical, upper thoracic, lower thoracic, lumbar and sacral (figure 4.67). Care was taken to ensure the tissue was not damaged. The outer-most membrane (the dura) around the CNS tissue was removed, and the inner two membranes (the arachnoid and the pia-mater) were left intact. A section was also removed from the centre of the left cerebral hemisphere and similarly placed in an NMR tube. Once the T measurements were completed, the tissue samples were placed in 10% 2  144  Chapter 4. In Vitro Quantitative Relaxation Measurements  C UT LT  1  L  S  LH  B (  T tissue  f nmr tube  glass rod  Figure 4.67: A . Diagram showing division of spinal cord into five sections denoted by their approximate level in the spinal column. B . The tissue was gently rolled onto the small diameter glass rod and placed in the bottom of the 10 mm N M R tube. C - cervical, U T - upper thoracic, L T - lower thoracic, L - lumbar, S - sacral and L H - left hemisphere. i  neutral buffered formalin; after at least one week of fixation, the T measurements were 2  repeated on the fixed tissue.  The fixed samples were then embedded in paraffin and  stained with hematoxylin and eosin (H & E) [116]. Samples were then graded according to table 4.22.  E A E in Primates T data were obtained from the tissue samples taken from case 8 in chapter 3. 2  Once  the studies were completed, the tissue was fixed in 10% neutral buffered formalin in preparation for any histology required.  Fixed MS Tissue In order to determine accurate values for the tissue aleady studied using quantitative M R I methods, samples were taken from two of the cases described in chapter 3; case 1 and case  Chapter 4. In Vitro Quantitative Relaxation Measurements  Perivascular Inflammation in the Subarachnoid Space (PVI-SS) + 1 or 2 perivascular cuffs ++ 3-6 vessels involved with widening of subarachnoid space +++ Many vessels involved and subarachnoid space widened and filled with inflammatory cells Perivascular Inflammation in the Virchow-Robin Spaces ( P V I - V R ) + 1 or 2 perivascular cuffs ++ 3-6 vessels involved ++-f  Many vessels showing perivascular cuffing  Parenchymal Cellular Infiltration (PCI) + A few cells invading parenchyma from 1 or 2 cuffs -f + Invasion of parenchyma by cells from several cuffs +++ Many cells infiltrating parenchymal, but more normal than abnormal tissue ++-f-+ Extensive cellular infiltration into parenchyma, more abnormal than normal tissue Demyelination (D) + Possible ++ Probable +++ Definite Edema (E) + Possible ++ Probable +++ Definite Table 4.22: Pathological scoring for guinea pig CNS tissue.  145  Chapter 4. In Vitro Quantitative Relaxation Measurements  146  grey  Figure 4.68: Diagram showing the right cerebral hemisphere of the superior surface of brain slice 4 i n case 1. This is the opposite face to that shown in figure 2.29, page 38. Sections were taken from areas corresponding to lesions 8 and 10. Samples of normal appearing grey and white matter were also removed. 4.  These two cases contained lesions which varied in the degree of demyelination and  of gliosis. Samples were taken from the opposite face to that from which the block was taken for histologic examination. For case 1, samples were removed from the superior surface of slice 4 as shown in figure 4.68. Tissue samples were also removed from 2 slices in case 4. The sections taken are shown infigure4.69.  Chapter 4. In Vitro Quantitative Relaxation Measurements  Figure 4.69: A . Diagram showing the superior surface of slice 7 in case 4, the opposite face to that shown in figure 2.18B, page 52. Sections were taken from the areas corresponding to lesions 1 and 3 as indicated. Samples of normal appearing grey and white matter, as indicated on the M R I , were also removed. B . Diagram showing the superior surface of slice 6 in case 4, the opposite face to that shown in figure 2.18A, page 52. Sections were taken from the areas corresponding to lesion 2 as indicated.  Chapter 4. In Vitro Quantitative Relaxation Measurements  4.4 4.4.1  148  Results Guinea Pig Tissue  Fresh Tissue The data from normal and abnormal tissue using minuit to analyse the data, are given in the following tables.  The data were divided into the part of the spinal cord and  the cerebral hemisphere, and the tissue was examined and given a pathological score according to table 4.22. The histologic appearance of each section was graded according to three degrees of inflammation (figures 4.70, 4.71 and 4.72) and for the presence of demyelination and edema.  H k, E is not the ideal stain for identifying demyelination,  but where possible it was noted. The pathology for each section is given for each sample after the C P M G data for the same tissue section.  Chapter 4. In Vitro Quantitative Relaxation Measurements  149  Figure 4.70: Histologic section stained with H & E , from the sacral spine of guinea pig 14, showing a large cuff of perivascular inflammation in the subarachnoid space.  Figure 4.71: Histologic section stained with H & E , from the upper thoracic spine of guinea pig 13, showing a perivascular cuff in a Virchow-Robin space.  Chapter 4. In Vitro Quantitative Relaxation Measurements  150  Figure 4.72: Histologic section stained with H & E, from the lower thoracic spine of guinea pig 16, showing extensive parenchymal cellular infiltration. Comparison of this section with the one shown infigure4.72 shows the hypercellularity in the tissue around the vessel.  GP 1 2 3 7 7 9 18 19  T  2a  8.68 11.25 10.61 10.04 9.98 10.08 8.41 10.27  % 11.99 13.31 11.77 12.85 12.18 13.04 9.67 11.48  T  2b  71.51 82.22 82.23 79.43 76.40 74.93 70.99 75.91  % 56.02 61.70 61.91 61.39 59.65 59.56 58.75 59.11  T  2c  201.27 230.02 222.02 233.93 211.37 224.73 190.49 192.66  % 31.98 24.99 26.32 25.76 28.17 27.39 31.58 29.41  Table 4.23: Results from minuit analysis of CPMG data obtained from the cervical spine of normal control guinea pigs. T (ms) values are given together with the percentage contribution of each component to the signal observed. Pathological examination of this tissue confirmed that each section was normal. 2  151  Chapter 4. In Vitro Quantitative Relaxation Measurements  GP 4 8 8 11 12 12 13 14 15 16 17  T  2a  9.04 8.69 8.69 8.51 10.79 10.86 8.84 8.96 8.89 9.69 9.75  % 11.21 12.01 11.74 8.21 10.44 11.01 8.64 9.37 9.69 9.60 9.21  %  T  2b  73.34 60.56 74.87 63.60 72.24 60.33 67.38 49.44 78.73 66.79 81.10 66.91 72.57 56.39 81.85 65.34 75.09 59.51 76.59 59.74 77.59 64.26  T  2c  %  209.13 28.23 213.97 24.39 189.87 27.93 162.64 42.35 209.79 22.77 212.56 22.08 169.49 34.97 219.27 25.28 204.08 30.79 186.26 30.65 199.38 26.52  Table 4.24: Results from minuit analysis of C P M G data obtained from the cervical spine of guinea pigs inoculated to produce E A E . T (ms) values are given together with the percentage contribution of each component to the signal observed. 2  152  Chapter 4. In Vitro Quantitative Relaxation Measurements  PVI-SS PVI-VR  GP 4 11 12 13 14 16  + +++ +++ +++ +++ ++  ++ ++ 4-+  +++ ++  PCI  D  E  + + ++ + +  -  -  ++ ++ + -  -  +  -  Table 4.25: Pathology from the cervical spine of guinea pigs inoculated to produce E A E . PVI-SS - perivascular inflammation in the subarachnoid space, PVI-VR - perivascular inflammation in the Virchow-Robin spaces, PCI - parenchymal cellular infiltration, D demyelination and E - edema.  GP  T  2a  1 8.15 2 10.40 3 10.37 9 7.72 18 9.32 19 9.82  % 12.69 12.71 12.02 10.68 10.71 11.28  T  2b  70.03 75.90 79.69 67.81 75.95 72.02  % 52.15 57.01 59.93 53.47 60.89 56.72  T  2c  204.23 214.26 219.05 198.19 199.96 189.54  % 35.15 30.28 28.04 35.85 28.39 31.99  Table 4.26: Results from minuit analysis of CPMG data from the upper thoracic spine of normal control guinea pigs. T (ms) values are given together with the percentage contribution of each component to the total signal observed. 2  Chapter 4. In Vitro Quantitative Relaxation Measurements  GP 4 8 11 12 13 14 15 16 17  T  2a  9.01 7.78 9.81 9.90 7.58 10.48 11.26 9.74 10.79  % 12.27 10.48 9.82 11.10 7.58 9.92 11.46 10.31 10.22  %  T  2b  72.35 68.29 78.22 73.30 62.72 82.60 76.89 74.68 79.09  %  T  2c  58.35 57.27 60.52 55.70 47.26 63.08 60.31 52.56 62.42  153  198.71 179.35 192.88 178.99 154.39 212.01 206.07 177.31 194.78  29.38 32.25 29.66 33.2 45.16 26.99 28.23 37.13 27.36  Table 4.27: Results from minuit analysis of C P M G data from the upper thoracic spine of guinea pigs inoculated to produce E A E . T (ms) values are given together with the percentage contribution of each component to the total signal observed. 2  GP 4 11 12 13 14 16  PVI-SS  + ++ + +++ +++ ++  PVI-VR  +^ +4-  ++ + ++  PCI  D  E  + + + + +  -  -  + -  4-  Table 4.28: Pathology from the upper thoracic spine of guinea pigs inoculated to produce E A E . PVI-SS - perivascular inflammation in the subarachnoid space, P V I - V R perivascular inflammation in the Virchow-Robin spaces and P C I - parenchymal cellular infiltration.  Chapter 4. In Vitro Quantitative Relaxation Measurements  GP 1 2 3 7 9 18 19  2a  %  10.38 11.17 11.61 11.44 11.04 11.87 9.07  14.59 15.66 14.36 15.19 15.11 14.23 12.12  T  %  T  2b  70.40 74.23 79.69 75.06 75.39 81.66 67.42  48.39 52.32 56.14 54.61 55.76 61.37 53.83  154  2c  %  206.19 223.15 228.66 234.90 227.55 235.33 196.44  37.01 32.02 29.49 30.20 29.13 24.39 34.05  T  Table 4.29: Results from minuit analysis of C P M G data from the lower thoracic spine of normal control guinea pigs. T (ms) values are given together with the percentage contribution of each component to the total signal observed. 2  GP 4 11 12 13 14 15 16 17  T  2a  9.95 9.67 10.20 11.03 11.64 11.27 10.44 10.44  % 13.13 10.73 11.60 8.98 12.62 12.54 6.99 11.27  %  T  2h  73.71 73.44 74.10 80.99 87.77 74.87 102.04 77.63  53.36 53.89 57.10 51.27 64.75 55.67 69.52 58.58  T  2c  210.76 182.34 182.7 178.25 232.94 202.36 256.75 203.85  % 33.52 35.38 31.2 39.74 22.63 31.79 23.49 30.15.  Table 4.30: Results from minuit analysis of C P M G data from the lower thoracic spine of guinea pigs inoculated to produce E A E . T (ms) values are given together with the percentage contribution of each component to the total signal observed. 2  155  Chapter 4. In Vitro Quantitative Relaxation Measurements  GP 4 11 12 13 14 16  PVI-SS  PVI-VR  PCI  D  E  ++ + +++ +++ +++ +++  + +++ +++ + ++.+  -  -  -  ++ ++ +++ +++ + ++++ +++  ++ -  +++  Table 4.31: Pathology from the lower thoracic spine of guinea pigs inoculated to produce E A E . PVI-SS - perivascular inflammation in the subarachnoid space, P V I - V R perivascular inflammation in the Virchow-Robin spaces and P C I - parenchymal cellular infiltration.  GP  T  2 3 7 9 18 19  2a  11.17 10.36 10.96 10.76 11.19 10.99  % 14.79 12.42 15.43 15.18 13.42 13.19  T  2b  74.92 76.83 74.89 77.21 84.23 71.71  % 54.16 54.86 53.46 55.93 63.07 53.26  T  2c  223.87 220.38 226.13 235.22 232.79 196.61  % 31.04 32.72 31.10 28.88 23.51 33.55  Table 4.32: Results from minuit analysis of C P M G data from the lumbar spine of normal control guinea pigs. T (ms) values are given together with the percentage contribution of each component to the total signal observed. 2  156  Chapter 4. In Vitro Quantitative Relaxation Measurements  T  GP  2a  4 8 11 12 13 14 15 16 17  10.48 10.43 10.16 9.20 10.22 10.07 10.94 9.05 11.91  %  T  %  T  %  12.87 10.93 11.51 10.05 7.12 10.40 10.85 6.60 10.80  80.33 80.93 81.19 66.50 96.22 78.74 78.55 95.02 84.87  55.86 59.89 60.67 48.98 68.80 58.26 60.72 66.95 62.05  223.76 185.69 201.41 160.98 220.50 203.48 210.55 210.41 205.72  31.27 29.19 27.82 40.95 24.07 31.34 28.43 26.45 27.15  2b  2c  Table 4.33: Results from minuit analysis of C P M G data from the lumbar spine of guinea pigs inoculated to produce E A E . T (ms) values are given together with the percentage contribution of each component to the total signal observed. 2  GP 4 11 12 13 14 16  PVI-SS  PVI-VR  ++ ++ +++ +++ +++ +++  + + +++ +++ ++ +++  PCI  D  E  + + ++ ++ ++ + ++ + +++ +++  Table 4.34: Pathology from the lumbar spine of guinea pigs inoculated to produce E A E . PVI-SS - perivascular inflammation in the subarachnoid space, P V I - V R - perivascular inflammation in the Virchow-Robin spaces and P C I - parenchymal cellular infiltration.  157  Chapter 4. In Vitro Quantitative Relaxation Measurements  T  GP 1 2 3 7 9 18 19  2a  10.08 10.38 11.41 10.35 8.85 10.15 9.11  % 10.48 11.35 10.82 12.91 10.79 9.34 9.77  T  2i  %  75.19 54.48 78.78 61.06 85.03 61.22 81.19 61.71 82.63 65.34 80.95 60.33 73.69 57.17  T  2c  198.19 211.19 215.81 232.92 246.76 201.99 190.68  % 35.04 27.59 27.96 25.38 23.86 30.33 33.06  Table 4.35: Results from minuit analysis of C P M G data from the sacral spine of normal control guinea pigs. T (ms) values are given together with the percentage contribution of each component to the total signal observed. 2  GP 4 8 11 12 13 14 15 16 17  T  2a  9.30 8.73 9.36 9.09 8.04 9.96 9.72 8.76 9.55  % 10.71 10.15 8.28 8.73 6.76 8.53 7.96 6.95 8.12  T  2b  83.70 73.19 72.91 76.60 74.53 81.66 74.49 77.68 82.50  % 60.75 61.96 52.78 61.65 55.28 60.37 52.54 50.99 61.01  T  2c  224.66 183.71 168.15 181.29 168.97 191.51 176.23 158.48 190.00  % 28.54 27.89 38.94 29.60 37.96 31.10 39.49 42.05 30.86  Table 4.36: Results from minuit analysis of C P M G data from the sacral spine of guinea pigs inoculated to produce E A E . The T (ms) values are given together with the percentage contribution of each component to the total signal observed. 2  158  Chapter 4. In Vitro Quantitative Relaxation Measurements  GP  PVI-SS  + ++ +++ +++ ++ +++  4 11 12 13 14 16  PVI-VR  PCI  + ++ ++  + + ++ +  -  -  +++  D  E  -  -  ++ ++  Table 4.37: Pathology from the sacral spine of guinea pigs inoculated to produce E A E . PVI-SS - perivascular inflammation in the subarachnoid space, P V I - V R - perivascular inflammation in the Virchow-Robin spaces and P C I - parenchymal cellular infiltration.  GP 4 7 9 11 13 14 16 17 18  T  2a  9.42 7.00 8.17 15.64 13.11 7.31 14.93 16.33 14.80  %  T  26  95.65 88.50 92.28 96.73 91.55 96.31 96.47 94.03 94.39  4.23 4.15 5.03 4.03 5.01 3.01 5.15 5.67 4.03  % 95.78 95.85 94.97 95.96 94.99 96.99 94.84 94.33 95.97  Table 4.38: Results from minuit analysis of C P M G data from the left cerebral hemisphere of guinea pigs inoculated to produce E A E . T (ms) values are given together with the percentage contribution of each component to the total signal observed. 2  159  Chapter 4. In Vitro Quantitative Relaxation Measurements  GP  PVI-SS  4 11 13 14 16  ++ -  + -  PVI-VR  + + + ++  PCI  D  + + + + -  E  -  -  Table 4.39: Pathology from the left hemisphere of guinea pigs inoculated to produce E A E . PVI-SS - perivascular inflammation in the subarachnoid space, P V I - V R - perivascular inflammation in the Virchow-Robin spaces, and P C I - parenchymal cellular infiltration.  Chapter 4. In Vitro Quantitative  A  Relaxation  160  Measurements  7 0  •0 80 Faroant 40 SO SO  u  n  R  SO  B  100  ISO  200  250  70  •0 50 40 Feroant 30 20 10 0  -rf^—  r-j  —  t*  ^  -© SO  c :  •  100  150  200  250  3£  Farcant  SO  100  150  200  250  300  70 •0 50 Farcant  _•  40 30 20 10 0  E  ^  1—1  •  -6—  70  60 50 Faroant 40 30 20 10 00  so  100  150  200  250  100  150  tOO  250  i t  SO  *2 <•»>  Figure 4.73: Summary of minuit data for spinal cord. Plot of percentage contribution to signal versus T value for guinea pig spinal cord. A . Cervical. B . Upper thoracic. C. Lower thoracic. D . Lumbar. E . Sacral. • tissue from normal controls, • abnormal tissue. 2  Chapter 4. In Vitro Quantitative Relaxation Measurements  161  o NNLS  Examples of the results of NNLS analysis of C P M G data from fresh guinea pig tissue are given in figures 4.74 and 4.75. The rest of the data can be found in appendix B .  Chapter 4. In Vitro Quantitative Relaxation Measurements  -I  I  I  I  162  IXJU  o a-, =3  E 1H  10  _A_  - T — i — i — i — i i 11 10"  T  2  (s)  10  to  Figure 4.74: Plot of amplitude versus T value for the lower thoracic spine of guinea pig 16. 2  Chapter 4. In Vitro Quantitative Relaxation Measurements  163  1.0 €> 0 .8 0.6 Q. E 0.4 < 0.2 0.0 10"  3  1.0 11 « 0.8 J 0.6 O. E 0.4 < 0.2 Il 0.0 • 10-3 IO" 2  1.0 -• 0.8 | 0.6 £ 0.4  I 10" 1  10°  16  <  0.2 0.010"  3  | IO"  2  10" 1  10°  Figure 4.75: Plot of amplitude versus T value for the lumbar spines of guinea pigs 2, 7 and 9 (normal) and 13, 16 and 11 (abnormal). 2  Chapter 4. In Vitro Quantitative Relaxation Measurements  164  F i x e d G u i n e a P i g Tissue Minuit  Examples of the data from fixed normal and abnormal lumbar and lower thoracic spine are shown i n tables 4.40 and 4.41.  NNLS  Figure 4.76 shows the results of NNLS analysis of fixed lumbar spine for normal and abnormal tissue.  4.4.2  P r i m a t e Tissue  Minuit  The results obtained after analysis using minuit are shown i n table 4.42.  NNLS  Typical examples of the results obtained after NNLS of the primate tissue C P M G data are shown in the figures 4.77-4.79.  165  Chapter 4. In Vitro Quantitative Relaxation Measurements  GP 2 4 7 9 11 12 13 14 16 18 19  T  2a  24.82 25.89 26.73 12.00 22.19 20.25 27.91 19.98 22.12 16.86 15.20  % 26.33 26.46 26.81 12.49 24.15 22.12 15.93 22.09 10.34 22.97 18.02  T  2h  98.74 92.75 95.33 84.09 76.68 76.80 111.96 71.65 94.51 57.99, 53.16  % 45.63 45.89 30.78 63.23 59.92 66.52 61.08 66.93 77.38 66.58 68.54  T  2c  774.77 723.33 839.70 492.00 561.89 551.99 868.28 402.43 405.39 268.22 213.50  % 28.04 27.65 42.42 24.28 15.92 15.92 22.99 10.98 12.27 10.45 13.44  Table 4.40: Results obtained from minuit analysis of C P M G data from fixed guinea pig lumbar spine. T values are given together with the percentage contribution of each component to the signal observed. 2  166  Chapter 4. In Vitro Quantitative Relaxation Measurements  GP 2 4 7 9 11 12 13 14 16 19  T  2a  25.21 27.09 24.14 11.65 23.09 20.58 26.47 17.90 22.82 16.95  % 27.88 28.24 32.12 13.35 26.50 23.28 16.67 20.34 12.22 23.08  %  T  2b  98.95 94.48 85.76 79.37 78.16 76.56 105.85 65.74 101.65 55.39  43.34 39.29 38.29 67.70 59.58 65.50 50.22 68.07 75.78 65.36  %  T  2c  767.69 28.78 754.20 32.46 852.08 29.59 488.42 17.95 558.28 13.92 560.32 11.22 869.56 33.11 326.11 11.59 511.59 11.99 253.67 11.55  Table 4.41: Results obtained from minuit analysis of C P M G data from fixed guinea pig lower thoracic spine. T values are given together with the percentage contribution of each component to the signal observed. 2  Tissue WM1 WM2 GM GM/WM LI L2 L3 LMB  T  2a  13.34 13.11 6.31 17.04 8.50 13.10 14.20 11.80  % 18.30 17.08 5.04 7.97 10.40 18.40 13.68 15.97  T  2b  102.60 88.42 93.80 90.90 93.30 95.50 110.40 92.84  % 81.70 82.90 93.90 92.00 89.6 81.60 86.30 84.00  X  2  3848 25011 1157 7871 2796 9864 17116 6258  Table 4.42: Results from minuit analysis of C P M G data from primate case 8. T (ms) values are given together with the percentage contribution of each component to the total signal observed. W M - white matter, G M - grey matter, L - lesion, L M B - left midbrain. 2  Chapter 4. In Vitro Quantitative Relaxation Measurements  167  8000 •§ 6000 i . 4000 E < 2000  _  „ 3000 J  2000  Fl 1 L  i  CL  | 1000 10"  3  10"  10"  2  2  T3  |  10°  10  1  10  1  F16L  x> D 4000 "Q.  §" 1 0 0 0 <  1  (s)  6000  F9L  2000 1500  1  500  I  \  10~ I O " 3  E < 2000  2  10"  1  10°  10  0 10"  1  i (s)  1 10-2 10" T, (s)  1  10°  Figure 4.76: Plot of amplitude versus T value (sec) for fixed guinea pig lumbar spine. Guinea pig number is given in each spectrum; numbers 2, 7 and 9 are normal, and 13, 16 and 11 are abnormal. The peak around 1 sec corresponds to the T of free water in the formalin solution around the tissue. The magnitude and T of this peak varied, depending on the amount of water present. 2  2  2  Chapter 4. In Vitro Quantitative Relaxation Measurements  168  Figure 4.77: Plot of amplitude versus T value for normal appearing white matter in primate case 8. 2  Figure 4.78: Plot of amplitude versus T value for lesion 3 in primate case 8. 2  Figure 4.79: Plot of amplitude versus T value for the left midbrain in primate case 8. 2  169  Chapter 4. In Vitro Quantitative Relaxation Measurements  CASE  TISSUE  !  WM GM WM/GM 8 10 WM WM GM 1 2 3  1 1 1 1 4 4 4 4 4 4  T  2a  17.66 56.06 48.43 15.75 9.03 4.22 4.75 57.54 10.69 16.35 40.21  % 32.65 81.08 59.64 4.08 1.48 6.28 6.40 69.87 6.64 4.64 7.67  T  2b  53.66 177.52 108.72 96.44 112.03 25.29 26.41 133.50 70.40 100.96 150.36  % 67.35 18.92 40.35 79.71 73.92 67.96 70.90 30.13 69.86 78.57 71.76  %  T  2c  -  j  -  1280 256.80 72.06 86.83  16.22 25.91 25.75 22.69  179.94 251.51 392.33  23.50 16.79 20.57  Table 4.43: T values obtained using minuit to analyze C P M G data from fixed M S tissue. Lesions are numbered as they were in chapter 2, and W M - normal appearing white matter, G M - normal appearing grey matter and G M / W M - mixture of grey and white matter. Two values are given for case 4 W M , the data were obtained six months apart. 2  4.4.3  F i x e d M S Tissue  Minuit  The T data obtained after analysis of C P M G data from the fixed MS fixed are presented 2  in table 4.43.  NNLS  The T and X i values for the fixed MS tissue were analysed using the NNLS program. 2  The results are given in figures 4.80-4.92.  Chapter 4. In Vitro Quantitative Relaxation Measurements  T,  170  (s)  Figure 4.80: Plot of amplitude versus T value for normal appearing white matter in case 1. 2  T,  (s)  Figure 4.81: Plot of amplitude versus T value for normal appearing grey matter in case 1. 2  171  Chapter 4. In Vitro Quantitative Relaxation Measurements -J  • • '' • * 1  X  -8 H  -I  1  1—I—I I I I I  T, (s)  1F  I  l  r  l—T  1 t  10  Figure 4.82: Plot of amplitude versus T value for normal appearing grey/white matter mixture in case 1. 2  50  • • •I  1 I—I  I  1—1—' I I I 111  1—1—I I I I 11  40 H  o X 30  T3  E 10  -1 1 1 ! I I f l' | K>"  1 11 I I | 10 '  10  T, ( s )  Figure 4.83: Plot of amplitude versus T value for lesion 8 i n case 1. 2  172  Chapter 4. In Vitro Quantitative Relaxation Measurements _i  i  i i ' < < • _j  i  i  i • ' • ''  J  I  I  I  '  '  ' ' '  12 -  x a> -a  4  10  10 '  -T  10"  1 1 1—I—I—I  10  T, (s)  Figure 4.84: Plot of amplitude versus T value for lesion 10 in case 1. 2  1 1—I—I—1 I  _J 5  —I  1 J  I  I ' ' '  -  -S  1  E2  A  •A-  T  2  -i—i—i—i—i r  (s)  Figure 4.85: Plot of amplitude versus T value for normal appearing white matter in case 4. 2  173  Chapter 4. In Vitro Quantitative Relaxation Measurements 12  _]  I I L—I ' ' '  _l  • • 1 L__I—L-  I I 1 I l_  O  X  T3 =3 CL E 4  10  10  10  10  T, (s)  Figure 4.86: Plot of amplitude versus T value for normal appearing grey matter in case 4. 2  10  _i  i i i i • * •'  —i  1—i i i i i  11  _i  1 ii i > • •  o X  6  CD  CX.  E 2 H  T  2  10  (s)  I' ' 1 1 III!*] 10"  Figure 4.87: Plot of amplitude versus T value for lesion 1 in case 4. 2  174  Chapter 4. In Vitro Quantitative Relaxation Measurements -I  i i I  I—1—1—1—1-  _i  i  1—J-  12  O X 8  H  ZD6 CL  E  2  H • 1 It 1 I I |  —1 1 1 1 I 1  T, ( s )  Figure 4.88: Plot of amplitude versus T value for lesion 2 in case 4. 2  15  -  m  CD  X  Figure 4.89: Plot of amplitude versus T value for lesion 3 i n case 4. 2  Chapter 4. In Vitro Quantitative Relaxation Measurements  175  R e l a x a t i o n T i m e T1 ( s e c )  Figure 4.90: Plot of amplitude versus T i value for normal appearing white matter in case 1.  10  to  R E L A X A T I O N TIME T 1 ( S E C )  Figure 4.91: Plot of amplitude versus T i value for lesion 1 in case 1.  B.  A.  otloo Mmm T l <B*C)  Roboxotion T i m * T l (aec)  Figure 4.92: A . Plot of amplitude versus T i value for normal appearing white matter in case 4. B . Plot of amplitude versus T i value for lesion 1 i n case 4.  176  Chapter 4. In Vitro Quantitative Relaxation Measurements  4.5  Discussion  The results show that T values from tissue are reproducible, with ~ 5% error. Using 2  minuit, three  exponentials provided the best fit to spinal cord data, and two exponentials  to brain data. A two exponential fit to brain tissue was not an adequate fit to the data (figure 4.93), but three exponentials provided numbers which could not be resolved, being separated by less than a factor of three. For the best fit, \ should be equal, on average, 2  to the number of data points; the number of data points for fresh tissue was usually around 120, and \ varied from 188 to 1738. The high values of x were only obtained 2  2  for two animals; 65% of the samples studied were in the range 200-400. Much better fits were obtained using the NNLS analysis, where the x  2  w  a  s  approximately equal to the  number of points. It is not possible to determine the reproducibility of the abnormal T  2  values, since no two samples are absolutely identical. The data will now be considered in turn, followed by a general summary at the end of the discussion. The results obtained from the fresh guinea pig tissue, showed that the lesions were primarily inflammatory; perivascular inflammation could be seen in the subarachnoid space and in the Virchow-Robin spaces. In a few animals, extensive parenchymal cellular infiltration was also seen (figure 4.72). The type of models which have been invoked to explain the relaxation data obtained from tissue were described in the introduction; using a similar approach, possible explanations for the data presented above will now be considered. The minuit data demonstrated that three exponentials provided an adequate fit to the spinal cord T data; the question then is what do these three exponentials 2  represent? As already indicated, literature relaxation models refer to bound water, intracellular and extracellular water as different compartments; since there is no extracellular space in normal CNS tissue, the source of the third exponent is not clear. From a purely anatomical standpoint, there are only two possible signal sources, the central canal of  Chapter 4. In Vitro Quantitative Relaxation Measurements  177  Figure 4.93: Plot of log of signal amplitude versus time after the 90° pulse in a CPMG sequence, demonstrating non- exponential behaviour of brain water proton relaxation. A. Manganese chloride solution. B. Brain tissue data fitted to one exponential. C. Brain tissue data fitted to two exponentials. D. Brain tissue datafittedto three exponentials.  178  Chapter 4. In Vitro Quantitative Relaxation Measurements  the spinal cord or the subarachnoid space; both these areas contain CSF. The T value 2  observed for the third component was i n the range 189 — 246 ms for normal guinea pig spine; i n abnormal tissue, the range was 154 — 257 ms.  The variations seen in the  long T component did not coincide with the samples studied late in the experimental 2  procedure; had this been the case, then the differences could be attributed, at least in part, to sample dehydration. If the source of the third exponent is taken to be the spinal canal, then this value would remain constant, since inflammatory cells do not infiltrate this area. However, if the source of this exponent is taken to be the subarachnoid space, the larger range observed in abnormal tissue could be explained as follows; perivascular inflammation in the subarachnoid space releases exudate from the blood vessels into the surrounding CSF; the exudate contains inflammatory cells and proteins resulting in a decrease in the observed T value due to the increase in the protein concentration [154,155]. 2  This interpretation then implies the middle component is intracellular water; this model works for normal tissue, but in the abnormal tissue, the middle component changes with parenchymal cellular infiltration. The value of T and the percentage contribution to the 2  signal both increase; hypercellularity in the tissue would lead to an increase in the signal contribution from intracellular water, but does not readily explain the increase observed in T . Such a change i n T would require changes in cell diameter, thereby increasing 2  2  the proportion of free water to bound water, or a decrease in the bound water fraction. CNS cells vary in size from 6 — 8 fim to 60 — 80 pm, the largest cells being the Purkinje cells in the cerebellum and the anterior horn cells of the spinal cord [156]. The cells that infiltrate the area are actually smaller than neurons, therefore this simple explanation is insufficient to explain the observations. The self diffusion coefficient of free water at 37°C is 3.05 x 1 0 m s _ 9  2  _ 1  [157]; the mean squared average distance travelled during time  179  Chapter 4. In Vitro Quantitative Relaxation Measurements  r i n three dimensions is given by equation 4.54 [158]; r = 6DT  (4.54)  2  where D is the self diffusion coefficient.  For r = T , a water molecule would travel 2  1.83 x 1 0 m ; this corresponds to at least 40 pm in one dimension. - 9  2  The CNS cell  diameters are of the order of tens of microns; thus compartmentalization cannot be explained i n terms of intracellular structures.  Relaxation studies on protein and other  macromolecular solutions show a single exponential decay; likewise the studies on red blood cell ghosts show that the cell membrane is permeable to water on the time scale of the M R experiment; thus, rather than intra- and extracellular water, a range of relaxation times will be observed due to the water sampling the different cells and extracellular space created in abnormal tissue. The rate of diffusion will be affected when the water molecules cross membranes and when they come into contact with macromolecules, both resulting i n slower diffusion; this could lead to multi-exponential relaxation behaviour. [159]. Differences were observed in the minuit data between animals having primarily perivascular inflammation in the subarachnoid and Virchow-Robin spaces, and those also having extensive parenchymal infiltration. If the two extremes are considered (figure 4.94), the normal tissue T data he between them. Thus, the T values must pass through normal 2  2  as the tissue becomes more inflamed, and the abnormal area becomes infitrated with inflammatory cells.  This supports observations made by Karlick et al., that normal  values of T were seen with parenchymal cellular infiltration [160]; i n that study, data 2  were fitted to a single exponential. When tissue was homogenized, a three-exponential fit the data was no longer acceptable, since the T values of the two slower relaxing com2  ponents were separated by less than a factor of three (table 4.44).  Even though the  fit is questionable, it is interesting that the bound fraction does not change even when  Chapter 4. In Vitro Quantitative Relaxation Measurements  180  P l o t of Percentage Contribution to Signal versus T2 f o r Normal and Abnormal Guinea P i g CNS 70 60 50 40 Percent 30 20  1• Y  •  •  • -ft  50  250  200  150  100  300  T2 (ms)  Figure 4.94: Plot of percentage contribution to signal versus T for normal and abnormal guinea pigs. Ik predominantly perivascular inflammation in the subarachnoid and Virchow-Robin spaces. D extensive parenchymal cellular infiltration. • normal tissue. 2  Tissue B C/T C/T  %  T  2a  4.0 9.8 9.5  2.8 10.9 11.3  T  2b  83.4 70.6 70.6  % 80.7 60.1 61.4  T  2c  133.9 176.3 169.7  % 16.5 28.8 27.3  Table 4.44: Results from minuit analysis of C P M G data from the homogenized brain and spine of a normal guinea pig. T values (ms) are given together with the percentage contribution of each component to the total signal observed. B - brain tissue and C / T cervical/thoracic spine. Two sets of data are given for C / T , determined 4 hours apart. 2  Chapter 4. In Vitro Quantitative Relaxation Measurements  the tissue is homogenized.  181  It provides further evidence that this component is real,  and future studies must include experiments to identify the source of this proton signal; literature data do not explain what the bound water represents. The homogenized tissue data also shows that multi- exponential relaxation behaviour does not arise from cellular organization.  The minuit data are sensitive to the pathological changes in the tissue,  but i n terms of understanding the relaxation of water protons i n tissue, is an inadequate model, particularly i n this instance where the samples consisted of normal and abnormal tissue. The NNLS analysis provided better fits to the data, but the small changes in the Ti and in the amplitude of each component made it difficult to interpret the discrete fits. Two peaks were often seen on either side of the Ti value obtained using minuit (figure 4.74 and 4.75, page 160); this suggests a distribution of T values rather than a discrete 2  value.  It was also noted, that abnormal tissue consisted of less peaks i n the discrete  fit; this is seen clearly if the data from the lumbar spine data from three normal and three abnormal guinea pigs are compared (figure 4.75). A smooth fit was applied to the lumbar spine of three abnormal guinea pigs, and three normal pigs. The data are shown in figure 4.95. The data shows an apparent decrease i n the bound water in the abnormal samples; this is more likely to be an increase in the longer components.  In addition,  the ratio of the two distributions is 6 in normal tissue (GP7L) and increases to 15 in abnormal tissue (GP13L); this is a significant change that could be used to differentiate the normal from the abnormal tissue.  The distribution of the longer components is  also narrower in abnormal tissue, just as was suggested by the discrete fit to the data. This can be explained in terms of probability distributions; in abnormal tissue, at any given time, the probability of finding a water molecule in a long Ti compartment is much higher than i n normal tissue.  This is due to the destruction of tissue, thereby forming  extracellular space; the water molecules in abnormal tissue will also cross less boundaries  Figure 4.95: Plots of amplitude versus T value for a smooth fit to guinea pig lumbar spinal cord C P M G data. The figure on each spectrum denotes the guinea pig number; numbers 2,7 and 9 are normal and 13, 11 and 16 are abnormal. 2  Chapter 4. In Vitro Quantitative Relaxation Measurements  183  1.01  0.9  0.8 c  c 0.7  _o o o k_ u.  _ 0.6 D  6  6  2  7  "o V  CL  00  0.5  0.4  9 Pig  13  16  11  Number  Figure 4.96: Plot of maximum and minimum amplitude versus guinea pig lumber showing differences between normal and abnormal tissue. in a given time than in normal tissue, for the same reason.  In addition, the cell type  in an abnormal area will tend to be more homogeneous in size than in normal tissue; all of these changes will lead to an observed increase in T and to a narrower distribution 2  of T values. The smooth fits shown here are similar to those obtained by Kroeker and 2  Henkelmen [146]. The ability to use the appearance of the smooth fit to differentiate normal from abnormal tissue was further tested; a program which maximises or minimises the observed amplitude in specified regions was applied to the data. The lumbar data from guinea pigs 2, 7 and 9 (normal) and guinea pigs 13, 11 and 16 (abnormal) were maximized in the region 65 — 150 ms and minimized in the region 30 — 300 ms; the results are shown in figure 4.96. The data shown are the maximum and minimum values of signal amplitude in these two regions, determined with 95% confidence limits. It was  184  Chapter 4. In Vitro Quantitative Relaxation Measurements  not possible to convert the normal into the abnormal, or vice versa; this shows that the differences between the normal and abnormal tissue are real.  The fact that the data  from guinea pig 11 overlaps the normal data is not unexpected since this tissue was only 10% abnormal, compared to 50 — 60% for 13 and 16; this demonstrates very clearly the problem of volume averaging, and provides a method for determining at what ratio of normal to abnormal tissue it is can no longer possible to differentiate normal from pathology. The NNLS analysis provides a model which can explain the results observed, and just as for minuit, this method is sensitive to T differences between normal and 2  abnormal tissue.  The spectra obtained from the guinea pig tissue showed no evidence  of a signal contribution from mobile lipid protons (figure 4.97). The appearance of the spectra was determined by the position of the tissue in the N M R tube.  The shoulder  observed in the spectra shown appears to be due to magnetic susceptibility; positioning the tissue parallel to the direction of the magnetic field gave rise to a water peak with no shoulder. The fixed guinea pig tissue relaxation data are more complicated; this is due to the presence of free formalin in many of the samples, resulting in a long T component not 2  observed in normal tissue.  A three parameter fit to minuit is therefore insufficient to  differentiate normal from abnormal tissue.  The bound water fraction appears to be  different from that seen in fresh tissue, but this may simply be due to the different fit obtained because of the free formalin.  NNLS analysis of a few normal and abnormal  samples showed that the information obtained from the fresh tissue is still observed in the fixed tissue.  The discrete fit to fixed tissue data also showed a tendency to be less  complex i n the abnormal tissue (figure 4.76).  This is an important observation, since  literature data have shown that fixed tissue relaxation measurements cannot be used as an indicator of relaxation behaviour in vivo or in vitro. These data show that this is not the case; even in the presence of the contaminating formalin solution, the NNLS program  Chapter 4. In Vitro Quantitative Relaxation  Measurements  185  B.  A.  Pr«qua&07 (kHz)  Frequency (kHz)  Figure 4.97: A . Proton spectrum from the lumbar spine of guinea pig number 13. B . Proton spectrum from the fixed lumbar spine of guinea pig 13. In both spectra tau = 400 its. could still access the information on the tissue water. The bound water fraction is still present in the fixed tissue, and varies from 8 — 18 ms. Comparison of the NNLS analysis for data from fresh and fixed left hemisphere shows a remarkable similarity between the two (figure 4.98). The data obtained from the primate tissue in case 8 in chapter 2, showed little difference between the normal and abnormal samples; this was true for both minuit and NNLS analyses. The most likely explanation for this is that the tissue is in fact mostly normal; cutting fresh tissue is difficult, and although attempts were made to correlate the samples taken with abnormal areas on the M R I , it is possible that the lesions were missed.  This data does however provide supporting evidence for using a distribution  of relaxation times rather than a discrete exponential fit; the x obtained using minuit 2  was high compared to the number of data points, but increasing the number of exponentials to three was not possible, because the separation between components was too  Chapter 4. In Vitro Quantitative Relaxation Measurements  Q)  .•3  186  12000 --  P9LH  8000  Q.  E < 10"  3  10"  2  4000 0 10~  1CT 10° T (s) 1  10"  3  2  ZD  ! . 2000 < '1000  6000  F7LH  •S 3000  V X> ZD +; 4000 "Q.  -  E < 2000  -  10~  II 3  IO"  2  I  10" T (s)  1  10°  1  10°  10  1  2  2  4000 -  10" T (s)  . 10  F9LH  -  0 10"  1  A 3  I A,  .  IO"  2  10" T (s)  1  10°  10  1  2  2  Figure 4.98: Plot of amplitude versus T value for fresh and fixed left hemisphere. The numbers denote the guinea pig and F implies fixed tissue. The peak around 1 sec in the fixed tissue data corresponds to the T value for free water in the formalin solution around the tissue. The magnitude and T of this peak varied, depending on the amount of water present. 2  2  2  Chapter 4. In Vitro Quantitative Relaxation Measurements  187  .6 -  E  .1 .O  Rolaxcrtion  Time  T2  Figure 4.99: Plot of amplitude versus T value for a smooth fit to normal appearing white matter in primate case 8, showing two distributions of T 2  2  small.  The x obtained using NNLS varied from 99 to 438, and the number of points 2  was approximately 112 for each sample.  A smooth fit to the data using boundaries  as described above for the guinea pig data, gives a ratio of two distributions like that seen for guinea pig tissue, as shown in figure 4.99. The fixed M S tissue relaxation data showed marked differences between normal and abnormal tissue, using both minuit and NNLS analyses.  The minuit data shows fixation to give rise to a short T component 2  that is much longer than that observed in fresh tissue.  However, the NNLS indicates  that a component around 12 ms is still observed in both normal and abnormal fixed tissue; just as for fresh tissue, two exponentials are insufficient to describe the data, but a three exponent fit gives T values that are only separated by a factor of 2. 2  The T\  data from the same tissue also shows a distribution of relaxation times. In lesion 8, case 1, a long T of 1.2 sec was observed; this lesion had cystic spaces, which are holes in the 2  tissue filled with fluid, and would be expected to give rise to a longer T\ and T . 2  The  fully demyelinated and gliotic lesions had longer T\ and T components than the partially 2  demyelinated and gliotic lesions from case 4; this supports the trend seen in chapter 2, where it appeared that the most gliotic lesions had the highest T\ values and from the signal intensity on the images seemed to also have higher T values. 2  The results from  188  Chapter 4. In Vitro Quantitative Relaxation Measurements  normal appearing white matter in case 4 suggest that the tissue was not in fact normal; increased water content and other changes have been observed in normal appearing white matter from M S patients [161]; i n addition, some authors have reported a small increase in the observed T\ relaxation time of white matter in M S patients compared to normal controls [162]. The results observed suggest that M R is sensitive to these changes, which may be the first stages i n the disease process, or a secondary process; further study is necessary in this area to interpret the data. A recent publication by Kroeker and Henkelman [163] describes the problem of spinlocking using the C P M G pulse sequence; this results in the measurement of T\ (the p  rotating frame spin-lattice relaxation time) rather than T . 2  The fixed tissue data pre-  sented here showed no tau dependence, and a tau value of 400 ps is longer than the correlation times of both solid and mobile protons; this indicates that T is the param2  eter being measured.  It should also be noted however, that even if T\ were being p  measured, if the data are reproducible and differentiates between the pathology observed in M S , then this would still be an acceptable measurement to use in the study of M S patients.  In summary, it is clear from the above results, that relaxation measurements  can be used to differentiate not only normal from abnormal tissue, but the different types of pathology present in the tissue. Although the differences observed between the guinea pigs with varying degrees of inflammation were not large, it must be remembered that for the majority of samples studied the tissue was mainly normal.  Larger differences  would be expected if pure pathological tissue were studied; this comes back to the same problem of volume averaging described in chapters 2 and 3. It must be stressed again that quantitative studies should be carried out on as small a volume of tissue as possible, and preferrably from the centres of lesions. The measurements on fixed tissue appear to show the same trends as those seen in vitro; this means that the differences observed in the T values obtained from the fixed M S tissue should be indicative of the changes that 2  \  Chapter 4. In Vitro Quantitative Relaxation Measurements  189  would be seen in fresh M S tissue; both minuit and NNLS analysis of relaxation data can then be used to differentiate differing degrees of demyelination and gliosis. In addition, the inflammatory lesions in the guinea pig tissue showed different relaxation behaviour from demyelinated and gliotic lesions (figure 4.100).  It is clear from these in depth  T studies that it is possible to use quantitative M R measurements to characterize MS 2  pathology; further studies must be completed before this goal can be realized. Recent technical improvements in commercial M R I systems allow C P M G data to be collected from a localized volume; it is also possible to obtain multi-echo images of the same slice, from which the relaxation decays of chosen R O I can be determined; these data can then be analyzed using the NNLS programs employed in the present studies.  Relaxation  studies on animal models for M S in vivo, with follow-up in vitro studies, would confirm the validity of the in vitro results described in this thesis. Other investigators have concluded that in vitro studies are indicative of the changes observed in vivo [164], although small differences are seen between the in vivo and in vitro relaxation times.  190  Chapter 4. In Vitro Quantitative Relaxation Measurements  •  c  E i  T,  :  (s)  i InftoMwmitw y  1  ... K  K  L*alon  "  Figure 4.100: Plot of amplitude versus T value for normal and abnormal fixed tissue. A . Normal spinal cord. B . Normal white matter from M S patient. C . Inflammatory lesion from guinea pig spine. D . Partially demyelinated and gliotic lesion from M S patient. 2  Chapter 5  Conclusions and Suggestions for Further Work  5.1  Conclusions  1. MRI at 0.15 Tesla is an accurate indicator of disease extent in both MS and EAE in primates.  Comparison of the abnormal areas on the SE MRI with the actual  pathology seen on the fixed brain slices themselves showed differences in individual lesion areas of 5 — 60%, and for total brain lesion area, 20 — 30%. These differences are attributed to the effects of volume averaging over the 10 mm slice thickness used to obtain the images; in addition, the margins of periventricular and subcortical lesions are difficult to define, contributing a further source of error. MRI can detect lesions in EAE before the onset of clinical signs and allows the disease progression to be followed in vivo. For the first time, it was demonstrated that spontaneous relapsing-remitting EAE can be induced in a primate following a single inoculation and without treatment; the course of the disease in this animal followed that seen in serial MRI studies in MS patients. 2. The results show that quantitative MRI using the two- point computational method available on the Picker International system provides reproducible values of the spin-lattice and spin-spin relaxation times (T\ and T respectively) for normal ap2  pearing white matter in both MS and EAE.  For MS, typical values obtained  ' post mortem are Ti = 381 ± 12 ms and T = 77 ± 9 ms; for fixed MS tissue 2  Ti = 207 ± 14 ms and T = 65 ± 5 ms. For EAE, typical in vivo values of normal 2  191  Chapter 5.  192  Conclusions and Suggestions for Further Work  appearing white matter relaxation times are T\ = 3 9 5 ± 2 5 ms and T = 7 7 ± 5 ms. 2  The values obtained from primate tissue are essentially equal to those obtained from human tissue, as expected.  It is clear from the in vitro studies that the  imaging pulse sequences used to determine T are only detecting the most mobile 2  protons; that is, the protons with a T value around 100ms. The fast relaxing 2  component around 10ms is not observed.  These quantitative M R I methods are  sensitive to differences in pathology in both M S and E A E ; the changes observed were dependent on the degree of gliosis present in MS tissue and on the severity of inflammation, demyelination and hemorrhagic necrosis in E A E . The most gliotic MS lesion present in three slices caused a 120% increase in T i and a 130% increase in T compared to normal post mortem white matter; the same lesion fixed gave 2  rise to a 160% in T i and a 230% change in T . 2  In primate E A E the maximum  changes observed were 41% in Ti and 133% in T . Volume-averaging and the effects 2  of multi-slice interference resulted in lower than expected T i and T values from 2  small lesions observed in only one slice image. The relaxation data obtained using this two-point method are useful for comparative purposes, but are not sensitive enough to provide the discrimination necessary to differentiate between acute MS lesions and more chronic demyelinating and gliotic lesions. 3. In vitro quantitative M R measurements on CNS tissue show that multi-exponential data analyses of C P M G data can discriminate between differing degrees of inflammation in guinea pigs inoculated to produce E A E , even when a maximum of only 50—60% of the sample is affected. Three exponentials provided the best fit to spinal cord C P M G data, and two exponentials the best fit to brain tissue C P M G data; analysis of fresh brain tissue data from a primate showed similar findings. Analysis of fixed guinea pig tissue using a fixed number of exponentials was complicated by  Chapter 5.  Conclusions and Suggestions for Further Work  193  the presence of free formalin solution; it was not possible to distinguish normal from abnormal tissue. Analysis of fixed tissue data using a continuous distribution of relaxation times provided better fits than the fixed number of exponentials; this analysis method could also discriminate between normal and abnormal fixed tissue. Relaxation studies on fixed M S tissue showed that both minuit and NNLS methods of analyses could distinguish between normal and abnormal tissue, and between differing degrees of gliosis and demyelination.  Once again, analysis assuming a  distribution of relaxation times provided the best fit to the data. In all the tissue studied, it was clear that two distributions were present; one centred around 10 ms and the other around 100 ms.  The source of the short component is unknown,  but the latter distribution probably arises from exchange between different cell environments and, in abnormal tissue, also between intra- and extra-cellular water. The ratio of these distributions changes from 6 to 15 in the presence of extensive parenchymal cellular infiltration; the T distribution at longer times also narrows. 2  These changes could occur due to an increase in the probability of finding water molecules in long T environments, and the decrease in the boundaries that the 2  water molecules have to cross in abnormal tissue. In addition, the cellular environment is more homogeneous in abnormal tissue. The technological advances being made in the field of M R I will enable the in vitro studies described in this thesis to be reproduced in vivo. Although these experiments must first be carried out on animal models, there is little doubt that the evolution of the disease processes, that are known as M S , can be followed using M R I .  This information could lead  to an understanding of the mechanisms involved in the disease progression, which would lead to better treatments and ultimately to understanding the cause(s) of the disease.  Chapter 5.  5.2  194  Conclusions and Suggestions for Further Work  Suggestions for Further Work  One of the main questions arising from the results reported here is the source of the two distributions observed for T in CNS tissue. While it seems likely that the distribution at 2  longer times arises from intra- and extra- cellular water exchange, the source of the bound water is not so obvious. One way to determine how real this fast relaxing component is, would be to study its behaviour with the amount of solid material present in the sample. Quantitative M R studies on wood [165] show that the moisture content can be determined from the F I D by extrapolation to zero, and subsequent determination of the solid components proton contribution to the signal.  Similar analysis could be carried  out on tissue; this would show whether the solid component changes, and/or whether the fast relaxing water component is constant or variable.  It would also be possible  to determine whether there is exchange between the solid protons and the bound water molecules using T\ measurements. Studies on rocks using a pulse gradient M R method, show that the range of pore sizes available to the water protons can be determined [166]. Such a study on tissue would provide supportive evidence for the increase in cellular homogeneity in the abnormal tissue, which may be one of the causes for the changes observed i n the guinea pig data. This is complicated by the rate of diffusion of water molecules in tissue; however, such pulsed gradient methods also allow the determination of the self diffusion coefficient, which would be expected to vary between normal and abnormal tissue due to changes in the number of boundaries encountered by the water molecules. A n obvious extension to this work is to increase the T\ data base for both fresh and fixed tissue.  Just as for T this would be carried out initially in vitro, with eventual 2  application in vivo. The guinea pig data presented here were primarily from inflammatory tissue; it is possible to produce more chronic forms of E A E in the strain 13 guinea  Chapter 5.  Conclusions and Suggestions for Further Work  195  pig. The pathology observed in these animals more closely resembles the demyelinating lesion seen in M S patients; a colony of these animals has been set up at U . B . C . to carry out such a study.  This would provide information on the effects of demyelination on  relaxation measurements separate from the effects of gliosis. This information could not be obtained from the studies presented here. As indicated in the discussion in chapter 4, the work presented here must be validated by an equivalent study in vivo, on E A E in guinea pigs for example. Spectroscopic studies in vitro and in vivo may also provide further information to aid in the interpretation of the relaxation data from CNS tissue.  This again would  involve the use of animal models for M S , but could also include a parallel study on M S patients.  Such a study would show whether mobile lipid protons contribute to the M R  signal observed, particularly during periods of relapse, which may be unaccompanied by clinical signs. The data presented in this thesis required the interaction of people in a number of different fields. Medically oriented projects, such as this, require a merging of science and medicine for them to succeed. The author has found this an exciting project, particularly when the long term implications for M S are considered; it is not the usual project carried out by chemistry graduate students, but the most interesting observation the author has made is that each area of medicine encountered (neurology, immunology, pathology, histology, anatomy, pharmacology and physiology) all have their basis in molecular science. This has meant that a chemistry background was probably the most useful for such a venture, and made the understanding of these different areas that much simpler.  The  near future promises to be an exciting time for M R scientists in medicine, and the author looks forward to seeing a solution to the M S puzzle in the not too distant future.  Appendix A  Extraction of Myelin Basic Protein from C N S Tissue  Myelin basic protein was prepared from primate brain tissue as follows; one whole primate brain was homogenized in 4 litres of a 2:1 mixture of chloroform and methanol. This was left to stir overnight at 4°C. The homogenate was filtered and washed twice with the 2:1 chloroform/methanol mixture and suspended in 2 litres of cold acetone. The residue was once again filtered and suspended in 4 litres of cold water, and left to stir for just under 1 hour. The suspension was filtered and the residue suspended in 0.03M HC1 (2 vol per gram of tissue). 1.0M HC1 was added as required to reduce the pH to 3. The suspension was stirred for 1 hour, then filtered. To the filtrate was added enough 8 M urea solution to dilute it to 2 M urea. The resulting solution was then titrated to pH 9 by adding cellulose DE-52. Having completed this stage, the mixture was allowed to stir for 30 min. The suspension was filtered under vacuum through Whatman 41 filter paper, and the clear foamy filtrate placed in dialysis tubing (Spectrapor tubing, 32 mm diameter. Mol. wt. cutoff 6,000-8,000), and dialyzed overnight in 4 litres of water at 4°C with two water changes. The solution was then freeze-dried and subjected to polyacrylamide gel electrophoresis ( P A G E ) to check the purity.  196  Appendix B  Results of N N L S Analysis of Guinea Pig C N S Tissue  The plots of amplitude versus T are shown in the following pages. The pig numbers 2  given in each spectrum, and the sample region specified above the plots.  197  endix B- Results of NNLS Analysis of Guinea Pig CNS Tissue  (duLiy  «_>  fdusy  t "* 0  (diuy  198  Appendix 8. Results of NNLS Analysis of Guinea Pig CNS Tissue  J 8 8 i£i m ^ r i  888 8 o  8  i_ _ -  8 8 8 8 8°  S o  IOUJV  M  loury  10 U J *  8 8 8 8 8° 8 5° SS  199  References  [1] Poser, C M . , Paty, D.W., Scheinberg, L., et al. The Diagnosis of Multiple Sclerosis. Edited by C M . Poser. New York: Thieme-Stratton (1984). [2] Alvord, Jr. E . C Acute Disseminated Encephalomyelitis and Allergic Neuroencephalopathies. In: Handbook of Clinical Neurology, Volume 9. Edited by P.J. Vinken and G.W. Bruyn. Amsterdam: North-Holland, pp. 500-571 (1970). [3] Confraveux, C , Aimard, G., Devic, M . Course and Prognosis of Multiple Sclerosis by the Computerized Data Processing of 349 Patients. Brain. 103, 281-300 (1980). [4] Kurtze, J.F., Hyllested, K . Multiple Sclerosis in the Faroe Islands II. Clinical Update, Transmission and the Nature of MS. Neurol. 36, 307-328 (1986). [5] Prineas, J.W., Connell, F. The Fine Structure of Chronically Active Multiple Sclerosis Plaques. Neurol. New York 28, 68-75 (1978). [6] Prineas, J.W. The Neuropathology of Multiple Sclerosis. In: Handbook of Clinical Neurology, Volume 3(47). New York: Elsevier Science, pp213-257 (1985). [7] Ebers, G . C , Vinuela F.V., Feasby T., Bass, B . Multifocal CT. Enhancement in Multiple Sclerosis. Neurol. 34, 341-346 (1984). [8] Young, I.R., Pallis, C.A., Bydder, G.M. et al. Nuclear Magnetic Resonance Imaging of the Brain in Multiple Sclerosis. Lancet II, 1063-1066 (1981). [9] Keeler, E.K., Coyle, P.K. Contrast Enhancement in NMR by Selection of Radiofrequency Pulse Sequence and Time: Examples in Demyelinating Disease. Physiol. 200  References  201  Chem. Phys. Med. N M R . 15, 235-237 (1983). [10] Runge, V . M . , Price, A . M . , Kirshner, H.S. et al. Magnetic Resonance Imaging of Multiple Sclerosis: A Study of Pulse Sequence Efficacy. A . J . R . 143, 1015-1026 (1984). [11] Young, I.R., Randall, C P . , Kaplan, P . W . et al. Nuclear Magnetic Resonance (NMR) Imaging in White Matter Disease of the Brain using Spin-Echo Sequences. J . Comput. Assist. Tomogr. 7(2), 290-294 (1983). [12] Bailes, D . R . , Young, I.R., Thomas, D . J . at al. NMR Imaging of the Brain using Spin-Echo Sequences. Clin. Radiol. 33, 395-414 (1982). [13] Paty, D . W . , Oger, J.J.F., Kastrukoff, L . F . et al. MRI in the Diagnosis of MS: A Prospective Study with Comparison of Clinical Evaluation, Evoked Potentials, OligoclonalBanding, and CT. Neurol. 38, 180-185 (1988). [14] Farlow, M . R . , Markand, O . N . , Edwards, M . K . at al. Multiple Sclerosis: Magnetic Resonance Imaging, Evoked Responses and Spinal Fluid Electrophoresis. Neurol. 36, 828-831 (1986). [15] Cutler, J.R., Aminoff, M . J . , Brant-Zawadski, M . Evaluation of Patients with Multiple Sclerosis by Evoked Potentials and Magnetic Resonance Imaging: A Comparative Study. A n n . Neurol. 20, 645-648 (1986). [16] Haughton, V . M . , Rimm, A . A . , Sobocinski, K . A . et al. A Blinded Clinical Comparison of MR Imaging and CT in Neuroradiology. Radiol. 160, 751-755 (1986). [17] Kirshner, H.S., Tsai, S.I., Runge, V . M . , Price A . C .  Magnetic Resonance Imaging  and other Techniques in the Diagnosis of Multiple Sclerosis. 859-863 (1985).  Arch. Neurol. 42,  References  202  [18] Reese, L . , C a n , T.J., Nicholson, R . L . , Lepp, E . K . Magnetic Resonance Imaging for Detecting Lesions of Multiple Sclerosis: Comparison with Computed Tomography and Clinical Assessment C . M . A . J . 135, 639-643 (1986). [19] Honig, L.S., Siddharthan, R., Sheremata, W . A . et al. Multiple Sclerosis: Correlation of Magnetic Resonance with Cerebrospinal Fluid Findings. J . Neurol. Neurosurg. Psych. 51, 277-280 (1988). [20] Uhlenbrock, D . , Seidel, D . , Gehlen, W . MR Imaging in Multiple Sclerosis: Comparison with Clinical, CSF and Visual Evoked Potentials Findings. A . J . N . R . 9, 59-67 (1988). [21] Papadopoulos, N . M . , McFarlin, D.E., Patronas, N . J . et al. A Comparison between Chemical Analysis and Magnetic Resonance Imaging with the Clinical Diagnosis of Multiple Sclerosis. A m . J . Clin. Pathol. 88, 365-368 (1987). [22] Farlow, M . R . , Edwards, M . K . , Kolar, O.J. et al. Magnetic Resonance Imaging in Multiple Sclerosis: Analysis of Correlations to Peripheral Blood and Spinal Fluid Abnormalities.  Neurol. 37, 1527-1530 (1987).  [23] Isaac, C , L i , D., Genton, M . at al. Multiple Sclerosis: A Serial study using MRI in Relapsing Patients. Neurol. In press. [24] Koopmans, R., L i , D . K . B . , Oger, J . J . F . K . et al.  Chronic Progressive Multiple  Sclerosis: Serial Magnetic Resonance Imaging over Six Months.  A n n . Neurol. In  press. [25] Willoughby, E.W., Growchowski, E . , L i , D . K . B . et al. Serial Magnetic Resonance Scanning in Multiple Sclerosis: A Prospective Study in Relapsing Patients. Ann. Neurol. In press.  References  203  [26] Johnson, M . A . , L i , D . K . B . , Bryan, D. J., Payne, J.A. Magnetic Resonance Imaging: Serial Observations in Multiple Sclerosis. A . J . N . R . 5, 495-499 (1984).  [27] Kastrukoff, L . F . , Hashimoto, S.A., Oger, J.J. et al. A Trial of Namalwa Interferon in the Treatment of Chronic Progressive (CP) Multiple Sclerosis (MS). I. Clinical  and MRI Analysis. Can. J . Neurol. Sci. 14, 242 (1987). [28] Ormerod, I.E.C., Miller, D . H . , McDonald, W.I. et al. The Role of NMR Imaging in the Assessment of Multiple Sclerosis and Isolated Neurological Lesions. A  Quantitative Study. Brain. 110,1579-1616 (1987). [29] Gebarski, S.S., Gabrielsen, T.O., Gilman, S. et al. The Initial Diagnosis of MS: Clinical Impact of MRI.  Ann. Neurol. 17, 469-474 (1985).  [30] Wider, G., Macura, S., Kumar, A . at al. Homonuclear Two-Dimensional H 1  NMR  of Proteins. Experimental Procedures. J . Magn. Reson. 56, 207-234 (1984). [31] Nelson, D.J., Opella, S.J., Jardetzsky, O.  1 3  C Nuclear Magnetic Resonance study  of Molecular Motions and Conformational Transitions in Muscle Calcium Binding  Parvalbumins. Biochem. 15(25), 5552-5560 (1976). [32] Fraser, P . E . , Deber, C M . Structure and Function of the Proline-Rich Region of  Myelin Basic Protein. Biochem. 24, 4593-4598 (1985). [33] Damadian, R., Zaner, K . , Hor, D. Human Tumors by NMR. Physiol. Chem. Phys.  5 , 381-402 (1973). [34] Bottomley, P.A., Foster, T . H . , Argersinger, R . A . , Pfeifer, L . M .  Review of Nor-  mal Tissue Hydrogen NMR Relaxation Times and Relaxation Mechanisms from 1-100 MHz: Dependence on Tissue Type, NMR Frequency, Temperature, Species,  Excision and Age. Med. Phys. 11(4), 425-448 (1984).  204  References  [35] Bottomley, P.A., Hardy, C.J., Argersinger, R . E . , Allen-Moore, G . A Review of^H Nuclear Magnetic Resonance Relaxation in Pathology: Are T\ and T Diagnostic! 2  Med. Phys. 14(1), 1-37 (1987). [36] Fullerton, G.D., Potter, J.L., Dornbluth, N . C . NMR Relaxation of Protons in Tissues and other Macromolecular Water Solutions. Magn. Reson. Imag. 1, 209-228  (1982). [37] Smith, A . S . , Weinstein, M . A . , Modic, M . T . at al. Magnetic Resonance with Marked T - Weighted Image. Improved Demonstration of Brain Lesion, Tumors and Edema. 2  A . J . R . 145, 945-955 (1985). [38] Buonanno, F.S., Pykett, I.L., Brady, T . J . at al. Proton NMR Imaging in Experimental Ischemic Infarction. Stroke. 14(2), 178-184 (1983). [39] Brant-Zawadski, M . , Enzmann, D . R . , Placone, Jr. R . C . at al. NMR Imaging of Experimental Brain Abscess: Comparison with CT. A . J . N . R . 4, 250-253 (1983).  [40] Brant-Zawadski, M . , Fein, G., Van Dyke, C. at al. MR Imaging of the Aging Brain: Patchy White Matter Lesions and Dementia. A . J . N . R . 6, 675-682 (1985).  [41] Langfitt, T . W . , Obrist, W . D . , Alavi, A . at al. Computerized Tomography, Magnetic Resonance Imaging and Positron Emission Tomography in the Study of Brain  Trauma. J . Neurosurg. 64, 760-767 (1986). [42] Guilleux, M - H . , Steiner, R . E . , Young, I.R. MR Imaging of Progressive Multifocal Leucoencephalopathy. A . J . N . R . 7, 1033-1035 (1986). [43] Paty, D . W . , L i , D . K . B .  Neuroimaging in Multiple Sclerosis. In: Clinical Neu-  roimaging. New York: Alan R. Liss Inc. pp. 249-278 (1988).  References  205  [44] Sze, G . , Knol, G . , Olsen, W . L . , Harper, P.S. at al. Hemorrhagic Neoplasms: MR Mimics of Occult Vascular Malformations. A . J . R . 149(6), 1223-1230 (1987). [45] Hunter, S.B., Ballinger, W . E . , Rubin, J.J. Multiple Sclerosis Mimicking Primary Brain Tumor. Arch. Path. Lab. Med. I l l , 464-468 (1987). [46] Larsson, H . B . W . , Frederiksen, J., Kjoer, L . at al. In Vivo Determination of Ti and T in the Brain of Patients with Severe but Stable Multiple Sclerosis. Magn. 2  Reson. Med. 7, 43-55 (1988). [47] Farrar, T.C., Becker, E.D.  Pulse and Fourier Transform NMR. New York: Aca-  demic Press (1971). [48] Abragam, A . Principles of Nuclear Magnetism.  Edited by W . C . Marshall and  D . H . Wilkinson. New York: Oxford University Press (1971). [49] Slichter, C.P. Principles of Magnetic Resonance.  2nd. Edition. Edited by M .  Cardona, P. Fulde and H.-J. Quiesser. Berlin: Springer-Verlag (1978) [50] Sanders, J . K . M . , Hunter, B . K . Modern NMR Spectroscopy: A Guide for Chemists. New York: Oxford University Press (1987). [51] Bracewell, R. The Fourier Transform and its Applications. New York: McGrawHill (1965). [52] Stark, D.D., Bradley Jr., W . G .  Magnetic Resonance Imaging. Edited by G . Sta-  mathis. St. Louis, Missouris: C.V. Mosby Company (1988). [53] Mansfield, P., Morris, P.G. NMR Imaging in Biomedicine. Edited by J.S. Waugh. New York: Academic Press (1982).  206  References  [54] Lerski, R . A . , Vennart, W . , Taylor, D . G . et al. Physical Principles and Clinical Applications of nuclear Magnetic Resonance. Edited by R . A . Lerski.  London:  The Hospital Physicists' Association (1985). [55] Kumar, A . , Welti, D . , Ernst, R . R . NMR Fouier Zeugmatography. J . Magn. Reson.  18, 69-83 (1975). [56] Bax A . Two-Dimensional Nuclear magnetic resonance in Liquids. Delft, Holland:  Delft University Press (1982). [57] Hall, L . D . , Morris, G . A . , Sukumar, S.  Resolution and Assignment of the 270  MHz Proton Spectrum of Cellobiose by Homo- and Heteronuclear Two-Dimensional  NMR.  J . A m . Chem. Soc. 102, 1745-1747 (1980).  [58] Talagala, S.L. Aspects of NMR imaging and In Vivo Spectroscopy. P h . D . Thesis.  University of British Columbia, Vancouver (1986). [59] Hahn, E . L . Spin Echoes. Phys. Rev. 80, 580-594 (1950). [60] Koenig, S.H., Brown, R . D . Relaxation of Solvent Protons by Paramagnetic Ions and its Dependence on Magnetic Field and Chemical Environment. Implications  for NMR Imaging. Magn. Reson. Med. 1, 478-495 (1984). [61] Weinmann, H.-J., Brasch, R . C , Press, W . R . et al. Characteristics of GadoliniumDTPA: A Potential NMR Contrast Agent A . J . R . 142(1-3), 619-624 (1984). [62] Gadian, D . G . , Payne, J . A . , Bryant, D . J . et al. Gadolinium-DTPA  as a Contrast  Agent in MR Imaging: Theoretical Projections and Practical Observations. J . Corn-  put. Assist. Tomogr. 9(2), 242-251 (1985).  References  207  [63] Strich, G . , Hagan, P . L . , Gerber, K . H . , Slutsky, R . A .  Tissue Distribution and  Magnetic Resonance Spin-Lattice Relaxation Effects of Gadolinium-DTPA.  Radiol.  154, 723-726 (1985). [64] Niendorf, H.P., Semmler, W . , Laniado, M . , Felix, R. Some Aspects of the use of Contrast Agents in Magnetic Resonance Imaging. Diagn. Imag. Clin. Med. 55,  25-36 (1986). [65] Wolf, G . L . , Joseph, P . M . , Goldstein, E . J . Optimal Pulse Sequences for MR Con-  trast Agents. A . J . R . 147, 367-371 (1986). [66] Kilgore, D.P., Bregar, R . K . , Daniels, D . L . et al. Cranial Tissues: Normal Appearance after Intravenous Injection of Gd-DTPA. Radiol. 160, 757-761 (1986).  [67] Valk, J., de Slegte, R . G . M . , Crezee, F . C . et al. Contrast Enhanced Magnetic Resonance Imaging of the Brain using Gadolinium-DTPA.  Acta. Radiol. 28(6), 659-665  (1987). [68] Grossman, R.I., Gonzalez-Scarano, F., Atlas, S.W. et al.  Multiple Sclerosis:  Gadolinium Enhancement in MR Imaging. Radiol. 161, 721-725 (1986). [69] Gonzalez-Scarano, F., Grossman, R . L , Galetta, S. et al. Multiple Sclerosis Disease Activity Correlates with Gadolinium-Enhanced Magnetic Resonance Imaging. Ann.  Neurol. 21, 300-306 (1987). [70] Miller, D . H . , Rudge, P., Johnson, G . at al. Serial Gadolinium Enhanced Magnetic Resonance Imaging in Multiple Sclerosis. Brain. I l l , 927-939 (1988). [71] Lynch, L . J . , Webster, D.S.  Proton Spin-Lattice Relaxation of Water Associated  with Brown Coal. J. Magn. Reson. 40, 259-272 (1980).  208  References  [72] McDonald, G . G . , Leigh, Jr. J.S. A New Method for Measuring Longitudinal Relaxation Times. J . Magn. Reson. 9, 358-362 (1973). [73] Freeman, R., H i l l , H . D . W . Fourier Transform Study of NMR Spin-Lattice Relaxation by "Progressive Saturation". J . Chem. Phys. [74] Levy, G.C., Peat, L R .  54(8),  3367-3377 (1971).  The Experimental Approach to Accurate Carbon-lS Spin-  Lattice Relaxation Measurements. J . Magn. Reson. 18, 500-521 (1975). [75] Sass, M . , Ziessow, D . Error Analysis for Optimized Inversion-Recovery Spin-Lattice Relaxation Measurements. J . Magn. Reson. 25, 263-276 (1977). [76] Carr, H . Y . , Purcell, E . M .  Effects of Diffusion on Free Precession in Nuclear Mag-  netic Resonance Experiments. Phys. Rev. 94, 630-638 (1954). [77] Meiboom, S., G i l l , D . Modified Spin-Echo Method for Measuring Nuclear Relaxation Times. Rev. Sci. Instr. 29, 688-691 (1958). [78] Freeman, R., H i l l , H . D . W . High Resolution Study of NMR Spin-Echoes: "J Spectra". J . Chem. Phys. 54, 301-313 (1971). [79] Condon, B . , Patterson, J . , Wiper, D . et al. Comparison of Calculated Relaxation Parameters between an MR Imager and Spectrometer Operating at Similar Frequencies. Magn. Reson. Imag. 4, 449-454 (1986). [80] L i n , M.S., Fletcher, J . W . , Herbig, F . K . , Donati, R . M .  Numerical T  x  Computation  from NMR Intensity Ratios. Magn. Reson. Imag. 4, 311-319 (1986). [81] Graumann, R., Fisher, H . , Oppelt, A . A New Pulse Sequence for Determining Ti and T Simultaneously. Med. Phys. 2  13(5),  644-647 (1986).  References  209  [82] MacFall, J . M . , Wehrli, F . W . , Breger, R . K . , Johnson, G . A . Methodology for the Measurement and Analysis of Relaxation Times in Proton Imaging. Magn. Reson.  Imag. 5(3), 209-220 (1987). [83] Conturo, T . E . , Price, R . R . , Beth, A . H . et al.  Improved Determination of Spin  Density, T i and T from a Three-Parameter Fit to Multiple-Delay-Multiple-Echo 2  (MDME) NMR Images. Phys. Med. Biol. 31(12), 1361-1380 (1986). [84] Stewart, W . A .  Quantitative NMR Imaging and its Applications In Vivo. M.Sc.  Thesis. University of British Columbia, Vancouver (1986). [85] Kandel, E . R . , Schwartz, J.S. Principles of Neural Science. Edited by E . R . Kandel  and J.H. Swartz. New York: Elsevier North Holland, Inc. (1982). [86] DeArmond, S.J., Fusco, M . M . , Dewey, M . M .  Structure of the Human Brain. A  Photographic Atlas, 2nd Edition. New York: Oxford University Press (1976). [87] Gilman, S., Winans, S.S. Manter and Gatz's Essentials of Clinical Neuroanatomy  and Physiology, 6th Edition. [88] Shiraki, H . , Otani, S.  Philadelphia: F . A . Davis Company (1982).  Clinical and Pathological Features of Rabies Post-Vaccinal  Encephalomyelitis in Man. In: Allergic Encephalomyelitis. Edited by M . W . Kies and E . C . Alvord Jr. Springfield: Charles C. Thomas, pp. 58-129 (1959) [89] Levine, S. Relationship of Experimental Allergic Encephalomyelitis to Human Dis-  eases. In: Immunological Disorders of the Nervous System. Edited by L.P. Rowland.  Research Publications, Association for Research in Nervous and Mental  Disease. 49, 33-49 (1971). [90] Zimmerman, H . M . , Netsky, M . G .  The Pathology of Multiple Sclerosis. Res. Publ.  Assoc. Res. Nerv. Ment. Dis. 28, 271-312 (1950).  210  References  [91] Greenfield, J . G . Demyelinating Diseases. In: Neuropathology.  Edited by J . G .  Greenfield, W . Blackwood, W . H . McMenemey, A . Meyer and R . M . Norman. London:Edward Arnold, pp. 441-474 (1958). [92] Greenfield, J . G . , Norman, R . M . Demyelinating Diseases. In: Greenfield's Neuropathology 2nd Edition. Edited by W . Blackwood, W . H . McMenemey, A . Meyer, R . M . Norman and D.S. Russell. London: Edward Arnold, pp. 475-519 (1963). [93] Lumsden, C.E. The Neuropathology of Multiple Sclerosis. In: Handbook of Clinical Neurology. Vol. 9, Multiple Sclerosis and Other Demyelinating Diseases.  Edited  by P . J . Vinken, G . W . Bruyn. Amsterdam: North-Holland, pp 217-309 (1970). [94] Dawson, J.W. The Histology of Multiple Sclerosis. Trans. R. Soc. Edinburgh, Vol. L pp 517-740 (1916), reproduced by the Montreal Neurological Institute, 1973. [95] Adams, R . D . , Kubik, C.S.  The Morbid Anatomy of the Demyelinating Diseases.  A m . J. Med. 12, 510-546 (1952). [96] Adams, C . W . M .  The General Pathology of Multiple Sclerosis: morphological and  clinical aspects of the lesions. In: Multiple Sclerosis. Pathology, Diagnosis and Management.  Edited by J.F. Hallpike, C . W . M . Adams and W . W . Tourtellotte.  Baltimore:WiUiams and Wilkins, pp. 203-240 (1983). [97] Hallpike, J.F., Adams, C . W . M . , Bayliss, O . B . Histochemistry of Myelin. VIII. Proteolitic activity around multiple sclerosis plaques. Histochem. J . 2, 199-208  (1970). [98] Perier, O., Gregoire, A .  Electron Microscopic Features of Multiple Sclerosis Le-  sions. Brain. 88, 937-952 (1965).  211  References  [99] Boyazis, R . M . , Martin, L . , Bouteille, M . et al. Images histochemiques et ultrastructurales dans un cas de sclerose en plaques associee a un spongioblastoma R i v .  Patol. Nerv. Ment. 88, 1-20 (1967). [100] Suzuki, K . , Andrews, J . M . , Waltz, J . M . , Terry, R . D .  Ultrastructural studies of  multiple.sclerosis. Lab. Invest. 20, 444-454 (1969). [101] Gonatas, N . K . Ultrstructural observation in a case of multiple sclerosis (Abstract).  J. Neuropathol. Exp. Neurol. 29, 49 (1970). [102] Lhermitte, F . , Escourolle, R., Cathala, F . , Hauw J.-J., Marteau R. Etude neuropathologique d'un cas de sclerose en plaques. Rev. Neurol. 129, 3-19 (1973).  [103] Tanaka, J . , Garcia, J . H . , Khurana, R. at al. Unusual demyelinating disease. A Form of diffuse-disseminated sclerosis. Neurol. 25, 588-593 (1975). [104] Reikkinen, P. J., Rinne, U . K . , Arstila, A . V . Neurochemical and morphological studies on demyelination in multiple sclerosis with special reference to etiological as-  pects. Z. Neurol. 203, 91-104 (1972). [105] Prineas, J . Pathology of the Early Lesions in Multiple Sclerosis. Hum. Pathol. 6,  531-554 (1975). [106] Prineas, J.W., Connell, F . Remyelination in multiple sclerosis. A n n . Neurol. 5,  22-31 (1979). [107] Raine, C.S., Scheinberg, L . , Waltz, J . M . Multiple Sclerosis. Oligodendrocyte survival and proliferation in an active established plaque. Lab. Invest.  (1981).  45(6), 534-546  References  212  [108] Prineas, J . W . , Kwon, E . E . , Cho, E.-S., Sharer, L . R . Continual breakdown and regeneration of myelin in progressive multiple sclerosis plaques. A n n . N . Y . Acad. Sci. 436, 11-32 (1984). [109] Zimmerman, J.R., Brittin W . E . Nuclear Magnetic Resonance Studies in Multiple Phase Systems: Lifetime of a Water Molecule in an Adsorbing Phase on Silica Gel. J . Phys. Chem. 6 1 , 1328-1333 (1957). [110] Richards, T . L . , Kenney, J.S., Rose, L . M . et al. Detection of Brain Lipids in Primates with Experimental Allergic Encephalomyelitis by In Vivo Proton Spectroscopy and In Vitro Histochemistry. Seventh Annual Meeting, S . M . R . M . San Francisco, August (1988). [ I l l ] Gyorffy-Wagner, Z., Englund, E . , Larsson, E . - M . et al. Proton Magnetic Resonance Relaxation Times Ti and T Related to Post Mortem Interval. An investigation on 2  porcine brain tissue. Acta. Radiol. 27, 115-118 (1986). [112] Kamman, R . L . , Go, K . G . , Stomp, G.P. et al.  Changes of Relaxation Times T\  and T in Rat Tissue After Biopsy and Fixation. Magn. Reson. Imag. 3, 245-250 2  (1985). [113] Chang, D . C . , Hazefwood, C . F . , Woessner, D . E . The Spin-Lattice Relaxation Times of Water Associated with early Post Mortem Changes in Skeletal Muscle. Biochim. Biophys. Acta. 437, 253-258 (1976). [114] Thickman, D . L , Kundel, H . L . , Wolf, G . Nuclear Magnetic Resonance Characteristics of Fresh and Fixed Tissue: The Effects of Elapsed Time. Radiol. 148, 183-185 (1983).  213  References  [115] Nelson, T.R., Tung, S.M. Temperature Dependence of Proton Relaxation Times In Vitro. Magn. Reson. Imag. 5, 189-199 (1987). [116] Kiernan, J . A . Histological and Histochemical Methods: Theory and Practice. Oxford: Pergamon Press (1981). [117] Raine, O S . Analysis of Autoimmune Demyelination: Its Impact on Multiple Sclerosis. Lab. Invest. 50(6), 608-635 (1984). [118] Tourtellotte, W . W . , Parker, J . A . Some Spaces and Barriers in Postmortem Multiple Sclerosis. In: Progress in Brain Research, Volume 29. Edited by A . Lajthan and D . H . Ford. New York: Elsevier Science, pp. 493-525 (1968). [119] Alvord Jr., E . C . Disseminated Encephalomyelopathies:  Its Variations in Form  and their Relationship to other Disease of the Nervous System  In: Handbook of  Clinical Neurology, Volume 3(47). Edited by P.J. Vinken, G . W . Bruyn and H . L . Klawans. New York: Elsevier Science, pp. 467-502 (1985). [120] Alvord Jr., E . O , Shaw, C , Hruby, S. Myelin Basic Protein Treatment of Experimental Allergic Encephalomyelitis in Primates. [121] Deibler, G . E . , Boyd, L . F . , Kies, M . W . Basic Protein.  A n n . Neurol. 6, 469-473 (1979).  Proteolytic Activity with Purified Myelin  In: Experimental Allergic Encephalomyelitis: A Useful Model for  Multiple Sclerosis. Edited by E . C . Alvord Jr., M . W . Kies and A . J . Suckling. New York: Alan R. Liss, pp. 249-256 (1984). [122] Alvord Jr., E . C , Shaw, C M . , Hruby, S. et al.  The Onset of Experimental Al-  lergic Encephalomyelitis as Defined by Clinical, Electrophysical and Immunological Changes. In: Experimental Allergic Encephalomyelitis: A Useful Model for Multiple Sclerosis. Edited by E . C . Alvord Jr., M . W . Kies and A . J . Suckling. New York:  214  References  Alan R. Liss, pp. 461-466 (1984). [123]  Kjos, B.O., Ehmen, R . L . , Brant-Zawadski, M . et al. Reproducibility of Relaxation Times and Proton Density Calculated from Routine MR Imaging Sequences: Clin-  ical Study of the CNS. A . J . R . 144, 1165-1170 (1985). [124]  Stewart, W . A . , Alvord Jr., E.C., Hruby S., et al. Early Detection of EAE in a Primate using NMR Imaging. Lancet II, 895 (1985).  [125] O'Brien, J.T., Noseworthy, J.H., Gilbert, J.J., Karlick, S.J.  NMR Changes in Ex-  perimental Allergic Encephalomyelitis: NMR Changes Precede Clinical and Patho-  logical Events. Magn. Reson. Med. 5, 109-117 (1987). [126]  Stewart W . A . , Alvord Jr., E.C., Hruby, S. et al. Magnetic Resonance Imaging (MRI) of Experimental Allergic Encephalomyelitis (EAE) in Primates. Submitted  for publication. [127]  Stewart, W . A . , Barwick, S.E., Whittall, K . P . et al. In Vitro Multi-Exponential Relaxation In Multiple Sclerosis and Experimental Allergic Encephalomyelitis. Works  In Progress, Seventh Annual Meeting, S . M . R . M . , San Francisco, August (1988). [128] Bakker, C . J . G . , Vriend, J . Multi-exponential Water Proton Spin-Lattice Relaxation in Biological Tissues and its Implications for Quantitative NMR Imaging. Phys.  Med. Biol. 29(5), 509-518 (1984). [129]  Kamman, R . L . , Bakker, C . J . G . , Van Dijk, P. et al.  Multi-exponential Relax-  ation Analysis with MR Imaging and NMR Spectroscopy Using Fat- Water Systems.  Magn. Reson. Imag. 5(5), 381-392 (1987). [130]  Koenig S.H., Brown R.D., Adams D . et al. Magnetic Field Dependence of l/T^ of Protons in Tissue. Invest. Radiol. 19, 76-81 (1984).  References  [131] Beall P.T.  215  Practical Methods for Biological NMR Sample Handling. Magn. Reson.  Imag. 1, 165-181 (1982). [132] Woessner D . E . Spin-Relaxation Processes in a Two-Proton System Undergoing  Anisotropic Reorientation J . Chem. Phys. 30(1), 1-4 (1962). [133] Bloemgergen, N . , Purcell, E . M . , Pound, R . V .  Relaxation Effects in Nuclear Mag-  netic Resonance Absorption. Phys. Rev. 73, 679-712 (1948). [134] Martin, M . L . , Delpuech, J.J., Martin, G.J. Practical NMR Spectroscopy. London:  Heyden (1980). [135] Akber, S.F. Comments on "A Review of H Nuclear magnetic Resonance Relax1  ation in Pathology; Are T and T Diagnostic"? x  2  Med. Phys. 14(6), 1090- 1091  (1987). [136] Harvey, S.C., Hoekstra, P.  Dielectric Relaxation Spectra of Water Absorbed on  Lysozyme. J . Phys. Chem. 76(21), 2987-2994 (1972). [137] Grosch, L . , Noack, F . NMR Relaxation Investigation of Water Mobility in Aqueous  Bovine Serum Albumin Solutions. Biochim. Biophys. Acta. 453, 218-232 (1976). [138] Fullerton, G.D., Cameron, I.L., Hunter, K . , Fullerton, H . J .  Proton Magnetic Re-  osnance Relaxation Behaviour of Whole Muscle with Fatty Inclusions.  Radiol.  155(3), 727-730 (1985). [139] Fullerton, G . D . , Cameron, I.L., Ord, V . A .  Frequency Dependence of Magnetic  Resonance Spin-Lattice Relaxation of Protons in Biological Materials. Radiol. 151,  135-138(1984).  216  References  [140] Fullerton, G.D., Cameron, I.L. An Evaluation of the Hydration of Lysozyme by an NMR Titration Method. Biochim. Biophys. Acta. 869, 230-246 (1986). [141] Karplus, M . , Rossky, P. Solvation: A Molecular Dynamics Study of a Dipeptide in Water. In: American Chemical Society No. 127. Edited by S.P. Rowland. Washington D . C . , pp. 23- 42 (1980). [142] Edzes, H . T . , Smulski, E . T . The Measurement of Cross Relaxation Effects in the Proton NMR Spin-Lattice Relaxation of Water in Biological Systems: Hydrated Collagen and Muscle. J. Magn. Reson. 31, 207-229 (1978). [143] Hallenga, K . , Koenig, S. Protein Rotational Relaxation as Studied by H and H 1  Magnetic Relaxation  2  Biochem. 15, 4255-4264 (1976).  [144] Lynch, L . J . , Marsden, K . H . , George, E.P. NMR of Absorbed System, I. A Systematic Method of Analyzing NMR Relaxation Time Data for a Continuous Distribution of Correlation Times. J . Chem. Phys. 51(12), 5673-5680 (1969). [145] Resing, H . A . Apparent Phase-Transition Effect in the NMR Spin-Spin Relaxation Time caused by a Distribution of Correlation Time. J . Chem. Phys. 43, 669-678 (1965). [146] Kroeker R . M . , Henkelman R . M . Analysis of Biological NMR Relaxation Data with Continuous Distributions of Relaxation Times. J . Magn. Reson. 69, 218-235 (1986). [147] Peemoeller, H . , Pintar, M . M . , Kydon, D . W . Nuclear Magnetic Resonance Analysis of Water in Natural and Deuterated Mouse Muscle Above and Below Freezing. Biophys. J . 29, 427-435 (1980). [148] Peemoeller H . , Pintar M . M .  Nuclear Magnetic Resonance Multiwindow Analysis  References  217  of Proton Local Fields and Magnetization Distribution in Natural and Deuterated  Mouse Muscle. Biophys. J . 28, 339-356 (1979). [149] Menon, R.S., Allen P.S. Proton Relaxation in Cellular Model Edema Systems.  Abstract, Seventh Annual Meeting, S . M . R . M . , San Francisco, August (1988). [150] James F . , Roos M . Minuit-A system for Function Mfnization and Analysis of the  Parameter Errors and Correlations. Comput. Phys. Commun. 10, 343-367 (1975). [151] Provencher S.W. A Constrained Regularization Method for Inverting Data Represented by Linear Algebraic or Integral Equation. Comput. Phys. Commun. 27,  213-227 (1982). [152] Whittall, K . P . , Mackay, A . L . Quantitative Interpretation of NMR Relaxation Data.  J. Magn. Reson. In press, July (1989). [153]  Lawson, C . L . , Hanson, R . J .  Solving Least Squares Problems. Engelwood Cliffs,  N . J . : Prentice-Hall (1974). [154]  Gaber, B.P., Brown R.D., Koenig S.H., Fee, J.A.  Nuclear Magnetic Relaxation  Dispersion in Protein Solutions. V. Bovine Erythrocyte Superoxide Dismutase.  Biochem. Biphys. Acta. 271, 1-5 (1972). [155] Koenig S.H., Schillinger W . E .  Nuclear Magnetic Relaxation Dispersion in Protein  Solutions. II. Transferrin. J . Biol. Chem. 244, 6520-6526 (1969). [156]  Raine C S . Neurocellular Anatomy. In: Basic Neurochemistry 3rd Edition. Edited  by G.J. Siegel, R . W . Albers, B . W . Agranoff and R. Katzman. Brown and Company pp. 21-47 (1981).  Boston: Little,  218  References  [157] Mills, R. Self Diffusion  in Normal  and Heavy  Water  in the Range  l-456o.  J. Phys.  Chem. 7 7 , 685-689 (1973). [158] Davidson, N . Statistical  New York: McGraw-Hill, pp.284-285 (1962).  Mechanics.  [159] Brownstein, K . R . , Tarr, C E . Importance of Water  in Biological  of Classical  Diffusion  in NMR  Phys. Rev. 1 9 ( 6 ) , 2446-2453 (1979).  Cells.  [160] Karlik, S.J., Strejan, G . , Gilbert, J.J., Noseworthy, J . H . NMR perimental  Allergic  Parenchymal  Encephalomyelitis  Cellular  infiltration.  (EAE):  Normalization  Nervous  Studies  in Ex-  of T\ and T  with  2  Neurol. 3 6 , 1112-1114 (1986).  [161] Cuzner, M . L . , Davidson, A . N . , Thompson R.H.S. the Central  Studies  The Demyelinating  Diseases  of  In: The Molecular Basis of Neuropathology. Edited  System  by A . N . Davidson and R.H.S. Thompson. London: Edward Arnold L t d . Chap. 15, pp. 384-411 (1981). [162] Lacomis, D., Osbakken, M . , Gross, G . SpinCerebral  White  Matter  in Multiple  Sclerosis.  Lattice  Relaxation  in Tissue  Resulting  from  Spin  of  Magn. Reson. Med. 3 , 194-202 (1986).  [163] Santyr, G.E., Henkelman, R . M . , Bronskill M . J . Variation Relaxation  (T\) Times  Locking  with  in Measured the CPMG  Transverse  Sequence.  J.  Magn. Reson. 7 9 , 28-44 (1988). [164] Kroeker, R . M . , McVeigh, E.R., Hardy, P., Bronskill, M . J . , Henkelman, R . M . In vivo Measurements  of NMR  Relaxation  Times.  Magn. Reson. Med. 2, 1-13 (1985).  [165] Menon, R.S., Mackay, A . L . , Hailey, J.R.T. et al. An NMR Determination Physiological  1-15 (1987).  Water  Distribution  in Wood During  Drying.  of the  J . Appl. Poly. Sci. 29,  219  References  [166] Lipsicas, M . , Banavar, J.R., Willemsen, J. Surface Relaxation and Pore Sizes in Rocks-a Nuclear Magnetic Resonance  1546 (1986).  Analysis.  Appl. Phys. Lett. 4 8 ( 2 2 ) , 1544-  

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.831.1-0059449/manifest

Comment

Related Items