UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Empiric risk to first-degree relatives of Alzheimer disease patients Hirst, Clinton G. 1993

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

Item Metadata


831-ubc_1993_spring_hirst_clinton.pdf [ 4.82MB ]
JSON: 831-1.0086117.json
JSON-LD: 831-1.0086117-ld.json
RDF/XML (Pretty): 831-1.0086117-rdf.xml
RDF/JSON: 831-1.0086117-rdf.json
Turtle: 831-1.0086117-turtle.txt
N-Triples: 831-1.0086117-rdf-ntriples.txt
Original Record: 831-1.0086117-source.json
Full Text

Full Text

We accept this thesis as conformingto the^standardEMPIRIC RISK TO FIRST-DEGREE RELATIVESOF ALZHEIMER DISEASE PATIENTSbyClinton Grant HirstB.Sc., The University of British Columbia, 1989THESIS SUBMITTED IN PARTIAL FULFILLMENT OFTHE REQUIREMENTS FOR THE DEGREE OFMASTER OF SCIENCEinTHE FACULTY OF GRADUATE STUDIES(Genetics Graduate Program)THE UNIVERSITY OF BRITISH COLUMBIAMarch 1993©Clinton Grant HirstIn presenting this thesis in partial fulfilment of the requirements for an advanceddegree at the University of British Columbia, I agree that the Library shall make itfreely available for reference and study. I further agree that permission for extensivecopying of this thesis for scholarly purposes may be granted by the head of mydepartment or by his or her representatives. It is understood that copying orpublication of this thesis for financial gain shall not be allowed without my writtenpermission.(Signature)Department of  G enet cS G ruduet ate- ^rc-t tv‘The University of British ColumbiaVancouver, CanadaDate Apr , 1 14^19913DE-6 (2/88)AbstractKaplan-Meier risk estimates were generated and evaluated for 1867 first-degreerelatives of 338 Alzheimer Disease (AD) index cases and 1873 first-degree relativesof 351 non-cognitively impaired, non-demented elderly controls. A cumulativelifetime risk of 26.73 ± 4.42% for first-degree relatives of AD cases, whilesignificantly higher than the cumulative lifetime risk of 7.26 ± 2.74% found for first-degree relatives of controls, does not suggest that all cases of AD are due to a singleautosomal dominant gene(s), but is evidence that a genetic component to the etiologyof the disease exists. Female and male first-degree relatives of AD cases as well asparents and sibs of AD cases showed higher cumulative risks throughout theirlifetimes when compared to their analogous control subgroups. Equal cumulativelifetime risks between first-degree female and male relatives (28.97 ± 3.60%, 22.03 ±8.29%) and between parents and sibs (27.67 ± 4.50%, 28.87 ± 8.92%) of AD cases,but significantly different lifetime risk curves suggest that non-genetic factors mayaffect the age-specific expression of AD in individuals with an identical geneticpropensity. Equal risk between first-degree relatives of early-onset and late-onset ADcases does not support the suggestion that relatives of early-onset cases share ahigher risk to develop the disease. These results are interpreted as being suggestiveof a complex etiology with both genetic and environmental factors, as well asinteractions between the two playing a role in AD expression.iiTable of ContentsAbstract ^  iiTable of Contents ^  iiiList of Tables  viiList of Figures ^  ixAcknowledgements  X1 Introduction1.1 Background ^  11.2 Diagnosis  21.3 Etiological Heterogeneity1.31 Risk Factors1.31.1 Background ^  51.31.2 History of Head Trauma ^  61.31.3 History of Aluminum Exposure  81.31.4 History of Thyroid Disease ^  91.31.5 Female Gender ^  101.31.6 Other Risk Factors  121.32 FAD ^  131.33 Molecular Genetic Investigations ^  141.34 Twin Studies ^  161.35 Association Between AD and Down Syndrome ^ 18iii1.4 Overview of Previous Studies Examining Empiric Risk to First-Degree Relatives of AD Cases and Controls Using SimilarMethodologies ^  191.5 Canadian Study on Health and Aging ^  251.6 The Modified Mini-Mental State Exam  271.7 Clinic for Alzheimer Disease and Related Disorders, UniversityHospital-UBC Site ^  281.8 Relevance of Study1.81 Magnitude of AD on the Canadian Population and HealthCare System ^  291.82 Familial Risks  302 Materials and Methods2.1 Cases ^  322.2 Relatives of Cases ^  322.3 Controls ^  352.4 Relatives of Controls ^  362.5 Analysis2.51 Overview ^  362.52 Computational Formulas ^  392.6 Risk Comparison2.61 Overview ^  422.62 Criteria and Analysis for FAD ^  43iv2.63 Risk Comparisons for Sample Subgroups ^ 443 Results3.1 Sample Analyzed3.11 Cases ^  473.12 Controls  473.2 Risk Estimates for First-Degree Relatives of Cases andControls ^  503.3 Risk Estimates for FAD ^  503.4 Risk Estimates for Female and Male First-Degree Relatives ofCases and Controls3.41 Risk Estimates for Female First-Degree Relatives ofCases and Controls ^  583.42 Risk Estimates for Male First-Degree Relatives ofCases and Controls ^  583.43 Risk Estimates for Female and Male First-DegreeRelatives of Cases ^  673.44 Risk Estimates for Female and Male First-DegreeRelatives of Controls ^  673.5 Risk Estimates for Parents and Sibs of Cases andControls3.51 Risk Estimates for Parents of Cases and Controls ^ 703.52 Risk Estimates for Sibs of Cases and Controls ^ 70v3.53 Risk Estimates for Parents and Sibs of Cases ^ 803.54 Risk Estimates for Parents and Sibs of Controls ^ 803.6 Risk Estimates for First-Degree Relatives of Early-Onset andLate-Onset Cases ^  803.7 Comparison of Previous and Present Samples ^ 853.8 Effect of Using Missing Ages-of-Onset  854 Discussion4.1 First-Degree Relatives of Cases and Controls ^ 924.2 Gender-Specific Risks ^  994.3 Generational-Specific Risks  1014.4 First-Degree Relatives of Early-Onset and Late-Onset Cases ^ 1044.5 Methodology Caveats ^  1064.6 AD Models ^  1085 Conclusion ^  112References  114Appendix A ^  130viList of TablesTable 1.1 Major Causes of Dementia ^  4Table 1.2 Commonly Reported Risk Factors for Alzheimer Disease ^ 7Table 1.3 Overview of Previously Reported Risk Estimates Using SimilarMethodologies ^  22Table 2.1 First-Degree Relative Subgroup Comparisons ^  46Table 3.1 Diagnosis for 883 Alzheimer Clinic Patients After Evaluation ^ 48Table 3.2 Number and Percent of Affected First-Degree Relatives in EachCase Subgroup^  48Table 3.3 Number and Percent of Affected First-Degree Relatives in EachControl Subgroup ^  49Table 3.4 Comparison of Cases and Controls ^  49Table 3.5 Age Specific Risks for First-Degree Relatives of Cases andControls ^  51Table 3.6 Age Specific Risks for First-Degree Relatives from FADFamilies ^  54Table 3.7 Age Specific Risks for Female First-Degree Relatives of Casesand Controls ^  60Table 3.8 Age Specific Risks for Male First-Degree Relatives of Casesand Controls ^  62Table 3.9 Age Specific Risks for Parents of Cases and Controls ^ 71viiTable 3.10 Age Specific Risks for Sibs of Cases and Controls ^ 74Table 3.11 Age Specific Risks for First-Degree Relatives of Early-Onsetand Late-Onset Cases ^  81Table 3.12 Summary of Results  91Table 4.1 Cumulative Risk Calculations and Number of First-Degree Relatives"at-Risk": A Literature Review ^  93viiiList of FiguresFigure 3.1 Risk to First-Degree Relatives of Cases and Controls ^ 56Figure 3.2 Risk to "FAD-Only" First-Degree Relatives of Cases  57Figure 3.3 Risk to Female First-Degree Relatives of Cases and Controls ^ 65Figure 3.4 Risk to Male First-Degree Relatives of Cases and Controls ^ 66Figure 3.5 Risk to Female and Male First-Degree Relatives of Cases ^ 68Figure 3.6 Risk to Female and Male First-Degree Relatives of Controls ^ 69Figure 3.7 Risk to Parents of Cases and Controls ^  76Figure 3.8 Risk to Sibs of Cases and Controls  77Figure 3.9 Risk To Parents and Sibs of Cases ^  78Figure 3.10 Risk to Parents and Sibs of Controls  79Figure 3.11 Risk to First-Degree Relatives of Early and Late-OnsetCases^  84Figure 3.12 Comparison of Previous and Present Samples ^  86Figure 3.13 Effect of Using Unknown Ages-of-Onset  90ixAcknowledgementsThe data on "controls" reported in this thesis were collected as part of the CanadianStudy of Health and Aging This was funded by the Senior Independence ResearchProgram, administered by the National Health Research and Development Program(NHRDP) of Health and Welfare Canada. The study was coordinated through theUniversity of Ottawa and the Federal Government's Laboratory Center for DiseaseControl.Specific Funding for this project was from the Alzheimer Society of Canada, BritishColumbia Division. NHRDP Studentship funding was from the Alzheimer Society ofCanada.I would like to acknowledge the help and assistance of Dr. Hossein Ameli, BevCapper, Nancy Greig, Dr. John Petkau, Jean Turnbull, and Irene Yee.I would also like to thank Dr. Dessa Sadovnick for her guidance and supportthroughout the last three years and acknowledge that without her assistance this workwould not be possible.x1 INTRODUCTION1.1 BackgroundAlzheimer Disease (AD) is a progressive degenerative brain disease believed tobe the most common cause of dementia. AD affects between 5 to 11% of thepopulation aged 65 and older, and up to 47% of those aged 85 and above (Evans etal., 1989; Pfeffer et al., 1987; Schoenberg et al., 1985). AD was first described by theGerman psychiatrist Alois Alzheimer, who presented the case of a 51 year old womanwith behavioral symptoms and memory problems. When a then, newly developedsilver stain technique (Bielschowsky method) was used at autopsy, abnormal nervecells containing tangles of nerve fibres (neurofibrillary tangles) and collections ofdegenerating nerve ends (neuritic plaques) were found in the affected woman's brain(Alzheimer, 1907).Blessed et al. (1968) first recognized that all individuals with AD hadcharacteristic histopathologic changes. Further study showed that thesehistopathologic changes were the result of the abnormal deposition of intracellularcytoskeletal filaments in neurofibrillary tangles, and the accumulation of extracellulardeposits of B-amyloid protein collecting in the form of plaques in the hippocampus,cortex, and temporal lobe areas of the brain (Glenner and Wong, 1984; Terry andKatzman, 1983). The major component of these senile plaques is a 39-43-amino-acidpeptide (BA4) which is part of a larger molecule encoded by the amyloid-precursor-1protein (APP) gene mapped to the long-arm of chromosome 21 (Goldgaber et al.,1987; Kang et al., 1987; Robakis et al., 1987; Tanzi et al., 1987a).1.2 DiagnosisDementia is now recognized to result from several different causes includingvascular disease, multiple infarcts (strokes) as well as AD (Table 1.1). Dementia isdefined in the Diagnostic and Statistical Manual of Mental Disorders, third edition,revised (DSM-III-R) (American Psychiatric Association, 1987) as "the impairment ofcognition severe enough to interfere with daily functioning". This impairment canaffect different areas of cognition such as short-term memory (e.g. the ability to learnnew information); long term memory (e.g. the ability to remember birthplace);abstract thinking (ability to understand conceptual information); judgement (e.g.ability to make rational plans); aphasia (language disorder); apraxia (impaired motoractivity); agnosia (impaired recognition of objects despite having intact sensoryfunction); constructional ability (e.g. ability to draw a 3-dimensional object); andpersonality The National Institute of Neurological and Communicative Disordersand Strokes (NINCDS) and the Alzheimer Disease and Related DisordersAssociation (ADRDA) Work Group Criteria (McKhann et al., 1984) considerbehaviour in addition to cognitive impairment in the definition of AD. NINCDS-ADRDA criteria define dementia as a decline of memory and other cognitiveabilities from a previous level of functioning. This definition allows the inclusion of2individuals with both high and low levels of intellectual functioning, including thosewith mental handicap (e.g. Down syndrome).NINCDS-ADRDA diagnostic criteria (McKhan et al., 1984) for "definite" AD,"probable" AD, and "possible" AD are as follows:i) 'Definite" AD: This diagnosis can only be made after clinical criteria for"probable" AD are met (see below) and histopathologic material from biopsy orautopsy is examined for the classic neuropathologic signs. Large numbers ofneuritic plaques and neurofibrillary tangles in the hippocampus and temporallobes would be definitive signs of AD, however the accumulation of a fewplaques and tangles is within the range of normal aging.ii) "Probable" AD: This is the most definitive clinical diagnosis and criteria includedementia established by clinical examination with an insidious onset, aprogressive decline in two or more areas of cognition including memory and theabsence of all other conditions which could impact an individuals cognitivefunctioning (see Table 1.1).iii) "Possible" AD: This diagnosis requires that an individual meet the clinical criteriafor AD (i.e. "probable" AD) which is confounded by a concurrent conditionpossibly affecting cognitive functioning.Clinically, the disease is best characterized by an insidious onset, a progressivecourse and deterioration in two or more areas of cognition. AD generally has itsonset in the sixth or seventh decade of life, but individuals can become affected asearly as the third or fourth decade. Based on clinical examinations, these NINCDS-3ADRDA diagnostic criteria provide an inter-clinician reliability of approximately 95%(Forette et al., 1989). Studies using strict diagnostic criteria find good to excellentagreement between clinical and histological diagnoses (70%-100%) (Burns et al.,1990; Jellinger, 1990; Tierney et al., 1988). Adherence to strict diagnostic criteriaresults in an acceptable level of reliability for research purposes.Table 1.1 Major Causes of Dementia.1. Cerebral Neuronal DegenerationsAlzheimer DiseasePick DiseaseParkinson DiseaseHuntington DiseaseProgressive Supranuclear Palsy2. Acquired Cerebral Disorders (some potentially reversible)Vascular Dementia: Multi-Infarct Dementia, Binswanger DiseaseMultiple SclerosisIntracranial NeoplasmsTrauma (including Subdural Hematoma)HydrocephalusTransmissible Spongiform Encephalopathies (e.g. Creutzfeld-Jakob Disease)3. Other Potentially Reversible DisordersMetabolic Disorders: Hypothyroidism, Renal DialysisToxic/Nutritional Disorders: e.g. Chronic Drug Intoxication, Alcoholism,Malnutrition (e.g. vitamin B12 deficiency)Infections: e.g. HIV, neurosyphillis, tuberculous or bacterial meningitis,cryptococcosis, acute viral encephalitisMajor Depression'This is not meant to be a comprehensive list, but is presented to illustrate theetiologic heterogeneity of dementia.Revised from Morris and Rubin, 1991.4Clinical (phenotypic) heterogeneity exists in AD. The presentation of thecognitive syndrome can vary and include any or all of the following: amnesia;aphasia; apraxia; agnosia. A wide variation in ages-of-onset occurs both withinmulticase families and among "sporadic" (non-familial) cases (Bird et al., 1983).Historically, AD has been separated into early-onset (onset before age 65), and late-onset (onset at or after age 65) subgroups. However, other than age-of-onset, nophenotype has been definitively recognized to differentiate early and late-onset formsof AD. Variation in myoclonus, extrapyramidal signs, brain cholinergic activity, andneuropathological changes are also evidence of diverse clinical expression of thedisease (Chui et al., 1985; Freidland et al., 1988; Mayeux et al., 1985).In summary, it has not yet been possible to correlate the wide variation inclinical phenotype with subgroups of AD (eg. early-onset versus late-onset; familialversus sporadic).1.3 Etiological Heterogeneity1.31 Risk Factors1.31.1 BackgroundHeterogeneity in AD is not only restricted to the presentation of clinicalsymptoms. The etiology of AD is also now believed to be heterogeneous. A widerange of risk factors have been reported for AD. These include both genetic andenvironmental factors as well as interactions between the two. The most commonlyreported risk factors are listed in Table 1.2. The singular most definitive risk factor5appears to be age. Jorm et al. (1987) reviewed 22 studies and found that theprevalence of AD doubles approximately every 4.5 years until age 90. It is unclearwhether the trend for increasing risk continues indefinitely with age. One studyreported that the incidence of dementia actually decreased after age 90 in an isolatedSwedish population (Hagnell et al., 1981). To date no age-related molecular levelchanges have been identified to explain the neuropathological changes seen in ADbrains (reviewed by McLachlan et al., 1991).Other than age, a family history of dementia is the only other widely recognizeddefinitive risk factor. The elevated risk to first-degree and second-degree relatives ofAD probands provides evidence for a genetic contribution to the disease etiology.Further evidence of a genetic component to the etiology of AD is provided by apossible increased risk to develop AD for individuals with a family history of Downsyndrome. Both familial risks and the association between Down syndrome and ADwill be discussed in detail later (1.32, 1.35).1.31.2 History of Head TraumaHead trauma has been reported as a risk factor for AD, but not all studiesagree on the significance of a history of head injury. Head trauma may furtherdeplete an individuals' natural age-related decline in the number of neurons. Thisloss of neurons may result in the decrease of brain function below the level requiredfor normal activity (reviewed by McLachlan et al., 1991). A head injury may alsodamage the blood-brain barrier, resulting in the brain being exposed to environmental6Table 1.2: Commonly Reported Risk Factors for Alzheimer DiseaseFactor^ ReferenceAge Reviewed by Jorm et al., 1987Family History of Dementia^Breitner and Folstein, 1984; Broe et al.,1990; Chandra et al., 1987; Graves et al.,1990; Li et al., 1992; Mendez et al.,1992; Mortimer, 1983Parental Age^ Urakami et al., 1989; Rocca et al., 1991Family History of Down's Syndrome^Broe et al., 1990; Heston et al., 1981;Heyman et al., 1983;Family History of Parkinsonsim^Heyman et al., 1983; Hofman et al., 1989Thyroid Disease^ Ewins et al., 1991; Heyman et al., 1984;Li et al., 1992; Mortimer 1989Depression^ French et al., 1985; Shalat et al., 1987;Head TraumaAluminum ExposureAmaducci et al., 1986; Heyman et al.,1984; Mortimer et al., 1985; Sullivan etal., 1987Freed and Kandel, 1988; Hein et al.,1990Henderson, 1988; Jorm et al., 1987;Rocca et al., 1986Flaten, 1987; Leventhal, 1986; Martyn etal., 1989; Still and Kelley, 1980; Vogt1986Organic Solvent ExposureFemale Gender7toxins and viruses normally present in the body, but from which the brain wouldotherwise have been protected. The destruction of the blood-brain barrier could alsoallow an abnormal interaction between the immune system and the central nervoussystem (Mortimer et al., 1985). Together, these data emphasize the complexinteractions which may be important in the etiology of AD.1.31.3 History of Aluminum ExposureA history of aluminum exposure has been cited as a risk factor for AD.However, this is not a universal observation (Markesbery et al., 1981; 1983;McDermott et al., 1979). An elevated level of aluminum has been found in the coreof senile plaques in the brains of AD patients (Crapper et al., 1973). A study of ADneocortex samples found the level of aluminum in a DNA fraction enriched for genesrepressed in AD patients is nine times that of controls (Crapper et al., 1980;McLachlan et al., 1989). One hypothesis is that the increased electrostatic binding ofaluminum to DNA-associated proteins results in altered chromatin structure and adecrease in gene transcription (reviewed by McLachlan et al., 1991).Another possible mechanism for aluminum's putative effect on a susceptibleindividual's brain is the association between aluminum and stress response proteins.It has been reported that the level of heat-shock protein 17 (a stress responseprotein) is increased with temperature in the hippocampus of adult rats, but not agedrats. It has also been demonstrated that exposing neuroblastoma cells to aluminumcan induce the production of heat shock proteins (reviewed by Lake et al., 1991).8Another interesting link between these pieces of evidence is in a recent report bySchellenberg et al. (1992) who reported a candidate gene for AD found in a subset ofearly-onset multicase families to be HSPA2 (a 70-kd heat shock protein) located onchromosome 14. It is hypothesized that HSPA2 could potentially be involved inprotein assembly and degradation, thus affecting the accumulation of 13-amyloidprotein in the brain. Further investigation is needed to determine the role of thesemechanisms in AD neurodegeneration. However, these observations do serve toillustrate that complex interactions may play a part in the disease etiology.1.31.4 History of Thyroid DiseaseA history of thyroid disease has also been reported as a risk factor for AD(Ewins et al., 1991; Heyman et al., 1984; Li et al., 1992; Mortimer 1989), but thisassociation again has not been confirmed by all studies (Ammaducci et al., 1986;Henderson et al., 1986; Rocca et al., 1986). Several lines of evidence suggest thatthyroid hormone which has as essential role in brain function, may play a part in thedeterioration of cognitive functioning. Adults deficient in thyroid hormone exhibitslowing of intellectual function, loss of initiative, confusion, disorientation, memoryloss, slowing of speech, and lethargy (Jellinek, 1962; Karnosh and Stout, 1935). Oneform of thyroid hormone receptor (T3 receptor) is normally found in highconcentration in the cortex and hippocampal regions of the brain (Dussault and Ruel,1987). These same regions are deteriorated in AD patients. It is therefore9hypothesized that altered or decreased response to thyroid hormone may be involvedin the etiology of AD (reviewed by McLachlan et al., 1991).Antibodies to thyroglobulin (an inactive supply form of thyroid hormone) havebeen found in the cerebrospinal fluid of AD patients (McRae-Degueruce et al.,1988). These antibodies may further decrease thyroid hormone levels, aggravatingthe age-related deterioration in cognitive functioning associated with AD (reviewedby McLachlan et al., 1991; Reinisch et al., 1991).1.31.5 Female GenderA female preponderance (up to twice the rate for males) has been reported inAD (Bachman et al., 1992). This sex-specific difference in risk may be related todifferences in the expression of male and female hormones (androgens andestrogens). Several morphologic sex differences in the human brain have beenidentified. These include the size of the corpus callosum and hypothalamus as wellas hemispheric brain organization (reviewed by Reinisch et al., 1991).Studies on the central nervous systems of other mammals suggest that sexdifferences in brain structure and function are direct results of the earlyorganizational influence of sex hormones. Areas affected include the nuclei of thehippocampus, cerebellum, amygdala and cerebral cortex. Differences in themorphologic dimensions of the brain including synaptic connections, dendritic fieldpatterns and the size of brain nuclei have all been reported (reviewed by Reinisch etal., 1991). Hormones also exert an effect on the brain at a cellular level, impacting10cell number and size. Androgens (male hormones) may increase the survival ofparticular spinal cord cells early in human development (Toran-Allerand, 1986).These hormonal effects on the brain are not limited to the prenatal period, and cancontinue into adulthood (Greenough, 1986).Biochemical and physiological functions such as hormone regulation, glucosemetabolism and neurotransmitter function (including cholinesterase activity) areaffected by androgens and estrogens (reviewed by Reinisch et al., 1991). Thedeficiency of the neurotransmitter choline acetyltransferase has been reported in ADpatients (reviewed by Beattie, 1992) . Several studies provide evidence that adepressed level of choline acetyltransferase is a central factor in the disease process(reviewed by Beattie, 1992). These hormones are also known to control theexcitability of brain cells by modulating the action of neurotransmitters. Increasedcholinergic and catecholinergic activity are noted in males while serotinergic activityis elevated in females (reviewed by Reinisch et al., 1991).It is known that raised estrogen and progesterone levels correlate with epilepticseizures. Further, postmenopausal women with decreased estrogen levels suffermood changes which can be alleviated by estrogen supplements, suggesting thatnaturally occurring hormone level fluctuations are sufficient to affect neuronalactivity. Decreased estrogen levels may remove the protective effect this hormone isthought to exert on the brain (reviewed by McLachlan et al., 1991). Raisedglutamate levels in the brain cause overexcitation of the neurons and eventually leadto cell death. Estrogen regulates glutametergic activity, but whether the decreased11level of estrogen found in post menopausal women is adequate to sufficiently raiseglutamate levels to cause this overexcitation and disruption of neuronal activity is, asyet, unknown. Altered glutamate activity in the cerebral cortex and hippocampalregions may be associated with impaired short term memory usually found in ADpatients.1.31.6 Other Risk FactorsOther risk factors which have been reported include consanguineous marriages(Lowenberg and Wagoner, 1934), and ethnic and racial differences (Bird et al., 1988;Goudsmit et al., 1981, Hendirie et al., 1989). A recent case-control study done on aChinese population found both left handedness/ambidexterity and a family history ofpsychotic disorders as well as previously more commonly reported factors increase anindividuals risk for AD (Li et al., 1992). Advanced maternal age has been widelyreported as a risk factor (Ammaducci et al., 1986; Rocca et al., 1991; Urakami et al.,1989), and provides one link for a possible association between AD and Downsyndrome (see section 1.35). Another risk factor providing additional evidence of alink between AD and autoimmune thyroid disease is a high incidence of autoimmunethyroid disease among affected relatives of FAD families (Ewins et al., 1991). Ahigher incidence of autoimmune thyroid disease in individuals suffering from Downsyndrome and also been previously reported (Burgio et al., 1966; Lobo et al, 1980).As it is known that the underlying cause of Down syndrome is trisomy 21, theseresults lend further support to the possible genetic contribution to both autoimmune12thyroid disease and AD, with the possibility that the latter diseases are linked tochromosome 21 (see sections 1.31.4, 1.35). The large number of possible risk factorshas lead some researchers to postulate polygenic interaction to explain clustering ofAD in some families (Sjogren et al., 1952; Sulkava et al., 1979; Whalley et al., 1982).1.32 FADA portion of individuals with AD are from families whose disease aggregationof AD suggests an autosomal dominant mode of transmission. Estimates of theproportion of the AD population showing this type of transmission range from 5%(Appel, 1981) to 100% (Breitner et al., 1988) (see Section 1.4). Such families arereferred to as "Familial Alzheimer Disease" (FAD) (Feldman et al., 1963; St. GeorgeHyslop et al., 1989). It has been hypothesized that an autosomal dominant gene(s) isresponsible for all cases of the disease, but potentially "affected" individuals die fromcompeting causes before the age-of-onset for AD. This results in a lower percentageof observed cases within families than predicted by the autosomal dominant model(ie. 50%). To date, phenotypic heterogeneity has not been shown to exist betweenFAD and non-FAD cases (Haupt et al., 1992; Swearer et al., 1992). This wasinterpreted as further evidence that "sporadic" cases of AD actually represent reducedpenetrance of a single autosomal dominant gene with non-expression being due to"unaffected" family members not living long enough to exhibit symptoms (St. GeorgeHyslop et al., 1989). The difficulty in assessing families affected by a disease with13such a late-onset led one researcher to comment "... perhaps Alzheimer's diseasewould be simpler to understand if we all lived to be 150 years old" (Davies, 1986).1.33 Molecular Genetic InvestigationIt was originally reported in 1987 that DNA markers on the proximal long armof chromosome 21 segregated with AD in four early-onset FAD families (St. George-Hyslop et al., 1987). This led to a series of studies examining genetic linkagebetween a possible defect in the APP gene, already mapped to the same region onchromosome 21 and implicated in AD pathogenesis, and FAD. Further studydetected several obligate crossovers between FAD and APP, suggesting the possibilityof other FAD gene loci (Tanzi et al., 1987b, Van Broeckhoven et al., 1987).Continued study has not supported the hypothesis that all FAD families are the resultof an autosomal dominant segregating mutation linked to the proximal long arm ofchromosome 21 (Goate et al., 1991; St. George-Hyslop et al., 1990; St. George-Hyslop et al., 1987). Although a small subset of early-onset FAD families appear tobe linked to chromosome 21, the majority of early-onset families and all late-onsetfamilies do not support proximal 21q linkage (Pericak-Vance et al., 1988;Schellenberg et al., 1991; St. George-Hyslop et al., 1990). Recently (October 1992) astudy found evidence for an FAD locus on chromosome 14 in a subset of early-onsetFAD families (Schellenberg et al., 1992). This finding has been confirmed by threelater studies reporting linkage of a subset of early-onset FAD families to the long14arm of chromosome 14 (Mullan et al., 1992; St George-Hyslop et al., 1992; VanBroeckhoven et al., 1992).It has been suggested that FAD is more often found among early-onset ADthan late-onset AD (reviewed by Nalbantoglu et al., 1990). However, several studieshave reported linkage of a late-onset subset of FAD families to the proximal longarm of chromosome 19 (Pericak-Vance et al., 1991a, 1991b; Roses et al., 1990;Schellenberg et al., 1992). These data provide evidence that the expression of FADis not restricted to individuals exhibiting an early-onset form of the disease. Theyalso further support nonallelic genetic heterogeneity in AD.Further evidence of genetic heterogeneity is provided by investigations ofseveral different APP mutations which have been reported to segregate with FAD(Chartier-Harlin et al., 1991; Goate et al., 1991; Karlinsky et al., 1992; Murell et al.,1991). The 13-amyloid sequence is coded by parts of exons 16 and 17 of the APPgene. These mutations reside within codon 717 in exon 17 and involve changingvaline to isoleucine, phenylalanine or glycine. Investigations have also revealedfurther mutations in the APP gene outside codon 717 apparently resulting in the ADphenotype (Hendriks et al., 1992; Mullan et al., 1992). One involves a substitution ofglycine for the native alanine at codon 692, which is also coded for in exon 17 of theAPP gene. The other is a double mutation at codon 670 and 671 coded for in exon16 of the APP gene. A more recent investigation reported a double mutation atcodon 715 and 713 in one apparently sporadic case of AD and in four of her six15unaffected sibs (two of which have passed the probands age-of-onset) as well as anunaffected aunt (Carter et al., 1992).This evidence suggests that only approximately 3% of FAD cases are a result ofmutations in the APP gene (Chartier-Harlin et al., 1991; Schellenberg et al., 1991;Van Duijn et al., 1992), and that additional factors (environmental or genetic) may benecessary for manifestation of the AD phenotype by the end of the human lifespan.In conclusion, the results indicating that no sporadic and only an extremelysmall proportion of FAD cases segregate with these known mutations strongly pointto heterogeneity in the etiology of AD.1.34 Twin StudiesThe study of twins lends additional evidence to the supposed etiologicheterogeneity of AD. If AD were purely a genetic disease, concordance rates amongmonozygotic twins (genetically identical individuals) would approach 100%. Further,if AD were a single gene trait, concordance rates for dizygotic twins (individuals whoshare 1/2 of their genes, as do non-twin siblings) would approach 50%.A concordance rate of 42% for monozygotic twins and 8% for dizygotic twinswas reported by Kallman et al. (1956). A follow up study found a concordance ratefor monozygotic twins to be approximately 50% (Jarvik et al., 1971). However boththese studies suffered from design problems such as insufficient follow-up time for adisease known to be characterized by widely variable ages-of-onset, as well as theabsence of strict diagnostic criteria (these studies were prior to the NINCDS-16ADRDA criteria). A more recent study found concordance rates of approximately40% for both monozygotic and dizygotic twins (Nee et al., 1987). To date, casereports of 40 twin pairs (either one or both individuals affected) have been published.Sixteen of 35 monozygotic twin pairs were concordant; 19 monozygotic twin pairswere discordant, even after follow-up of 20 years in some cases (Hunter et al., 1971;Karlinsky, 1993; Kumar et al., 1991; Nee et al., 1987; Renvoize et al., 1986). The fivereported dizygotic twin pairs are all concordant (reviewed by Karlinsky, 1993).These twin data represent many of the biases in twin studies that are notpopulation based. These include over-ascertainment of monozygotic pairs (ie. 87% oftwins reported were monozygotic, whereas in the North American and Europeanpopulations from which these cases were drawn, only approximately 30% of twins aremonozygotic) (Ebers et al., 1986; McFarland, 1993; Vogel and Motulsky, 1979). Non-population based series are also biased towards concordant pairs as seen by the factthat 100% of the published dizygotic twin pairs are concordant.These studies in AD clearly demonstrate that monozygotic twin pairs are not100% concordant and concordant pairs can have widely different ages-of-onset. Ithas also been reported that an increased familial rate for AD among first-degreerelatives of concordant monozygotic twin pairs compared to first-degree relatives ofdiscordant monozygotic twin pairs exists, lending further evidence that a singledisease etiology is unlikely (Rappaport, 1991).171.35 Association Between AD and Down SyndromeThe putated association between Down syndrome (DS) and AD further supportsthe genetic contribution to AD etiology. DS results from the duplication of a sectionof chromosome 21 containing a specific portion of the chromosomal material, usuallythrough meiotic nondisjunction. Interestingly, chromosome 21 is the samechromosome to which the FAD linked APP gene mutations have been mapped. Theresulting DS phenotype is presumed to be a result of an overproduction of the geneproducts coded by the duplicated genes (reviewed by Potter, 1991).All individuals affected by DS show the neurohistologic signs of AD by age 40and a proportion also exhibit clinical signs of dementia (reviewed by Breitner andFolstein, 1984; McLachlan et al., 1991). It was this association that led to the originalinvestigations linking FAD and chromosome 21 (see section 1.33).DS has been reported to occur in the families of AD probands more often thanexpected (Heston et al., 1981; Heyman et al., 1984), but more recent studies whichstrictly control for maternal age do not confirm this association (Sadovnick et al.,1992).It is important to note that although all DS affected individuals who live longenough have been found to have plaques and tangles characteristic of AD on autopsy,only 15% to 30% (or even 45%) will develop dementia (reviewed by Ball, 1987).These data emphasize the relationship between the neuropathology and dementia, ie.,AD neuropathology is associated with dementia, but may not account for thedementia.18Sanford et al. (1991) reported that a similar defect in DNA repair may causethe neurodegeneration seen in individuals with both AD and DS. Additionalevidence of a link between AD and DS comes from the shared risk factor ofadvanced maternal age (Ammadduci et al., 1986; Rocca et al., 1991; Urakami et al.,1989), although this has not always been confirmed for AD (Hofman et al., 1990,Jouhan-Flahault et al., 1989). The increased risk due to advanced maternal age canbe explained in DS by the dramatic increase of non-disjunction in the aged ova of themother, resulting in a greater proportion of trisomic offspring (reviewed byMcLachlan et al., 1991). Urakami et al. (1989) proposed that the associationbetween AD and DS may be due to similar cytogenetic events occurring in bothdisease processes. Potter (1991) hypothesized that the accumulation of trisomy 21cells developing over time due to unequal segregation during mitosis leads to AD bythe same mechanism that DS individuals acquire the disease.1.4 Overview of Previous Studies Examining Empiric Risk to First-Degree Relativesof AD Cases and Controls Using Similar MethodologiesAge specific risks for AD to first-degree relatives of AD cases and controls canbe calculated into the eighth and ninth decades of life if the study population issufficiently large. Assuming a fully penetrant autosomal dominant model ofinheritance, the risk to first-degree relatives of AD cases should approach 50% by theninth decade of life. Assessment of lifetime risk to first-degree relatives is onemethod to evaluate the contribution of an autosomal dominant gene(s) to the19etiology of AD (Breitner et al., 1988; Martin et al., 1988; Mohs et al., 1987).However, a risk approaching 50% does not unambiguously prove that a singleautosomal dominant gene is responsible for the disease. Molecular genetic studies inAD support etiologic (genetic) heterogeneity.Table 1.3 summarizes studies in the literature which examined the risk to first-degree relatives of AD cases and non-demented, cognitively unimpaired controls,using similar methodologies to those in this study. With the exception of one earlystudy (Brietner and Folstein, 1984), all studies listed in Table 1.3 used NINCDS-ADRDA criteria (or very similar criteria) for the diagnosis of AD cases. Moststudies find a cumulative lifetime risk to first-degree relatives (or parents and sibsonly, as including children in the analysis generally lends nothing to the final estimatesince few have reached the "at risk" onset age) of approximately 50% by the eighth orninth decade of life, as predicted by an autosomal dominant model of inheritance(Breitner and Folstein, 1984; Breitner et al., 1988; Huff et al., 1988; Martin et al.,1988; Mayeux et al., 1991; Mohs et al., 1987). However, two studies did not find this.Farrer et al. (1989), using a method which weighted the accuracy of diagnosis in theaffected relatives, found a lower risk to all first-degree relatives of AD cases. Theyreported a maximum risk of 39 ± 10%, but felt a value weighted for certainty ofdiagnosis, 24%, was probably closer to the true risk. Sadovnick et al. (1989) studied151 AD cases and their relatives and found a cumulative lifetime risk to parents andsibs of AD cases of 23.3 ± 3.8%, a figure very close to the weighted value of 24%reported by Farrer et al. (1989). Although the cumulative lifetime risks differ20between these two studies and the bulk of other analyses, all studies found a sharpincrease in risk with age at approximately age 70.Cumulative lifetime risks have been compared for different subgroups of first-degree relatives, but to date no significant differences have been found. The twomajor studies which examined the risk to female and male first-degree relatives ofAD cases found no significant differences (Brietner et al., 1988; Farrer et al., 1989),however Breitner et al. (1988) suggested that the "tendency" for female first-degreerelatives of AD cases to develop the disease at an earlier age was "notable". Severalstudies examined cumulative lifetime risk to parents and sibs of AD cases andreported no significant differences (Brietner et al., 1984; Breitner et al., 1988; Farreret al., 1989). Finally, no significant differences in cumulative lifetime risk betweenfirst-degree relatives of early-onset and late-onset AD cases have been reported,although relatives of early-onset cases reportedly have a non-significant tendency todevelop the disease at an earlier age (Breitner et al., 1988; Farrer at al., 1989; Huffet al., 1988;).The cumulative lifetime risk estimates to first-degree relatives of non-demented,cognitively unimpaired controls were found to be significantly lower than those tofirst-degree relatives of AD cases in all studies, however these estimates vary widely(8.07 ± 5.97% to 23 ± 10.8%). An important factor in weighing the significance ofthese results is the number of first-degree relatives surviving to the last ages of life.At these late ages (age 80 to 90 years old), the standard error of the estimated risktends to become quite substantial. Studies investigating significantly larger samples21Table 1.3: Overview of Previously ReportedRisk Estimates Using Similar MethodologiesCasesReference Basis ofDiagnosisNo. Subgroup RiskEstimate (%)Breitner and Folstein, Clinical Exam 39 Siblings and 55.3 ± 15'1984 (agraphia*,aphasia*)Children 57.6 ± 22*Mohs et al, 1987 NINCDS- 50 First-Degree 45.9 ± 9ADRDA RelativesBreitner et al, 1988 NINCDS- 79 First-Degree 49.3 ± 8ADRDA RelativesParents 43 ± 8Siblings 61.0 ± 24Males 34 ± 13Females 54 ± 1027 Relatives of 56 ± 14Early-OnsetCases52 Relatives of 38 ± 9Late-OnsetCasesHuff et al, 1988 NINCDS- 50 Parents/Sibs 45 ± 11ADRDA Relatives of 49 ± 16Early-OnsetCasesRelatives of 40 ± 14Late-OnsetCasescontinued22CasesReference Basis ofDiagnosisNo. Subgroup RiskEstimate (%)Martin et al, 1988 NINCDS- 22 First-Degree 40.8 ± 9ADRDA RelativesFarrer et al, 1989w NINCDS- 114 First-Degree 24 to 39 ± 10ADRDA RelativesParents 21 to 35 ± 6Sibs 35 to 59 ± 29Females 35 to 56 ± 14Males 12 to 18 ± 457 Relatives of 45 ± 15Early-OnsetCases57 Relatives of 37 ± 14Late-OnsetCasesSadovnick et al, 1989 NINCDS- 151 Parents/Sibs 23.3 ± 3ADRDAMayeux et al, 1991 NINCDS- 110 First-Degree 48.7 ± ?'ADRDA RelativesLRisk estimates calculated by a weighting method. Estimates represent weighted riskestimates and maximum risk estimates.Standard error not reported.continued23ControlsReference^Composition^No.^Subgroup^RiskEstimate (%)Breitner and^Non-AD^33^First-^8.1 ± 6Folstein, 1984 Institut- Degreeionalized^RelativesResidentMohs et al, 1987^Spouse/^45^First-Degree^12.1 ± 7Volunteer RelativesBreitner et al,^Spouse/^61^First-Degree^9.8 ± 61988^Volunteer RelativesHuff et al, 1988^Spouse^47^First-Degree^11 ± 4RelativesMartin et al,^Spouse^24^First-Degree^23 ± 101988 RelativesMayeux et al,^Volunteer^59^First-^20 ± r1991^ DegreeRelativesI/ Risk estimates calculated by a weighting method. Estimates represent weightedrisk estimates and maximum risk estimates.Standard error not reported.will facilitate more accurate risk estimation.241.5 Canadian Study on Health and AgingIn response to advice from experts in various disciplines, Health and WelfareCanada provided funding for a country-wide collaborative study on AD in theCanadian population. The objectives of this study were: (i) to estimate theprevalence of dementia among elderly Canadians; (ii) to determine risk factors forAD; (iii) to determine the weight of responsibility and the need for support bycaregivers of demented patients; (iv) to establish a database for future studies.The study of the prevalence of AD in the Canadian population (age ? 65 years),provided the unique opportunity to screen a large sample of elderly individuals livingin British Columbia for cognitive decline. This screening process identified elderlycognitively unimpaired individuals; "control" index cases for the study of familial risksfor AD.The Canadian Study for Health and Aging (CSHA) was designed to screen arepresentative sample from the population aged > 65, including those living in thecommunity ("community sample") and in care facilities ("institutionalizedpopulation"). In British Columbia (B.C.), the sampling frame for the "communitysample" was based on the B.C. Provincial Health Care Registry compiled through theBritish Columbia Provincial Health Care Plan. The sample was divided into threeage group strata (65-74; 75-84; 85 and above), and weighted using the optimalallocation method (reviewed by Armitage, 1971). The following procedure to selectthe sample was in accordance with guidelines established for the entire country:25(A) Community Sample1. Select individuals age 65 and over as of October 1, 1990.2. Exclude individuals living in institutions.3. Use postal codes to select individuals living in designated study areas.4. Create sequential sampling frames alphabetically for each CentralMetropolitan area and Urban Area using postal codes.5. Sort the sample into the Central Metropolitan Areas and Urban Areas usingpostal codes.6. Sort the above samples into three age strata: 65-74, 75-84, 85 + .7. Randomly select the number (2N) subjects required for each stratum. Byselecting twice the required number per stratum, replacement cases areavailable.8. Within each stratum, the odd numbers constitute the sample, and the evennumbers, the replacement sample.(B) Institutionalized Sample1. Use the Central Registry of Continuing Care Facilities to obtain a list of allinstitutions within Vancouver, Victoria, Kelowna, Kamloops, Matsqui, Mission,Chilliwack and Nanaimo. Geographic location was identified by postal codes.2. Obtain a list of all individuals age >65.3. Sort the list into three strata based on size (maximum number of beds)of institution: small (4-25 beds), medium (26-100 beds) and large ( > 100 beds).4. From each stratum collect a random sample of approximately 200 (to allow for26a final sample of 84 individuals per stratum with replacements).1.6 The Modified Mini -Mental State ExamThe Modified Mini-Mental State Exam (3MS) (Teng and Chui, 1987) wasselected as the screening procedure for the CSHA. The 3MS has been widely usedfor several years. Its administration is more standardized and the scoring system ismore refined compared to the original Mini-Mental State Exam (MMSE) on whichthe 3MS is based. The 3MS assesses a subject's cognitive functioning, examininghis/her orientation to time and place, instant recall, short term memory, andarithmetic ability (Teng and Chui, 1987). The test is scored by summing the pointsassigned to each task. The test-retest reliability of the original MMSE is excellent,with most studies finding 24 hour retest scores of approximately 0.85 to 0.90(Anthony et al., 1982; Folstein et al., 1983; Uhlman et al., 1987) which suggests themore refined 3MS has adequate reliability for research purposes.The 3MS has a sensitivity of 0.94; superior to the score of 0.88 reported for theMMSE. A cut-off point of 80 out of a possible 100 on the 3MS is roughly equivalentto a cut-off point of 24 out of a possible 30 on the MMSE. A score of 23 or less onthe MMSE is believed to be an indication of some form of cognitive impairment(Kay et al, 1985). Scoring 80 or above on the 3MS should therefore identify aconservative number of individuals free of cognitive loss or dementia.271.7 Clinic for Alzheimer Disease and Related Disorders, University Hospital-UBCSiteThe Clinic for Alzheimer Disease and Related Disorders, University Hospital -UBC Site ("Alzheimer Clinic") serves the entire population of British Columbia. Theclinic's main roles are: (i) to provide a diagnosis (often a second opinion) forindividuals referred with memory or other cognitive impairment; (ii) to counsel familymembers of patients with respect to prognosis for the patient, and familial risk; (iii)to conduct research on AD and other dementing illnesses. The B.C. Health CarePlan gives all patients essentially equal financial access to the clinic's services, thusavoiding many socioeconomic biases which can exist in other countries.Each patient is assessed by the multidisciplinary Clinic Team, composed of aninternist/geriatrician, a neurologist, a neuropsychologist, a social worker, a geneticist,and a language pathologist. Each clinic patient also has an extensive laboratoryscreen. Diagnoses are assigned according to NINCDS-ADRDA Criteria (McKahn etal., 1984). Patients not meeting the criteria for a diagnosis of dementia arereassessed at regular intervals. Longitudinal follow-up of patients with dementia isdone as appropriate.A wide battery of tests are used to assess each patient, including acomprehensive array of blood tests and routine diagnostic procedures such as a CThead scan, chest x-ray, and electrocardiogram. Detailed neuropsychological tests arealso given. These include: (i) the Multi-focus Assessment Scale which measuressocial behaviour, auditory and visual receptive language skills, expressive language28skills, orientation, mood and accessibility for testing (Coval et al., 1985); (ii) theWechsler Adult Intelligence Scale (WAIS) Information, Digit Span, Similarities andBlock Design subscales; age corrected scores (Wechsler, 1955); (iii) the ControlledOral Word Association Test or Word Fluency test which measures the ability togenerate words in response to stimulus letters (Benton 1968); (iv) Finger TappingTest and Dynamometer, measures of fine motor speed and grip strength; (v) Item 227(Visual Recall) from the Luria Nebraska Neuropsychological Battery, modified toinclude copy and delayed recall components (Golden et al., 1984); and the CuedRecall procedure for memory assessment (Buschke, 1984).The Alzheimer Clinic provides an excellent resource for family studies in AD.As family histories are routinely collected for all consecutive unrelated patients(cases), many biases inherent in family studies where cases are ascertained by thesolicitation of volunteers are avoided. Over the years a number of families have beenreferred to the Alzheimer Clinic specifically because of their family history. Thesecases have been identified and are not included in this population based database.1.8 Relevance of Study1.81 Magnitude of AD on the Canadian Population and Health Care SystemThe demographic change in North America is towards an increasingly agedpopulation. In 1988, it was estimated that health care costs for institutionalized ADpatients in the United States exceeded $50 billion (Yankner and Mesulam, 1991). Inaddition to these direct monetary costs, the impact on family caregivers must also be29considered. The impact of the "baby boom" generation approaching age 65 could bea health care problem of serious consequence. In 1900 approximately 5% of theCanadian population was over the age of 65. It is estimated that approximately 12%of the Canadian population will be aged 65 and over by the year 2000. It is presentlyestimated that at least 250,000 Canadians suffer from dementia, with a projectedannual incidence of 25,000 cases (Feher, 1992).1.82 Familial RisksRelatives of individuals with AD are becoming increasingly concerned abouttheir own risk to develop the disease. Given the extremely small proportion ofindividuals for whom autosomal dominantly segregating genetic mutations have beenshown to be the underlying cause of the disease, the exact etiologic role of geneticfactors must still be resolved (see Section 1.3). At present, the most appropriatemethod of counselling concerned relatives of AD individuals with respect to theirown risks to develop AD is to provide the most up to date empiric age-specific riskdata. This study is the largest of its kind to date, is relatively unbiased in case andcontrol ascertainment, and will provide the most representative risk data for the B.C.population.Investigations into the epidemiology of AD may lead to the clarification of therelative roles of "nature" and "nurture". The calculation of age-corrected familialrisks to first-degree relatives of AD patients is one way to evaluate the geneticcontribution to the etiology of the disease. The goal of these investigations is to gain30sufficient insight into this debilitating and costly disease to enable treatment andprevention.312 MATERIALS AND METHODS2.1 CasesData on cases were collected from all consecutive unrelated persons (patients)attending the Alzheimer Clinic during the period August, 1985 to March, 1992.NINCDS-ADRDA criteria were used to assign a diagnosis of the likelihood of AD.As described previously (section 1.2), a diagnosis of "definite" AD can only be madeafter examination of histopathologic material (eg. at autopsy or in rare instances,biopsy). The most definitive clinical diagnosis is "probable" AD. The importance ofneuropathological examination is discussed with clinic patients and their relatives.Many consent to this procedure at the appropriate time.The analyses presented in this thesis include only cases with a diagnosis of"probable" or "autopsy confirmed". Ideally, only "autopsy confirmed" AD casesshould be used. However, this is not feasible as sufficient numbers of cases neededto allow for meaningful analysis would be extremely difficult to obtain.2.2 Relatives of CasesStudies investigating the role of genetic factors in the etiology of a diseaserequire detailed information about each case's relatives. Family history data werecollected using modifications of the "family history" method. At least two, and oftenmore, knowledgable informants were interviewed by the clinic's geneticist whocollected and verified family data. The use of multiple informants has been shown to32greatly reduce errors from underestimation of the number of affected relatives(Andreasen et al., 1977; Silverman et al., 1986).The "family history" method was used over the "family study" method whichrequires all family members to be assessed directly for several reasons. In a lateonset disease such as AD, many of the case's relatives are deceased at the time ofstudy. Given the demographics of the B.C. population, it is rare that family memberslive in geographic proximity. For these reasons, a family study could not be done.The methods used in this study have been proven reliable for the well recognizedgenetic database for the Multiple Sclerosis Clinic at the University Hospital-U.B.C.Site (Ebers et al., 1986; Sadovnick et al., 1992).The "family history" method was also used over the study of health recordsalone because many important and updated details proffered by knowledgeablefamily informants are not noted on health records. Early clinicians were inclined toascribe dementia to undocumented sources such as arteriosclerosis or even failed torecognise dementia as a clinical condition, calling it normal aging.Since AD can affect the cases' ability to accurately recall family history data, thestudy required co-informants (in addition to the case) with knowledge of all aspectsof the cases' life. If the disease was obviously impeding the case's ability to recallcertain aspects of their family history an additional co-informant was sought. Spousesand siblings of cases were preferred co-informants rather than children as the lattermay lack detailed knowledge of older relatives (ie. parents and sibs of case).Identification of a co-informant was not a problem as all patients attending the33Alzheimer Clinic were requested in advance to attend the interview with anindividual knowledgable about the patient's family history.Whenever possible, reportedly affected relatives of cases were assessed in aspecialized dementia clinic. Arrangements were made for neuropathologicalexamination at a later date whenever possible. If the relative was deceased and/orassessment could not be arranged, medical/autopsy records were obtained (withappropriate consent) and reviewed by appropriate members of the Alzheimer ClinicTeam to determine the most likely diagnosis, taking into account current criteria.This study had the advantage that the Canadian Provincial psychiatric facilities, wheremany relatives of cases have been institutionalized keep detailed records indefinitely,and with appropriate consent, release these to the Alzheimer ClinicTo increase the reliability of the "family history" method, a standardized "FamilyHistory Questionnaire" originally devised by Silverman et al. (1987) was given toinformants in the case of an "affected" relative being unavailable for examination (seecopy in Appendix A). The "Family History Questionnaire" was used in conjunctionwith medical records to determine possible causes for the reported dementia. Theuse of a structured questionnaire allows more uniform assessment of the nature andprogression of symptoms of dementia (Silverman et al., 1987). Questionnaires onreportedly affected relatives were completed by appropriate next of kin for eachrelative, and usually not the co-informant for the case. Our experience has been thatfamily members are very cooperative in aiding these inquiries.34For the purpose of the analyses, a family member was considered "unaffected" ifthere was any possible explanation for the dementia other than AD (see Table 1.1).Relatives for whom no other identifiable cause of dementia could be determined, andwhose dementia was reported to be irreversible and progressive were considered"affected". In the absence of clinical documentation, the Dementia Questionnaireand informant interview(s) were carefully considered in assigning "affected" or"unaffected" status.Offspring of cases were not included in the lifetime risk analysis, although theinformation they proffer can be useful in determining whether a family representsFAD (see section 2.62). Only parents and full sibs (common mother and father)were included in the present analysis.2.3 ControlsThe combined "community" and "institution" sample from the CSHA wasscreened using the 3MS (see section 1.6). Using a high cut-off point of 80, aconservative number of elderly subjects (age > 65) showing no signs of cognitive lossand/or dementia were identified. A proportion of these individuals were thenreferred for a full clinical evaluation at the Alzheimer Clinic.Since controls were collected from the CSHA, they should be relatively free ofbiases inherent by the solicitation of volunteers. The CSHA controls did not includeindividuals showing mild forms of cognitive loss and/or those in the very early stagesof dementia which would be included if a screening process were not used. The35present study only included controls for whom data collection was complete at thetime of analysis for this study (April 1992). Follow-up of CSHA controls and theirrelatives is still in progress.2.4 Relatives of ControlsControl family history data were collected in a similar manner to case familyhistory data, ie. a direct interview using standardized questionnaires, multipleknowledgable informants and followed-up by obtaining medical records whenpossible.2.5 Analysis2.51 OverviewThe family history information collected for studies of this type representslifetime data. In the majority of situations where the risk to develop a disease isbeing evaluated among family aggregates, the genotype of an unaffected individual israrely in question. However, for a late-onset disease such as AD, some data must be"right-censored". For AD the "right-censored" data represent those individuals whoare potentially susceptible but who either die before their predicted age-of-onset, orare studied before their age-of-onset. In other words, being unaffected does notnecessarily reflect the genotype, but rather the timing of chance events, ie. time ofstudy or age-of-death. The data cannot be properly analyzed without taking intoaccount these "right-censored" individuals.36One method that has been widely used to calculate lifetime risk for diseaseswhere the disease onset is relatively late in life is the Kaplan-Meier method. Thismethod uses information from all individuals to estimate age-specific risks (Kaplanand Meier, 1958).At the time the family histories were collected, some susceptible individuals(first-degree relatives) will already exhibit symptoms of AD, allowing the true riskfunction to be calculated. Individuals who represent "right-censored" data are alsoessential for the calculation of this risk function as they contribute to the age-specificrisk estimates by representing the decreasing numbers of individuals "at risk" withincreasing age. The risk of an unaffected, susceptible individual becoming affectedrepresents the conditional probability for an individual to succumb to the disease at aparticular age. A chain multiplication of conditional probabilities is necessary todetermine the true risk, due to the decreasing number of "at-risk" individuals withincreasing age. For example, under an autosomal dominant model of inheritance, a50% risk to first-degree relatives of AD cases would not be expected to beconsistently observed due to either deaths from competing causes prior to diseaseonset or study being "too early" for disease manifestation in some of these "at-risk"individuals.The Kaplan-Meier method has several advantages for this study. The first isthat it is a non-parametric risk estimation method, thus no assumption is made aboutthe age-of-onset distribution among affected relatives. It is known that for AD, intra-familial ages-of-onset tend to be more similar than inter-familial ages-of-onset (Bird37et al., 1989). However, making the assumption that the age-of-onset distributions ofall cases, all affected relatives of cases and all affected relatives of controls areidentical may be incorrect if factors other than a simple autosomal dominant gene(s)play a role in disease etiology as is suggested by the evidence for heterogeneity in theetiology of AD. Another advantage of the Kaplan-Meier method is that it allows fora direct comparison of lifetime risk estimates with the lifetime risk estimatescalculated by previous studies using similar/identical methods (Table 1.3).The major disadvantage of utilizing the Kaplan-Meier method results from theuse of a database constrained by the imprecision of assigning an age-of-onset for AD,a common problem with this disease because of the insidious, gradual onset. Age-of-onset is usually assigned retrospectively, after the repeated review of past medicalrecords and questioning of knowledgable informants, as is the practice at theAlzheimer Clinic (see Section 2.2). When this proves impossible, a conservativemethod for estimating age-of-onset is to consider the individual unaffected to thelatest possible age. This method has been successfully utilized previously (Sadovnicket al., 1989). Possible implications of this approximation are considered later in thisthesis (see Section 3.8).The Kaplan-Meier method of risk estimation should yield meaningful riskestimates based on the assumption that the sample analyzed contains an affectedindividual with an age-of-onset at the limit of the age-of-onset distribution (ie. thesample contains an individual becoming affected at the oldest age possible. The riskcalculated for this oldest age-of-onset represents cumulative lifetime risk). The large38size of the sample in this current study minimizes the likelihood of violating thisassumption.2.52 Computational FormulasThe SAS statistical package (SAS, 1985) was utilized for data analyses. Theage-specific risk estimate can be defined in the following terms: the data set containsobservations on N relatives with M ordered, distinct onset times denoted as(t1 < t2 < ... < tm). The possibility of more than one onset at each t m is allowed with dmrepresenting the number of onsets at t m. There are also a number of censoring times,em, for individuals who do not display symptoms at the time of study or have diedbefore their age-of-onset. Since there may be some censored observations in [0, t 1 ],the total sample size is N = e 0 + E m=1 (dm + em). The total number of relatives at riskat tm is nm^(di + ei). The product-limit survivor function, or empirical survivorfunction, S(t), is therefore defined as:S(tm) = T17 -n1 -d1-njwhere ni-di/ni represents the fraction of those relatives at risk in each interval who donot become affected. If a censoring time em, and an onset time tm are recorded asequal, the censoring times are adjusted an infinitesimal amount to the right so that e mis considered to be infinitesimally larger than t m. In other words, any relative with acensoring age equal to t m is included in the set of n m relatives at risk at tm, as are39relatives who become affected at t m . The empirical cumulative distribution function,F(t), is then F(t)=1-S(t).The estimated variance a 2 of S(tm) can be calculated using Greenwood'sformula (Kaplan and Meier, 1958), defined as:_ 2 = t-.2tt^d.am^j=1Differences between independently estimated lifetime risk curves will beassessed for significance using the Mantel-Haenszel test, called a "log-rank" test in thesurvival context. This test has been used in several previous studies, and was selectedfor this study primarily because it places more weight on older onset ages. The log-rank test can be defined as follows. Assume two risk curves are defined for subgroup0 and subgroup 1. Let (t 1 <^< tm) be the distinct, ordered ages-of-onset for thetwo subgroups together. Let dmk be the number of onset times equal to t m in thesubgroup k, k =0, 1, and define emk as the number of observations censored by deathor study occurring before the age-of-onset from sample k in the interval (t m, tm , i )•The number at risk from each subgroup is then n mk^(djk elk). The totalnumber of observations in each subgroup is Nk = eok^(clik eik)}. As well, thequantities dm = dmo + dmi , nm =nmo +nml, and N = N0 + N1 for the combined subgroups aredefined. The observations at each t m can then be displayed in a 2 x 2 contingencytable with expressions representing those individuals at risk from each sample, andwhether they become affected or remain unaffected:40At tmSubgroup Total^Total^Total AtOnsets Unaffecteds Riskdm° nmo dm()dmi Ilm1 drilldm^nm - dmThe significance of an assumed constant odds ratio of relative risk fromindependent 2 x 2 tables can then be assessed. For example, the Oml observednumber of onsets in subgroup 1 at tm is dmi whereas the expected number of onsets,assuming the risk is equal between the two subgroups, is E mi = (nm, x dm/nm) (thenumber at risk from the first sample multiplied by the risk estimated from thecombined sample). The durations in these predictions are then summarized byadding over the tables for the distinct ages-of-onset t m, the difference between theobserved, Oml , and expected, Emi, number of disease onsets to obtain 0 1-E1=-E m=1 (Omi-Emi)• The log rank test statistic is then (0 1 - Ei)2/V, with the variance,V, defined as:V =E m= 1d; (n ;-d ;)(n i)2 (n 1 -1)the log-rank test, under the assumption of equal risk in the two subgroups, isdistributed approximately as Chi-Square with one degree of freedom (df). Note thatthis statistic allows for differences between the lifetime risk curves in either direction(reviewed by Crowley, 1984).01nmoIlm1nm41Differences in the estimated cumulative lifetime incidence were assessed using adifference of proportions test. This difference can be assessed by the followingstatistic:1 132 - Pi I Z— f(a1)2 + (a2)2} 1/2..^.,where p 1 and p 2 are the two sample proportion estimates, and (a i)2 and (a 2)2 arethe estimated sample proportion variances. If Z exceeds the critical value obtainedfrom the standard normal curve for a specified significance level a, the proportions Piand 132 are considered to be unequal (reviewed by Fleiss, 1981).2.6 Risk Comparison2.61 OverviewThe comparison of lifetime risk curves and cumulative lifetime risks for varioussubgroups of first-degree relatives of cases and controls should provide insight intothe genetic etiology of AD. If an age dependant autosomal dominant gene(s) whichis fully penetrant by the end of the human lifespan is responsible for all cases of ADas has been suggested (see Table 1.3), cumulative lifetime risks and lifetime riskcurves for female and male first-degree relatives, as well as parents and sibs of ADcases should be equal and should approach 50% by the oldest ages (age 80-90).These risk curves should resemble a sigmoid curve asymptotic to a final lifetime riskof 50% if: (i) the gene(s) is fully expressed within the age range of the sample and(ii) age-of-onset follows a normal distribution (St. George-Hyslop et al., 1989).42Under the same model, first-degree relatives of controls should have a risk to developthe disease of approximately 0% (assuming the gene is rare). This 0% risk is purelyhypothetical since the collection of a totally homogeneous potentially unaffectedgroup of controls is highly unlikely. Alternatively, if AD is not strictly inherited in anautosomal dominant fashion, but a genetic component to the etiology of the diseasedoes exist, then the risk to first-degree relatives of cases should be higher (but lessthan 50%) than that to first-degree relatives of controls.2.62 Criteria and Analysis for FADThe following criteria used by this study and the Alzheimer Clinic foraccepting a family as representing FAD are recognized to be quite stringent and thusidentify a relatively conservative group as FAD. This was done in an attempt toexclude those families in which non-genetic clustering occurs (ie. phenocopies), andappears to follow an autosomal dominant model of inheritance. The criteria are asfollows:1. Detailed family history must be available for at least the case's generation andthe previous generation.2. Good clinical documentation of dementia in relatives, preferably from at leasttwo separate sibships within the family, must be available; and there must beno other possible explanation for the dementia (eg. infarcts, strokes,alcoholism, head injury, etc.).3. Neuropathologic documentation of AD must be available for at least one43member of the family, but preferably for two or more.4. Accurate information on ages of death and/or present ages of relatives mustbe available so that it is possible to assess the "significance" of being clinicallyunaffected.Data on first-degree relatives of cases will also be analyzed according to how"affected" relatives are included. Families meeting the criteria for FAD will beincluded in the overall risk calculations for first-degree relatives of cases in order todirectly compare risk estimates with those calculated for previous studies (see Table1.3). These families will also be analyzed separately to assess whether the criteriaused to distinguish FAD families results in a subset of AD families for which thelifetime risk is in fact compatible with an autosomal dominant model of transmission.Identifying these families is increasingly important as molecular genetic studiespursue linkage analyses.2.63 Risk Comparisons for Sample SubgroupsThe data on first-degree relatives represent relatively unbiased samples from theentire B.C. population. Therefore, results of analyses offer important informationabout lifetime risks to first-degree relatives of elderly individuals either affected orunaffected with AD. Subgroups of first-degree relatives of cases and controls can beidentified and allow meaningful analyses due to the relatively large sample for thisstudy. This sample represents the largest of its kind reported to date, and as a result,44estimation of risk to various subgroups will not have the large standard error of mostprevious studies (see Table 1.3).Table 2.1 shows the subgroups for which age-specific and cumulative lifetimerisks will be calculated and compared using the Kaplan-Meier estimates. Differencesin lifetime risk curves will be assessed for significance using the log-rank test withdifferences being considered significant at a =.05. Differences in cumulative lifetimerisk estimates will be assessed for significance using a difference of proportions test.Results will also be considered significant at a =.05.Based on previous estimates, this study should provide adequate sensitivity indetecting significant differences between the cumulative lifetime risks to varioussubgroups of first-degree relatives. If it is assumed that the risk estimates for presentdata set will approximate those from the earlier study (23.3 ± 3.8% to first-degreerelatives of AD cases) (Sadovnick et al., 1989), a difference in cumulative lifetimerisk of approximately 15% will be detected at a significance level of a =.05 and apower of 1-13=.80. This level of sensitivity is believed to be more than adequate todetect differences between a hypothesized 50% cumulative lifetime risk to first-degree relatives of AD probands under an autosomal dominant transmission modeland the previously calculated 23.3% risk (1-13=.999 at a significance level of a =.05).Since subgroups of first-degree relatives represent a smaller sample size than theentire sample of first-degree relatives, the detectable differences in cumulativelifetime risk between subgroups will be greater than the approximately 15%detectable when comparing the entire sample.45Table 2.1: First-Degree Relative Subgroup Comparisons1) First-Degree Relatives of Cases vs First-Degree Relatives of Controls2) Female First-Degree Relatives of Cases vs Female First-Degree Relatives ofControls3) Male First-Degree Relatives of Cases vs Male First-Degree Relatives of Controls4) Female First-Degree Relatives of Cases vs Male First-Degree Relatives of Cases5) Female First-Degree Relatives of Controls vs Male First-Degree Relatives ofControls6) Parents of Cases vs Parents of Controls7) Sibs of Cases vs Sibs of Controls8) Parents of Cases vs Sibs of Cases9) Parents of Controls vs Sibs of Controls10) First-Degree Relatives of Early-Onset Cases (onset age <65) vs First-DegreeRelatives of Late-Onset Cases (onset age? 65)463 RESULTS3.1 Sample Analyzed3.11 CasesTable 3.1 summarizes the diagnosis of all 883 Alzheimer Clinic cases after fullclinical assessment. Informative and complete family histories were available forthree hundred thirty eight (38%) cases diagnosed as either "probable" AD (N=316)or subsequent to clinic assessment as "autopsy-confirmed" (N=22) AD. Table 3.2gives the number and percentage of affected relatives in each subgroup of first-degreerelatives of cases. The case family histories represent data on 1867 first-degreerelatives (parents = 676, sibs =1191) of 338 index cases.Eighteen of the families (5.36%) met the Alzheimer Clinic's criteria for FADand are included in the risk calculations for the entire sample of first-degree relativesof cases as well as separately under the "FAD only" criteria. Under the "FAD only"criteria, 30 of 106 first-degree relatives are considered affected.3.12 ControlsInformative and complete family histories were available for 351 consecutive,unrelated controls and 1873 of their first degree relatives (parents =702, sibs =1171)from the CSHA at the time of analysis. Table 3.3 shows the number and percentageof affected first-degree relatives of controls in each subgroup. Of the 351 controls,only two were identified from the institution sample (0.57%). Eighteen of 35147Table 3.1: Diagnosis for 883 Alzheimer Clinic Patients After Evaluation*Clinic Diagnosis No. % of TotalDemented, Alzheimer Unlikely 67 7.59Demented, Possible Alzheimer 230 26.04Demented, Probable Alzheimer 357 40.43Autopsy Confirmed Alzheimer 30 3.40Not Demented 199 22.53Total 8831 100.00*Seen during period 1985 to 19911Total number of patients for whom diagnosis was made at time of analysesTable 3.2: Number and Percent of AffectedFirst-Degree Relatives in Each Case SubgroupRelative Total No. No. Affected % Affected*Females 937 84 8.96Males 930 30 3.23Parents 676 72 10.64Sibs 1191 42 3.53Relatives of 648 34 5.25Early-OnsetCasesRelatives of 1219 80 6.56Late-OnsetCasesAll Relatives 1867 114 6.11Numbers and percentages are not age-corrected risk.48controls, (5.13%), were given a full clinical examination as part of the CSHA afterscoring 80 or above on the 3MS. None of these 18 individuals showed any signs ofcognitive impairment and/or dementia after full clinical assessment.Table 3.3: Number and Percent of Affected First-Degree Relatives in Each Control SubgroupRelative Total No. No. Affected % Affected*FemalesMales9559181771.781.78ParentsSibs702117113111.85.94All Relatives 1873 24 1.28Numbers and percentages are not age-corrected risk.Table 3.4: Comparison of Cases and ControlsCases ControlsNumber of Females 219 195Number of Males 119 156Mean Age* 72 ± 7.9* 74.5 ± 6.3Mean Age-of-Onset of Dementia 67 ± 8.4* N/A*Standard Error.*Mean age does not differ significantly (t=.25, df=687, p>.05).49Table 3.4 shows a comparison of selected characteristics of cases and controls.3.2 Risk Estimates for First -Degree Relatives of Cases and ControlsThe Kaplan-Meier age specific risk estimates for first-degree relatives (parentsand full sibs) of cases and controls are given in Table 3.5. The cumulative lifetimerisk to first-degree relatives of cases is 26.73 ± 4.42% compared with 7.26 ± 2.74%for first-degree relatives of controls. This difference is statistically significant whenevaluated with a difference of proportions test (z = 3.75, p < .001). The lifetime riskcurve for first-degree relatives of cases is significantly higher than that of first-degreerelatives of controls after assessment by a log-rank test (Log Rank Chi-Square = 87.13,p < .001). The lifetime risk curves for both groups of first-degree relatives are plottedin Figure 3.1. The risk for developing AD increases with age in both subgroups.Risk begins at an earlier age, and increases more rapidly for first-degree relatives ofcases compared with controls, but the risk increases at greater rate at the later agesfor both groups of relatives.3.3 Risk Estimates for FADTable 3.6 gives the age specific lifetime risk estimates for first-degree relativesof cases under the "FAD only" criteria. These values are plotted in Figure 3.2. Thecumulative lifetime risk for "FAD only" first-degree relatives increases more rapidlythan when the total group of all first-degree relatives of cases is considered. The riskfor the former reaches the expected 50% risk compatible with autosomal dominant50Table 3.5: Age Specific Risks for First-Degree Relatives of Cases and ControlsCases ControlsAge Risk (%) S.E. Risk (%) S.E.35 0.00 0.00 0.00 0.0036 0.06 0.06 0.00 0.0037 0.06 0.06 0.00 0.0038 0.12 0.08 0.00 0.0039 0.18 0.11 0.00 0.0040 0.24 0.12 0.00 0.0041 0.24 0.12 0.00 0.0042 0.24 0.12 0.06 0.0643 0.24 0.12 0.06 0.0644 0.31 0.14 0.06 0.0645 0.31 0.14 0.06 0.0646 0.31 0.14 0.06 0.0647 0.31 0.14 0.06 0.0648 0.31 0.14 0.06 0.0649 0.31 0.14 0.06 0.0650 0.37 0.15 0.06 0.0651 0.37 0.15 0.06 0.0652 0.44 0.17 0.06 0.0653 0.57 0.19 0.06 0.0654 0.57 0.19 0.06 0.0655 0.57 0.19 0.06 0.0656 0.57 0.19 0.06 0.0657 0.71 0.21 0.06 0.0658 0.78 0.23 0.06 0.0659 0.86 0.24 0.13 0.0960 1.08 0.27 0.13 0.0961 1.16 0.28 0.13 0.0962 1.39 0.31 0.13 0.09continued51Cases ControlsAge Risk (%) S.E. Risk (%) S.E.63 1.47 0.32 0.13 0.0964 1.64 0.34 0.13 0.0965 1.89 0.37 0.13 0.0966 2.08 0.40 0.13 0.0967 2.36 0.42 0.13 0.0968 2.46 0.44 0.22 0.1369 2.67 0.46 0.40 0.1870 3.43 0.54 0.50 0.2071 3.68 0.56 0.60 0.2372 4.30 0.63 0.70 0.2573 5.11 0.70 0.82 0.2874 5.97 0.78 0.94 0.3075 7.03 0.87 0.94 0.3076 7.57 0.92 1.22 0.3677 8.16 0.97 1.37 0.3978 9.00 1.05 1.37 0.3979 9.97 1.14 1.56 0.4380 13.03 1.41 1.56 0.4381 14.86 1.56 1.56 0.4382 15.86 1.65 2.80 0.7083 16.59 1.71 2.80 0.7084 17.01 1.75 3.10 0.7685 18.99 1.97 3.10 0.7686 18.99 1.97 3.54 0.8787 19.87 2.14 4.06 1.0188 23.07 2.74 4.06 1.0189 23.07 2.74 4.76 1.2290 23.07 2.74 4.76 1.2291 23.07 2.74 4.76 1.22continued52Cases^ ControlsAge Risk (%) S.E. Risk (%) S.E.92 23.07 2.74 4.76 1.2293 23.07 2.74 4.76 1.2294 23.07 2.74 4.76 1.2295 26.73 4.42 7.26 2.7453Table 3.6: Age Specific Risks for First-Degree Relatives of Cases from FAD FamiliesAge^Risk (%)^S.E.35 0.00 0.0036 1.01 1.0037 1.01 1.0038 2.04 1.4339 3.07 1.7540 4.10 2.0141 4.10 2.0142 4.10 2.0143 4.10 2.0144 5.16 2.2545 5.16 2.2546 5.16 2.2547 5.16 2.2548 5.16 2.2549 5.16 2.2550 5.16 2.2551 5.16 2.2552 5.16 2.2553 6.34 2.5154 6.34 2.5155 6.34 2.5156 6.34 2.5157 8.91 3.0358 10.25 3.2759 11.61 3.4960 14.33 3.8861 14.33 3.8862 15.81 4.0863 15.81 4.0864 17.34 4.28continued54Age Risk (%) S.E.65 17.34 4.2866 18.99 4.5167 22.51 4.9568 24.27 5.1469 24.27 5.1470 29.95 5.7171 31.95 5.8872 31.95 5.8873 34.22 6.1174 34.22 6.1175 36.57 6.3376 36.57 6.3377 41.86 6.8178 41.86 6.8179 41.86 6.8180 41.86 6.8181 51.04 7.5282 51.04 7.5283 51.04 7.5284 55.12* 7.93* Cumulative lifetime risk reaches predicted 50% by age 84.Figure 3.1: Risk to First-DegreeRelatives of Cases and Controls56Figure 3.2: Risk to First-DegreeRelatives of FAD Families57transmission by the late ages of life (age 84).3.4 Risk Estimates for Female and Male First-Degree Relatives of Cases andControls3.41 Risk Estimates for Female First-Degree Relatives of Cases and ControlsAge specific risks for female first-degree relatives of cases and controls aregiven in Table 3.7. Female first-degree relatives of cases show a significantly highercumulative lifetime risk than female first-degree relatives of controls (28.97 ± 3.60%,6.14 ± 1.78%, z =5.72, p < .001). Figure 3.3 plots the lifetime risk curves for femalefirst-degree relatives of cases and controls. The curves show that the risk for femalefirst-degree relatives of cases is significantly higher than that to first-degree femalerelatives of controls throughout their lifetime (Log Rank Chi-Square = 68.56, df= 1,p < .001). The risk to female first-degree relatives of cases begins at an earlier ageand increases more rapidly than the risk to first-degree relatives of controls.3.42 Risk Estimates for Male First -Degree Relatives of Cases and ControlsThe first-degree male relative subgroups of cases and controls also show anincrease in risk with age. Male first-degree relatives of cases show a much higher, yetstatistically non-significant cumulative lifetime risk when compared to male first-degree relatives of controls (22.03 ± 8.29%, 9.44 ± 6.79%, z =1.18, p < .23), as shownin Table 3.8. The large difference in cumulative risk estimates between the two malefirst-degree relative subgroups may be significant. However, due to the large58standard errors assumed by the cumulative lifetime risk estimates, even in a sampleof this size, the possible significance in risk may not be detectable. This standarderror reflects the extremely low number of males alive at the extreme older ages,which will, of course, result in very low number of affected males at these same ages.Canadian census data show 49.45% of the total Canadian population is male whilethis percentage drops rapidly to 44.08% by ages 70-74, 41.20% by ages 75-79, 37.61%by ages 80-84 and 30.32% by ages ? 85 (Statistics Canada, 1986). Figure 3.4 showsthat from age 82 to age 95, no male first-degree relatives of controls are consideredto become affected. One individual has an age-of-onset at age 95, but this is aninstance of using an individual's last known age when definitely affected in place ofan unknown age-of-onset (see section 2.51). This single age-of-onset results in a verylarge increase in risk (2.47% to 9.44%) over a 1 year time interval. It would be safeto assume that this individual had a true age-of-onset before age 95. This wouldresult in a decrease in the final cumulative lifetime risk estimate. For example, if thissingle individual had an age-at-onset 8.5 years before the family history was collected(the average for first-degree relatives of controls), the final cumulative lifetime riskfor male first-degree relatives of controls could be calculated at 3.93 ± 1.81%. Thislower cumulative lifetime risk estimate for male first-degree relatives of controlswould be significantly lower than that for male first-degree relatives of cases (z = 2.13,p < .04).Although the final cumulative lifetime risk for the two male subgroups was notfound to differ significantly in this sample, the lifetime risk to first-degree59Table 3.7: Age Specific Risks for FemaleFirst-Degree Relatives of Cases and ControlsCases ControlsAge Risk (%) S.E. Risk (%) S.E.37 0.00 0.00 0.00 0.0038 0.12 0.12 0.00 0.0039 0.24 0.17 0.00 0.0040 0.24 0.17 0.00 0.0041 0.24 0.17 0.00 0.0042 0.24 0.17 0.00 0.0043 0.24 0.17 0.00 0.0044 0.36 0.21 0.00 0.0045 0.36 0.21 0.00 0.0046 0.36 0.21 0.00 0.0047 0.36 0.21 0.00 0.0048 0.36 0.21 0.00 0.0049 0.36 0.21 0.00 0.0050 0.48 0.24 0.00 0.0051 0.48 0.24 0.00 0.0052 0.61 0.27 0.00 0.0053 0.86 0.33 0.00 0.0054 0.86 0.33 0.00 0.0055 0.86 0.33 0.00 0.0056 0.86 0.33 0.00 0.0057 0.99 0.35 0.00 0.0058 1.13 0.38 0.00 0.0059 1.13 0.38 0.13 0.1360 1.42 0.43 0.13 0.1361 1.42 0.43 0.13 0.1362 1.86 0.49 0.13 0.1363 2.01 0.52 0.13 0.13continued60Cases ControlsAge Risk (%)^S.E. Risk (%) S.E.64 2.32 0.56 0.13 0.1365 2.64 0.60 0.13 0.1366 2.81 0.62 0.13 0.1367 3.16 0.67 0.13 0.1368 3.34 0.69 0.29 0.2069 3.54 0.72 0.62 0.3170 4.54 0.84 0.79 0.3671 4.76 0.87 0.97 0.4072 5.44 0.94 0.97 0.4073 6.88 1.10 1.18 0.4574 7.92 1.20 1.40 0.5075 9.00 1.31 1.40 0.5076 9.96 1.40 1.65 0.5577 10.64 1.47 1.91 0.6178 11.37 1.55 1.91 0.6179 13.01 1.72 1.91 0.6180 17.26 2.10 1.91 0.6181 19.73 2.31 1.91 0.6182 20.80 2.40 3.10 0.9183 21.95 2.50 3.10 0.9184 22.59 2.56 3.57 1.0385 25.59 2.87 3.57 1.0386 25.59 2.87 4.27 1.2387 25.59 2.87 5.07 1.4688 28.97 3.60 5.07 1.4689 28.97 3.60 6.14 1.7961Table 3.8: Age Specific Risks for Male First-Degree Relatives of Cases and ControlsCases^ ControlsAge Risk (%) S.E. Risk (%) S.E.35 0.00 0.00 0.00 0.0036 0.12 0.12 0.00 0.0037 0.12 0.12 0.00 0.0038 0.12 0.12 0.00 0.0039 0.12 0.12 0.00 0.0040 0.25 0.18 0.00 0.0041 0.25 0.18 0.00 0.0042 0.25 0.18 0.13 0.1343 0.25 0.18 0.13 0.1344 0.25 0.18 0.13 0.1345 0.25 0.18 0.13 0.1346 0.25 0.18 0.13 0.1347 0.25 0.18 0.13 0.1348 0.25 0.18 0.13 0.1349 0.25 0.18 0.13 0.1350 0.25 0.18 0.13 0.1351 0.25 0.18 0.13 0.1352 0.25 0.18 0.13 0.1353 0.25 0.18 0.13 0.1354 0.25 0.18 0.13 0.1355 0.25 0.18 0.13 0.1356 0.25 0.18 0.13 0.1357 0.40 0.23 0.13 0.1358 0.40 0.23 0.13 0.1359 0.56 0.28 0.13 0.1360 0.71 0.32 0.13 0.1361 0.88 0.36 0.13 0.1362 0.88 0.36 0.13 0.13continued62Cases ControlsAge Risk (%)^S.E. Risk (%) S.E.63 0.88 0.36 0.13 0.1364 0.88 0.36 0.13 0.1365 1.06 0.40 0.13 0.1366 1.26 0.45 0.13 0.1367 1.47 0.49 0.13 0.1368 1.47 0.49 0.13 0.1369 1.70 0.54 0.13 0.1370 2.18 0.64 0.13 0.1371 2.44 0.69 0.13 0.1372 3.00 0.79 0.37 0.2773 3.00 0.79 0.37 0.2774 3.64 0.91 0.37 0.2775 4.66 1.07 0.37 0.2776 4.66 1.07 0.69 0.4277 5.12 1.16 0.69 0.4278 6.12 1.34 0.69 0.4279 6.12 1.34 1.14 0.6280 7.38 1.60 1.14 0.6281 8.17 1.77 1.14 0.6282 9.04 1.95 2.47 1.1183 9.04 1.95 2.47 1.1184 9.04 1.95 2.47 1.1185 9.04 1.95 2.47 1.1186 9.04 1.95 2.47 1.1187 11.37 2.99 2.47 1.1188 14.23 4.03 2.47 1.1189 14.23 4.03 2.47 1.1190 14.23 4.03 2.47 1.1191 14.23 4.03 2.47 1.11continued63Cases^ ControlsAge Risk (%) S.E. Risk (%) S.E.92 14.23 4.03 2.47 1.1193 14.23 4.03 2.47 1.1194 14.23 4.03 2.47 1.1195 22.03 8.29 9.44 6.7964Figure 3.3: Risk to Female First-DegreeRelatives of Cases and Controls65Figure 3.4: Risk to Male First-DegreeRelatives of Cases and Controls66male relatives of cases begins at an earlier age than the risk to first-degree malerelatives of controls and remains significantly higher (Log Rank Chi-Square =19.61,df= 1, p<.001).3.43 Risk Estimates for Female and Male First -Degree Relatives of CasesFemale and male first-degree relatives of cases show equal cumulative lifetimerisks (28.97 ± 3.60%, 22.03 ± 8.29%, z=.78, p > .42). Figure 3.5 plots the risks tospecific ages for both subgroups. The risk to female first-degree relatives of casesand male first-degree relatives of cases begins at approximately the same age,however the risk to female first-degree relatives increases much more rapidly thanthe risk to male first-degree relatives. The risk to female first-degree relatives thenlevels off at an earlier age, yielding a significantly different lifetime risk curve (LogRank Chi-Square = 16.23, df =1, p <001).3.44 Risk Estimates for Female and Male First -Degree Relatives of ControlsThe cumulative lifetime risk to female and male first degree relatives ofcontrols does not differ significantly (6.14 ± 1.78%, 9.43 ± 6.79%, z=.47, p>.62).The lifetime risk curves show similar increases in risk with age as well (Log RankChi-Square =1.07, df =1, p > .22). Figure 3.6 plots the lifetime risk curves for the twosubgroups. While the risk to first-degree male relatives of controls begins at anearlier age, this risk is extremely low, and remains comparatively low untilapproximately age 70 when both subgroups begin to show a steady increase in risk.67Figure 3.5: Risk to Female and MaleFirst-Degree Relatives of Cases68Figure 3.6: Risk to Female and MaleFirst-Dedree Relatives of Controls69The small number of affected male first-degree relatives of controls as well as the useof the previously described procedure to deal with unknown age-of-onsets reduces thepower of the tests comparing both cumulative lifetime risk, and lifetime risk curves.3.5 Risk Estimates for Parents and Sibs of Cases and Controls3.51 Risk Estimates for Parents of Cases and ControlsThe age specific risks for parents of cases and controls are given in Table 3.9.The risk to both subgroups of parents increases with age. Parents of cases show asignificantly higher cumulative lifetime risk than parents of controls (27.67 ± 4.50%,6.86 ± 2.92%, z =3.91 p<.001). The lifetime risk curves for the two subgroups ofparents are plotted in Figure 3.7, demonstrating the risk to parents of cases begins torise at an earlier age, and increases more rapidly than the lifetime risk to parents ofcontrols (Log Rank Chi-Square = 60.09, df =1, p < .001).3.52 Risk Estimates for Sibs of Cases and ControlsTable 3.10 gives the age specific risks for sibs of cases and controls. The risk toboth subgroups shows an increase with age as expected. The cumulative lifetime riskfor sibs of cases is significantly higher than that to sibs of controls (28.87 ± 8.92%,7.44 ± 3.51%, z =2.23, p < .03). Figure 3.8 plots the lifetime risk curves for sibs ofcases and controls. The risk to sibs of cases begins at an earlier age and rises muchmore rapidly than does the risk to sibs of controls (Log Rank Chi-Square = 27.16,df= 1, p<.001).70Table 3.9: Age Specific Risks for Parents of Cases and ControlsCases^ ControlsAge Risk (%) S.E. Risk (%) S.E.35 0.00 0.00 0.00 0.0036 0.16 0.16 0.00 0.0037 0.16 0.16 0.00 0.0038 0.16 0.16 0.00 0.0039 0.16 0.16 0.00 0.0040 0.16 0.16 0.00 0.0041 0.16 0.16 0.00 0.0042 0.16 0.16 0.00 0.0043 0.16 0.16 0.00 0.0044 0.16 0.16 0.00 0.0045 0.16 0.16 0.00 0.0046 0.16 0.16 0.00 0.0047 0.16 0.16 0.00 0.0048 0.16 0.16 0.00 0.0049 0.16 0.16 0.00 0.0050 0.34 0.24 0.00 0.0051 0.34 0.24 0.00 0.0052 0.52 0.30 0.00 0.0053 0.89 0.40 0.00 0.0054 0.89 0.40 0.00 0.0055 0.89 0.40 0.00 0.0056 0.89 0.40 0.00 0.0057 1.08 0.44 0.00 0.0058 1.08 0.44 0.00 0.0059 1.27 0.48 0.00 0.0060 1.66 0.55 0.00 0.0061 1.85 0.58 0.00 0.0062 2.25 0.64 0.00 0.00continued71Cases ControlsAge Risk (%)^S.E. Risk (%) S.E.63 2.45 0.67 0.00 0.0064 2.66 0.67 0.00 0.0065 3.27 0.78 0.00 0.0066 3.71 0.84 0.00 0.0067 3.71 0.84 0.00 0.0068 3.71 0.84 0.00 0.0069 3.94 0.87 0.00 0.0070 4.67 0.96 0.21 0.2171 4.92 0.99 0.43 0.3072 5.43 1.05 0.43 0.3073 5.98 1.11 0.66 0.3874 7.37 1.26 0.66 0.3875 8.80 1.39 0.66 0.3876 9.12 1.42 1.19 0.5377 9.80 1.49 1.46 0.6078 11.21 1.62 1.46 0.6079 12.71 1.76 1.46 0.6080 16.61 2.07 1.46 0.6081 17.49 2.14 1.46 0.6082 18.87 2.25 336 1.0283 19.88 2.33 3.36 1.0284 20.44 2.38 3.36 1.0285 21.75 2.51 3.36 1.0286 21.75 2.51 3.36 1.0287 22.81 2.69 3.36 1.0288 24.05 2.92 3.36 1.0289 24.05 2.92 4.20 1.3190 24.05 2.92 4.20 1.3191 24.05 2.92 4.20 1.31continued72Cases^ ControlsAge Risk (%) S.E. Risk (%) S.E.92 24.05 2.92 4.20 1.3193 24.05 2.92 4.20 1.3194 24.05 2.92 4.20 1.3195 27.67 4.50 6.86 2.9273Table 3.10: Age Specific Risks for Sibs of Cases and ControlsCases^ ControlsAge Risk (%) S.E. Risk (%) S.E.37 0.00 0.00 0.00 0.0038 0.10 0.10 0.00 0.0039 0.29 0.17 0.00 0.0040 0.29 0.17 0.00 0.0041 0.29 0.17 0.00 0.0042 0.29 0.17 0.10 0.1043 0.29 0.17 0.10 0.1044 0.39 0.19 0.10 0.1045 0.39 0.19 0.10 0.1046 0.39 0.19 0.10 0.1047 0.39 0.19 0.10 0.1048 0.39 0.19 0.10 0.1049 0.39 0.19 0.10 0.1050 0.39 0.19 0.10 0.1051 0.39 0.19 0.10 0.1052 0.39 0.19 0.10 0.1053 0.39 0.19 0.10 0.1054 0.39 0.19 0.10 0.1055 0.39 0.19 0.10 0.1056 0.39 0.19 0.10 0.1057 0.50 0.22 0.10 0.1058 0.61 0.25 0.10 0.1059 0.61 0.25 0.21 0.1560 0.74 0.28 0.21 0.1561 0.74 0.28 0.21 0.1562 0.87 0.31 0.21 0.1563 0.87 0.31 0.21 0.1564 1.00 0.34 0.21 0.15continued74Cases ControlsAge Risk (%)^S.E. Risk (%) S.E.65 1.00 0.34 0.21 0.1566 1.00 0.34 0.21 0.1567 1.50 0.44 0.21 0.1568 1.68 0.47 0.37 0.2269 1.87 0.51 0.70 0.3270 2.67 0.64 0.70 0.3271 2.90 0.68 0.70 0.3272 3.63 0.80 0.90 0.3673 4.68 0.95 0.90 0.3674 4.98 0.99 1.15 0.4575 5.62 1.08 1.15 0.4576 6.42 1.21 1.15 0.4577 6.89 1.29 1.15 0.4578 6.89 1.29 1.15 0.4579 6.89 1.29 1.61 0.6480 8.34 1.63 1.61 0.6481 12.20 2.45 1.61 0.6482 12.20 2.45 1.61 0.6483 12.20 2.45 1.61 0.6484 12.20 2.45 2.68 1.2485 15.93 3.49 2.68 1.2486 15.93 3.49 4.79 2.4287 15.93 3.49 7.44 3.5188 28.87 8.92 7.44 3.5175Figure 3.7: Risk to Parentsof Cases and Controls76Figure 3.8: Risk to Sibsof Cases and Controls77Figure 3.9: Risk to Parentsand Sibs of Cases78Figure 3.10: Risk to Parentsand Sibs of Controls793.53 Risk Estimates for Parents and Sibs of CasesThe cumulative lifetime risk for parents and sibs of cases do not differsignificantly (z=.11, p>.90). However, the lifetime risk curve for parents of casesdoes differ significantly from that for sibs of cases (Log Rank Chi-Square =4.83, df= 1,p<.03). Figure 3.9 demonstrates that the risk to both parents and sibs of cases beginat approximately the same age. The risk to parents remains higher at every age untilthe very late ages of life where it levels off. The risk to sibs then continues toincreases at a much more rapid rate until the risk to both parents and sibs becomeapproximately equal.3.54 Risk Estimates for Parents and Sibs of ControlsThe lifetime risk curves for parents and sibs of controls are plotted in Figure3.10. There is no significant difference between the two risk curves (Log Rank Chi-Square =.24, df= 1, p>.63). The risk to sibs begins at an earlier age but remainsextremely low until the later ages of life at which both risk curves show approximatelyequivalent increases until equal cumulative lifetime risks are reached (z=.13, p>.84).3.6 Risk to First -Degree Relatives of Early -Onset and Late-Onset CasesThe age specific risks for first-degree relatives of early-onset cases (onset beforeage 65) and for first-degree relatives of late-onset cases (onset at or after age 65) aregiven in Table 3.11. Both subgroups show the expected increase in risk with age.Cumulative lifetime risk for the two groups are approximately equal when evaluated80Table 3.11: Age Specific Risks for First-DegreeRelatives of Early-Onset and Late-Onset CasesEarly-Onset Cases Late-Onset CasesAge Risk (%) S.E. Risk (%) S.E.35 0.00 0.00 0.00 0.0036 0.17 0.17 0.00 0.0037 0.17 0.17 0.00 0.0038 0.34 0.24 0.00 0.0039 0.51 0.30 0.00 0.0040 0.68 0.34 0.00 0.0041 0.68 0.34 0.00 0.0042 0.68 0.34 0.00 0.0043 0.68 0.34 0.00 0.0044 0.86 0.38 0.00 0.0045 0.86 0.38 0.00 0.0046 0.86 0.38 0.00 0.0047 0.86 0.38 0.00 0.0048 0.86 0.38 0.00 0.0049 0.86 0.38 0.00 0.0050 1.05 0.43 0.00 0.0051 1.05 0.43 0.00 0.0052 1.24 0.47 0.00 0.0053 1.44 0.51 0.10 0.1054 1.44 0.51 0.10 0.1055 1.44 0.51 0.10 0.1056 1.44 0.51 0.10 0.1057 1.87 0.59 0.10 0.1058 1.87 0.59 0.21 0.1559 2.10 0.63 0.21 0.1560 2.34 0.67 0.42 0.2161 2.60 0.72 0.42 0.21continued81Early-Onset Cases Late-Onset CasesAge Risk (%) S.E. Risk (%) S.E.62 3.13 0.81 0.53 0.2463 3.13 0.81 0.53 0.2664 3.42 0.86 0.76 0.3165 4.03 0.95 0.88 0.3666 4.03 0.95 1.14 0.4267 4.03 0.95 1.53 0.4468 4.03 0.95 1.66 0.4969 4.03 0.95 1.95 0.6070 4.45 1.04 2.83 0.6271 4.92 1.13 3.00 0.7272 4.92 1.13 3.83 0.8473 4.92 1.13 4.92 0.9574 4.92 1.13 6.08 1.0675 5.51 1.27 7.30 1.1476 5.51 1.27 8.04 1.2277 5.51 1.27 8.85 1.2878 7.04 1.65 9.44 1.3579 8.85 2.05 10.10 1.6980 10.77 2.42 13.58 1.8581 13.03 2.83 15.26 1.9082 15.35 3.20 15.73 2.0283 15.35 3.20 16.77 2.0884 15.35 3.20 17.34 2.2585 19.11 4.01 18.68 2.2586 19.11 4.01 18.68 2.5487 19.11 4.01 19.93 3.3088 22.34 4.99 23.14 3.3089 22.34 4.99 23.14 3.3090 22.34 4.99 23.14 3.30continued82Early-Onset Cases^Late-Onset CasesAge Risk (%) S.E. Risk (%) S.E.91 22.34 4.99 23.14 3.3092 22.34 4.99 23.14 3.3093 22.34 4.99 23.14 3.3094 22.34 4.99 23.14 3.3095 22.34 4.99 28.26 5.5383Figure 3.11: Risk to First-DegreeRelatives of Early and Late-Onset Cases84with a difference of proportions test (22.34 ± 4.98%, 28.26 ± 5.83%, z=.75, p > .42).The lifetime risk curves for the two subgroups of case relatives are plotted in Figure3.11. The risk to first-degree relatives of early-onset cases shows a non-significant butnotable tendency to develop at an earlier age after which both subgroups show anequal increase of risk with time (Log Rank Chi-Square =.21, df= 1, p >.65).3.7 Comparison of Previous and Present SamplesThe earlier study on the predecessor of this sample ("original" sample)examined the risks to develop AD to 840 first-degree relatives of 151 AD cases(Sadovnick et al., 1989) and, using the Kaplan-Meier method, found a cumulativelifetime risk of 22.0 ± 3.6% by age 88. The risk to the consecutively and similarlyascertained group ("replication" sample) of 1043 first-degree relatives of 195 cases (8cases from the "original" sample were rediagnosed) found a cumulative lifetime riskof 22.58 ± 3.77% by age 88 and 28.53 ± 6.70% by age 95. No significant differencesin lifetime risk between the "original" and "replication" studies were found (z = 0.86,p > .37; Log Rank Chi-Square =1.32, df =1, p >.25). The total sample reported here("combined" sample) represents a 2.2 fold increase in size over the "original" sample.Figure 3.12 plots the lifetime risk for all first-degree relatives from the "original" and"replication" samples.3.8 Effect of Using Missing Ages -of-OnsetThe use of an affected relatives' last known age when definitely affected as their853025 "Original" Sample"Replication" Sample,,,,,^i20-.,P10-5-_ _ ,..."0 I^1^1^I^I^I^1 I^1^1^1^1^1^1^1^II^I^I^I I^I^11^I^1^1^I^1^I^1^1^1^I^11^I^11^I^I^11^I^I^1^I^1^1^1^I IIFigure 3.12: Comparison ofPrevious and Present Samples35 40 45 50 55 60 65 70 75 80 85 90 95AGE (years)86true age-of-onset in the instance of an unknown age-of-onset is a conservativeapproach previously used (Sadovnick et al., 1989) to treat data influenced by theimperfect recall of a family member giving the family history. The precise effect ofthis transformation remains unknown as the true age-of-onset for some individualscan never be known. However, several different estimates used in place of anunknown age-of-onset suggest that overall, no statistically significant differences in thelifetime risk curves will result from utilizing this procedure.An illustrative example of the effect of using a different estimate in place of anaffected relative's unknown age-of-onset is to use the average age-of-onset for theentire sample. This was calculated as 70.13 (maximum value =85, minimumvalue = 32, standard deviation =10.50). This estimate obviously has limitations; themost prominent being the wide range of values and the fact that using this estimatewill definitely result in some relatives being considered affected long before their trueage-of-onset, while others will not be considered affected long after their true age-of-onset. Nevertheless, this procedure was used only to illustrate the possibleimplications of the procedure described in section 2.51. The resulting Kaplan-Meierrisk estimates are shown in Figure 3.13. A comparison of the lifetime risk curvecalculated using the average age-of-onset in place of an unknown age-of-onset doesnot differ significantly from the risk curve calculated using the last known age whenalive for an unknown age-at-onset using a log-rank test (Log Rank Chi-Square =.022,df= 1, p > .88). The lifetime cumulative risk does differ significantly (z = 2.15, p <.04),but, as will be discussed later in this section, does not have a significant effect on87testing the hypothesis of an autosomal dominant model of transmission. It is notedthat these comparisons are not performed on strictly independent samples, but werecompleted only to illustrate the effect of the age-of-onset estimation.Another comparison can be done using a "worst case" and "best case" scenario.This is accomplished by considering an affected relative with unknown age-of-onset tobe affected at the earliest ("worst-case") or latest ("best-case") recorded onset agesdocumented in the sample (age 36 and 85). Using the "best case" scenario results insome relatives having onset ages after their recorded age at death, but the analysiswas completed as if the relative lived at least to the onset age of 86. Using theseestimates, as with the one previously described results in the improbable situation ofa large group of relatives becoming affected at exactly the same age, thereforeresulting in a large increase in risk at that particular age. These results are alsoshown in Figure 3.13, and as in the previous comparison, the lifetime risk curves showno statistically significant differences using the log-rank test (for age 36: Log RankChi-Square = .06, df =1, p > .81; for age 85: Log Rank Chi-Square = .11, df =1, p > .74).The cumulative lifetime risk estimate using age 36 differs significantly from theestimate used in section 3.2 (z = 2.56, p < .012), while using age 85 does not result in asignificantly different lifetime cumulative risk (z=.22, p > .84).Using the procedure described in section 2.51 will result in a slightoverestimation of the true cumulative lifetime risk due to the fact that some affectedrelatives will be considered to have an age-of-onset after their true age-of-onset.However, this method is considered reasonable as all affected relatives will definitely88suffer from AD at the time they are considered affected. The highest possible riskestimate calculated using an improbable age-of-onset approximation ("best case"scenario), results in a cumulative lifetime risk far below the 50% risk which would beconsistent with an autosomal dominant model of transmission. This result shows thatthe method used will have no significant effect on hypothesis testing, and will notresult in any statistically significant changes in lifetime risk curves.89Figure 3.13: Effect of UsingUnknown Ages-of-Onset90Table 3.12: Summary of ResultsSubgroup^ Cumulative Lifetime Risk Estimate (%)Cases^ControlsAll First-Degree Relatives^26.73 ± 4.42^7.26 ± 2.74Female First-Degree Relatives 28.97 ± 3.60^6.14 ± 1.79Male-First Degree Relatives 22.03 ± 8.29^9.44 ± 6.79Parents^ 27.67 ± 4.50^6.86 ± 2.92Sibs 28.87 ± 8.92^7.44 ± 3.51First-Degree Relatives of Early-Onset Cases^22.34 ± 4.99^N/AFirst-Degree Relatives of Late-Onset Cases^28.26 ± 5.53^N/ADifference ofProportions p^Log Rank pSubgroup Comparison^ Value^ValueFirst-Degree Relatives of Cases vs Controls^<.001^<.001Female First-Degree Relatives of Cases vs^< .001 < .001ControlsMale First-Degree Relatives of Cases vs^.23^<.001ControlsFemale vs Male First-Degree Relatives of^.42^< .001CasesFemale vs Male First-Degree Relatives of^.62^.22ControlsParents of Cases vs Controls^<.001^<.001Sibs of Cases vs Controls .03 <.001Parents vs Sibs of Cases .90^.03Parents vs Sibs of Controls^ .63 .83First-Degree Relatives of Early vs Late-^.42^.65Onset Cases914 DISCUSSION4.1 First-Degree Relatives of Cases and ControlsThis study group is the largest of its kind reported to date, and represents anincrease of up to 15 fold in the number of cases compared to other studies in theliterature (see Table 1.3). The number of cases and the resultant number of "at-risk"first-degree relatives is an important factor in studies of late onset diseases wherethere is a rapid decrease in the number of first-degree relatives surviving to theeighth or ninth decades of life. Table 4.1 shows the numbers of relatives "at-risk" tovarious ages for this study compared with other studies for which these numbers aswell as the standard errors assumed by each age-specific risk estimate were available.The standard errors assumed by some previously reported risk estimates becomerather large as the numbers of individuals "at-risk" decrease in the last decades of life(see Table 1.3, Table 4.1).The risk estimates calculated for this sample represent important data for riskcounselling of first-degree relatives of AD patients. As media reports on risk factorsfor AD increase, these relatives are more concerned about their own risks to developAD.The significantly higher lifetime risk curve and 3.7 fold higher lifetimecumulative risk to first-degree relatives of cases compared to first-degree relatives ofcontrols agree with the results of other studies. However, the estimated cumulativelifetime risk for first-degree relatives of both cases and controls is substantially lower92Table 4.1: Cumulative Risk Calculations and Numbers ofFirst-Degree Relatives "at-Risk" : A Literature Review.No. Relatives^CumulativeReferences^Age^At Risk^Incidence (%)Mohs et al., 1987^60 173.5 .58 ± .5865^139.5^2.71 ± 1.3470 91 7.71 ± 2.5375 53 14.62 ± 3.8080^26.5^24.06 ± 5.6886 8.5 45.87 ± 9.82Breitner et al.,^60^255^1.03 ± .601988^65 200 2.93 ± 1.1170 135.5^4.94 ± 1.5875^88 11.07 ± 2.6880 49 19.48 ± 4.1385 21^36.54 ± 6.4187^12 49.33 ± 8.38Martin et al.,^65 71^1.39 ± 1.381988^70^50 9.74 ± 3.8174 33 16.93 ± 5.3181 14^35.44 ± 8.6083^11 40.82 ± 9.42Present Study 65 1158^1.89 ± .3770^889 3.34 ± .5475 622 7.03 ± .8780 352^13.03 ± 1.4185^167 18.99 ± 1.9790 55 23.07 ± 2.7495 20^26.73 ± 4.42'Numbers of first-degree relatives at specific ages are shown for studies which gavethis information. For total number of cases used in previous studies see Table 1.393than that estimated by most other studies (Breitner and Folstein, 1984; Breitner etal.,1988; Huff et al., 1988; Martin et al., 1988; Mayeux et al., 1991; Mohs et al., 1987).Notable exceptions are Farrer et al. (1989) and Sadovnick et al. (1989).Previously published cumulative lifetime risk estimates for AD to first-degreerelatives of non-demented controls range widely from 8.1 ± 6.0% to 29 ± 7%. Allestimates assume extremely large standard error (Table 1.3). The estimated lifetimecumulative risk to first-degree relatives of controls found for this sample is not widelydisparate from those estimates in the lower range of previously reported risks, butobviously differ greatly from the estimates in the higher range. The cumulativelifetime risk calculation for first-degree relatives of controls should approximate thelifetime risk of the general population, and it will be of interest to compare theseresults to those of the CSHA once it is completed. A study investigating AD in aCanadian population (Gautrin et al., 1990) used data from two Finnish studies(Molsa et al., 1982; Sulkava et al., 1985) to infer prevalence rates in the generalpopulation. The projected rates seem to agree with the results from this studyconsidering the standard error of the risk estimates. Gautrin et al. (1990) projectedprevalence rates of 1% for 65-74 year olds compared with .94 ± .30% to age 74 forthis study; 4% for 75-84 year olds compared with 3.10 ± .76% to age 84 for thisstudy; and 10.5% to those 85 years and over compared with 7.26 ± 2.74% to age 95for this study. The data on which Gautrin et al. (1990) base their estimates varywidely with prevalence rates for 85 years old and above ranging from 6.30% (Molsa94et al., 1982) to 14.80% (Sulkava et al., 1985). This wide variation in estimates shouldbe taken into account when considering the final overall prevalence estimates.Although the cumulative lifetime risk to first-degree relatives of AD casescalculated for this study ("combined" sample), is much lower than that calculated formost other studies (Table 1.3), it agrees well with the cumulative lifetime riskestimate calculated for the predecessor of this sample ("original" sample). Sincecumulative lifetime risks and lifetime risk curves between the "original" sample andthe consecutively and similarly ascertained sample ("replication" sample) of 195 casesand 1043 of their first degree relatives are also in agreement it is unlikely that resultsrepresent sampling error. The results of analyses on the "replication" sample areslightly (though non-significantly) higher than the "original" sample. This can beexplained by the fact that the "replication" sample contains a very long-livedindividual with an unknown age-of-onset who is considered to become affected at thelast possible age (age 95). Since this individual probably had a true age-of-onset atan earlier age, the cumulative lifetime risk estimate could presumably be slightlylower (see explanation in section 3.32). The similar results between the "original" and"replication" samples further suggests that the structure of the sample has remainedrelatively constant throughout the seven years of data collection.One previous study (Farrer et al., 1989) estimated a lifetime risk for AD tofirst-degree relatives of AD cases of 24% by age 93. This estimate is much lowerthan that calculated by the majority of other studies, and similar to the one calculatedfor this study. Although Farrer et al. (1989) used the Kaplan-Meier method to95generate their risk estimate, this risk may not be directly comparable to other studiesas it used an original weighting method to estimate the likelihood of a correctdiagnosis in affected relatives.There are several possible factors which may influence the lower risk estimatecalculated for this, as well as the "original" sample. One reason may be the attemptto be as rigorous and conservative as possible in labelling a relative "affected". Carewas taken to rule out dementia due to other causes whenever possible (see Table1.1). This results in a sample relatively "clean" of confounding factors which mayaccount for a relative being misclassified as "affected". In a prevalence of dementiastudy, Bachman et al. (1992) reported that only 55.6% of all cases of dementia werediagnosed as "probable" AD. The effect of including all reportedly senile ordemented relatives as "affected" would be to artificially increase the lifetime riskestimates.Secondly, the definition for onset may also impact the risk calculations. Onsetis defined in this study as "the first detectable symptoms of cognitive embarrassment"rather than using the stricter definitions of "the first definite symptoms of dementia"or "evidence of progressive dementia" (Breitner et al., 1989). Although evidence ofprogressive and irreversible loss of cognitive function is one criteria used to consideran individual affected, the age-of-onset is considered to occur when these deficits firstbecome apparent. Using "caseness" instead of "onset" as the definition for age-of-onset would result in relatives considered to become affected at a later age. Since alower number of total relatives "at-risk" survive to each successive onset age, this96decreased number of "at-risk" relatives would result in an increase to risk estimates.Breitner et al. (1989) found that using the definition of "onset" rather than "caseness"may result in an estimate only 60% of that calculated for the latter.A third factor which could explain the difference in risk estimates is thecomposition of the study sample. This sample is from a population to whichuniversal health care is available, unlike the situation for studies from United States.Although the information is not available, the ethnic mix and education level ofsamples may differ as well.The method in which cases for this study were ascertained should limit bias ofthe sample toward a population which was identified through a positive familyhistory. The cumulative risk to families meeting the FAD criteria shows that theover-representation of these FAD aggregates in a sample would result in an upwardshift in the lifetime risk curve as well as an increase in cumulative lifetime risk closerto 50%. Mayeux et al. (1991) found that in certain studies, possible selection andinformation biases may result in an increase in the risk of dementia to first-degreerelatives of AD patients that is not specific, and may exist in first-degree relatives ofpatients suffering from other neurological disorders as well.A lifetime cumulative risk approaching 50% would be consistent with anautosomal dominant model of inheritance as supported by several previous studies.However, this risk estimate would not be proof that a single autosomal dominantgene is responsible for all cases, a point which is further supported by moleculargenetic studies suggesting etiologic heterogeneity. The cumulative lifetime risk97calculated for this sample supports the previous finding of Sadovnick et al. (1989) andsuggests that an autosomal dominant model of inheritance with full penetrance by theend of the human lifespan is not responsible for all cases of AD. Nevertheless, theelevated risk in first-degree relatives of AD index cases offers further evidence thatfamilial factors do have a role in AD etiology (Amaducci et al., 1986; Chandra et al.,1987; Heyman et al., 1984; Shalat et al., 1986). The evidence presented byepidemiological studies suggesting a wide range of possible environmental risk factorscombined with the molecular genetic studies seem to suggest etiologicalheterogeneity. Although a polygenic model cannot be ruled out, it seems likely thatboth genetic and environmental factors may combine to play a significant role in ADetiology. An interesting repercussion of such a hypothesized interaction betweengenetic and environmental factors in at least a proportion of AD cases is thatinterventions delaying the onset of AD in a genetically susceptible individual for onlya few years, past the age-of-death, could greatly decrease the incidence of the disease.The cumulative lifetime risk estimate for FAD in first-degree relatives of casesfrom such families confirms that using the criteria outlined in the section 2.62identifies families which show transmission of AD conforming to an autosomaldominant model of transmission. If etiologic heterogeneity exists then a proportionof cases in FAD families would be expected to be due to causes other than anautosomal dominant gene(s). The results permit a slight increase over the expected50% estimate (55.12 ± 7.93%) which would allow for such non-genetic cases(phenocopies) to occur within FAD families. Using the criteria outlined in section982.62, only approximately 5.36% of all Alzheimer Clinic cases of AD represent FAD.The true proportion of FAD cases may be higher, but the criteria used in this studywere designed to identify an extremely rigorously defined group of FAD families onwhom linkage analysis studies could be performed.4.2 Gender-Specific RisksThe results show male and female first-degree relatives of AD cases bothdisplay a higher lifetime risk when compared to their analogous control subgroups.Although the difference in cumulative lifetime risk between male subgroups in thissample was found to be significant only at a level of p < .25, the risk to male first-degree relatives of controls could be, credibly, much lower (see section 3.42). Thelow number of male first-degree relatives surviving until the eighth decade of life isreflected in the large standard error reported for the cumulative lifetime riskestimates reported for males (Table 1.3). Farrer et al. (1989) found very few malesin their sample surviving to the late ages of life, which presenting a greater problemsince their total sample size was approximately half of this sample. Nevertheless, anelevated risk, regardless of gender, in first-degree relatives of AD cases lends furthersupport to the contribution of genetic factors to AD etiology.A female preponderance in the prevalence and incidence of AD compared withmales has been reported (Akesson 1969; Bachman et al, 1992; Broe et al., 1976;Hagnell et al., 1991; Schoenberg et al., 1985; Sturt, 1986; Sulkava et al., 1985; Treveset al., 1986). Since estimated prevalence rates of a disease are influenced by both its99duration and incidence, it has been suggested that increased prevalence of AD amongfemales could result from of an increased duration among females which has beenpreviously reported (Breitner et al., 1988; Gruenberg, 1978).The findings suggesting female first-degree relatives of AD cases show astatistically greater increase in their cumulative risk at an earlier age, but a similarcumulative lifetime risk when compared to male first-degree relatives of AD casesaffirms those results reported by Breitner et al. (1988). They reported a non-significant, but "notable" higher risk curve among female first-degree relatives of ADindex cases. The present results also support the suggestion by Breitner et al. (1988)that the sex specific differences presented by the earlier epidemiologic studies mirrorthe differential age-specific expression of an identical propensity in both female andmale first-degree relatives of AD cases. These findings are by no means universal, asothers have asserted that no difference in risk exists and the increased incidence infemales is due to an earlier censoring age in males (Farrer et al., 1989). Theseresults also discount an X-linked gene(s) playing a significant role in AD etiology.The differential age-specific expression found in this study would seem to beparticular to female and male first-degree relatives of AD cases as female and malefirst-degree relatives of controls show an equal increase in cumulative risk as well asequal cumulative lifetime risks. The results seem to suggest that individuals in whomfamilial recurrence risk is high (first-degree relatives of AD cases) show a greater sex-specific expression, while individuals in whom this risk is lower (first-degree relativesof controls) are not as sensitive to this sex-specific difference.100A potential mechanism for this difference in the sex-specific expression of ADamong genetically susceptible individuals is the difference in the expression of femaleand male hormones. The details of these differences are discussed in Section 1.31.5.Possible interaction of environmental and genetic factors affecting AD expressionhave been proposed previously, but to date, no suggestion of the effect such aninteraction might have on the sex-specific expression of AD has been made.4.3 Generational -Specific RisksThe increased risk to parents and sibs of AD cases compared to their respectivecontrol subgroups confirms that the increased familial risk among first-degreerelatives of cases is not restricted to any particular subgroup. These results lendfurther support the supposition of a genetic contribution to the etiology of thedisease.The results of the comparison of risk to parents and sibs of cases is more difficultto interpret. Parents of cases appear to have a higher risk compared to sibs of casesthroughout most of their lifetime. In the extremely late ages of life however, thecumulative lifetime risks appear to become equal, much as female first-degreerelatives of cases show a higher lifetime risk compared to male first-degree relativesof cases throughout most of their lifetimes until the very late ages.Previous studies present varied results. An earlier study done on a smaller butcomparable sample found a higher, but non-significant, cumulative lifetime risk insibs (Breitner et al., 1984). It was felt that poor or incomplete information on long101dead parents accounted for this difference. In a later study, the poor-qualityinformation on the parental generation was discarded from analyses, resulting insimilar risk calculations for both parents and sibs (Breitner et al., 1988). This resultis also supported by another investigation which found a much lower but statisticallysimilar lifetime risk among parents and sibs (Farrer et al., 1989). Other studiesdiscount these findings and report a lower lifetime risk among sibs. These resultsmay not be directly comparable however, as the first (Heyman et al., 1983) reportsrisk to only age 75, and the second (Heston et al., 1981) used only index casesdiagnosed by examination of histopathologic material without benefit of a clinicalhistory. It is important to note that the information regarding sibs may be morereliable than that for parents. Since sibs are contemporaries of cases, currentknowledge is available when assessing medical histories. This is not the case for mostof the parental generation who will have been deceased for at least 25 or 30 years atthe time the family histories were collected making it more likely that someinformation will be forgotten or lost.The results presented by this study support the majority of evidence and suggestthat lifetime cumulative risk for parents and sibs are equal, thus providing evidenceagainst an autosomal recessive gene(s) playing a part in the etiology of the majorityof AD cases. Results show equal cumulative lifetime risks, but significantly differentage-specific risk curves between parents and sibs of AD cases and an identical riskamong parents and sibs of controls. Taken together, this evidence might suggest that102the differential age-specific expression in parents and sibs of cases is due to the effectof non-genetic factors on individuals with an identical genetic predisposition.A recent investigation provides an example of the magnitude of anenvironmental agent's effect changing through time. A study on the epidemiology ofasthma found the incidence rate of the disease was doubling and tripling amongchildren and adolescents during the period from 1964 to 1983, whereas the incidencerates for adults and infants had remained constant (Yunginger et al., 1992). Theauthors suggested this increase in incident rates could be partly due to "unidentifiedenvironmental factors that preferentially exert their effects upon the lower airways ofgrowing children...". They further suggest that the changes in the air exchange ratesof newer, more energy efficient homes built in the last twenty years has resulted inincreased levels of indoor allergens; the predominant cause of asthma. One mightpropose an analogous environmental factor exerting a greater impact on the parentalgeneration resulting in the differential age-specific expression in identical geneticallysusceptible individuals, while having a diminished effect on the age-specific expressionin less genetically susceptible individuals.A recent workshop examining the role of environmental factors in the etiologyof neurodegenerative disorders came to the consensus that a wide variety of agentsincluding food additives, natural food contaminants, and atmospheric contaminantsplay a role in the etiology of some of these disorders (Henneberry and Spatz, 1990).Epidemiological studies examining exposure to all known environmental risk factorsincluding aluminum, viruses and organic solvents may help to elucidate a possible103cause for the observed difference in risk. As previously suggested, if a multifactorialetiology plays a major role in AD, altering a susceptible individuals exposure toenvironmental risk factors could result in the delayed expression of the disease, pastthe age-of-death, a possible means of decreasing the incidence of AD exists.4.4 First-Degree Relatives of Early-Onset and Late-Onset CasesThe partioning of the first-degree relatives of AD cases into early-onset andlate-onset subgroups based on the age-of-onset of the proband, ( <65, >65), whilesomewhat arbitrary, follows the convention of several previously published reports(Huff et al, 1987; Selzer and Sherwin, 1983). It should be noted that the division offamilies based solely on the proband's onset age may not necessarily reflect theoverall age-of-onset pattern within a family. It has been suggested that division offamilies based on the age-of-onset of proband is a method by which thisheterogeneous disorder could be successfully partioned. It was further suggested thatearly-onset cases contained a greater familial component than late-onset cases(reviewed by Nalbantoglu et al., 1990).The lifetime cumulative risks and lifetime risk curves to first-degree relatives ofearly-onset and late-onset cases appear to be similar. Although the lifetime riskcurves are shown to be homogeneous overall by a log-rank test, risk to first-degreerelatives of early-onset cases is elevated at an earlier age. Two previous studies(Farrer et al., 1989; Huff et al., 1988) present results which suggest that while nodifference in cumulative lifetime risk exists between the two relative subgroups, a104difference in lifetime risk curves does exist. Similar age-of-onsets within familieshave been previously reported (Heston et al., 1981; Larsson et al., 1963; Powell andFolstein, 1984) leading one group to theorize the possible existence of an allelic, ortightly linked modifier gene of an AD gene (Huff et al., 1988). The results presentedhere, as well as those of several other studies (Breitner et al., 1988; Heyman et al.,1983; Heston et al., 1981) do not find any evidence of a modifier gene having asignificant effect in the majority of cases, and do not support the proposal thatindividuals with an affected first-degree relative suffering from an early-onset form ofthe disease have a greater lifetime risk to develop AD.A study on kindreds suffering from FAD by Farrer et al. (1990) found a higherrisk in first-degree relatives of late-onset FAD families (86%) compared to first-degree relatives of early-onset FAD families. This suggests that two or moremechanisms may be involved in the disease etiology, with the possibility of bothgenetic and non-genetic cases (phenocopies) being expressed within the same family.Taking into account these results, as well as the studies linking a late-onsetFAD subset of individuals to chromosome 21 as well as chromosome 19, suggests thatetiological heterogeneity exists, and partioning families by age-of-onset of theproband is not necessarily a way of increasing the chances of identifying FADfamilies1054.5 Methodology CaveatsAlthough this is the largest study of its kind to date, and rigorous methods wereused to document reportedly affected relatives, investigations of this type do havelimitations, many of which are unavoidable. The most obvious drawback is thegreatly diminished number of individuals surviving to the later ages of life. Thedecreasing numbers of individuals surviving into the eighth and ninth decades of lifeare reflected in the standard error of the risk estimates to these ages. In the presentstudy this effect is most conspicuously evident in the risk calculations to male first-degree relatives of both cases and controls. This restriction will decrease somewhatas clinics continues to document family histories and the samples increase to a size towhich minimal standard errors will be assumed by the cumulative lifetime riskestimates.Since both the case and control samples were population based, ascertainmentbias was minimal. Both samples are representative of the entire population of B.C..Although the literature does not provide evidence of significant differences in theoverall prevalence rates of AD within different ethnic groups, the infrequentoccurrence of FAD clustering within certain ethnic groups has been documented(Bird et al., 1988; Goudsmit et al., 1981). Since FAD cases represent onlyapproximately 5% of cases diagnosed with AD at the "Alzheimer Clinic" and since noovert "clustering" of ethnic groups has been noted for patients attending the"Alzheimer Clinic", any effect that the differing ethnicity of the case sample may haveon results should be minimal. A detailed examination of the ethnicity of the elderly106B.C. population in comparison to the ethnicity of patients diagnosed with AD at the"Alzheimer Clinic" would be one way to conclusively rule out any ethnic differencesbetween case and control samples.An issue which is harder to address is that of collecting controls. Although themean age of the control sample is at the age where most cases of AD should beevident, obviously this group may contain "at-risk" individuals who have not yetreached their age-of-onset. At present there is no method to determine which ofthese controls is susceptible. Even the use of family histories collected from non-demented, cognitively unimpaired elderly deceased individuals would not solve thisproblem as the age-of-death could occur before the age-of-onset in some individuals.The result of this dilemma may be the dilution of the homogeneity of the controlproband sample. However, due to the advanced age of the sample used in this studythis problem may be minimal.The clinical diagnosis of cases gives approximately 90% accuracy in assigningdiagnosis (Tierney et al., 1988) which, though excellent, will still result in the dilutionof the homogeneity of the case proband sample. This problem can be minimized asthe number of "autopsy-confirmed", definite AD, case probands which have beendiagnosed based on clinical and neuropathological examinations also increases asclinics continue to collect data.Another issue which is difficult to address is the form of the data, and themethod of analyses. The data on individual first-degree relatives does not representa collection of independent observations, but rather "clusters" of non-random107observations (ie. family histories). To date, no adequate means of taking this"clustering" into account has been used in studies of this type. The resolution of thisquandary can only be accomplished by the future development of suitable analyticalmethods by statisticians.Finally, the difficulty in determining the unknown age-of-onset for a long deadrelative will remain intrinsic in studies utilizing the "family-history" method. Onesolution is the use of a recently developed maximum-likelihood method utilizing theEM algorithm (Cupples et al., 1991). This method uses a maximum-likelihoodtechnique to estimate the age of onset distribution among relatives with both knownand unknown ages-of-onset. This method was not used for the present study due tothe authors having some reservations as to the application of the program in its'current form to the problem of lifetime risk estimation (personal communication).However, with further refinements, the use of this method should facilitate moreaccurate risk estimate calculations in data sets containing some missing information.4.6 AD ModelsEvidence to date suggests that the etiology of AD is complicated, and theprospect of discovering a single cause for the disease now seem nonexistent. Anearly model proposed by Bird et al. (1989) which draws analogies to the complexetiology of atherosclerosis is one possibility that seems to encompasses the body ofevidence to date. Atherosclerosis, like AD, is a very common disease with a complexmultifaceted etiology. Atherosclerosis has a final common pathway with arterial fatty108plaque deposits and coronary artery disease. The factors leading to this pathway aremany and varied, with evidence pointing at environmental factors (including serumlipids as influenced by diet, blood pressure and smoking) several different monogenicfactors, and multifactorial combinations of genetic and environmental factors. Adefinite familial risk has been recognized and numerous mutations have beenidentified in several genes including the LDL receptor as well as applipoproteins A-I,C-III, A-IV, and A-II (reviewed by Breslow et al., 1989). This model would appear tohave parallels to the presumed cascade of events leading to AD. Whether geneticand environmental influences interact on a common pathway, or act separately toinfluence the disease outcome remains unknown and continuing studies are needed toclarify these interactions. However, this model provides a useful framework forfurther research into the etiology of AD.Several recent proposals would seem to fit into this general model framework.Hardy and Higgins (1992) hypothesize that any number of causes can initially triggerthe final cascade of 13-amyloid protein. Their hypothesis further proposes that thedeposition of 13-amyloid protein is the final pathway leading to the pathologic changescharacteristic of AD. Potter (1991) offered a hypothesis that can also be placed inthe framework of this general model. One of the "triggers" proposed by Hardy andHiggins (1992) could be the accumulation of trisomy 21 cells developed throughabnormal segregation during mitosis, which leads to AD through the samemechanism by which trisomy 21 DS patients develop the disease (Potter, 1991). Themodel also proposes this abnormal segregation could be caused by an inherited109genetic mutation near or at the centromere of chromosome 21, as well as throughexposure to environmental agents like aluminum. Another model which fits into thisoverall framework is that of Tanzi and Bradley (1991) who hypothesize thatmutations in the APP gene disrupt a translational regulatory stem-loop structure inthe APP messenger RNA. The regulatory loop contains a consensus sequencecharacteristic of the "iron responsive elements", which would allow for both geneticand environmental factors to effect APP production, and eventually lead to theabnormal deposition of 13-amyloid protein. More recently Spurr et al. (1992) foundthat the frequency of mutations in cytochrome P450's and glutathione S-transferase,genes having a central role in the metabolism and detoxification of drugs as well asenvironmental and endogenous chemicals, occur more frequently in AD cases whencompared to controls. This evidence lead the researchers to conclude that "geneticsusceptibility to the environment may explain the majority of disease occurrence".Placed into the context of this general model, the results presented by this studyshow that the less than 50% increased risk to all first-degree relatives of cases overfirst-degree relatives of controls is evidence of etiologic heterogeneity. A smallproportion of these cases follow a model of transmission compatible with autosomaldominant inheritance, however the cumulative lifetime risk to all first-degree relativesof AD cases suggests that an autosomal dominant gene(s) is not responsible for allcases of AD, and that multifactorial inheritance may play a major role in the diseaseetiology. Although a polygenic model cannot be ruled out, given the preponderanceof evidence from epidemiological studies, it seems likely that environmental agents110may play a part in the disease etiology. These same non-genetic agents may also playa part in the differential age specific expression seen between females and males aswell as between parents and sibs.1115 CONCLUSIONThe results presented in this study add to a growing body of evidence thatsuggest that AD is a etiologically heterogeneous disorder. The early hypothesisadvocating all cases of AD are due to a fully penetrant autosomal dominant gene(s)are no longer supported by this investigation.The elevated risk to all first-degree relatives of AD cases over all first-degreerelatives of controls as well as the results of risk analysis to "FAD only" first-degreerelatives are evidence that a genetic component to the etiology of AD exists. Thelower lifetime risk estimates presented in this investigation agree with those reportedfor the "original" sample (Sadovnick et al., 1989), and probably reflect the stringentcriteria used in accepting a relative as "affected", as well as the minimalascertainment bias in identifying cases. Confidence in results is increased due to therelatively large size of the sample investigated and the agreement in results betweenthe "original" sample and "replication" samples.The increased risk to all subgroups of first-degree relatives of cases compared tothe analogous subgroups of first-degree relatives of controls suggests that the geneticcontribution to the etiology of AD is not generational or gender specific. However,although an equal cumulative lifetime risk for female and male first-degree relativesof cases exists, female first-degree relatives seem to be at a higher risk throughoutmost of their lifetime suggesting a differential age-specific expression of an analogousgenetic predisposition exists. A similar age-specific differential expression appears to112exist between parents and sibs of cases with parents showing a higher risk throughoutmost of their lifetime. Possibly, differences in hormonal or environmental exposuremay account for the differences in the expression of AD. If exposure to such anenvironmental risk factor plays a major role in the age-specific expression of AD, thepossibility of delaying the onset past the age-of-death as a means of decreasing theoverall incidence of the disease exists. The identification of possible risk factors istherefore vitally important, and the continued conduct of large case-control studies isof major import.The separation of families into early-onset and late-onset aggregates based onlyon the age-of-onset of the proband as a means of identifying a larger proportion offamilies which exhibit an autosomal dominant model of disease transmission seemsunfounded based on the results of this investigation. Further, the molecular geneticstudies identifying a subset of late-onset families with an autosomal dominantlysegregating FAD gene discounts the supposition that families exhibiting an early-onset are a means by which to identify FAD familiesAn important step in the process of unravelling the complex etiology of AD isto collect and document those families meeting a strict criteria for FAD. As previousinvestigations have shown, using these families in linkage analysis studies allows theidentification of purely genetic etiologies of AD. A clinical marker making a simpledefinitive diagnosis of AD is another necessary step in improving studies of this type,hopefully leading to the final goal of finding therapies to extend the age-of-onset pastthe age-of-death, or ultimately finding a cure.113ReferencesAkesson, H.O. (1969) A Population Study of Senile and Arterosclerotic Psychosis. Hum.Hered., 19:546-563.Alzheimer, A. (1907) fiber Eine Eigenartige Erkangung der Hirnrinde. Allg. Z. furPsychiatrie, 64:146-48.Amaducci, L.A., Fratiglioni, L, Rocca, W.A., et al. (1986) Risk Factors for ClinicallyDiagnosed Alzheimer's Disease: A Case Control Study of an Italian Population.Neurology, 36:922-931.American Psychiatric Association (1987) Diagnostic Criteria from DSM-II-R.Washington, D.C.: American Psychiatric Association.Andreasen, N.C., Endicott, J., Spitzer, R.L. (1977) The Family History Method UsingDiagnostic Criteria: Reliability and Validity. Arch. Gen. Psychiatry, 34:1220-1235.Anthony, J.C., LeResche, L., Niaz, U., et al. (1982) Limits of the Mini-Mental State as aScreening Test for Dementia and Delerium Among Hospital Patients. Psychol. Med.,12:397-408.Appel, S. (1981) A Unifying Hypothesis for the Cause of Amytrophic Lateral Sclerosis,Parkinsonism, and Alzheimer Disease. Ann. Neurol., 20:499-505.Armitage, P. (1971) Statistical Methods in Medical Research. London: BlackwellScientific Publications, 171-172.Bachman, D.L., Wolf, P.A., Linn, R., et al. (1992) Prevalence of Dementia and ProbableSenile Dementia of the Alzheimer Type in the Framingham Study. Neurology,42:115-119.Ball, M.J. (1987) Pathological Similarities Between Alzheimer Disease and DownSyndrome: Is There a Genetic Link? Integr. Psychiatry, 5:159-170.114Beattie, B.L. (1992) Dementia and Alzheimer's Disease: Theory and Reality. Can. J.CME, 3(2):43-53.Benton, A.L. (1968) Differential Behavioural Effects in Frontal Lobe Disease.Neuropsychologia, 6:53-60.Bird, T., Strahahan, S., Sumi, S.M., et al. (1983) Alzheimer Disease: AcetyltransferaseActivity in Brain Tissue from Clinical and Pathological Subgroups. Ann. Neurol.,14:284-293.Bird T., Lampe, T.H., Nemens, E., et al. (1988) Familial Alzheimer's Disease inAmerican Descendants of Volga Germans: Probable genetic founder effect. AnnNeurol., 23:25-31.Bird, T., Sumi, S.M., Nemens, E.J., et al. (1989) Phenotypic Heterogeneity in FAD: AStudy of 24 Kindreds. Ann Neurol., 25:12-25.Blessed, G., Tomlinson, B.E., Roth, M. (1968) The Association of Quantitative Measuresof Dementia and of Senile Change in the Cerebral Grey Matter of Elderly Subjects.Br. J. Psychiatry, 114:797-811.Breitner, J.C.S., Folstein M.F. (1984) Familial Alzheimer's Dementia: A PrevalentDisorder with Specific Clinical Features. Psychol. Med., 14:63-80.Breitner, J.C.S., Silverman, J.M., Mohs, R.C., et al. (1988) Familial aggregation inAlzheimer's Disease: Comparison of Risk Among First Degree Relatives of Earlyand Late-Onset Cases, and Among Male and Female Relatives in a SuccessiveGeneration. Neurology, 38:207-212.Breitner J.C.S., Magruder-Habib, K.M. (1989) Criteria for Onset Critically Influence theEstimation of Familial Risk in Alzheimer's Disease. Genetic Epidem., 6:663-669.Breslow, J., Deeb, S., Lalouel, J., et al. (1989) Genetic Susceptibility to Atherosclerosis.Circulation 80(3):724-728.115Broe, G.A., Akhter, A.J., Andrews, G.R., et al. (1976) Neurological Studies of theElderly at Home. J. Neurol. Neurosurg. Psychiatry, 39:362-366.Burns, A., Luthert, P., Levy, R., et al. (1990) Accuracy of clinical diagnosis ofAlzheimer's Disease. Br. Med. J., 301:1026.Burgio, G., Severi, F., Rossoni, R., et al. (1966) Autoantibodies and Downs Syndrome.American Journal of Pathology, 73:457-476.Bushke, H. (1984) Cued Recall in Amnesia. J. Clin. Exp. Neuropsych., 6:433-440.Carter, D.A., Desmarias, E, Bellis, M., et al. (1992) More Missense in Amyloid Gene.Nature, 2:255-256.Chandra, V., Phillipose, V., Bell, P., et al. (1983) Case-Control Study of Late-OnsetProbable Alzheimer's Disease. Neurology, 37:590-597.Chartier-Harlin, M.C., Crawford, F., Houlden, H., et al. (1991) Early-Onset Alzheimer'sDisease Caused by a Mutation at Codon 717 of the B-Amyloid Precursor ProteinGene. Nature, 353:884-886.Chui, H.C., Teng, E.L., Henderson, V.W., et al. (1985) Clinical Subtypes of Dementia ofthe Alzheimer Type. Neurology, 35:1544-1550.Coval, M., Crockett, D., Holliday, S., et al. (1985) A Multi-Focus Assessment Scale forUse with Frail Elderly Populations. Can. J. Aging, 4:101-109.Crapper, D., Krishnan, S., Dalton, A. (1973) Brain Aluminum Distribution in AD andExperimental Neurofibrillary Degeneration. Science, 180:5111-5113.Crapper, D., Quittkat, S., Krishnan, S., et al., (1980) Intranuclear Aluminum Content InAlzheimer Disease, Dialysis, Encephalopathy, and Experimental AluminumEncephalopathy. Acta Neuropathol., 50:19-24.116Crowley, J. (1984) Statistical Analysis of Survival Data. Ann. Rev. Public Health, 5:385-411.Cupples, L., Risch, N., Farrer, L., et al. (1991) Estimation of Morbid Risk and Age atOnset with Missing Information. Am. J. Hum. Genet., 49:76-87.Davies, P. (1986) The Genetics of Alzheimer's Disease: A Review and a Discussion ofthe Implications. Neurobiol. Aging, 7:459.Dussault, J.H., Ruel, J. (1987) Thyroid Hormones and Brain Development. Ann. Rev.Physiol., 49:321-334.Ebers, G.C., Bulman, D.E., Sadovnick, A.D., et al. (1986) A Population-Based TwinStudy in Multiple Sclerosis. N. Engl. J. Med., 315:1638-1642.Evans, D.A., Funkenstein, H.H., Albert, M.S., et al. (1989) Prevalence of AlzheimerDisease in a Community Population of Older Persons. JAMA, 262:2551-2556.Ewins, D.L., Rossor, M.N., Butler, J., et al. (1991) Association Between AutoimmuneThyroid Disease and FAD. Clinical Endocrinology, 35:93-96.Farrer, L.A., O'Sullivan, D.M., Cupples, L.A., et al. (1989) Assessment of Genetic Riskfor Alzheimer's Disease Among First-Degree Relatives. Arm. Neurol., 25:485-493.Feher, E.P. (1992) Establishing the Limits of the MMS Examination of Subjects.Archives of Neurol., 49(1):87-92.Feldman, R.G. (1963) Familial Alzheimer's Disease. Neurology, 13:811-824.Fleiss, J.L. (1981) Statstical Methods for Rates and Proportions. New York, New York:Wiley and Sons.Folstein, M.F. (1983) The Mini-Mental State Examination. In: Crook, T., Ferris, S.,Bortus, R. (eds) Assessment in Geriatric Psychopharmacology. New Canaan, Conn:Mark Powley, 47-49.117Forette, F., Henry, J., Orgogozo, J., et al. (1989) Reliability of Clinical Criteria for theDiagnosis of Dementia. Arch. Neurol., 46:646-648.Freidland, R.P. (Moderator) (1988) NIH Conference. Alzheimer's Disease•Clinical andBiological Heterogeneity. Ann. Intern. Med., 109:298-311.Gautrin, D., Froda, S., Tetreault, H., et al. (1990) Canadian Projections of CasesSuffering From Alzheimer's Disease and Senile Dementia of Alzheimer Type Overthe Period 1986-2031. Can. J. Psychiatry, 35:162-165.Glenner, G.G., Wong, C.W. (1984) Alzheimer's Disease: Initial Reports of thePurification and Characterization of a Novel Cerebrovascular Amyloid Protein.Biochem. Biophys. Res. Commun., 120:885-890.Goate, A., Chartier, M., Mullan, M., et al. (1991) Segregation of a Missense Mutation inthe Amyloid Precursor Protein Gene with Familial Alzheimer's Disease. Nature,349:704-706.Goldgaber, D., Lerman, J.I., McBride, 0.W., et al. (1987) Charecterization andChromosomal Location of a cDNA encoding Brain Amyloid of Fibral Protein.Science 235:877-880.Golden, C.J., Hammeke, T.A, Purisch, A.D. (1980) Luria-Nebraska NeuropsychologicalBattery. Los Angeles, Cal: Western Psychological Services.Goudsmit, J., White, B.J., Weitkamp, L.R., et al. (1981) Familial Alzheimer's Disease inTwo Kindreds of the Same Geographic and Ethnic Origin: A Clinical and GeneticStudy. J. Neurol. Sci., 49:79-89.Greenough, W.T. (1986) What's Special About Development? Thoughts on the Bases ofExperience-Sensitive Synaptic Plasticity. In: Greenough, W.T., Juraska, J.M. (eds)Developmental Neuropsychobiology. New York, New York: Academic Press, 387-408.118Gruenberg, E.M. (1978) Epidemiology. In: Katzman, R., Terry, R.D., Bick K.L. (eds)Alzheimer's Disease: Senile Dementia and Related Disorders. New York, New York:Raven Press, 323-326.Hagnell, 0., Franck, A., Greisbeck, A., et al. (1981) Senile Dementia of the Alzheimer'sType in the Lunmby Study. Eur. Arch. Psychiatry Clin. Neurosi., 241:159-164.Hardy, J.A., Higgins, G.A. (1992) Alzheimer's Disease: The Amyloid CascadeHypothesis. Science, 256:184-185.Haupt, M., Kurz, A., Pollman, S., et al. (1992) Alzheimer's Disease: Identical Phenotypeof Familial and Non-Familial Cases. J. Neurol., 239:248-250.Hendrie, H., Hall, K., Austrom, M. (1989) Some Thoughts on Cross Cultural Prevalenceof Dementia. In: Carstensen, L., Neale, J. (eds) Mechanisms of PsychologicalInfluences on Physical Health with Special Attention to the Elderly. New York, NewYork: Plenum Press, 161-184.Hendriks, L., van Duijn, C., Cras, P., et al. (1992) Presenile Dementia and CerebralHaemorrhage Linked to a Mutation at Codon 692 of the B-amyloid PrecursorProtein Gene. Nature, 1:218-221.Henneberry, R., Spatz, L. (1990) The Role of Environmental Factors inNeurodegenerative Disorders. Neurobiology of Aging, 12:75-79.Heston, L.L., Mastri, A.R., Anderson, V.E., et al. (1981) Dementia of the Alzheimer'sType: Clinical Genetics, Natural History, and Associated Conditions. Arch. Gen.Psychiatry, 38:976-981.Heyman, A., Wilkinson, W.E., Hurwitz, B.J., et al. (1984) Alzheimer's Disease: A Studyof Epidemiological Aspects. Aim. Neurol., 15:335-341.Hofman, A., van Duijn, C.M., Schulte, W., et al. (1990) Is Parental Age Related to theRisk of Alzheimer's Disease. Br. J. Psych., 157:273-275.119Huff, F.J., Aurebach, J., Chakravarti, A., et al. (1988) Risk of Dementia in Relativesof Patients with Alzheimer's Disease. Neurology, 38:786-790.Huff, F.J., Growden, J.H., Corkin, S., et al. (1987) Age at Onset and Rate of Progressionof Alzheimer's Disease. J. Am. Geriatr. Soc., 35:27-30.Hunter, R., Dayan, A.D., Wilson, J. (1971) Alzheimer's Disease in One MonozygoticTwin. J. Neurol. Neurosurg. Psychiatry, 35:707-710.Jarvik, L.F., Altshuler, K.Z., Kato, T., et al. (1971) Organic Brain Syndrome andChromosome Loss in Aged Twins. Dis. Nerv. Sys., 32:154-169.Jellinek, E.H. (1962) Fits, Faints, Coma and Dementia in Myxedema. Lancet, 2:1010-1012.Jellinger, K. (1990) Morphology of Alzheimer's Disease and Related Disorders. In:Maurer, K., Beckman, H., Riederer, P., (eds) Alzheimer's Disease. Epidemiology,Neuropathology, Neurochemistry, and Clinics. Vienna: Springer, 61-77.Jorm, A.F., Korten, A.E., Henderson, A.S. (1987) The Prevalence of Dementia: AQuantitative Integration of the Literature. Acta. Psychiat. Scand. 76:465-479.Jouan-Flahault, C., Seroussi, M., Colvez, A. (1989) Absence de Liason Entre DemenceSenile et Age Parental. Revue d'Epidemiologe et de Sante Publique, 37:73-76.Kallman, F.J. (1956) Genetic Aspects of Mental Disorders in Later Life. In: Kaplan, O.J.(ed) Mental Disorders in Later Life. Stanford, Con: Stanford University Press.Kang, J., Mullan, M., Chartier-Harlin, M.C., et al. (1987) The Precursor of Alzheimer'sDisease Amyloid A4 Protein Resembles a Cell-Surface Receptor. Nature 325:733-736.Kaplan, E.L., Meier, P. (1958) Non-Parametric Estimations from IncompleteObservations. JASA, 53:457-581.120Karlinsky, H., MacDonald, A.M., Berg, J.M. (1993) Primary Degenerative Dementia ofthe Alzheimer Type in Twins: Initial Findings from the Maudsley Hospital TwinRegister. Int. J. Geriatric Psych., (In Press).Karlinsky, H., Vaula, G., Ridgley, C., et al. (1992) Molecular and Prospective PhenotypicCharecterization of a Pedigree with FAD and a missense Mutation in Codon 717 ofthe B-Amyloid Precursor Protein Gene. 42nd Annual Meeting of the CanadianPsychistric Association, 20-21.Karnosh, L.J., Stout, R.E. (1935) Psychoses of Myxedema. Amer. J. Psychiat., 91:1263-1274.Katzman, R., Jackson, J.E. (1991) Alzheimer disease: Basic and Clinical Advances.JAGS, 39:516-525.Kay, D.W.K., Henderson, A.S., Scott, R., et al. (1985) Dementia and Depression Amongthe Elderly Living in the Hobart Community: The Effect of the Diagnostic Criteriaon the Prevalence Rates. Psychol. Med., 15:771-788.Kumar, A., Schapiro, M.B., Grady, C.L., et al. (1991) Anatomic, Metabolic,Neuropsychological and Molecular Genetic Studies of Three Pairs of Identical TwinsDiscordant for Dementia of the Alzheimer's Type. Arch. Neurol., 48:160-168.Lake, S., Wingblad, B., Iqbal, K. (1991) Second International Conference on AlzheimersDisease and Related Disorders. Neurobiol. Aging, 12:1609-1612.Larsson, T., Sjogren, T., Jacobsen G. (1963) Senile Dementia: A Clinical Sociomedicaland Genetic Study. Acta. Psychiatr. Scand., 39:1-39.Li, G., Shen, Y.C., Li, Y.T., et al. (1992) A Case-Control Study of Alzheimer Disease inChina. Neurology, 42:1481-1488.Lobo, E., Khan, M., Tew, J. (1980) Community Study of Hypothyroidism in DownsSyndrome. Paediatrics, 63:80-87.121Lowenberg, K., Waggoner R.W. (1934) Familial Organic Psychosis (Alzheimer's Type).Arch. Neurol. Psychiatry 31:337754.Markesbery, W.R., Ehman, W.D., Alauddin, M. (1983) Bulk Brain Concentrations inAlzheimer Disease. In: Wills, M.R., Savoy J. (eds) Aluminum Analysis in BiologicalMaterials. Charlottesville: University of Virginia Press.Markesbery, W.R., Ehmann, W.D., Hussein, T.I.M., et al. (1981) Instrumental NeutronActivation Analysis of Brain Aluminum In Alzheimer's Disease and Aging. ArmNeurol, 10:511-516.McDermott, J.R., Smith, A.L., Iqbal, K., et al. (1979) Brain and Aluminum in Agingand Alzheimer Disease. Neurolgy, 29:809-814.McFarland, H. (1992) Twin Studies in Multiple Sclerosis. Ann. Neurol., 32:722-723.McKahnn, G., Drachman, D., Folstein, M., et al. (1984) Clinical Diagnosis ofAlzheimer's Disease: Report of the NINCDS-ADRDA Work Group under theAuspices of the Department of Health and Human Services Task Force onAlzheimer's Disease. Neurology, 34:939-944.McLachlan, D.R.C., Lukiw, W.J., Mizzen, C.A., et al. (1989) Chromatin Structure inAlzheimer Disease: Effect on 5' Leader Sequence for NF-L Gene and Role ofAluminum. In: Iqbal, K., Wisniewski, H., Winblad, B. (eds) Alzheimer's Disease andRelated Disorders. New York, New York: Alan R. Liss Inc.McLachlan, D.R.C., Rupert, J.L., Sutherland, M.K., et al. (1991) The Etiology ofAlzheimer Disease. In: Berg, J., Karlinsky, H., Lowy, F. (eds) Alzheimer's DiseaseReasearch: Ethical and Legal Issues. Vancouver, B.C.: Carswell, 93-119.McRae-Degueurce, A., Haglid, K., Rosengren, L., et al. (1988) Antibodies RecognizingCholinergic Neurons and Thyroglobuline Are Found in the Cerebrospinal Fluid of aSubgroup of Patients with Alzheimer's Disease. Drug Develop. Res., 15:153-163.122Mayeux, R., Stern, Y., Spanton, S. (1985) Heterogeneity in Dementia of the AlzheimerType: Evidence of Subgroups. Neurology, 35:453-461.Mayeux, R., Sano, M., Chen, J., et al. (1988) Risk of Dementia in First-Degree Relativesof Patients with Alzheimer's Disease and Related Disorders. Arch. Neurol.,48:269-273.Martin, R.L., Gerteis, G., Gabrielli, W.F. (1988) A Family-Genetic Study of Dementia ofthe Alzheimer Type. Arch. Gen. Psychiatry, 45:894-900.Mohs, R.C., Breitner, J.C.S., Silverman, J.M., et al. (1987) Alzheimer's Disease. MorbidRisk Among First-Degree Relatives Approximates 50% by 90 Years of Age. Arch.Gen. Psychiatry, 44:405-408.Molsa, P.K., Marttila, R.J., Rinne, U.K. (1982) Epidemiology of Dementia in a FinnishPopulation. Acta. Neurol. Scand., 65:541-552.Morris and Rubin (1991) Clinical Diagnosis and Course of Alzheimer Disease. In:Davidson (ed) The Psychiatric Clinics of North America: Alzheimer's Disease.Montreal, Que: W.B. Saunders Co., 14(2):223-236.Mortimer, J.A., French, L.R., Hutton, J.T., et al. (1985) Head Injury as a Risk Factor forAlzheimer's Disease. Neurology, 35:264-267.Mortimer, J.A. (1989) Genetic and Environmental Risk Factors for Alzheimer's Disease:Key Questions and New Approaches. In: H. Altman (ed) Alzheimer's Disease andDementia: Problems, Prospects and Perspectives. New York, New York: PlenumPress, 85-100.Mullan, M., Crawford, F., Axelman, K., et al. (1992) A Pathogenic Mutation for ProbableAlzheimer's Disease in the APP Gene at the N-Terminus of 13-amyloid. Nature,1:345-347.123Mullan, M., Houlden, H., Windelspecht, M., et al. (1992) A Locus for Familial Early-Onset Alzheimer's Disease on the Long Arm of Chromosome 14, Proximal to thea 1-Antichymostrypsin Gene. Nature, 2:340-342.Murrell, J., Farlow, M., Ghetti, B., et al. (1991) A mutation in the Amyloid PrecursorProtein Associated with Hereditary Alzheimer's Disease. Science, 254:97-99.Nalbantoglu, J., Locoste-Royal, G., Gauvreau, D. (1990) Genetic Factors in Alzheimer'sDisease. JAGS, 38:564-568.Nee, L.E., Eldridge, R., Suderland T., et al. (1987) Dementia of the Alzheimer Type:Clinical and Familial Study of 22 Twin Pairs. Neurology, 37:359-363.Pericak-Vance, M., Yamaoka, L., Haynes, C. (1988) Genetic Linkage Studies inAlzheimer's Disease Families Exp. Neurol., 102:271-279.Pericak-Vance, M., Bebout, J., Gaskell, P., et al. (1991a) Linkage Studies in FamilialAlzheimer's Disease: Evidence for Chromosome 19 Linkage. Am. J. Hum. Genet.,48:1034-1050.Pericak-Vance, M., Haines, J., St. George-Hyslop, P., et al. (1991b) CollaborativeLinkage Analysis of Chromosome 9 and 21 Loci in Familial Alzheimer Disease. Am.J. Hum. Genet., 49[Suppl]:A1992.Pfeffer, R.I., Afifi, A.A., Chance, J.M. (1987) Prevalence of Alzheimer's Disease in aRetirement Community. Am. J. Epidemiol., 125:420-435.Potter, H. (1991) Alzheimers Disease and Downs Syndrome. Chromosome 21Nondisjunction May Underlie Both Disorders. Am. J. Hum. Genet., 48:1192-1200.Powell, D., Folstein, M.F. (1984) Pedigree Study of Familial Alzheimer's Disease. J.Neurogenet., 1:189-197.124Rappaport, S., Pettigrew, K., Schapiro, M. (1991) Discordance and Concordance ofDementia of the Alzheimer Type in Monozygotic Twins Indicate Heritable andSporadic Forms of Alzheimer Disease. Neurology, 41:1549-1553.Reinisch, J., Ziemba-Davis, M., Sanders, S.A. (1991) Hormonal Contributions to SexuallyDimorphic Behavioural Development in Humans Psychoneuroendocrinology, 16:213-278.Renvoize, E.B., Mindham, R.H.S., Stewart, M., et al. (1986) Identical Twins Discordantfor Presenile Dementia of the Alzheimer Type. Br. J. Psychiatry, 149:509-512.Robakis, N.K., Ramakrishna, N., Wolfe, G., et al. (1987) Molecular Cloning andCharacterization of a cDNA Encoding the Cerebrovascular and the Neuritic PlaqueAmyloid Peptides. Proc. Natl. Acad. Sci. USA, 84:4190-4194.Rocca, W.A., Amaducci, L.A., Schoenberg, B.S. (1986) Epidemiology of ClinicallyDiagnosed Alzheimer's Disease. Ann Neurol., 19:415-424.Rocca, W., Van Duijn, C., Clayton, V. (1991) Maternal Age and Alzheimers Disease: ACollaborative Re-Analysis of Case-Control Studies. International Journal ofEpidemiological Association, 20(2):521-527.Roses, A.D., Pericak-Vance, M.A., Clark, C.M., et al. (1990) Linkage Studies of Late-Onset Familial Alzheimer's Disease. Adv Neurol., 51:185-196.St George-Hyslop, P., Tanzi, R., Polinsly, P., et al. (1987) The Genetic Defect CausingFamilial Alzheimer's Disease Maps on Chromosome 21. Science, 235:885-889.St George-Hyslop, P., Meyers R., Haines, J., (1989) Familial Alzheimer's Disease:Progress and Problems. Neurobiol. Aging, 10:417-425.St George-Hyslop, P., Haines, J., Farrer, L., et al. (1990) Genetic Linkage StudiesSuggest that Alzheimer's Disease is not a Single Homogeneous Disorder. Nature,347:194-197.125St George-Hyslop, P., Haines, J., Rogaev, E., et al. (1992) Genetic Evidence for a NovelFamilial Alzheimer's Disease Locus on Chromosome 14. Nature, 2:330-334.Sadovnick, A.D., Irwin, M.E., Baird, P.A., et al. (1989) Genetic Studies on an AlzheimerClinic Population. Genetic. Epidem., 6:663-643.Sadovnick, A.D., Yee, I.M.L., Hirst, C. (1992) The Rate of Down Syndrome amongOffspring of Women with AD. (Submitted).Sadovnick, A.D., Armstrong, H., Rice, G.P.A. (1992) A Population-Based Twin Study ofMultiple Sclerosis in Twins: Update. Ann Neurol., (In Press).Sanford, K., Price, F., Shapiro, M., et al. (1991) X-Ray induced Chromatid Damage inCells from Down Syndrome and Alzheimer Disease Patients in Relation to DNArepair and Cancer Proness. Am. J. Hum. Genetics, (Suppl.) 4:61.SAS User's Guide: Statistics.5th ed. (1985) Cary, NC: SAS Institute.Selzer, B., Sherwin, I. (1983) A Comparison of Clinical Features in Early-Onset PrimaryDegenerative Dementia: One Entity or Two? Arch. Neurol., 40:143-146.Schellenberg, G.D., Bird, T.D., Wijisman, E.M., et al. (1992) Genetic Linkage Evidencefor a Familial Alzheimer's Disease Locus on Chromosome 14. Science, 258:668-671Schellenberg, G.D., Anderson, L., O'Dahl, S., et al. (1991) APP717, APP693 and PRIPgene Mutations Are Rare in Alzheimer's Disease. Am. J. Hum. Genetics, 49(3):511-517.Schoenberg, B.S., Anderson, D.W., Haerer, A.F. (1985) Severe Dementia: Prevalenceand Clinical Features in a Biracial US Population. Arch. Neurol., 42:740-743.Shalat, S.L., Seltzer, B., Pidcick, C., et al. (1986) A Case-Control Study of Medical andFamily History and Alzheimer's Disease. Am. J. Epidemiol., 124:540.126Shalat, S.L., Seltzer, B., Pidcork C., et al. (1987) Risk factors for Alzheimer's Disease.Neurol, 37:1630-1633.Silverman, J.M., Brietner, J.C.S., Mohs, R.C., et al. (1986) Reliability of the FamilyHistory Method in Genetic Studies of Alzheimer's Disease and Related Dementias.Am. J. Psychiatry, 43:1279-1282.Sjogren, T., Sjogren, H., Lindgren, A. (1952) Morbus Alzheimer and Morbus Pick:Genetic, Chemical, and Pathoanatomical Study. Acta. Psychiatr. Neurol. Scand.[Suppl] 829:1-152.Spurr, N., Gough, A., Smith C., et al. (1992) Increased Frequency of Mutations inParkinson's Disease and Certain Cancers using DNA polymorphisms of DrugMetabolising Enzymes as Markers. Am. J. Hum. Genet., 51:A25;Statistics Canada (1985) Causes of Death. Ottawa, Ont: Minister of Supply and ServicesCanada.Sturt, E. (1986) Application of Survival Analysis to the Inception of Dementia. Psychol.Med., 16:583-593.Sulkava, R., Wikstrom, J., Aromaa A., et al. (1985) Prevalence of Severe Dementia inFinland. Neurology: 1025-1029.Swearer, J. M., O'Donnell, B.F., Drachman, D.A. (1992) Neuropsychological Features ofFamilial Alzheimer's Disease. Ann. Neurol., 32:687-694.Tanzi, R.E., Gusella, J.F., Watkins, P.C., et al. (1987a) Amyloid Beta Protein Gene:cDNA , mRNA Distribution, and Genetic Linkage Near the Alzheimer Locus.Science, 235:880-884.Tanzi, RE., St George-Hyslop, P.H., Haines, J.L., et al. (1987b) The Genetic Defect inFamilial Alzheimer's Disease is not Tightly Linked to the Amyloid Beta ProteinGene. Nature, 329:156-157.127Tanzi, R.E. (1991) Gene Mutations in Inherited Amyloidpathies of the Nervous System.Am. J. Hum. Genet., 49:507-510.Tanzi, R.E., Hyman, B.T. (1991) Alzheimer's Mutations. Nature, 350:564.Teng, E.L., Chui, H.C. (1987) The Modified Mini-Mental State (3MS) Examination. J.Clin. Psychiat., 48:314-318.Terry, R.D., Katzman, R. (1983) Senile Dementia of the Alzheimer's Type. Neurology,14:497-506.Tierney, M.C., Fisher, R.H., Lewis, A.J., et al. (1988) The NINCDS-ADRDA WorkGroup Criteria for the Clinical Diagnosis of Alzheimer's Disease: AClinicopathologic Study of 57 cases. Neurology, 38:359-364.Toran-Allerand C.D. (1986) Sexual Differentiation of the Brain In: Greenough W.T.,Juraska, J.M. (eds) Developmental Neuropsychobiology. Orlando, Fla: AcademicPress, 175-211.Treves, T., Korczyn, A.D., Zilber, N., et al. (1986) Presenile Dementia in Israel. Arch.Neurol., 43:26-29.Uhlman, R.F., Larson, E.B., Buchner, D.M. (1987) Correlations of Mini-Mental Stae andModified Dementia Rating Scale to Measures of Transitional Health Status inDementia. J. Gerontol., 42:33-36.Urakami, K., Adashi, Y., Takahashi, K. (1989) A Community Based Study of ParenatalAge in Alzheimer Type Dementia in Western Japan. Arch. Neurol., 43:375.Van Broeckhoven, C., Genthe, C.A., Vandenberghe, B., et al. (1987) Failure of FamilialAlzheimer's Disease to Segregate with the A-4 Amyloid Gene in Several EuropeanFamilies Nature, 329:153-155.Van Broeckhoven, C., Backhovens, H., Cruts, M., et al. (1992) Mapping of a GenePredisposing to Early-Onset Alzheimer's Disease to 14q24.3. Nature, 2:335-339.128Van Duijn, C.M., Clayton, D., Chandra, V., et al. (1991) Familail Aggregation ofAlzheimer's Disease and Related Disorders: A Collaborative Re-Analysis of Case-Control Studies. Int. J. Epidemiol., 20:13-20.Vogel, F., Motulsky, A.G. (1979) Human Genetics. Problems and Approaches. Berlin:Springer-Verlag, 173-186.Wechsler, D. (1955) Wechsler Adult Intelligence Scale Manual. New York, New York:Psychological Corporation.Whalley, L.J., Carothers, A.D., Collyer, S., et al. (1982) A Study of Familial Factors inAlzheimer's Disease. Br. J. Psychiatry, 140:249-256.Yankner, B.A., Mesulam, M.M. (1991) B-amyloid and the pathogenesis of Alzheimer'sDisease. New England Jour. Med., 325:1849-1857.Yunginger, J.W., Reed, C.E., O'Connell, E.J., et al. (1992) A Community-BAsed Study ofthe Epidemiology of Asthma. Am.Rev. Respir. Dis., 148:888-894.Zubenko, G.S., Huff, F.,J., Beyer, J., et al. (1988) Familial Risk of Dementia Associatedwith a Biological Subtype of Alzheimer's Disease. Arch. Gen. Psychiatry 45:889-893.129Appendix AFamily History Questionnaire130MEMORYDid (does) thesubject have anyproblems with:1) memory?Don'tYES^NO^Know Date2) rememberingpeoples names? 3) recognizingfamiliar faces?4) finding way aboutindoors?5) finding way onfamiliar streets?6) remembering ashort list of items?7) did trouble with memory begin suddenly ____ or slowly _?8) has the course of the memory problems been a steady downhill progression or havethere been abrupt declines _?9) Ever see a doctorfor memoryproblems?10) If yes, what was the cause given131EXPRESSION11) Ever havetrouble finding theright word orexpressing self?12) talking becomeless over time?13) tendency todwell in the past?DAILYFUNCTIONING14) Trouble withhousehold tasks?15) Handlingmoney?16) Graspingsituations orexplanations?17) Difficulty atwork? (N/A ___)18) Trouble dressingor caring for self?19) Trouble feedingself?13220) controllingbladder and bowels?21) agitation andnervousness?OTHER PROBLEMS22) High bloodpressure?23) Stroke?24) More than 1stroke?25) Is one side ofbody weaker thanother side?26) Parkinson'sdisease? (tremors,shuffling gait,rigidity of limbs)27) Injury to thehead resulting in aloss of consciousnessfor more than asecond or two?28) Seizure or fits?13329) Syphilis?30) Diabetes?31) Drinkingproblem? (ifalcoholismsuspected explorefurther SADS Sxs)32) Did memoryproblems coincidewith drinking?33) Ever depressedor sad for twoweeks or more? (ifdepressionsuspected explorefurther SADS Sxs)34) If yes, ever seektreatment?35) Ever very high,euphoric, top of theworld?36) If yes, ever seektreatment?37) Ever seekpsychiatric orpsychological helpfor any reason?13438) If yes, everhospitalized forpsychiatric illness?39) Down'ssyndrome?40) Other medicalproblems we havenot talked about?MEDICAL CONTACTS42) Name and address of first doctor seen for problems43) Ever receivemedications?44) Ever receive aneurological orpsychiatric exam?45) Ever receive aCAT scan?46) Ever in aninstitution?(name^)47) What was the diagnosis given for the problem?135RECOGNITION48) Who was the first person to notice something wrong?49) What was noticed?50) When was the last time (the subject) seemed to be really well, or his oldself?


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



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"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items