Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Development of an epoxy-based microfluidic device for automated circulating tumour cell separation Yan, Justin Phillip 2017

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

Item Metadata

Download

Media
24-ubc_2017_september_yan_justin.pdf [ 3.04MB ]
Metadata
JSON: 24-1.0354561.json
JSON-LD: 24-1.0354561-ld.json
RDF/XML (Pretty): 24-1.0354561-rdf.xml
RDF/JSON: 24-1.0354561-rdf.json
Turtle: 24-1.0354561-turtle.txt
N-Triples: 24-1.0354561-rdf-ntriples.txt
Original Record: 24-1.0354561-source.json
Full Text
24-1.0354561-fulltext.txt
Citation
24-1.0354561.ris

Full Text

DEVELOPMENT OF AN EPOXY-BASED MICROFLUIDIC DEVICE FOR AUTOMATED CIRCULATING TUMOUR CELL SEPARATION by Justin Phillip Yan  B.A.Sc., The University of British Columbia, 2014  A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF  MASTER OF APPLIED SCIENCE in THE FACULTY OF GRADUATE AND POSTDOCTORAL STUDIES (Mechanical Engineering)  THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver)  August 2017  © Justin Phillip Yan, 2017 ii  Abstract Circulating tumour cells (CTCs) are cancer cells shed from a primary tumour site into the bloodstream, where they have the potential to invade other tissues in the body, and thus become the seed of metastases. CTCs have great potential to monitor disease progression and guide cancer treatment, but a key technical challenge for their isolation and characterization is their extreme rarity in blood. CTCs are commonly enriched using immunoaffinity, which while being highly selective, may fail to capture cells that have weak antigen expression. The biophysical properties of CTCs offer a compelling alternative to immunoenrichment. CTCs are much larger in size than erythrocytes, but are similar to leukocytes. Owing to their epithelial origin however, CTCs are likely to be more rigid than leukocytes which allows for deformability based methods to separate these cells. Previously, our group has demonstrated the continuous flow microfluidic ratchet device for deformability based separation of CTCs. Here, an improved version of the device has been developed to be compatible with pre-enrichment methods, allowing for a dramatic increase in throughput. While similar in principle to the previous version, this work specifically improves the design of the sample infusion area to increase the points of contact between the sorting matrix and sample inlet, in order to prevent the accumulation of cell debris. Using this new design, epoxy resin devices and supporting instrumentation were developed to provide a pathway towards scale-up production and automation. These combined improvements allow biology laboratory technicians to enrich CTCs without significant training. The improved device is capable of capturing > 80% CTCs from whole blood at a throughput of 1 mL/hr, which when combined with a red blood cell lysis pre-enrichment step, increases to 8 mL/hr. Finally, devices were used to enrich CTCs from patients with metastatic castration-resistant prostate cancer. CTCs were found in 3 out of 11 patients, with an average count of 78. The enriched cells were further processed to perform single cell genomic sequencing where CTCs were found to contain driver mutations including those commonly associated with prostate cancer.   iii  Lay Summary The spread of cancer from its originating site to different parts of the body, known as metastasis, is the cause of nearly all cancer related deaths. During metastasis, cancer cells are shed from the primary tumour site into the bloodstream where they have the potential to invade other tissues. These transiting cancer cells in the bloodstream, known as circulating tumour cells (CTCs), potentially harbour valuable information for the diagnosis and treatment of each patient’s cancer. A key technical challenge in acquiring information from these cells is their extreme rarity in blood. This project aims to develop a simple and automated device for separating CTCs from blood that can be used by any biology lab technician without additional training. Cells isolated from patient blood samples can then be analyzed using a variety of techniques to obtain information about each patient’s disease.   iv  Preface Sections 1.5 and 1.6 describe the microfluidic ratchet mechanism for cell sorting. The initial mechanism was developed by Dr. Hongshen Ma and Quan Guo. Early versions of a cell sorting device were developed by Sarah McFaul, with further improvements provided by Chao Jin. Here, I have specifically improved on the device design by increasing the points of contact between the sorting area and sample inlet to prevent the accumulation of cell debris.  I primarily performed all device doping experiments in Section 3.1, with additional experiments performed with the assistance of Emily S. Park. I performed all of the validation experiments on the instrumentation described in Section 3.2. All software and firmware was written and tested by me, with additional debugging assistance provided by Richard R. Ang.  The single cell isolation workflow in Section 5.1 was developed by Emily S. Park, with confocal spectrum scanning analysis in Section 5.1.3 provided by Richard R. Ang. Immunofluorescence staining, LCM, cell enumeration, and qPCR were performed by Emily S. Park. I conducted patient sample enrichment and assisted with confocal scanning. Sequencing work in Section 5.1.5 was performed by the Wyatt Prostate Genomics Laboratory at the Vancouver Prostate Centre.   v  Table of Contents Abstract ........................................................................................................................................................ ii Lay Summary ............................................................................................................................................. iii Preface ......................................................................................................................................................... iv List of Tables ............................................................................................................................................ viii List of Figures ............................................................................................................................................. ix List of Abbreviations ............................................................................................................................... xiii Acknowledgements ................................................................................................................................... xv Dedication ................................................................................................................................................. xvi 1 Introduction .............................................................................................................................................. 1 1.1 Motivation ........................................................................................................................................... 1 1.2 Performance Metrics ........................................................................................................................... 2 1.3 Biophysical Properties of Cells ........................................................................................................... 3 1.4 Pre-enrichment Methods ..................................................................................................................... 3 1.5 Principle of the Microfluidic Ratchet .................................................................................................. 5 1.6 Microfluidic Ratchet Deformability Based Separation ....................................................................... 8 1.7 Goals of the Thesis ............................................................................................................................ 10 2 Background ............................................................................................................................................ 11 2.1 Competing Technology ..................................................................................................................... 11 2.1.1 ANGLE plc ................................................................................................................................ 11 2.1.2 ApoCell ...................................................................................................................................... 12 2.1.3 Clearbridge BioMedics .............................................................................................................. 14 2.1.4 Rarecyte ..................................................................................................................................... 15 2.1.5 ScreenCell .................................................................................................................................. 16 2.1.6 Comparison ................................................................................................................................ 17 3 Design & Validation ............................................................................................................................... 19 3.1 Device Design & Validation ............................................................................................................. 19 vi  3.1.1 Prevention of Cell Debris Accumulation ................................................................................... 19 3.1.2 Faster Filling Protocol ................................................................................................................ 23 3.1.3 Device Validation ...................................................................................................................... 25 3.1.4 Epoxy Based Device .................................................................................................................. 28 3.1.5 Autoclave-ability ........................................................................................................................ 30 3.2 Instrumentation Design & Validation ............................................................................................... 31 3.2.1 Mechanical ................................................................................................................................. 31 3.2.2 Electrical .................................................................................................................................... 34 3.2.3 Software ..................................................................................................................................... 37 3.2.4 System Validation ...................................................................................................................... 39 4 Methods ................................................................................................................................................... 43 4.1 Fabrication ........................................................................................................................................ 43 4.1.1 Fabrication of Silicon Masters ................................................................................................... 43 4.1.2 Fabrication of PDMS Masters .................................................................................................... 44 4.1.3 Fabrication of Polyurethane Moulds .......................................................................................... 44 4.1.4 Fabrication of PDMS Devices ................................................................................................... 45 4.1.5 Fabrication of Silicone Moulds .................................................................................................. 45 4.1.6 Fabrication of Resin Devices ..................................................................................................... 47 4.2 Experimental Setup ........................................................................................................................... 47 4.2.1 Sample Preparation .................................................................................................................... 47 4.2.2 Doping Experiments .................................................................................................................. 47 5 Separation of CTCs from Patients with mCRPC................................................................................ 49 5.1 Preparation and Analysis of Patient Samples.................................................................................... 50 5.1.1 Sample Collection and Preparation ............................................................................................ 50 5.1.2 CTC Separation and Immunofluorescence Staining .................................................................. 51 5.1.3 CTC Enumeration & Single Cell Identification ......................................................................... 52 5.1.4 Single Cell Extraction ................................................................................................................ 54 vii  5.1.5 Genomic Sequencing ................................................................................................................. 54 5.2 Results from mCRPC Patients .......................................................................................................... 55 5.3 Discussion ......................................................................................................................................... 57 6 Conclusion .............................................................................................................................................. 58 6.1 Summary of Results .......................................................................................................................... 58 6.2 Limitations ........................................................................................................................................ 58 6.3 Future Work ...................................................................................................................................... 58 References .................................................................................................................................................. 59    viii  List of Tables Table 2.1 Comparison of the performance of five different commercial CTC enrichment technologies ... 18 Table 3.1 Hydrodynamic resistances of the modified flow channels in the microfluidic ratchet device.... 23 Table 3.2 Experimental dry run control channel pressure readings ............................................................ 42    ix  List of Figures Figure 1.1 Average yield of five different pre-enrichment methods tested with cultured cancer cells. ....... 5 Figure 1.2 3D representation of a cell deforming through a funnel constriction. Pint denotes the internal cell pressure while ......................................................................................................................................... 5 Figure 1.3 Top-down view A) and cross-sectional view B) of a cell deforming through a microscale funnel constriction. ....................................................................................................................................... 6 Figure 1.4 Schematic of the geometry and forces acting on a curved membrane. A) Leading edge membrane. B) Trailing edge membrane. ...................................................................................................... 6 Figure 1.5 Deformation of a cell through a microscale funnel constriction in the forward A) and backward B) directions. ................................................................................................................................................. 8 Figure 1.6 Operation of a continuous flow microfluidic ratchet device. Cells are introduced at the bottom left and travel in a staircase diagonal path as the cells are oscillated forward and backward until they reach a funnel through which they can longer deform. Erythrocytes (red), leukocytes (blue), and CTCs (green) each follow a distinct path as they transit through the device. ...................................................................... 9 Figure 2.1 A) Schematic of the disposable cassette. B) Photograph of the Parsorter PR1 fluidic processor. C) Graphical representation of the stepped structure inside the cassette and the mechanism for cell sorting. (Adapted from Xu, L. et al.30 under CC-BY 4.0)........................................................................................ 12 Figure 2.2 A) Image of the ApoStream® instrument. B) 3D CAD model of the flow chamber. C) Overview of the enrichment of CTCs using DEP and buffer flow inside the flow chamber. (Adapted from Balasubramanian, P. et al.33 under CC-BY 4.0). ........................................................................................ 13 Figure 2.3 A) Photograph of the triple stacked microfluidic spiral CTChip® with two inlets and outlets. Fluid is pumped through the device using two separate syringe pumps. B) Illustrations of microchannel cross-sections which show the separation of whole blood (A-A) and focusing of CTCs (B-B) as a result of the Dean drag force and inertial lift force. (Adapted from Khoo, B.L et al.36 under CC-BY 4.0). ............ 14 Figure 2.4 A) Photograph of the separation tube after the initial centrifugation showing the split blood layers: plasma, buffy coat, and erythrocytes. B) Transfer tube after the final centrifugation with the buffy coat (highlighted) ready for extraction. C) The various components of the Accucyte® system; from left to right: the float, fully assembled system, EpiCollector®, transfer tube, and locking clip. (Adapted from Campton, D.E. et al.38 under CC-BY 4.0). ................................................................................................. 15 Figure 2.5 Overview of the ScreenCell® device which is comprised from top to bottom of: a filtration tank, filter, nozzle/holder, and waste container. The overall process to enrich CTCs is illustrated from blood collection to filter retrieval. (Adapted from Mu, Z. et al.44 under CC-BY 4.0). ............................... 17 Figure 3.1 Sample inlet region of the microfluidic ratchet device developed by Jin, C. et al.20 filled with cell debris. ................................................................................................................................................... 19 x  Figure 3.2 Accumulation of cell debris in a A) standard microfluidic ratchet device as compared to using a vibration motor with B) continuous, C) 10 Hz, and D) 1 Hz vibration waveforms applied for 10 seconds every 60 seconds. ........................................................................................................................................ 20 Figure 3.3 Representative image of the bifurcated sample inlet microfluidic device, showing the WBC and CTC outlets, forward and backward oscillation channel inputs, cross buffer flow input, and sample inlet. ............................................................................................................................................................ 21 Figure 3.4 Visualization of the flow (m/s) in the sample inlet as computed using COMSOL. .................. 22 Figure 3.5 Time required to fill a microfluidic ratchet device with PBS using different control channel configurations. ............................................................................................................................................ 24 Figure 3.6 Flow visualization of a bifurcated sample inlet microfluidic ratchet device using PBS dyed with food colour. ......................................................................................................................................... 25 Figure 3.7 Panoramic scan of the sample inlet area of a bifurcated sample inlet microfluidic ratchet device under a 4x objective showing A) fresh blood control with no accumulation and B) 3-day old blood with minimal cell accumulation. ......................................................................................................................... 26 Figure 3.8 Average yield of the base bifurcated sample inlet device as compared to the previous iteration microfluidic ratchet device developed by Jin, C. et al.20. ........................................................................... 27 Figure 3.9 Average yield of whole blood pre-enriched with G-Biosciences lysis buffer as compared to whole blood both processed through the bifurcated sample inlet microfluidic ratchet device. .................. 28 Figure 3.10 Disposable cartridge comprised of A) fluid reservoirs (top side) and B) microchannels (bottom side). .............................................................................................................................................. 29 Figure 3.11 Average yield of baseline PDMS microfluidic ratchet devices as compared to epoxy resin microfluidic ratchet devices. ....................................................................................................................... 29 Figure 3.12 Average yield of an autoclaved microfluidic device as compared to the base bifurcated sample inlet device. ..................................................................................................................................... 30 Figure 3.13 Completed support instrumentation for use with epoxy resin based microfluidic ratchet devices. ....................................................................................................................................................... 31 Figure 3.14 Lead screw assembly showing the pressure cap connected to the lead screw drive assembly via the lead nut with support provided by a linear guide. ........................................................................... 32 Figure 3.15 3D printed pressure cap with push-to-connect tube fittings and built in o-rings used to seal against epoxy devices. ................................................................................................................................ 33 Figure 3.16 High level pneumatic circuit diagram showing the major connections between the input pressure source and the output control solenoids. ....................................................................................... 33 Figure 3.17 Top view of the assembled custom PCB which contains the MCU and various support electronics used to control external peripherals connected via removable connectors. .............................. 34 xi  Figure 3.18 High level PCB overview showcasing the major components. ............................................... 36 Figure 3.19 Overview of the user interface broke up into 5 major sections: A) system status, B) digital IO control and feedback, C) settings, D) experimental buttons, and E) pressure clamp controls. ................... 38 Figure 3.20 Snapshot of a single cycle during the repeatability test of the pressure clamp system............ 40 Figure 3.21 Repeatability test of the pressure clamp system run till failure after ~230 cycles or ~4hrs. ... 41 Figure 3.22 Snapshot of the transition between the filling and running phases during the dry run test. .... 42 Figure 4.1 Overview of the fabrication process of silicon master. ............................................................. 44 Figure 4.2 Fabrication process of PDMS devices. ...................................................................................... 45 Figure 4.3 Silanization of a PDMS master. ................................................................................................ 46 Figure 4.4 Completed mould for epoxy resin devices made of two halves: the microchannel (bottom left) and reservoir halves (top and bottom right). The reservoir half is further split in two for easier removal of epoxy resin devices ..................................................................................................................................... 46 Figure 4.5 Pressure board with analog gauges and manual pressure regulators used for the control of PDMS microfluidic ratchet devices. ........................................................................................................... 48 Figure 5.1 Overview of the work flow for the separation and single cell analysis of CTCs from patient samples. ....................................................................................................................................................... 49 Figure 5.2 Representation of the gigapixel spectral image cube of patient VCCCTC023 comprised of 26 channels each 7680x7680. Channel peak emissions are listed in addition to the transmitted light quality control. ........................................................................................................................................................ 52 Figure 5.3 Idealized emission spectra of Hoechst (blue), EpCAM (green), and CD45 (red) used to identify CTCs and leukocytes. Cells which are Hoechst+, EpCAM+, and CD45- are classified as CTCs while cells that are Hoechst+, EpCAM-, and CD45+ are categorized as leukocytes...................................................... 53 Figure 5.4 Review software showing the results of a single well from patient VCCCTC023 overlaid with a markup to highlight key areas of the software. Regions of interest include the cell spectral graph, cell subimage, gel view, full resolution image, and selected cell coordinate list. ............................................. 53 Figure 5.5 Overview of the single cell selection process as validated using UM-UC13 cells. Steps include: A) cell identification, B) membrane cutting, and C) catapulting the excised sample to the D) collection tube cap. The cell is only visible under the E) EGFP filter and missing with the F) mCherry filter, indicating the retrieval of the correct target cell. ........................................................................................ 54 Figure 5.6  CTC enumeration values for mCRPC patients from 2 mL (VCCCTC010) and 8 mL (VCCCTC018 and VCCCTC023) of whole blood enriched using the bifurcated sample inlet microfluidic ratchet device. ............................................................................................................................................. 55 xii  Figure 5.7 Mutations detected through targeted next generation sequencing suggests the heterogeneity of CTCs. When compared to cfDNA, 3 mutations (TP53, PTEN, and FOXA1) are shared, while 20 mutations are only found only in CTCs. ..................................................................................................... 56    xiii  List of Abbreviations ADC  Analog-to-Digital Converter BLDC  Brushless DC Motor BUF  Buffer BWD  Backward CAD  Computer Aided Design CE  Conformité Européenne cfDNA  Cell-Free DNA CK  Cytokeratin CTC  Circulating Tumour Cell DAC  Digital-to-Analog Converter DEP  Dielectrophoresis EMT  Epithelial- Mesenchymal Transition EpCAM Epithelial Cell Adhesion Molecule FBS  Fetal Bovine Serum FDA  Food and Drug Administration FEA  Finite Element Analysis FWD  Forward I2C  Inter-Integrated Circuit IO  Input/Output IR  Infrared LCM  Laser Capture Microdissection mCRPC Metastatic Castration-Resistant Prostate Cancer MCU  Microcontroller Unit MOSFET Metal-Oxide-Semiconductor Field-Effect Transistor N/C  Nuclear to Cytoplasmic Ratio OD  Outer Diameter PBMC  Peripheral Blood Mononuclear Cell PBS  Phosphate-Buffered Saline PCB  Printed Circuit Board PDMS  Polydimethylsiloxane  PEN  Polyethylene Naphthalate PWM  Pulse Width Modulation qPCR  Quantitative Polymerase Chain Reaction xiv  RBC  Red Blood Cell SPL  Sample UART  Universal Asynchronous Receiver/Transmitter UI  User Interface USCI  Universal Serial Communication Interface UV  Ultraviolet WBC  White Blood Cell WGA  Whole Genome Amplification  xv  Acknowledgements I would like to thank all of my friends, family, and colleagues for their support in the pursuit of this project. In particular I would like to thank my supervisor, Dr. Hongshen Ma for his mentorship in providing new perspectives and ideas, and exposing me to a wide variety of projects. I would also like to thank all of the members of the Multi-Scale Design Laboratory for their assistance throughout my time as a part of the group; you have all been wonderful colleagues and friends.   xvi  Dedication      To Angela, For your continued support,  through everything   1  1 Introduction 1.1 Motivation When first discovered in 1869 by Thomas Ashworth, circulating tumour cells (CTCs) were described as a small number of cells in a patient’s blood resembling the cells of a primary tumour1. In recent years, the separation and analysis of CTCs has gained significant momentum in cancer research and treatment. The significance of CTCs comes from the hypothesis that CTCs are the seeds of metastases that are shed from the primary tumour and travel through the vasculature to create secondary tumours2. This hypothesis is supported, in part, by the concordance between the number of CTCs detected and the survival of patients with metastatic carcinomas3,4. Despite their potential utility, one key limitation in the separation and analysis of CTCs is their extreme rarity in the bloodstream5,6. The typical concentration is 1-10 CTCs per 1 mL of blood, or in terms of rarity to other blood cell types there is 1 CTC in 109 erythrocytes and 1 CTC in 107 leukocytes. Erythrocytes can be easily differentiated from both leukocytes and CTCs based on their biochemical and biophysical properties and are thus easy to separate. The major difficulty in CTC isolation thus comes from the biochemical and biophysical similarities between CTCs and leukocytes. Therefore, there is a need to effectively separate CTCs from leukocytes with a high level of discrimination.  Methods for separating CTCs from leukocytes and erythrocytes can be split into two major categories: biochemical and biophysical. Biochemical methods typically rely on a combination of affinity chromatography of cell membrane antigens to first enrich for potential candidate cells followed by fluorescent staining of intracellular markers for positive candidate cell identification. The Janssen Diagnostics CellSearch® system (Raritan, NJ, USA) is an example of a commercially available system which has received Food and Drug Administration (FDA) approval for clinical enumeration of CTCs using the biochemical approach. Firstly, the CellSearch® system enriches for candidate CTCs using the cell membrane antigen, epithelial cell adhesion molecule (EpCAM) which is a transmembrane glycoprotein that mediates homophillic cell-cell adhesion and is expressed by most epithelial tissues7. CTCs are then positively identified by fluorescent staining the proteins that make up the intracytoplasmic cytoskeleton found in epithelial tissue, which are known as cytokerations (CK). Lastly, leukocytes are then excluded by fluorescent staining for CD45 which is a receptor linked protein tyrosine phosphatase that is expressed by all leukocytes. One major limitation of the biochemical approach is that a subpopulation of CTCs will undergo an epithelial-mesenchymal transition (EMT) which results in a lower expression of epithelial markers like EpCAM and CK8. EMT is a biological process that allows epithelial cells to undergo biochemical changes to become mesenchymal cells which have increased mobility, 2  invasiveness, and apoptosis resistance9.  Consequently, this subpopulation may be impossible to isolate using systems based on EpCAM capture.  In contrast with biochemical methods, biophysical methods rely on the differences in cell morphology and other physical properties such as deformability, conductivity, magnetic susceptibility, and electrical polarizability. Since biophysical methods are label free, they are able to solve the problem present in biochemical methods where a lower epithelial marker expression caused by EMT prevents the capture of a particular cell subpopulation. In addition, biophysical isolation can often allow cells to retain their viability, enabling them to be compatible with downstream processes which require viable cells. A review of current technologies for the biophysical separation of CTCs follows in Chapter 2.  Chapter 1, Section 1.2 will first present the performance metrics used to compare and evaluate different label free separation techniques. Section 1.3 will go over the biophysical and biomechanical properties of CTCs. Section 1.4 will discuss various types of pre-enrichment methods and the selection of a suitable method. Section 1.5 will then review the working principle of the microfluidic ratchet for use with deformability based cell sorting. Finally, Section 1.6 will introduce continuous flow cell separation utilizing the microfluidic ratchet, while Section 1.7 presents the goals of the thesis.  1.2 Performance Metrics In order to compare and evaluate the performance of different CTC separation methods the metrics with which they will be compared by must first be defined. The yield of a separation method is the fraction of target cells captured at the output relative to the number of target cells that went into to it (Equation (1.1)). The yield is a particularly important metric for cell enumeration as it can be used to determine the initial concentration of target cells in the patient bloodstream. It is also sometimes referred to as the recovery rate or capture efficiency. The throughput describes the quantity of sample that a system is capable of processing per unit of time (Equation (1.2)). It is commonly expressed as either a volumetric flow rate or in terms of number of cells per unit of time.                                             (1.1)                                                           (1.2)  3  1.3 Biophysical Properties of Cells CTCs were historically identified by examining blood from metastatic cancer patients under a microscope. When first discovered in 1869, Ashworth described CTCs as cells that closely resembled those of the original tumour1. In the decades following Ashworth's discovery there were numerous studies run by Schleip, Aschoff, Marcus, Quensel, Pool, and Ward that reported on abnormal cells in the blood of a variety of cancer patients10–15. However, it wasn’t until George Papanicolaou’s development of the Papanicolaou stain, or Pap smear, and its subsequent widespread adoption in the late 1940s to 1950s that allowed CTCs to be identified with a much greater certainty16. With a new and more definitive method for identifying CTCs, a number of studies were conducted by Roberts, Grove, and Engell17–19. In these studies, erythrocytes and leukocytes were depleted by lysis or density gradient centrifugation, with the remaining cells being fixed and stained on a slide using the Papanicolaou method. Slides were then examined by experienced cytologists who identified CTCs by applying a set of criteria for malignancy which included the nuclear to cytoplasmic ratio (N/C), and variation in nuclear shape and size. The result of these historical studies demonstrated the key differences between CTCs and leukocytes in their size, density, and internal structure; properties that can be utilized in a label free separation method.  While CTCs have been successfully separated from other cells types based on their size alone, various studies have reported on the variation in CTC cell sizes ranging from 10-25 µm20,21. This large variation in size, in addition to the overlap in size with other cell types such as leukocytes, may inhibit the efficacy of cell separation and create other performance issues like clogging in size-only based enrichment methods. While deformability based methods are typically less susceptible to these issues, the data on the deformability of CTCs is limited. Deformability measurements require viable cells and the majority of enrichment methods fix and stain cells which results in a loss of viability. However, it is possible to infer deformability from the N/C ratio of a cell, where a larger N/C ratio indicates a less deformable cell. Studies have shown that high N/C ratios are correlated to poor outcomes and have a large intra-patient variability22,23. The differences in N/C ratios and the subsequent deformability of different cell types has been utilized by our group in the continual development of deformability based CTC separation20,24  1.4 Pre-enrichment Methods In order to increase the effective throughput of a device, other commercial systems typically include a pre-enrichment step to reduce the load on the sorting system. Erythrocytes make up the bulk of whole blood, so pre-enrichment methods commonly target erythrocytes for depletion. Common methods of pre-enrichment include cell lysis, density gradient centrifugation, and immunomagnetic depletion. A brief overview of each method is described herein. 4  Firstly, cell lysis uses a solution of salts to create a hypotonic environment which causes erythrocytes to burst due to an osmotic unbalance. Lysis buffer is added to whole blood and allowed to incubate for a set time. Following incubation, a sample is then centrifuged and the supernatant containing leukocytes and CTCs is aspirated. Density gradient centrifugation employs the use of specialized media that has a tuned density to separate out desired cells into bands after centrifugation. To separate cells, media is first added to whole blood and is then precisely layered over the density media prior to centrifugation. Cells are retrieved by removing the desired layer. This process requires a skilled hand to correctly aspirate all available cells. Finally, immunomagnetic depletion uses an antibody coated paramagnet which binds to specific antigens present on target cells. Antibodies are added to a blood cell sample and set to incubate for a set time. After incubation a sample is then placed inside of a permanent magnet which will draw target cells towards it, leaving all other cell types behind.  Previously, different pre-enrichment methods were tested with cultured cancer cells (Section 4.2.1), where a known concentration was spiked into whole blood and then subjected to a particular pre-enrichment method. As with the yield of a separation method, the yield of each pre-enrichment method is given by Equation (1.1). A lower pre-enrichment yield would result in fewer cells proceeding towards the cell separation step, where even fewer cells could be reclaimed at the end of the separation process. Figure 1.1 below shows the resulting yields of various pre-enrichment methods. EasySep is an immunomagnetic based method, Histopaque 1119 and RosetteSep are both density gradient based methods, and ammonium chloride and G-Biosciences are both lysis based methods. Histopaque 1119 and G-Biosciences perform the best with average yields of 77% and 74% each. While the intended use of these methods is to perform a single pre-enrichment, both lysis methods required double pre-enrichment to remove a sufficient red cell population to allow cultured cancer cells to be counted under microscope. As a result the yield from a single pre-enrichment should be greater than or equal to that of the twice pre-enriched sample. Therefore, the cell lysis method using the buffer from G-Biosciences was selected over the Histopaque 1119 density gradient method as its twice pre-enriched yield was nearly identical to that of the single pre-enriched Histopaque 1119. In addition, the workflow for the cell lysis only adds an additional 10 minutes to the overall enrichment process, while the density gradient separation requires 30 minutes.  5  EasySepHistopaque 1119RosetteSepAmmonium ChlorideG-Biosciences020406080100Yield (%) Figure 1.1 Average yield of five different pre-enrichment methods tested with cultured cancer cells.  1.5 Principle of the Microfluidic Ratchet The microfluidic ratchet mechanism previously demonstrated by Guo, Q. et al. uses the deformability of cells to irreversibly transport individual cells through a series of microscale funnel constrictions25. The tapered, or funnel shaped, constriction requires a smaller threshold pressure to transport the cell forward (Figure 1.2) through the constriction, as compared to backwards through the constriction. Using an oscillatory flow, cells can therefore be irreversibly transported along the direction of the constriction.   Figure 1.2 3D representation of a cell deforming through a funnel constriction. Pint denotes the internal cell pressure while  P1 and P2 denote the pressure at the leading and trailing edge, respectively. 6  To understand the deformation of a cell through a microscale funnel constriction, the forces required to transport a cell through the constriction are modelled using the Young-Laplace law and a solid liquid drop model26,27. In this model a cell is defined as a Newtonian fluid enclosed by a membrane with a constant and isotropic cortical tension Tc. As the cell deforms through a funnel constriction with a rectangular cross-section, it can be split up into two components: the leading edge, and the trailing edge. The internal pressure of the cell is denoted as Pint, while the outer pressures are denoted as P1 and P2 for the leading and trailing edges, respectively (Figure 1.2). As shown in Figure 1.3, each edge of the membrane has two radii of curvature, one which is constrained by the geometry of the funnel, and one which is not. Rc1 and Rc2 refer to the unconstrained radii, whereas Ra and Rb refer to the constrained radii.   Figure 1.3 Top-down view A) and cross-sectional view B) of a cell deforming through a microscale funnel constriction.   Figure 1.4 Schematic of the geometry and forces acting on a curved membrane. A) Leading edge membrane. B) Trailing edge membrane.  For the leading edge case, shown in Figure 1.4A, the normal components of the membrane tension forces FTc1 and FTa are equal to the force from the difference in pressure on either side of the membrane. 7                                  (1.3)  The membrane tension forces are then defined as the cortical tension TC multiplied by the length of membrane over which it acts.             (1.4)             (1.5)  Substituting Equations (1.4) and (1.5) into Equation (1.3) gives:                                  (1.6)  In addition, the membrane geometry yields:              (1.7)            (1.8)  Substituting Equations (1.7) and (1.8) into Equation (1.6) and dividing both sides by a factor of ac1 gives:                        (1.9)  Similarly, using the same method for the trailing edge case yields:                        (1.10)  To determine the relationship between the pressure difference and the cortical tension for the whole cell Equation (1.9) is subtracted from Equation (1.10) to give:                               (1.11)  The difference between Rc1 and Rc2 is assumed to be negligible and can be approximated as being equal, which reduces Equation (1.11) to the final form of:                (1.12)  8  As a cell deforms through the constriction, the pressure given by Equation (1.12) reaches a maximum. At this point the cell is rapidly pulled through the constriction in a phenomenon known as the Haines' jump instability28.  When the direction of flow is reversed to the backward case as shown in Figure 1.5B, the trailing edge radius Rb is much greater than the leading edge radius Ra. Following from Equation (1.12), the pressure difference required to transport the cell through the funnel must be larger than that of the forward case. Therefore, the application of an oscillating pressure enables the selective and irreversible transport of cells in a single direction.   Figure 1.5 Deformation of a cell through a microscale funnel constriction in the forward A) and backward B) directions.  1.6 Microfluidic Ratchet Deformability Based Separation Utilizing the microfluidic ratchet mechanism, a cell sorting device was developed by McFaul, S.M. et al. for batch processing of samples24. In this initial iteration of the device, the sorting area was comprised of a funnel matrix, where the pore size of funnels in each row was constant, but incrementally decreased going from bottom to top. Samples were run by first infusing cells into the sorting area from the inlet, and then applying an oscillatory flow to transport the cells through the funnels as described in Section 1.5. The more deformable cells would then be transported to the upper rows while the less deformable cells would be retained in the lower rows. Cells were then collected at the outlets and a new sample was infused to begin the cell sorting process once again. The use of the microfluidic ratchet allowed for a higher selectivity when compared to other sorting methods. However the two major disadvantages of the initial iteration were the extremely low throughput, 10,000 leukocytes per hour, and the complexity of using a control layer in device manufacture, meant that it would not be able to sufficiently handle the blood volume required in a clinical setting. 9  The next major iteration of the deformability based cell sorting device increased the throughput and reduced the complexity of the device. Developed by Jin, C. et al., the device retained the idea of the funnel matrix while increasing the number of columns and rows20. Instead of infusing a sample and then sorting, cells were continuously infused from the bottom left corner, with a horizontal buffer flow pushing the cells from left to right. In addition, an oscillatory flow was applied perpendicular to the buffer flow to transport the cells through the funnel matrix. Thus, the path of a cell through the device resembles that of a staircase as it makes its journey from inlet to outlet (Figure 1.6). Finally, cells were collected at two separate outlets dubbed the CTC and white blood cell (WBC) outlets, named by the expected cell type in each outlet. While the throughput of this iteration was greatly improved over the previous version at 1 mL/hr of whole blood, an improved version could still increase the throughput to better suit clinical applications. The use of a pre-enrichment method would then become the next logical step; however the use of a lysis buffer creates cell debris which can cause the device to become clogged and fail. Furthermore, the initial experimental setup and subsequent supervision requires a skilled technician which may not be available in a clinical setting.   Figure 1.6 Operation of a continuous flow microfluidic ratchet device. Cells are introduced at the bottom left and travel in a staircase diagonal path as the cells are oscillated forward and backward until they reach a funnel through which they can longer deform. Erythrocytes (red), leukocytes (blue), and CTCs (green) each follow a distinct path as they transit through the device.    10  1.7 Goals of the Thesis The goals of this thesis are to improve on the deficiencies identified in the previous iteration developed by Jin, C. et al., while continuing to meet the primary objective of separating CTCs from blood using the microfluidic ratchet mechanism. In particular the three areas of improvement are to:  Prevent the accumulation of cell debris at the sample inlet causing the device to clog and fail  Increase the throughput of blood to allow a clinically relevant sample size to be processed in a meaningful time frame  Increase automation to reduce the amount of supervision and technical skill required to successfully separate CTCs 11  2 Background 2.1 Competing Technology The ultimate goal of most academic research is the widespread adoption of their work, which typically takes the form of commercialization. While there is currently only a single company, Janssen Diagnostics (formerly known as Veridex), that offers an FDA approved method for CTC enrichment, various other companies are currently working on new technologies that in the future may become available for clinical use. As mentioned previously (Section 1.1), the Janssen Diagnostics CellSearch® method uses a biochemical approach to CTC enrichment and may therefore miss more aggressive cancer cells that have gone through EMT. Indeed, a study performed by our group showed that on average the biophysical microfluidic ratchet captured 175 CTCs per 7.5 mL whereas the biochemical CellSearch® system captured only 7 CTCs per 7.5 mL20. In addition to our group, various companies and research groups are exploring different biophysical CTC enrichment methods for use in the research and clinical space. Sections 2.1.1 to 2.1.5 will discuss the current developments of biophysical CTC enrichment technologies by different research groups and companies, and Section 2.1.6 will then summarize and compare the currently available technologies.  2.1.1 ANGLE plc ANGLE plc is a medical diagnostics company based in the United Kingdom whose flagship product the Parsortix™ cell separation system has attained a Conformité Européenne (CE) mark for regulatory approval in the clinical market in Europe and is currently seeking FDA approval in the United States. The Parsortix™ system is comprised of the disposable Cell Separation Cassette GEN3D10 (Figure 2.1A) which uses a size and deformability based microfluidic sorting scheme, and the accompanying fluidic processor the Parsorter PR1 (Figure 2.1B).  The GEN3D10 cassette is made of a plastic moulding which contains a stepped structure used to separate cells, and is enclosed with a second heat-bonded plastic cover. The gap between the stepped structure and the cover forms a critical separation gap that enables cell separation. Cells that are smaller and less rigid will be able to pass through the gap whereas CTCs which are larger and more rigid will become lodged. The stepped structure is folded in a serpentine structure across the width of the cassette to increase the active sorting area. Blood is infused from a single inlet and flows forward across the stepped structure to be sorted where all waste is collected in a single outlet. A backward flow is then applied to release the captured CTCs for collection (Figure 2.1C)29.  12  Prostate cancer cells lines PC3 and DU145, colorectal adenocarcinoma cell line HT29, and breast cancer cell line MCF-7 were used to characterize the GEN3D10 cassette and achieved an average yield of 54.4%, 56.6%, 69.0%, and 56.7%, respectively29,30. Further characterization was performed using 4 additional cell lines which gave a worst case average yield of 42% for the bladder carcinoma cell line T24, and best case average yield of 70% for the pancreas ductal adenocarcinoma cell line PANC-131. The use of a buffy coat (density centrifugation) pre-enrichment step allows for the enrichment of an equivalent 7.5 mL of whole blood in 84 minutes; otherwise a 10 mL whole blood sample is processed in 2.5 hours30. The Parsortix™ system was also used in various studies to enrich CTCs from metastatic and non-metastatic cancer patients, finding between 20-1474 CTCs in a study of small cell lung cancer patients29. Additionally, when compared to the FDA approved CellSearch® method, the Parsortix™ system statistically performed similar or better than CellSearch®29–31.    Figure 2.1 A) Schematic of the disposable cassette. B) Photograph of the Parsorter PR1 fluidic processor. C) Graphical representation of the stepped structure inside the cassette and the mechanism for cell sorting. (Adapted from Xu, L. et al.30 under CC-BY 4.0).  2.1.2 ApoCell ApoCell is a biomarker company based in Houston, Texas, operating clinical and research laboratories to enrich rare cell populations, including CTCs, using standard and proprietary technology. ApoCell’s proprietary CTC enrichment system, the ApoStream® (Figure 2.2A), uses dielectrophoresis (DEP) in combination with a microfluidic flow to separate CTCs from blood. This separation process occurs inside of a reusable flow chamber (Figure 2.2B). Currently the ApoStream® is only available for research use.  The isolation of CTCs from other cell types relies on the exploitation of the differences in dielectric properties of the different cell types. The balance of DEP, sedimentation, and hydrodynamic forces are used to position cells in a hydrodynamic flow profile in a method known as field-flow fractionation. In the ApoStream® system, cancer cells are attracted to the electrode plane by positive DEP forces while the 13  remaining blood cells are repelled away by negative DEP forces (Figure 2.2C). This is achieved by applying an AC voltage at a frequency between the crossover frequencies of CTCs and the peripheral blood mononuclear cell (PBMC) fraction. The crossover frequency is defined as the frequency where DEP forces transition from a negative to a positive force. CTCs that pass through the flow chamber are pulled by DEP forces and travel along the chamber floor where they go through a collection port for later retrieval, while all other cells are repelled and exit through a waste port32.  Ovarian cancer cell line SKOV3 and breast cancer cell line MDA-MB-231 were used to characterize the ApoStream® system in a series of doping experiments. Cells were spiked into the PBMC fraction obtained from 7.5 mL of Ficoll density gradient enriched whole blood. The recovery rate of cultured cancer cells was found to be 75.4% and 71.2% for SKOV3 and MDA-MB-231, respectively32. Additional characterization was performed by spiking 1000 or 50 cultured cancer cells per 1 mL of enriched PBMCs using the A549 cancer lung, ASPS-1 sarcoma, and MDA-MB-231 breast cancer cell lines. Average recovery rates across all three cell lines were 67.5% and 54.7% for the 1000 and 50 CTC spiking levels33. Total runtime to process samples in the ApoStream® system was approximately 60 minutes32. CTCs from patients with breast, ovarian, non-small cell lung adenocarcinoma, and non-small cell lung squamous cancers were enriched using the ApoStream® system finding average CTC counts of 9.2, 2.4, 139.4, and 2.0, respectively. The same study also showed that the ApoStream® system was able to enrich significantly more CTCs as compared to the FDA approved CellSearch® system34.   Figure 2.2 A) Image of the ApoStream® instrument. B) 3D CAD model of the flow chamber. C) Overview of the enrichment of CTCs using DEP and buffer flow inside the flow chamber. (Adapted from Balasubramanian, P. et al.33 under CC-BY 4.0). 14  2.1.3 Clearbridge BioMedics Clearbridge BioMedics is a Singapore based company that has developed the CTChip® FR and ClearCell® FX system to separate CTCs using inertial microfluidics. They originally began as a spin-off of the National University of Singapore as part of the Clearbridge accelerator, a high-technology incubator funded by the Singapore Government’s National Research Foundation. They are currently operating mainly in Asia, selling systems through distributors in Korea, Japan, China, and Australia.  The CTChip® FR utilizes inertial hydrodynamics to separate CTCs from background cells. Devices utilize a triple stacked spiralling microchannel (Figure 2.3A) with a rectangle cross section, where centrifugal forces in the radial direction generate two counter-rotating symmetric vortices commonly known as Dean vortices perpendicular to the primary flow direction. The vortices generate a drag force which drives cells along the vortices in a cyclic manner every characteristic length of microchannel, which is a function of the channel width and height. This translates to cells moving back and forth between the inner and outer microchannel walls, in a process referred to as Dean migration. The primary flow along the length of the microchannel is a Poiseuille flow, which has a parabolic flow velocity and exerts a shear-induced inertial lift force on the cells forcing them away from the central streamline towards the microchannel walls. The inertial lift force has a greater effect on cells which are similar in size to the microchannel dimensions. The combination of Dean migration and inertial focusing creates only two equilibrium positions, and allows for the precise separation of CTCs from erythrocytes and leukocytes (Figure 2.3B)35.    Figure 2.3 A) Photograph of the triple stacked microfluidic spiral CTChip® with two inlets and outlets. Fluid is pumped through the device using two separate syringe pumps. B) Illustrations of microchannel cross-sections which show the separation of whole blood (A-A) and focusing of CTCs (B-B) as a result of the Dean drag force and inertial lift force. (Adapted from Khoo, B.L et al.36 under CC-BY 4.0). 15  Whole blood is first pre-enriched using an ammonium chloride red blood cell (RBC) lysis buffer and concentrated by 2X in a process that requires 10 minutes. Breast adenocarcinoma (MCF-7) and bladder (T24) human cancer cells lines were used to characterize the CTChip® FR, which achieved a yield of 87.6% and 76.4%, respectively at a rate of 7.5 mL/10min37. Additionally, the CTChip® FR was validated clinically with CTCs enriched from patients (n=58) with metastatic breast and non-small cell lung cancer. CTCs were enriched in the ranges of 12-1275 CTCs/mL (median: 55 CTCs/mL) and 10-1535 CTCs/mL (median: 82 CTCs/mL), respectively36.  2.1.4 Rarecyte The Seattle, Washington based company Rarecyte developed the Accucyte® separation system for cell sorting which uses a multi-staged density centrifugation to progressively separate cells. The multi-staged centrifugation removes the need for a skilled user with precise pipetting abilities to isolate the buffy coat. The Accucyte® system is comprised of a separation tube, plastic cylindrical float, CyteSeal® sealing ring, transfer tube, and EpiCollector® (Figure 2.4C).    Figure 2.4 A) Photograph of the separation tube after the initial centrifugation showing the split blood layers: plasma, buffy coat, and erythrocytes. B) Transfer tube after the final centrifugation with the buffy coat (highlighted) ready for extraction. C) The various components of the Accucyte® system; from left to right: the float, fully assembled system, EpiCollector®, transfer tube, and locking clip. (Adapted from Campton, D.E. et al.38 under CC-BY 4.0).  To obtain CTCs, whole blood is first transferred from its collection tube into the Accucyte separation tube, the plastic float is then place inside the separation tube, and everything is then centrifuged. The denser erythrocytes will settle at the bottom of the separation tube, while the blood plasma will remain in the upper portion (Figure 2.4A). Between the boundary of the blood plasma and erythrocytes is the buffy coat 16  which contains all of the nucleated cells including leukocytes and CTCs. The buffy coat is retrieved using the EpiCollector® and transfer tube which is inserted into the separation tube following the aspiration of the blood plasma and addition of a high density retrieval fluid. A brass ring is fitted around the separation tube holding the float in place, and seals the erythrocytes to the bottom portion of the separation tube. Following a 2nd centrifugation, CTCs are moved to the transfer tube (Figure 2.4B) where they can be used in further downstream processes38,39.  Characterization of the Accucyte® system was performed using the prostate cancer cell lines LNCaP and PC3, lung carcinoma cell line A549, and breast adenocarcinoma cell line MCF-7, spiked into 7.5 mL of healthy donor blood. Mean yields of 90.0%, 90.2%, 90.5%, and 91.3% were achieved for each respective cell line, with an overall system average of 90.5%38,39. A small clinical study of 10 patients comprised of prostate, breast, and colorectal cancer patients found that the Accucyte system performed similarly to that of the FDA approved CellSearch® platform38. An additional study involving 30 prostate cancer patients further demonstrated the capabilities of the Accucyte® system finding CTCs in 21 patients, 13 of which had 10 or more CTCs40. Total processing time to enrich CTCs was 70 minutes, 10 of which require hands on work.  2.1.5 ScreenCell ScreenCell (France) uses size based micropore filtration to enrich CTCs using their ScreenCell® isolation devices. Currently approved in Europe with a CE mark, as well as in Canada with a Medical Devices Establishment License, ScreenCell sells three different types of ScreenCell® devices: ScreenCell® Cyto, ScreenCell® CC, and ScreenCell® MB for cytological studies, cell culture, and molecular biology, respectively (Figure 2.5). Cyto and CC devices are differentiated based on their pore size and filter/o-ring construction while MB devices are nuclease free and have a filter/capsule construction41.  CTCs are enriched based on their size, where they become trapped in micropores randomly distributed across a circular track-etched hydrophilic polycarbonate filter41. Firstly, 3 mL of whole blood is diluted in 4 mL of specialized buffers which are designed for fixed or live cells, depending on the desired application, in addition to lysing red blood cells42,43. Blood is then added to the filtration tank and processed through the microporous filter into a collection tube using a vacuum force provided by a 9 mL Vacutainer42. Filters are then released from the filtration tank using an actuated rod and placed in a tissue culture well or onto to a glass slide for fixation and enumeration (Figure 2.5)41.  17  Characterization of ScreenCell® devices was performed using the lung cancer cell line NCI-H2030. A series of experiments (n=25) were performed by spiking 5 and 2 CTCs per 1 mL into whole peripheral blood and processing the sample through the device. ScreenCell® devices achieved an average yield of 91.2% and 74.0% for 5 and 2 spiked CTCs, respectively41. The whole enrichment process including pre-processing takes around 10 minutes41–44. A study of 41 patients with low, moderate and, high-risk prostate cancer used ScreenCell® devices to enrich CTCs for characterization of nuclear size and volume, finding an average nuclear volume range of 392 to 1969 µm3 and an average nuclear diameter range of 9.1 µm to 15.6 µm43. Another study of 30 patients with metastatic breast cancer detected CTCs in 20 patients ranging from 1 to 347 CTCs using ScreenCell® devices44.   Figure 2.5 Overview of the ScreenCell® device which is comprised from top to bottom of: a filtration tank, filter, nozzle/holder, and waste container. The overall process to enrich CTCs is illustrated from blood collection to filter retrieval. (Adapted from Mu, Z. et al.44 under CC-BY 4.0).  2.1.6 Comparison An overview of each individual CTC enrichment method discussed in Sections 2.1.1 to 2.1.5 is presented below in Table 2.1. For systems that performed characterization experiments with different cancer cell lines, the average yield across all cell lines was taken and used as an overall system yield. The throughput of each device is calculated as per Equation (1.2) with the volume being equal to the equivalent amount of whole blood before any pre-enrichment, and duration given as the time required to complete enrichment from a whole blood sample to CTC enriched output. All throughputs are then scaled as a rate per hour to allow for direct comparisons.  18  Table 2.1 Comparison of the performance of five different commercial CTC enrichment technologies Company Device Type Pre-enrichment Average Yield Effective Throughput ANGLE plc Parsortix™ Deformability Density 69.7% 5.4 mL/hr ApoCell ApoStream® Electrical Signal Density 67.2% 7.5 mL/hr Clearbridge BioMedics ClearCell® FX Inertial RBC Lysis 82.0% 22.5 mL/hr Rarecyte Accucyte® Density n/a 90.5% 6.4 mL/hr ScreenCell ScreenCell® Size Dilution + Lysis 82.6% 18.0 mL/hr  With yields less than 70%, both the Parsortix™ and ApoStream® systems perform worse than the other available systems. Additionally, they do not offer an increased throughput as an acceptable trade off coming in below the top two performers. With an equivalent throughput of 22.5 mL/hr the ClearCell® FX system appears to stand above all other systems, however its high throughput is largely in part to its 3x multiplexed designed which increases the available sorting area. It may then be possible for the other microfluidic based systems to improve on their throughput by switching to a multiplexed configuration. While ScreenCell® has good performance in both yield and throughput; it has the distinct disadvantage of other filtration based techniques where the collected CTCs are stuck on the filter membrane instead of in suspension, which can make further downstream processes more complicated as CTCs must first be retrieved from the filter. Finally, Accucyte® is in the middle in terms of throughput, but has the greatest yield of all other methods. This performance advantage likely results from the lack of a pre-enrichment step, such as density centrifugation or RBC lysis used by other systems, where CTCs may be lost.  19  3 Design & Validation 3.1 Device Design & Validation Section 3.1.1 begins the chapter by walking through the design revision of the microfluidic ratchet device to alleviate the accumulation of cell debris at the sample inlet. Section 3.1.2 will then discuss the process to obtain a faster automated filling protocol, followed by the validation of the new device and its comparison to the previous iteration in Section 3.1.3. Finally, Section 3.1.4 will introduce the concept of an epoxy resin based device as a stepping stone towards commercialization, while Section 3.1.5 investigates the use of an autoclave for the sterilization of microfluidic devices.  3.1.1 Prevention of Cell Debris Accumulation One of the major limitations with the previous iteration of the microfluidic ratchet device was the accumulation of cell debris at the sample inlet (Figure 3.1). This behaviour is exhibited after using a cell lysis buffer, or processing aged blood which has begun to thicken and lyse on its own. Cell debris is likely composed of a combination of partial cell membranes, DNA fragments, and coagulation factors such as the Von Willebrand factor. The accumulation of cell debris would cause the device to clog up over time and subsequently fail, preventing any further cells from being sorted.    Figure 3.1 Sample inlet region of the microfluidic ratchet device developed by Jin, C. et al.20 filled with cell debris.  The geometry of the inlet is such that it has a single point of entry to infuse cells into the sorting region at the bottom left of the funnel matrix (Figure 3.1). Long pieces of cell debris wrap around funnels at the sample inlet, forming seeding pieces for smaller fragments of debris to become attached to over time as 20  the sample is infused from the sample inlet. Since there is no redundancy in the sample inlet it becomes a single point of failure, and when the accumulation of cell debris reaches a critical mass it then blocks the only interface between the sample inlet and the sorting matrix, causing the device to fail. One potential solution is then to attempt to break apart the cell debris to prevent it from building up to a critical mass.   Observation in previous experiments showed that physical agitation of the sample inlet area could allow a device which had become clogged to resume processing a sample for a short time longer. To consistently apply a force to break apart clumps of cells, a pan shaped vibration motor was embedded into a cavity formed in a PDMS device directly above the sample inlet area. Various vibration waveforms were tested including continuous, 10 Hz, and 1 Hz, where each waveform was applied for a period of 10 seconds every 60 seconds (Figure 3.2). While there was some reduction of cell debris accumulation in each case, devices would become clogged and fail when using aged blood. Additionally, the use of a vibration motor greatly increased the manufacturing and experimental complexity which is undesirable when attempting to simplify the overall process for a laboratory technician. Therefore, instead of attempting to break apart the accumulation of cells, the solution is to create redundancy at the sample inlet, allowing it to continue processing cells even if some of the interfaces between the sample inlet and the sorting matrix have become blocked.   Figure 3.2 Accumulation of cell debris in a A) standard microfluidic ratchet device as compared to using a vibration motor with B) continuous, C) 10 Hz, and D) 1 Hz vibration waveforms applied for 10 seconds every 60 seconds. 21  To increase the number of interfaces between the sample inlet and the sorting area, the sample inlet was relocated to be in parallel with the forward pressure inlet. The sample inlet is connected through a series of bifurcations to take up ¼ of the space at the bottom of the device, while the forward inlet occupies the remaining ¾ (Figure 3.3). All other flow channels, as well as the entire funnel matrix, were left identical to the previous device iteration to minimize the number of changes and maintain the device operational control and pressure optimizations previously determined. The funnel matrix is thus comprised of 2048 columns by 32 rows with a funnel pore size ranging from 18 µm to 2 µm. Pore size is constant within a row but decreases incrementally from top to bottom. Outer flow channels have a channel width ranging from 400 µm to 50 µm, decreasing as the channels become bifurcated. Finally, the funnel matrix region has channel height of 30 µm while the outer flow channels have a channel height of 40 µm.   Figure 3.3 Representative image of the bifurcated sample inlet microfluidic device, showing the WBC and CTC outlets, forward and backward oscillation channel inputs, cross buffer flow input, and sample inlet.  The tuning of hydrodynamic resistances RH was used to redesign the sample and forward inlet channels. Hydrodynamic resistance RH is analogous to a resistor in an electrical network and is a function of a microchannels height h, width w, characteristic length L, and coefficient of friction µ (Equation (3.1)). Specifically, the hydrodynamic resistances of the sample and forward inlet channels were designed to match as closely to the previous iteration (Equations (3.2) and (3.3)) as possible. In addition, their scaled resistances as determined by the fraction of the bottom inlet area they occupied were also designed to be 22  equal (Equation (3.4)). Initially resistances were estimated using Equation (3.1) before being calculated using finite element analysis (FEA). COMSOL 4.2a was used to simulate the flow in the forward pressure and sample inlet channels. The 2D channel geometry was drawn in SOLIDWORKS (Dassault Systèmes, Vélizy-Villacoublay, France) and imported into COMSOL where it was extruded to the proper channel height. The simulation was setup as a 3D stationary study, with a creeping flow, and no-slip wall condition. Water was used as the simulated fluid as it is a close approximation to phosphate-buffered saline (PBS). The model was meshed using a physics-controlled mesh and boundary conditions of 10,000 Pa at the inlet and 0 Pa at the outlet. Upon computation, flow through the sample inlet channel could be visualized (Figure 3.4) and the volumetric flow rate Q determined. The volumetric flow rate Q was then used to calculate the hydrodynamic resistance RH of the channel according to Equation (3.5) using the pressure difference ΔP boundary condition of 10 kPa. Using the same methodology, the resistance of the forward inlet channel was also found, and all resulting resistances are shown in Table 3.1 below.                     (3.1)                            (3.2)                              (3.3)                                    (3.4)        (3.5)   Figure 3.4 Visualization of the flow (m/s) in the sample inlet as computed using COMSOL. 23  Table 3.1 Hydrodynamic resistances of the modified flow channels in the microfluidic ratchet device Channel Previous (Pa•s/m3) Estimate (Pa•s/m3) COMSOL (Pa•s/m3) RH,Sample 2.695E+13 2.812E+13 2.854E+13 RH,Forward 9.259E+12 9.379E+12 9.510E+12 1/4RH,Sample=3/4RH,Forward n/a 7.030E+12 7.134E+12  3.1.2 Faster Filling Protocol Before a device can be used to sort cells it must first be vacated of air to allow the unrestricted flow of fluid throughout the device. Any trapped air bubbles, particularly in the sorting region, could have a detrimental effect on the operation of the device including flow instability, an increase in hydrodynamic resistance, and areas where cells could become blocked. Fluid is introduced into the device through a series of fittings and tubing, and pressurized using a custom pressure board. 10 cc syringe tubes (Nordson EFD, Westlake, OH) are used as fluid reservoirs and are connected to the pressure board using syringe barrel adapters (Nordson EFD, Westlake, OH). SafetyLok™ tube fittings (Nordson EFD, Westlake, OH) are used to connect reservoirs to 0.060” outer diameter (OD) Tygon tubing (Cole-Parmer, Vernon Hills, IL) and 3/32” OD Tygon tubing (McMaster-Carr, Elmhurst, IL). Tubing selection is dependent on the connecting channel, with the backward and forward channels using the larger tubing while all other channels are connected with the smaller tubing. Connections between devices and tubing are made using 0.025” OD needles (New England Small Tube, Litchfield, NH) and male Luer slip barb connector (Qosina, Ronkonkoma, NY) for small and large tubing, respectively. Finally, 1.5 mL Eppendorf collection tubes (Fisher Scientific, Hampton, NH) are connected to devices using the small Tygon tubing and needles. Devices are filled with PBS and 0.2% Pluronic F127 (Invitrogen, Carlsbad, CA) to prevent non-specific adsorption of cells to device walls.  In the previous iteration, buffer would first be added to all reservoirs and allowed to run through the length of tubing before the fitting was connected to the device. All control channels would then be pressurized and the device would begin to fill, which required ~15 minutes to completely vacate the device of any air. In order to reduce the runtime, excess buffer would then be removed from the sample reservoir before the blood sample was added to begin the experiment. In meeting the goal of improved automation and reduced technician input, a new filling protocol should be able to fill a device and begin an experiment without any outside intervention. Therefore, an improved filling scheme should 1) decrease the time required to fill a device, 2) remove the need for the operator to add the sample after filling, and 3) eliminate the requirement that all reservoirs and tubing be filled prior to connection to the device. Ideally, the operator would load the sample and all buffers into their respective reservoirs before initiating an experiment and walking away. 24  There are four control channels available for pressurization of the device: the buffer, sample, backward, and forward, abbreviated as BUF, SPL, BWD, and FWD, respectively. A commercial pressure controller (Fluigent, Villejuif, France) was used to test different fill combinations of control channels and was tested using a filling pressure of 11.5 psi, as determined by the maximum stable output the Fluigent could achieve on all 4 channels. To allow a blood sample to be loaded at the beginning of an experiment, fill combinations were not tested using a pressurized sample channel as doing so would allow cells to enter the sorting area before any air could be removed. Devices and tubing were first connected to their reservoirs, followed by adding PBS into each of the reservoirs. Control channels were then pressurized and the device was allowed to fill. Since the addition of fluid changes the diffraction of light within the device, it is easy to see by eye if a device still has air inside of it. Thus, a filled device is defined as one which no longer has any air pockets visible to the naked eye. The results of the filling experiments are shown below in Figure 3.5. The best performance of 70 seconds occurs when the FWD & BUF channels are pressurized simultaneously. In this case the air escapes through the SPL and BWD channels, completely filling up their respective tubing and ensuring that the device is ready to begin an experiment. Conversely, the worst case is the BUF pressurized alone, which after 600 seconds the BWD and FWD bifurcations are still entirely unfilled.  BUFBWDFWDBUF+FWDBUF+BWDBWD+FWDBUF+FWD+BWD0200400600Control  ChannelFill Time (s) Figure 3.5 Time required to fill a microfluidic ratchet device with PBS using different control channel configurations.  The high performance of the combination of the forward and buffer control channels may be the result of their positioning along the device. The buffer channel is located on the left while the forward channel is located on the lower right. Additionally, the flow patterns from each channel into the device are 25  perpendicular with respect to the other channel, which may also play a role in their fast filling time. It may then be possible to apply these criteria to other microfluidic devices in order to decrease their fill time.  Since the volume of air that needs to be vacated is dependent on the length of tubing used to connect the device to its respective reservoir, a fill time of 150 seconds was selected as a conservative value to ensure that the device and all tubing has had ample time to become completely filled with fluid. The use of the new filling scheme allows blood samples to be added directly, preventing the unnecessary steps of removing excess PBS before beginning an experiment. Additionally, the filling scheme can be applied to rigid plastic devices as air flows out through the unpressurized ports rather than through the membrane of the device, since rigid plastic devices are not gas permeable unlike PDMS.   3.1.3 Device Validation After fabricating PDMS devices (Section 4.1.4), initial validation was performed to verify the design changes and ensure proper fluidic flow throughout the device. This was achieved by filling a device with PBS dyed with green, red, and blue food colouring for the buffer, sample, and forward control channel fluids, to allow for proper visualization of the flow through the device. Using standard operating pressures (Section 4.2.2), it is expected that the fluid originating from the sample inlet should take a diagonal path through the device toward the backward inlet, whose sides are defined by the buffer and forward pressures. As shown in Figure 3.6 below, the fluidic flow through the device is as expected, validating both the design changes to the sample and forward inlets as well as the quality of the device manufacture.   Figure 3.6 Flow visualization of a bifurcated sample inlet microfluidic ratchet device using PBS dyed with food colour.  The primary motivator to modify the device design was to prevent the accumulation of cell debris at the sample inlet, causing it to clog and fail. To test whether devices were still susceptible to cell debris, whole 26  blood which had been left to age at room temperature for 3 days was processed through devices, while fresh whole blood was used as a control. Prior observations have shown that aged blood had a greater propensity to clump together in addition to the generation of cell debris and was selected over a cell lysis pre-enriched sample as a worst case scenario. Additionally, a study performed by Uyuklu, M. et al., showed that the deformability of erythrocytes stored at room temperature experienced significant changes after six hours45.  Whole blood was processed through a device as per the experimental procedure described in Section 4.2.2, without the addition of cultured cancer cells. Instead, devices themselves were examined under the microscope after blood had been completely processed. Previous generation devices developed by Jin, C. et al.20 would always become filled with cell debris and subsequently fail when 3-day aged blood was processed (Figure 3.1). In contrast, the bifurcated sample inlet microfluidic ratchet device (Figure 3.7) would show signs of minor debris accumulation when processing 3-day aged blood, but would never experience any slowdowns or failures. Thusly, the bifurcation of the sample inlet has successfully overcome the issue of cell debris accumulation which results in the subsequent clogging and failure of microfluidic ratchet devices.  While the accumulation of cell debris has been solved as demonstrated by the cell debris experiments, the use of a cell lysis pre-enrichment method (Section 1.4) may have the potential for performance degradation. Before testing the device performance when coupled with a pre-enrichment method, device performance was compared to the previous iteration microfluidic ratchet device developed by Jin, C. et al.20 to both establish a baseline, and ensure that design changes did not have a negative effect on device performance.   Figure 3.7 Panoramic scan of the sample inlet area of a bifurcated sample inlet microfluidic ratchet device under a 4x objective showing A) fresh blood control with no accumulation and B) 3-day old blood with minimal cell accumulation.  27  As per the experimental protocol outlined in Section 4.2.2, doping experiments were performed to validate the performance of the new device. Firstly, the performance of the bifurcated sample inlet microfluidic ratchet device was compared to the previous iteration microfluidic ratchet device developed by Jin, C. et al.20 using similar operating conditions. Using an unpaired t-test it was determined that there was no statistically significant difference in the average yield between the newly developed bifurcated sample inlet microfluidic ratchet device and the previous iteration microfluidic ratchet device developed by Jin, C. et al.20 (p-value = 0.1983). In all cases samples were processed at an average throughput of 1 mL per hour.  New Baseline Previous Device65758595Yield (%) Figure 3.8 Average yield of the base bifurcated sample inlet device as compared to the previous iteration microfluidic ratchet device developed by Jin, C. et al.20.  Next, using the G-Biosciences lysis buffer (Section 1.4), whole blood samples were pre-enriched and processed through the bifurcated sample inlet microfluidic device. Device performance was then compared to the baseline device performance where it was found that there was no significant difference (p-value = 0.1264) in the average yield (Figure 3.9). However, the equivalent throughput of blood was increased eight-fold from 1 mL per hour to 8 mL per hour. This increase in throughput, without the degradation of performance, allows for the processing of large volumes of blood to enrich CTCs from clinical patients for various downstream processes (Section 5).  28  Baseline Lysis65758595Yield (%) Figure 3.9 Average yield of whole blood pre-enriched with G-Biosciences lysis buffer as compared to whole blood both processed through the bifurcated sample inlet microfluidic ratchet device.  3.1.4 Epoxy Based Device While PDMS is one of the most common materials used for microfluidics in academia, it is unsuitable for the large scale production required of a commercial device. Commonly, injection moulding is used instead, to rapidly produce large quantities of products for low per unit cost. In this process, a thermoplastic or other such material is heated into a liquid and then rapidly forced into a die at high pressure. While the unit cost is typically quite low, the start up cost to create a die is substantial, so designs must be finalized before moving towards this final step. As a stepping stone along the pathway towards commercialization, the microfluidic ratchet device must first be tested using a rigid plastic, rather than the soft elastomer PDMS, to ensure that the final switch to injection moulding is feasible. Epoxy based resin (EpoxAcast® 690 (Coast Fiber-Tek, Burnaby, BC)) was chosen as the transitional material due to its optical clarity and fully cured properties, such as the ultimate tensile strength, tensile modulus, and Shore D hardness, being similar to injection moulded thermoplastics while not required the expensive tooling, heating, and pressurized machinery required to create the device. The use of epoxy resin in microfluidics has previously been demonstrated by Lim E. J. et al., with devices containing long straight microchannels with a square cross-section used to inertially focus bioparticles46.  To reduce the complexity of the experimental setup, fluid reservoirs are integrated into the device. The bottoms of each reservoir are designed to prevent the accumulation of blood within the reservoir itself. The bottoms of each cylindrical reservoir are rounded to form a half-sphere, which are connected from the center of each reservoir to the microfluidic channel layer. Additionally, the surface quality of the reservoir has an effect on cell retention within the fluid reservoir, and is therefore dependent on the 29  manufacturing method of the reservoir master. When using a manufacturing method such as 3D printing, striations are left behind as an artifact of the manufacturing process, providing small crevices where cells may become lodged. However, when using a subtractive machining method such as CNC milling, the surface can be made smoother thereby preventing cells from sticking inside the fluid reservoir. When fabricated, the reservoirs and microfluidic channels are cast in resin in a single piece and channels are sealed using another piece of resin in the shape of a glass slide (Section 4.1.6). This completed device forms a disposable cartridge (Figure 3.10) than can be manufactured in advance, and can be used at a later time to perform CTC enrichment.   Figure 3.10 Disposable cartridge comprised of A) fluid reservoirs (top side) and B) microchannels (bottom side).   Figure 3.11 Average yield of baseline PDMS microfluidic ratchet devices as compared to epoxy resin microfluidic ratchet devices.  With the incorporation of fluid reservoirs into the device, a different type of support instrumentation is required to successfully process blood through the microfluidic channels. The development and validation Baseline Epoxy65758595Yield (%)30  of this new support instrument is described in Section 3.2. Using this new support instrument, epoxy resin based devices were validated by performing doping experiments as described by Section 4.2.2. Figure 3.11 above shows the average yield for both epoxy based devices and baseline PDMS devices. Performing an unpaired t-test shows that there is no statistically significant difference (p-value = 0.3057) in the yield of CTCs.  3.1.5 Autoclave-ability An autoclave is commonly used to sterilize consumables and tools used in both research and clinical environments. The sterility of these items is important as any outside contamination could be detrimental to any experiment or medical procedure. In order to maintain their sterility, items are typically stored in some type of sterile packaging prior to being used for their intended application. To sterilize items, an autoclave uses high pressure saturated steam at 121°C. However, not all items are able to be sterilized with an autoclave as the high temperature and pressure may potentially damage the item.  To verify whether the autoclaving procedure would cause an adverse effect on the performance of devices, PDMS devices were loaded into single use sterilizable pouches (FisherBrand, Waltham, MA) and placed inside an autoclave (Tuttnauer, Breda, Netherlands) at 121°C for 50 minutes. Following the autoclaving procedure, doping experiments were performed according to Section 4.2.2. An unpaired t-test was performed and found that there was no statistically significant difference in average yields (Figure 3.12) between the baseline microfluidic device, and a device which had been subjected to the autoclaving procedure (p-value = 0.0987).  Baseline Autoclave65758595Yield (%) Figure 3.12 Average yield of an autoclaved microfluidic device as compared to the base bifurcated sample inlet device.  31  3.2 Instrumentation Design & Validation The following sections describe the development and testing of a new instrument (Figure 3.13) for use with epoxy resin based devices. Section 3.2.1 outlines the overall mechanical and pneumatic design of the system while Section 3.2.2 describes the electrical design. An overview of the user interface is outlined in Section 3.2.3, and finally the overall system validation is presented in Section 3.2.4.   Figure 3.13 Completed support instrumentation for use with epoxy resin based microfluidic ratchet devices.  3.2.1 Mechanical The body of the instrument is composed of a T-slot aluminum extrusion frame (McMaster-Carr, Elmhurst, IL) and water jet cut aluminum mounting panels (Metal Supermarkets, Richmond, BC). The body houses all of the electronics and acts as the support structure for the other two major mechanical subsystems: the lead screw assembly, and the pneumatics. The primary function of the lead screw assembly is to apply a clamping force between a pressure cap and an epoxy device, while the pneumatics provide controlled pressure to drive the flow of cells through the microfluidic device. Initial prototyping of the clamping function showed that a manual linkage based clamp could be used, potentially compressed into a flexure based design. However, the ease of manufacture of a lead screw approach led to its selection over other methods. Furthermore, the directly coupled driver design was selected for its simplicity over other lead screw driver implementations. The lead screw assembly (Figure 3.14) is comprised of the main screw (MISUMI USA, Schaumburg, IL), a lead nut (MISUMI USA, Schaumburg, IL) which is fixed to the 3D printed pressure cap (Objet30, Stratasys, Eden Prairie, MN), the directly coupled drive motor (Maxon 32  Motor, Sachseln, Switzerland), and is supported with a miniature linear guide (MISUMI USA, Schaumburg, IL) for transverse loads, and a bearing for axial loads. O-rings (McMaster-Carr, Elmhurst, IL) are used to create the pressure seal between the cap and epoxy device and are fixed in place inside the pressure cap in half dovetail retaining grooves designed using the Parker O-Ring Handbook ORD 5700 (Figure 3.15). O-rings are made from a soft silicone, Durometer 50A, to minimize the force required by the drive motor to create a pressure seal with devices. The validation of this setup is described below in Section 3.2.4.   Figure 3.14 Lead screw assembly showing the pressure cap connected to the lead screw drive assembly via the lead nut with support provided by a linear guide.  The primary pneumatic components of the system (Figure 3.16) are a pump which provides a pressure source, and digital pressure regulators that provide a stable output pressure to each control channel. The pump also includes a combination filter/silencer (McMaster-Carr, Elmhurst, IL) on its intake to ensure the pump receives a clean supply, and to reduce the ambient noise that the pump produces. While the source of all regulators is a common supply, each regulators output is connected to its terminating port on the pressure cap through a solenoid switch (Pneumadyne, Plymouth, MN). This allows each channel to be independently pressurized or vented while maintaining a constant pressure output from the regulator. In addition, a normally open solenoid (McMaster-Carr, Elmhurst, IL) is connected between the regulators source and pump output, and will vent the system to the atmosphere unless it is actively being closed. This provides a mechanical safety and will not allow the system to remain pressurized in the event of a 33  power failure. All pneumatic interconnections are made through push-to-connect tube fittings and 1/4” or 1/8” firm plastic tubing (McMaster-Carr, Elmhurst, IL). The control of the solenoids, pump, and pressure regulators is described in Section 3.2.3 .   Figure 3.15 3D printed pressure cap with push-to-connect tube fittings and built in o-rings used to seal against epoxy devices.   Figure 3.16 High level pneumatic circuit diagram showing the major connections between the input pressure source and the output control solenoids. 34  3.2.2 Electrical In order to control and interface with all of the various peripherals in the instrument, a custom printed circuit board (PCB) was designed and manufactured (Figure 3.17). A high-level overview of the PCB is shown in Figure 3.18, which highlights the major components: the microcontroller unit (MCU), lead screw motor, solenoids, pump, pressure sensor, and pressure regulator modules.  An MSP430 (Texas Instruments, Dallas, TX) microcontroller is used to interpret commands received from a PC running custom software (Section 3.2.3) and interfaces with all of the external components. The MCU contains multiple peripherals including but not limited to: digital input/output (IO), timers, universal serial communication interfaces (USCIs), and analog-to-digital converters (ADCs). Each of these peripherals plays a role in the development of custom firmware needed to operate the overall system.   Figure 3.17 Top view of the assembled custom PCB which contains the MCU and various support electronics used to control external peripherals connected via removable connectors.  As described in Section 3.2.1, a lead screw with a direct coupled motor is used to drive the pressure clamp and create a seal between a device and the clamp. The minimum continuous torque required by the drive motor is determined by the torque needed to raise the pressure clamp with the lead screw. This is given by the raising torque TR lead screw equation as shown in Equation (3.6). The required torque is a function of the lead screw parameters including the mean diameter dm, coefficient of friction µ, thread angle α, screw lead l, and mean collar diameter dc. The mass being moved by the lead screw is also included, and is 35  expressed as a force F. Using Equation (3.6) the minimum continuous torque was determined to be 13.6 mN·m. To determine the minimum peak torque required by the drive motor, the lowering torque TL equation was used (Equation (3.7). All variables are the same as in Equation (3.6) except the force is now determined as the forced required to deform the o-rings in the pressure cap. Using the Parker O-Ring Handbook ORD 5700, the clamping force required to deform the o-rings was estimated to be 25 N, which gives a minimum clamping torque of 88.6 mN·m. Thus, a brushless DC (BLDC) with integrated electronics (Maxon Motor, Sachseln, Switzerland) capable of delivering the required torque was selected as the drive motor, for its ease of use, increased lifespan, and higher torque to size ratio, as compared to other motor types.                                       (3.6)                                      (3.7)  The integrated electronics on the lead screw motor provides 5 pinouts: supply voltage, ground, rotational direction, speed set value, and speed monitor. The direction of the motor is set using a digital pin with logic high spinning the motor clockwise, while a logic low spins the motor counter clockwise. The motor speed, ηset, is controlled using a pulse width modulation (PWM) signal where the average voltage, Vset, dictates the motor speed as determined using Equations (3.8) to (3.10). Using a fixed frequency and amplitude, the motor speed is thereby set by varying the duty cycle between 0-100%. Since PWM is implemented on the MCU using a timer which counts up to a particular value before toggling the output to produce a square wave, the defining parameters are the clock speed and count value which under the current configuration provides a speed set value resolution of 0.6 rpm. Finally, the encoder used for the speed monitor only provides a resolution of 6 counts per rotation which is insufficient for position control given the small travel distance of the lead screw. Instead, control of the lead screw is performed using current feedback with a hall-effect current sensor placed in series with the lead screw motor. To raise or lower the pressure cap, the motor is run in either the clockwise or counter clockwise direction until the pressure cap collides with the device or the upper mechanical stop. Once the pressure cap has collided with a surface the motor will attempt to continue driving the cap which will result in a large current draw as the motor begins to stall. Using the current sensor for feedback this condition is checked, and signals that the motor should be shut off leaving the pressure cap in either the clamped or unclamped state. The reliability of this method is described in Section 3.2.4 below.  36                                           (3.8)                                     (3.9)                            (3.10)   Figure 3.18 High level PCB overview showcasing the major components.  Solenoids (Pneumadyne, Plymouth, MN) are used to control the flow of air at the output of each pressure regulator and are used to vent and pressurize each control channel. These solenoids were selected for their fast response time of < 10 ms, and small mechanical footprint. The fast response time of the solenoids is critical for the operation of oscillation channels, as any delays in actuation could have an effect on the overall operation of the device. An additional solenoid (McMaster-Carr, Elmhurst, IL) is used as the supply shutoff and will vent the system to atmosphere in the event of a power failure. The MCUs digital IO actuates metal-oxide-semiconductor field-effect transistor (MOSFET) switches to control all of the solenoids as well as the supply pump.  To provide an adjustable and stable pressure output for each of the control channels, MPV series digital proportional pressure regulators (Proportion-Air, McCordsville, IN) were used. The MPV modules provide the high resolution (±0.005% full scale) pressure control required to accurately maintain pressures during an experiment. The proportional design of these modules allows them to react quickly to changes in pressure, while maintaining an accurate output, making them highly reliable compared to their analog counterparts. The MPV modules recommend an input pressure of 110% of full scale output or 16.5 37  psi. Thus, a BLDC diaphragm pump (KNF Neuberger, Trenton, NJ) which delivers a maximum pressure of 35 psi, was selected as the pressure source. The extra head room ensures that there is no pressure drop on the input, allowing the regulators to perform correctly. Each different simultaneous pressure requires the use of an additional MPV. Since the buffer and sample operate at the same pressure, a total of 3 MPVs are required. Each MPV module takes an analog voltage command signal, Vcommand, where the pressure output, Pset, is directly proportional to the percentage of its full-scale output, PFullScale, as set by the ratio of Vcommand and the maximum voltage, Vmax, as shown in Equation (3.11). Here, MPV modules with a PFullScale of 15 psi were used to minimize the overall pressure range and increase resolution, while including the high pressure fill setting of 11.5 psi. The command signal is set using a 12-bit digital-to-analog converter (DAC) with an amplifier circuit, which in turn is set by the MCU communicating with the DAC using the inter-integrated circuit (I2C) protocol. Additionally, each MPV provides a pressure sensor signal, given as an analog voltage. This is translated into a 10-bit number using the MCUs onboard ADC where it is then transmitted to the PC for interpretation. The MPV modules were verified using a known and calibrated digital handheld manometer (McMaster-Carr, Elmhurst, IL), which has ±0.5% accuracy, and were found to be within 1.58% of the set value on average. In addition to the integrated pressure sensor for each control channel, there is an additional standalone sensor (TE Connectivity, Schaffhausen, Switzerland) which measures the supply pressure. This sensor provides a direct digital output of both a 14-bit pressure value and a 10-bit temperature value transmitted over I2C.                               (3.11)  3.2.3 Software The two primary functions of the PC software are to provide a platform to control the machine and to display key information to the user. The user interface (UI) was therefore designed to showcase all key controls and information on a persistent main screen while providing all other options and settings through flyout context menus. The UI was written using the Windows Presentation Format which utilizes XAML, a type of markup language, to define interface elements and Visual C# for behavioural interactions. Key features of the software include the ability to run an automatic timed experiment and save data logs recorded during an experiment, which includes readings for pressure, temperature, and valve states.  The UI main screen can be broken up into 5 major sections: system status, digital IO control and feedback, experimental controls, clamp controls, and settings (Figure 3.19). The system status section includes the 38  four control channel pressures as well as the supply pressure and temperature sensor readings in addition to a status message and experimental runtime. The digital IO control and feedback section is comprised of toggle buttons which serve a dual purpose of controlling the actuation and displaying the status of a control as open/on or closed/off through a changing background colour and a  or  mark. The experimental controls section contains buttons to start and stop an experiment. It has options for both a manual and automatic mode, where the manual mode will allow for parameters such as oscillation timings or pressures to be adjusted on the fly and runs for an indeterminate time, while the automatic mode will run for a set time frame using preset timings and pressures. Next, the clamp controls are used to load and unload a new device and provides a software lockout to all other controls until a device has been successfully loaded. Finally the settings section is broken up into 4 categories: general, data logging, experimental, and serial connection and contain applicable settings for each.  Figure 3.19 Overview of the user interface broke up into 5 major sections: A) system status, B) digital IO control and feedback, C) settings, D) experimental buttons, and E) pressure clamp controls.  To communicate with the MSP430 MCU a serial universal asynchronous receiver/transmitter (UART) connection is used. Communication takes place using a series of 8-bit (1 byte) numbers forming a complete communication packet. As the name serial suggests, the communication operates on a first in first out basis. The contents of both the transmission and receiving packet are designed to allow for easy interpretation on both ends of the communication line. Here, the start of a packet in both directions is signalled by a so called start byte which is equal to a complete 8-bit number (255). All subsequent bytes in a packet are enforced to only be 7-bits. This ensures that upon reception of the start byte the MCU/PC can be certain that the bytes that follow are valid for the expected packet format. For the MCU to PC direction a complete packet is comprised of 12 bytes and consists of four 10-bit numbers for the pressure 39  readings from the pressure modules, a 10-bit number for the lead screw motor current, a 14-bit number for the supply pressure, a 11-bit number for the supply temperature, a 2-bit supply sensor status, and the start byte. In contrast, the PC to MCU packet takes more of a generalized approach and consists of the start byte, a command byte, and three data bytes. The context of each data byte is determined by the command byte and allows for a much shorter packet since the PC always wants all sensor readings but the MCU does not always need to change multiple peripherals. This prevents multiple unused bytes in the case a larger all encompassing packet was used. The different commands are to set the digital IO, bitwise AND the digital IO, bitwise OR the digital IO, set an MPV module pressure, set the lead screw motor speed and direction, and request a read from all sensors.   3.2.4 System Validation Before using the new system to perform cell based experiments, the operation of the system needed to first be validated to ensure that it operated as intended without any unforeseen issues. To accomplish this, validation tests were performed using a solid block of Delrin which was used as an idealized device analog. Data logs provided by the user interface (Section 3.2.3) were then used to verify the operation of the system.  To test the repeatability of the clamping mechanism, a cyclic loading and pressurization test was performed. The device analog was used and was not fixed in place by any external means and was left to rest on top of the level bottom machine surface for the pressure cap to seal against. A testing cycle begins with the clamp lowering down and the pressure cap sealing against the device analog. A filling pressure of 11.5 psi would then be applied to both the buffer and forward channels as per the filling protocol for 30 seconds. Pressure would then be lowered to 5.0 psi and all channels would then toggle on and off with a period of 2 seconds for an additional 30 seconds. Finally, the clamp would then be raised to complete one full testing cycle. If the pressure cap failed to seal against the device analog an air leak would be created and would result in a lower than expected pressure reading in the data logs. Even a small air leak would be evidenced by a low pressure reading, and if such a failure occurred before starting an experiment, the operation of the device could be compromised.  Figure 3.20 below shows the pressure and valve status of a typical cycle. Key points of interest are the fast rise and fall time of the pressures which are both approximately 150 ms. This is due to the double solenoid design of the MPV modules and allows for quick changes to the pressure. In addition, the lack of large pressure variations during the rapid cycling of all control channels shows the overall stability of 40  both the pressure cap seal and the MPV modules which are of particular importance for the oscillation control channels used for cell sorting.   Figure 3.20 Snapshot of a single cycle during the repeatability test of the pressure clamp system.  To finally determine the repeatability of the system, the cyclic loading and pressurization test was set to run for 8 hours. The test would be stopped prematurely if the pressure cap failed to seal. Figure 3.21 below shows the results of the repeatability test where the pressure cap failed to seal completely against the device analog after ~230 cycles or ~14,000s. The cumulative small movements of the unfixed device analog along the bottom plate as a result of being clamped and unclamped are the cause of the failed seal. This is shown in Figure 3.21 by the steep pressure drop off in the buffer and sample channels at the ~14,000s mark. The data points observed in between the two pressure set points prior to the 14,000s mark arise from the transitioning of the pressure between the high and low states, and are therefore an expected measure and not indicative of a system failure. The pressure drop off in the buffer and sample channels indicates the presence of an air leak, which matches the expected points of failure as the reservoir and device geometry has both of these channels closest to the inner surface of the pressure cap. Additionally, after resetting the device back to its initial position, the system was able to once again properly seal 41  against the device. With a typical cell experiment lasting between 1-1.5 hours the failure time is equivalent to running the system for a whole month assuming an 8 hour work day and 1 cycle equaling 1 cell experiment. However since each device is single use and each subsequent experiment would require the placement of a new device in the correct position, the clamping system has been shown to be reliable for the purposes of cell based experiments.   Figure 3.21 Repeatability test of the pressure clamp system run till failure after ~230 cycles or ~4hrs.  The second validation test performed was an experimental dry run. Here, the machine was operated according to the experimental settings in Section 4.2.2, including the 11.5 psi filling, 4.5 psi buffer and sample, 2.5 psi forward, and 2.0 psi backward pressure set points. Additionally, the test was performed with the device analog and without fluids. Figure 3.22 below shows a snapshot of the transition between the filling phase and the running phase of the microfluidic ratchet sorting device, while Table 3.2 shows the average and standard deviation of the pressures in each control channel during both the filling and running phases. As evidenced by both Figure 3.22 and Table 3.2 the system if capable of maintaining accurate pressure settings with minimal variation in both high and low pressure configurations, proving the stability of the system. 42   Figure 3.22 Snapshot of the transition between the filling and running phases during the dry run test.  Table 3.2 Experimental dry run control channel pressure readings  Channel Filling Average (psi) Filling STD (psi) Running Average (psi) Running STD (psi) Backward 11.493 0.114 1.997 0.045 Forward 11.502 0.200 2.496 0.042 Buffer 11.504 0.205 4.516 0.032 Sample 11.504 0.205 4.516 0.032  43  4 Methods 4.1 Fabrication The following sections describe the steps in the fabrication process beginning with computer aided design (CAD) in Section 4.1.1 to a complete functioning PDMS device in Section 4.1.4. In addition, a new process for the fabrication of hard epoxy resin devices is described in Sections 4.1.5 and 4.1.6.  4.1.1 Fabrication of Silicon Masters CAD software Draftsight (Dassault Systèmes, Vélizy-Villacoublay, France) is used to generate the device geometry and is translated onto two separate photomasks; an optical chrome mask for high resolution which contains the central funnel constrictions and a transparency photo mask which contains all of the flow geometry. A negative master is fabricated on a silicon wafer using standard photolithographic techniques, and is created in two photolithographic layers (Figure 4.1). To start, a new silicon wafer is taken from the wafer stock and is cleaned in series with acetone, methanol, and isopropyl alcohol followed by a quick dry with nitrogen. The wafer is then placed on a hotplate at 200°C for 5 minutes to ensure a completely clean and dry starting surface. The first layer is created by coating the silicon wafer with an epoxy based negative photoresist SU-8 3025 (Microchem, Newton, MA) and spinning it at 2500 rpm for 30 seconds. This first layer includes all of the funnel constrictions, makes up the central portion of the device, and has a designed height of 30 µm. The wafer is then soft baked on a hotplate for 10 minutes at 95°C followed by ultraviolet (UV) exposure through the optical photomask (Advance Reproductions, Andover, MA). The wafer is then post exposure baked for 65°C, 95°C, and 65°C for 1, 4, and 1 minutes, respectively. The wafer is developed using SU-8 developer (Microchem, Newton, MA) and then rinsed with isopropyl alcohol and dried with nitrogen. Next, using the same photoresist as before, the second layer is made by spinning the wafer at 2200 rpm for 30 seconds. This second layer includes all of the flow channel geometry and has a designed height of 40 µm. The wafer is soft baked for 1, 14, and 1 minute at 65°C, 95°C, and 65°C, respectively. UV exposure through the transparency photomask is performed by the Canon PLA-501F, after alignment marks created during the first layer are matched to their corresponding alignment marks in the transparency photomask. Once again the wafer is then post exposure baked for 65°C, 95°C, and 65°C for 1, 4, and 1 minutes respectively and then developed using SU-8 developer (Microchem, Newton, MA) and rinsed with isopropyl alcohol and dried with nitrogen. Finally, to set the features, the wafer is hard baked at a ramp of 50°C/hour to 165°C where it is held for 30 minutes, then ramped down at a rate of 120°C/hour. The final height of the first and second layer was measured to be 25.0 µm and 35.3 µm, respectively. 44   Figure 4.1 Overview of the fabrication process of silicon master.  4.1.2 Fabrication of PDMS Masters A PDMS device master is made by mixing Sylgard 184 (Dow Corning, Midland, MI) base and hardener together in a 10:1 ratio by weight. Initial mixing is performed by hand with a wooden tongue depressor followed by mixing and defoaming in a planetary centrifugal mixer (Thinky USA, Laguna Hills, CA) for 2.5 and 2 minutes, respectively. Aluminium foil is formed into a 5cm tall dish around the silicon wafer and pressed around the edges to create a seal to prevent PDMS from flowing under the wafer and encasing it. The mixed PDMS is poured over the wafer and placed inside a vacuum chamber for 20 minutes to degas and ensure all bubbles are evacuated from the micro features. Following a 2 hour bake at 65°C the PDMS is carefully separated from the silicon wafer and cut to size.  4.1.3 Fabrication of Polyurethane Moulds Since the silicon wafers are fragile, polyurethane moulds are used to mass produce devices for use in experiments. Smooth-Cast ONYX (Coast Fiber-Tek, Burnaby, BC) polyurethane base and hardener is measured out separately, 6:5 by weight. The two components are degassed in a vacuum chamber for 30 minutes, along with PDMS masters sitting feature side up inside of a premade silicone mould, and then mixed vigorously for 2 minutes. The mixed polyurethane is then poured over the PDMS masters and scrapped across the microchannels using a wooden tongue depressor to remove any trapped air bubbles. After curing for 1 hour at room temperature and 2 hours at 65°C the polyurethane is demoulded from the silicone, and the PDMS masters are subsequently removed from the polyurethane mould. 45  4.1.4 Fabrication of PDMS Devices Using the method described in Section 4.1.2 and substituting the silicon wafer for the polyurethane mould in Section 4.1.3, PDMS devices are made and removed from the polyurethane mould. After removal, inlets and outlets are punched using a 0.5 mm and 4 mm punch (Harris Uni-Core, Redding, CA), followed by bonding to a glass slide after plasma treatment in air plasma (Harrick Plasma, Ithaca, NY) for 75 seconds. Devices are then post baked for 15 minutes at 65°C to alleviate any trapped air and complete the bonding process.   Figure 4.2 Fabrication process of PDMS devices.  4.1.5 Fabrication of Silicone Moulds Silicone moulds are made in two halves: one with microchannels and the other with reservoirs. Additionally, the reservoir half is split into two pieces to allow for easier removal of epoxy devices. The double sided mould ensures that devices are square to allow the pressure cap to properly seal against the reservoirs. Firstly, a CNC manufactured reservoir positive is placed face up inside of a mould box made of LEGO® building blocks (The LEGO® Group, Billund, Denmark). Small LEGO® bricks form the resin flow channels at the top of the reservoir. Mold MAX® 14NV (Coast Fiber-Tek, Burnaby, BC) is mixed together 10:1 by weight and mixed by hand. After mixing, it is placed into the desiccator and degassed for 10 minutes before being poured into the mould around the reservoir edges, followed by a 4 hour cure at room temperature and pressure. This process is repeated once more to fill in the reservoirs 46  and form the complete reservoir mould halve. Mann Ease Release 200 (Coast Fiber-Tek, Burnaby, BC) is sprayed over the mould and is left to dry for 15 minutes between each step to prevent self adhesion. Separately, a PDMS master is first cleaned in air plasma for 75 seconds and then placed feature side up inside of a small Petri dish inside of a desiccator located within a fume hood. Within the fume hood, a plastic eye dropper is used to place 3 equidistant droplets of Trichloro(1H,1H,2H,2H-perfluorooctyl)silane (Sigma-Aldrich, St. Louis, MO) on two glass slides located on either side of the Petri dish (Figure 4.3). Following desiccation for 5 minutes, the pump is shut off and the chamber is closed to allow the PDMS master to remain in vacuum for 30 minutes to allow the silanization process to take place. The PDMS master is then removed from vacuum and fixed on top of the inverted reservoir mould halve using double sided tape. As before, a new layer of Mold MAX® 14NV (Coast Fiber-Tek, Burnaby, BC) is mixed, degassed, and poured over the PDMS master and left to cure for 4 hours. Finally, the completed silicone mould is removed from its mould box and the PDMS and reservoir masters are subsequently demoulded to complete a double sided silicone mould (Figure 4.4).  Figure 4.3 Silanization of a PDMS master.   Figure 4.4 Completed mould for epoxy resin devices made of two halves: the microchannel (bottom left) and reservoir halves (top and bottom right). The reservoir half is further split in two for easier removal of epoxy resin devices 47  4.1.6 Fabrication of Resin Devices EpoxAcast® 690 (Coast Fiber-Tek, Burnaby, BC) base and hardener is mixed together 10:3 by weight and mixed thoroughly by hand. The mixed resin is then degassed for 30 minutes before pouring into the silicone device mould (Section 4.1.5). After curing for ~24 hours devices are demoulded and holes are made using a 1/8” and 1/16” drill bit. The time for demoulding is critical as at the 24 hour mark, devices are mostly cured, but are still soft enough to be bonded to another piece of resin with heat and pressure. Devices and resin slides are cleaned in air plasma for 75 seconds and placed on top of each other in the 65°C oven for 15 minutes. After removal from the oven, light pressure is applied to the device and slide by hand to complete the bonding process. Completed devices are left for an additional 24 hours to fully cure and set.  4.2 Experimental Setup The following sections describe the preparation and experimental setup of a doping experiment. Section 4.2.1 details the preparation of cultured cells, while Section 4.2.2 describes the overall process of a doping experiment.  4.2.1 Sample Preparation Doping characterization experiments were performed with a human transitional cell carcinoma lymphatic metastasis cell line UM-UC13 provided by Bladder Cancer SPORE Pathology Core at the MD Anderson Cancer Center. Cells were cultured in MEM media supplemented with 10% fetal bovine serum (FBS), 1% l-glutamine, 1% non essential amino acids, 1% sodium pyruvate, and 1% penicillin streptomycin. Cells were incubated in a humidified environment at 37°C and 5% CO2. Before doping, cells were stained with Calcein AM (Invitrogen, Burlington, ON) for 30 minutes and washed with PBS.  4.2.2 Doping Experiments Each experiment performed requires the use of support infrastructure. Experiments performed on PDMS devices use a custom pressure board with manual pressure regulators, analog gauges, and WinForms C# control software (Figure 4.5), while experiments performed on epoxy resin devices use the instrument described in Section 3.2.  Fresh blood is first collected from a healthy donor with informed consent in an EDTA Vacutainer and left at room temperature. Cultured cancer cells (Section 4.2.1) are then placed in a haemocytometer to determine the cell concentration. An equal volume of cells is then spiked into a set volume of whole blood dependent on the experiment being performed as well as a 96-well plate for confirmation of the 48  initial dope amount. The spiked sample is then either directly loaded into the sample reservoir or pre-enriched using RBC lysis (Section 1.4) and then loaded into the sample reservoir. PBS with 0.2% Pluronic is added to the buffer, backward, and forward reservoirs.   Figure 4.5 Pressure board with analog gauges and manual pressure regulators used for the control of PDMS microfluidic ratchet devices.  Devices are then filled for 150 seconds at 11.5 psi as per the filling protocol outlined in Section 3.1.2. Next, cells are sorted using a 4.5 psi sample and buffer pressure, 2.5 psi forward pressure, and 2.0 psi backward pressure. The oscillation period is 4 seconds, with 3 seconds applying a forward pressure and 1 second applying a backward pressure. The buffer and sample pressures are left on continuously, except for every 10 oscillation cycles where the sample pressure is shut off during the backward pressure period. The experiment is run until all liquid has left the sample reservoir and air has begun to enter the device. The final runtime of the experiment is then recorded and used to calculate the effective throughput.  The CTC outlet tube is pipetted in 100 µL increments into a new well in the 96-well plate, which is then centrifuged at 400 xg for 5 minutes, and then taken to the microscope for scanning. Panorama images are collected with a 4x objective on an Eclipse Ti-E (Nikon, Tokyo, Japan) inverted microscope with an automated stage using a QIClick (QImaging, Surrey, BC) CCD camera for later enumeration. Panorama images are stitched together using Image Composite Editor (Microsoft, Redmond, WA), analyzed using the cell counter plugin in ImageJ (National Institutes of Health, Bethesda, MD), and cell counts for the initial dope concentration, and CTC outlet are determined. The yield performance metric is then calculated from the counted cell numbers. 49  5 Separation of CTCs from Patients with mCRPC Park, E.S., a member of our lab, developed a workflow (Figure 5.1) to obtain genomic sequencing data from patients with metastatic castration-resistant prostate cancer (mCRPC). This workflow uses the microfluidic ratchet device to enrich CTCs, in combination with a laser capture microdissection (LCM) extraction of single cells. The development of the process is particularly advantageous as cells isolated through this method are compatible with downstream processes like next generation sequencing.  Using this workflow, genomic sequencing data was obtained from patients with mCRPC. To ensure the quality of isolated cells, samples were processed from the enrichment step through to single cell selection within 24 hours of collection. Following single cell selection, whole genome amplification (WGA) was performed to extract and amplify genomic DNA. The quality of amplified DNA was assessed by quantitative polymerase chain reaction (qPCR). Cells which had a Ct value, or number of amplification cycles47, greater than 25 did not proceed onwards to sequencing. Using the bifurcated sample inlet microfluidic ratchet device, CTCs were enumerated in 3 out of 11 patients with an average count of 78. Sequencing results from a patient with paired CTC and cell-free DNA (cfDNA) data were found to contain TP53, PTEN, and FOXA1 mutations commonly associated with prostate cancer along with additional mutations not present in cfDNA, showcasing the potential value of CTCs over cfDNA.   Figure 5.1 Overview of the work flow for the separation and single cell analysis of CTCs from patient samples. 50  5.1 Preparation and Analysis of Patient Samples The following sections detail each of the steps in the workflow to obtain genomic sequencing data from patient samples. Section 5.1.1 describes the collection and preparation of patient blood for use with the bifurcated inlet microfluidic ratchet device. Section 5.1.2 then describes the post sorting immunofluorescence cell staining followed by the cell enumeration and single cell identification process in Section 5.1.3. Section 5.1.4 then briefly discusses the process for single cell extraction, and finally Section 5.1.5 provides a brief overview of cfDNA and next generation sequencing.  5.1.1 Sample Collection and Preparation Patients with mCRPC were recruited by the BC Cancer Agency with informed consent, and blood was drawn from a total of 11 patients during routine clinical visits. Two EDTA tubes, each containing 5 mL of whole blood, were collected from each patient. 8 mL of whole blood was pre-enriched using RBC lysis (Section 1.4) and processed with the microfluidic ratchet device, with the remaining 2 mL being used for a parallel quality control test.   Samples pre-enriched by RBC lysis were concentrated from 8 mL down to 600 µL and pipetted into the sample reservoir for processing. PBS with 0.2% (w/v) Pluronic was loaded into the remaining reservoirs, with 1 mL added into the buffer reservoir, and 3 mL added to each of the forward and backward reservoirs. One key observation during the experimental setup was the increase in viscosity of the blood of patients as compared to healthy donors. Patient blood was somewhat more difficult to pipette than healthy blood, which may be the result of the various cancer treatments undertaken. While this increase in viscosity did exhibit occasional slowdowns in the processing of blood as evidenced by a change in slope of the trapezoidal flow pattern, the improved bifurcated inlet design allowed the sample to recover and continue running.  The parallel quality control process enriched the remaining 2 mL of blood using density centrifugation to check for any CTCs. Cells enriched from this process were stained and scanned according to the same protocol outlined for cells which were enriched using the microfluidic ratchet device. Potential cells found using this method would be unsuitable for single cell selection, due to the increased amount of background cells, but could illustrate a potential device failure to successfully enrich CTCs. However, in all patients where CTCs were not detected by the microfluidic ratchet enrichment process, the parallel quality control process also did not detect CTCs. Therefore, there is a high level of confidence that a lack of CTCs detected by the microfluidic ratchet device indicates the absence of CTCs in a patient, and that it is not a malfunction or failure of the device. 51  5.1.2 CTC Separation and Immunofluorescence Staining CTCs were separated from RBC lysis pre-enriched whole blood using PDMS based bifurcated sample inlet microfluidic ratchet devices. New 10 cc syringe tubes (Nordson EFD, Westlake, OH) were connected to fresh microfluidic devices through a series of fittings and tubing. 1.5 mL Eppendorf tubes (Fisher Scientific, Hampton, NH) containing 100 µL of 3% (w/v) BSA in PBS were then connected to the CTC and WBC outlets through tubing. As described in Section 4.2.2, devices were actuated using a custom built pressure board with manual pressure regulators and solenoid valves. Sample and buffer pressures were set to 4.5 psi, while forward and backward pressures were set to 2.5 and 2.0 psi, respectively. An additional fixed regulator provided the 11.5 psi high pressure required to fill devices. Solenoid valves were controlled by a C# WinForms application with timing parameters set by the user.   Firstly, devices were filled for 150 seconds at 11.5 psi using the forward and buffer control channels (Section 3.1.2). During the filling step, devices were examined to ensure that fluid was exiting through all remaining inlets and outlets, and that all air bubbles were being purged from the devices. Air bubbles would typically exit through the backward and sample outlets, and could be observed transiting through the tubing towards their respective reservoirs. Following filling, devices were run using an oscillation period of 4 seconds, with forward pressure applied for the first 3 seconds, and backward pressure applied for the remaining 1 second. Throughout the oscillation, the buffer and sample pressures remained on except for a shut off of the sample pressure every 10 oscillation cycles during the backward pressure period. This periodic shutoff allows cells which may be held in place by the continual application of the sample pressure and are unable to deform through further funnel constrictions to be released and continue towards the CTC and WBC outlets. Samples were processed until no liquid remained in the sample reservoir and connecting tubing, and air began to fill the device.  After separation, cells collected from the CTC outlet were left to incubate on a shaker for 1 hour. Cells were then stained for identification using 1 µL of 100 µg/mL anti-EpCAM conjugated to Alexa 488, 1 µL of a 1/10 dilution of 10 mg/mL Hoechst DNA stain for the nucleus, and 3 µL of 9 µg/mL APC bound to CD45 to exclude leukocytes. After incubation for 2 hours at room temperature on a shaker, cells were washed once with PBS and then transferred to PDMS chambers on top of a polyethylene naphthalate (PEN) membrane coated glass slide.  Cells collected from the WBC outlet of the device were used to provide a control leukocyte for the genomic sequencing process. As with the CTC outlet, cells in the leukocyte outlet were stained using immunofluorescencent dyes. Although the CTC outlet is the desired harvesting location for CTCs, it is 52  possible that some CTCs may escape to upper rows of the funnel matrix and make their way into the WBC outlet. However, as in the case of the parallel quality control process, CTCs found in the WBC outlet would be ineligible for single cell selection due to the increased amount of background cells. While a few CTCs were located within the leukocyte outlet, it was observed that > 95% of CTCs were found in their desired location, the CTC outlet.   5.1.3 CTC Enumeration & Single Cell Identification To aid in the enumeration of CTCs enriched from patient samples, Ang, R.R., a member of our lab, developed a spectral image cytometry platform. This platform is composed of image acquisition, automated pre-screening, and an assisted manual review. To acquire images, sample wells were scanned on a Zeiss LSM 780 (Carl Zeiss, Oberkochen, Germany) confocal microscope using a 40x objective and automated stage to capture gigapixel images comprised of 15x15 stitched tiles. Each individual pixel within the gigapixel image (Figure 5.2) contains 26 channels worth of spectral information which can be easily separated into 5 distinct emission bands plus bright field.   Figure 5.2 Representation of the gigapixel spectral image cube of patient VCCCTC023 comprised of 26 channels each 7680x7680. Channel peak emissions are listed in addition to the transmitted light quality control.  Antibodies are conjugated to different fluorophores which have different absorption and emission wavelengths (Figure 5.3). Cells are classified based on the presence or lack of an emission peak, where a cell with a peak at 521 nm (EpCAM-Alexa 488) and without a peak at 660 nm (CD45-APC) would be considered to be a candidate CTC, whereas a cell with a peak at 660 nm (CD45-APC) would be excluded 53  as a leukocyte. In other words a candidate CTC is a cell which is Hoechst+, EpCAM+, and CD45-, while a leukocyte is a cell which is Hoechst+, EpCAM-, and CD45+.   Figure 5.3 Idealized emission spectra of Hoechst (blue), EpCAM (green), and CD45 (red) used to identify CTCs and leukocytes. Cells which are Hoechst+, EpCAM+, and CD45- are classified as CTCs while cells that are Hoechst+, EpCAM-, and CD45+ are categorized as leukocytes.   Figure 5.4 Review software showing the results of a single well from patient VCCCTC023 overlaid with a markup to highlight key areas of the software. Regions of interest include the cell spectral graph, cell subimage, gel view, full resolution image, and selected cell coordinate list.  Using the gigapixel spectral image cube and the emission peak criteria, the spectral image cytometry software identifies, analyzes, and ranks cells based on their spectral emission from the most likely to least likely to be a CTC. The software then provides a simple interface for users to manually review the emission spectrum for each individual cell (Figure 5.4)48. Using this software, individual cells were reviewed and CTCs were subsequently enumerated. Additionally, CTC locations were marked to generate a location map which was then used during the LCM single cell extraction process (Section 5.1.4). 54  5.1.4 Single Cell Extraction The extraction of single CTCs from scanned PDMS wells was performed using LCM, which is a method that is commonly used to harvest cells from small regions of interest contained in a larger tissue sample. LCM technology can be broken up into two major categories: Infrared (IR) and UV49. Here, the UV based Zeiss PALM Microbeam (Carl Zeiss, Oberkochen, Germany) was used to isolate single CTCs. Cells in suspension were stabilized in a hydrogel layer on a PEN membrane slide to make samples compatible with the LCM process. Briefly, slides were loaded into the Zeiss PALM Microbeam (Carl Zeiss, Oberkochen, Germany) and CTCs were located using fluorescent microscopy and the location map generated in Section 5.1.3. Single cells were then cut from the hydrogel/PEN membrane and transferred into 200 µL collection tube caps (Carl Zeiss, Oberkochen, Germany) using laser pressure catapulting (Figure 5.5).   Figure 5.5 Overview of the single cell selection process as validated using UM-UC13 cells. Steps include: A) cell identification, B) membrane cutting, and C) catapulting the excised sample to the D) collection tube cap. The cell is only visible under the E) EGFP filter and missing with the F) mCherry filter, indicating the retrieval of the correct target cell.  5.1.5 Genomic Sequencing After successfully capturing single cells in collection tube caps, DNA was extracted and amplified from single cells using the REPLI-g WGA kit (Qiagen, Hilden, Germany) according to the manufacturer’s protocol. DNA obtained from this process was then placed into storage at -20°C. The quality of DNA obtained from the WGA amplification was then assessed using qPCR. qPCR reactions were carried out in triplicate using the ViiA 7 Real-Time PCR system (Applied Biosystems, Foster City, CA) according to the manufacturer’s protocol. Cells which had Ct values greater than 25 were excluded from further downstream work, while all remaining cells proceeded to genomic sequencing. 55  Cells that successfully passed the qPCR quality control were sequenced by the Wyatt Prostate Genomics Laboratory at the Vancouver Prostate Center. Briefly, targeted next generation sequencing was employed using a custom NimbleGen SeqCap EZ Choice Library (Roche Sequencing, Pleasanton, CA) for sequencing a panel of 73 genes relevant to prostate cancer50. Mutations were called if a CTC sample had at least 5 supporting reads, and 20% of all reads supported the mutation. Additionally, the allele fraction in the CTC sample had to be at least 5 times greater than the allele fraction in the WBC sample, and the WBC sample had to have to at least 20 overlapping reads.   5.2 Results from mCRPC Patients Counts of CTCs enriched from mCRPC patients using the bifurcated sample inlet microfluidic ratchet device and enumerated using the confocal image cytometry platform developed by Ang, R.R. are shown below in Figure 5.6. CTCs were detected in 3 out of 11 patients with an average count of 78. As mentioned previously, there is a high degree of certainty that the other 8 patients did not possess CTCs as determined through the use of a parallel quality control test. Single cells isolated from patient VCCCTC023 were further processed using next generation sequencing in a paired data set with cfDNA samples processed by the Wyatt Prostate Genomics Laboratory.   Figure 5.6  CTC enumeration values for mCRPC patients from 2 mL (VCCCTC010) and 8 mL (VCCCTC018 and VCCCTC023) of whole blood enriched using the bifurcated sample inlet microfluidic ratchet device.  Briefly, cfDNA is fragments of tumour DNA that can be found in the plasma or serum of cancer patients, and are suggested to be the result of necrosis and apoptosis of tumour cells51. Like CTCs, cfDNA is being investigated as a minimally invasive biomarker for cancer research. The analysis of both CTCs and cfDNA, increasingly referred to as a liquid biopsy, are considered to be complementary technologies each with their advantages and disadvantages52–54. VCCCTC010VCCCTC018VCCCTC023050100150200CTC  Count56  Five CTCs were sequenced along with cfDNA in addition to a leukocyte to serve as a positive control. Mutations, as defined in Section 5.1.5, found in cfDNA and CTC through targeted next generation sequencing are shown below in Figure 5.7. CTCs and cfDNA were found to share common mutations associated with prostate cancer (e.g. TP53, PTEN, and FOXA1). Furthermore, individual CTCs exhibited additional mutations not found in cfDNA.   Figure 5.7 Mutations detected through targeted next generation sequencing suggests the heterogeneity of CTCs. When compared to cfDNA, 3 mutations (TP53, PTEN, and FOXA1) are shared, while 20 mutations are only found only in CTCs. Gene Effect cfDNA CTC1 CTC2 CTC3 CTC4 CTC5 WBCTP53 Missense p.Y104C, p.Y77C, p.Y197C, p.Y236CPTEN Splice siteFrameshift p.L379fsFrameshift p.P381fsMissense p.A380GFrameshift p.H377fsMissense p.H377DFrameshift p.P371fsFrameshift p.L379fsMissense p.L379VMissense p.H377QATR Missense p.E1061GKMT2C Missense p.T427IMissense p.V317MStopgain p.Y430XFANCD2 Missense p.C801YPIK3CB Frameshift p.L122fsATR Missense p.S1583PPIK3R1 Missense p.S142N, p.S235N, p.S205N, p.S505NCHD1 Frameshift p.L596fsMissense p.E145K, p.E135KMissense p.M329I, p.M347IFANCE Frameshift p.A308fsFrameshift p.P325fsMissense p.T992I, p.T1010IMissense p.I4628NMissense p.A1531VMissense p.G315SCLU Missense p.A183TNon-frameshift indel p.T39delinsRSMissense p.T39RFrameshift p.I40fsMissense p.H42QMissense p.L43PNon-frameshift indel p.E372delinsDIQMissense p.L373MMissense p.V374FZFHX3 Missense p.S945C, p.S1859CRNF43 Frameshift p.P65fs, p.P192fsERCC2 Missense p.R631CFOXA3 Missense p.G135SMissense p.N1381YFrameshift p.N1381fsFrameshift p.I1382fsFrameshift p.H1822fsERCC4MED12FANCCFOXA1APCMETKMT2CATM57  5.3 Discussion The prognostic capabilities of CTC enumeration has already been shown in the clinical setting55,56. However, cell genotyping could provide a more suitable clinical readout to help guide tailored cancer treatments. Although tumour biopsy is considered to be the gold standard for the characterization of cancer, it is highly invasive and may not always be suitable as determined by the type of cancer. The agreement of mutations between primary tumours and CTCs has been shown in studies with prostate and colorectal cancer, demonstrating the use of CTCs as a surrogate for tumour biopsy57,58. Furthermore, single cell sequencing of CTCs has been used to show genomic changes in response to cancer therapies59. Thus, investigation into CTC genotyping could reveal key information in the progression of cancer, and help guide treatment for each patient’s disease.  The paired sequencing results of CTCs and cfDNA showed that individual cells shared mutations common to the circulating tumour population, which is consistent with previous reports60. The TP53, PTEN, and FOXA1 mutations shared by CTCs and cfDNA are commonly associated with prostate cancer61–63. Furthermore, CTCs showed additional mutations not found in cfDNA, with individual cells expressing different mutations than one another, demonstrating the heterogeneity of the tumour cell population. These mutations may be clinically relevant, and the additional information showcases the potential value of CTCs over cfDNA.    58  6 Conclusion 6.1 Summary of Results This work has demonstrated an improved microfluidic ratchet device that can separate circulating tumour cells from whole blood with a > 80% yield with a high throughput. Devices are no longer susceptible to the accumulation of cell debris at the sample inlet, and are therefore now compatible with an RBC lysis pre-enrichment step. This pre-enrichment step increases the effective throughput to 8 mL of whole blood per hour. Additionally, devices are still capable of handling whole blood samples directly at a rate of 1 mL per hour. The experimental procedure has been further simplified and automated with the development of epoxy based devices and the accompanying support instrumentation. Hands-on work for cell separation now only requires the loading of a disposable cartridge with a blood sample and additional buffers, before loading into the instrument where it will automatically sort CTCs at the press of a button. As a result, CTC enrichment with microfluidic ratchet devices can now be performed by an ordinary biology laboratory technician without any specialized training. Finally, devices were used to enrich CTCs from patients with mCRPC where CTCs were detected in 3 out of 11 patients with an average count of 78. Single cells were extracted from enriched samples and sequenced, and in a patient with paired cfDNA data were found to contain TP53, PTEN, and FOXA1 mutations commonly associated with prostate cancer. Furthermore additional mutations not present in cfDNA were detected in CTCs, showcasing the potential value of CTCs over cfDNA.  6.2 Limitations One of the key limitations is the timescale of the epoxy device manufacturing pipeline. Currently, it relies on curing at room temperature for ~24 hours where it is then demoulded and bonded. This timeframe is critical as the epoxy will be in a semi-cured state where it is still soft enough that heat treatment can be used to bond it. To ramp up production, further investigation is required into either accelerating the curing process or determining a new bonding method.  6.3 Future Work The revised microfluidic ratchet device and instrumentation setup can be used as a work platform for various future post-separation studies. Particular topics of interest include drug treatment studies in mouse xenograft models, and growing CTCs in culture. The throughput of the microfluidic ratchet device could also be further increased by multiplexing the device, with one potential implementation stacking devices vertically to share common control channels.  59  References 1. Ashworth, T. A case of cancer in which cells similar to those in the tumours were seen in the blood after death. Aust. Med. J. 14, 146–7 (1869). 2. Gupta, G. P. & Massagué, J. Cancer Metastasis: Building a Framework. Cell 127, 679–695 (2006). 3. De Bono, J. S. et al. Circulating tumor cells predict survival benefit from treatment in metastatic castration-resistant prostate cancer. Clin. Cancer Res. 14, 6302–6309 (2008). 4. Cristofanilli, M. et al. Circulating Tumor Cells, Disease Progression, and Survival in Metastatic Breast Cancer. N. Engl. J. Med. 351, 781–791 (2004). 5. Allard, W. J. et al. Tumor cells circulate in the peripheral blood of all major carcinomas but not in healthy subjects or patients with nonmalignant diseases. Clin. Cancer Res. 10, 6897–6904 (2004). 6. Miller, M. C., Doyle, G. V & Terstappen, L. W. M. M. Significance of Circulating Tumor Cells Detected by the CellSearch System in Patients with Metastatic Breast Colorectal and Prostate Cancer. J. Oncol. 2010, 617421 (2010). 7. Litvinov, S. V., Velders, M. P., Bakker, H. A. M., Fleuren, G. J. & Warnaar, S. O. Ep-CAM: A human epithelial antigen is a homophilic cell-cell adhesion molecule. J. Cell Biol. 125, 437–446 (1994). 8. Gorges, T. M. et al. Circulating tumour cells escape from EpCAM-based detection due to epithelial-to-mesenchymal transition. BMC Cancer 12, 178 (2012). 9. Kalluri, R. & Weinberg, R. a. Review series The basics of epithelial-mesenchymal transition. J. Clin. Invest. 119, 1420–1428 (2009). 10. Schleip, K. Zur Diagnose von Knochenmarkstumoren aus dem Blutbefunde. Z. klin. Med 59, 261 (1906). 11. Aschoff, L. Ein Fall von Myelom. München. med. Wchnschr. 53, 337 (1906). 12. Marcus, H. Krebszellen im strömenden Blut? J. Cancer Res. Clin. Oncol. 16, 217–230 (1917). 13. Quensel, U. Zur Kenntnis des Vorkommens von Geschwulstzellen im zirkulierenden Blute. Upsala läkaref. förh 1, 26 (1921). 14. Pool, E. H. & Dunlop, G. R. Cancer Cells in the Blood Stream. Am. J. Cancer 21, 99–103 (1934). 15. Ward, G. THE BLOOD IN CANCER WITH BONE METASTASES. Lancet 181, 676–677 (1913). 16. Papanicolaou, G. N. The cell smear method of diagnosing cancer. Am J Public Heal. 38, 202–5 (1948). 17. ROBERTS, S., WATNE, A., McGRATH, R., McGREW, E. & COLE, W. H. Technique and results of isolation of cancer cells from the circulating blood. AMA. Arch. Surg. 76, 334–46 (1958). 18. Engell, H. C. Cancer Cells in the Blood. Ann. Surg. 149, 457–461 (1959). 19. Grove, W. J., Watne, A. A. & Jonasson, O. M. Vascular Dissemination of Cancer in Children. 60  AMA Arch. Surg. 78, 698–702 (1959). 20. Park, E. S. et al. Continuous Flow Deformability-Based Separation of Circulating Tumor Cells Using Microfluidic Ratchets. Small 12, 1909–1919 (2016). 21. Harouaka, R. a., Nisic, M. & Zheng, S. Y. Circulating tumor cell enrichment based on physical properties Ramdane. J Lab Autom. 18, 1–24 (2013). 22. Park, S. et al. Morphological differences between circulating tumor cells from prostate cancer patients and cultured prostate cancer cells. PLoS One 9, (2014). 23. Ligthart, S. T. et al. Circulating Tumor Cells Count and Morphological Features in Breast, Colorectal and Prostate Cancer. PLoS One 8, e67148 (2013). 24. McFaul, S. M., Lin, B. K. & Ma, H. Cell separation based on size and deformability using microfluidic funnel ratchets. Lab Chip 12, 2369–2376 (2012). 25. Guo, Q., McFaul, S. M. & Ma, H. Deterministic microfluidic ratchet based on the deformation of individual cells. Phys. Rev. E - Stat. Nonlinear, Soft Matter Phys. 83, 1–5 (2011). 26. Lim, C. T., Zhou, E. H. & Quek, S. T. Mechanical models for living cells - A review. J. Biomech. 39, 195–216 (2006). 27. Hochmuth, R. M. Micropipette aspiration of living cells. J. Biomech. 33, 15–22 (2000). 28. Haines, W. B. Studies in the physical properties of soil. V. The hysteresis effect in capillary properties, and the modes of moisture distribution associated therewith. J. Agric. Sci. 20, 97 (1930). 29. Chudziak, J. et al. Clinical evaluation of a novel microfluidic device for epitope-independent enrichment of circulating tumour cells in patients with small cell lung cancer. Analyst 141, 669–678 (2016). 30. Xu, L. et al. Optimization and evaluation of a novel size based circulating tumor cell isolation system. PLoS One 10, 1–24 (2015). 31. Hvichia, G. E. et al. A novel microfluidic platform for size and deformability based separation and the subsequent molecular characterization of viable circulating tumor cells. Int. J. Cancer 138, 2894–2904 (2016). 32. Gupta, V. et al. ApoStream, a new dielectrophoretic device for antibody independent isolation and recovery of viable cancer cells from blood. Biomicrofluidics 6, 1–14 (2012). 33. Balasubramanian, P. et al. Antibody-independent capture of circulating tumor cells of non-epithelial origin with the ApoStream® system. PLoS One 12, e0175414 (2017). 34. O’Shannessy, D. J., Davis, D. W., Anderes, K. & Somers, E. B. Isolation of circulating tumor cells from multiple epithelial cancers with ApoStream for detecting (or monitoring) the expression of folate receptor alpha. Biomark. Insights 11, 7–18 (2016). 61  35. Warkiani, M. E. et al. Ultra-fast, label-free isolation of circulating tumor cells from blood using spiral microfluidics. Nat. Protoc. 11, 134–148 (2015). 36. Khoo, B. L. et al. Clinical validation of an ultra high-throughput spiral microfluidics for the detection and enrichment of viable circulating tumor cells. PLoS One 9, 1–7 (2014). 37. Warkiani, M. E. et al. An ultra-high-throughput spiral microfluidic biochip for the enrichment of circulating tumor cells. Analyst 139, 3245 (2014). 38. Campton, D. E. et al. High-recovery visual identification and single-cell retrieval of circulating tumor cells for genomic analysis using a dual-technology platform integrated with automated immunofluorescence staining. BMC Cancer 15, 360 (2015). 39. Campton, D. et al. High-recovery multiplex analysis of circulating tumor cells by density-based enrichment, automated platform immunofluorescence staining, and digital microscopy. Cancer Res. 74, 3072 (2014). 40. Kaldjian, E. et al. Multi-level analysis of circulating tumor cells in advanced prostate cancer using AccuCyte® – CyteFinder®. Proc. 22nd Annu. Prostate Cancer Found. Sci. Retreat 7 (2015). 41. Desitter, I. et al. A new device for rapid isolation by size and characterization of rare circulating tumor cells. Anticancer Res. 31, 427–441 (2011). 42. Freidin, M. B. et al. An assessment of diagnostic performance of a filter-based antibody-independent peripheral blood circulating tumour cell capture paired with cytomorphologic criteria for the diagnosis of cancer. Lung Cancer 85, 182–185 (2014). 43. Awe, J. A., Saranchuk, J., Drachenberg, D. & Mai, S. Filtration-based enrichment of circulating tumor cells from all prostate cancer risk groups. Urol. Oncol. Semin. Orig. Investig. 1–10 (2017). doi:10.1016/j.urolonc.2016.12.008 44. Mu, Z. et al. Detection and Characterization of Circulating Tumor Associated Cells in Metastatic Breast Cancer. Int. J. Mol. Sci. 17, 1665 (2016). 45. Uyuklu, M. et al. Effects of storage duration and temperature of human blood on red cell deformability and aggregation. Clin. Hemorheol. Microcirc. 41, 269–278 (2009). 46. Lim, E. J. et al. Inertio-elastic focusing of bioparticles in microchannels at high throughput. Nat. Commun. 1–9 (2014). doi:10.1038/ncomms5120 47. Heid, C., Stevens, J., Livak, K. & Williams, P. M. Real time quantitative PCR. Genome Res. 6, 986–994 (1996). 48. Ang, R. R. Spectral image cytometry for circulating tumor cell identification. (University of British Columbia, 2017). doi:10.14288/1.0340303 49. Espina, V. et al. Laser-capture microdissection. Nat. Protoc. 1, 586–603 (2006). 50. Annala, M. et al. Treatment Outcomes and Tumor Loss of Heterozygosity in Germline DNA 62  Repair-deficient Prostate Cancer. Eur. Urol. 72, 34–42 (2017). 51. Schwarzenbach, H., Hoon, D. S. B. & Pantel, K. Cell-free nucleic acids as biomarkers in cancer patients. Nat. Rev. Cancer 11, 426–437 (2011). 52. Pantel, K. & Alix-Panabières, C. Real-time liquid biopsy in cancer patients: Fact or fiction? Cancer Res. 73, 6384–6388 (2013). 53. Calabuig-Fariñas, S., Jantus-Lewintre, E., Herreros-Pomares, A. & Camps, C. Circulating tumor cells versus circulating tumor DNA in lung cancer—which one will win? Transl. Lung Cancer Res. 5, 466–482 (2016). 54. De Mattos-Arruda, L. et al. Circulating tumour cells and cell-free DNA as tools for managing breast cancer. Nat. Rev. Clin. Oncol. 10, 377–389 (2013). 55. Zhang, L. et al. Meta-analysis of the prognostic value of circulating tumor cells in breast cancer. Clin. Cancer Res. 18, 5701–5710 (2012). 56. Cohen, S. J. et al. Relationship of circulating tumor cells to tumor response, progression-free survival, and overall survival in patients with metastatic colorectal cancer. J. Clin. Oncol. 26, 3213–3221 (2008). 57. Lohr, J. G. et al. Whole-exome sequencing of circulating tumor cells provides a window into metastatic prostate cancer. Nat Biotechnol 32, 479–484 (2014). 58. Heitzer, E. et al. Complex tumor genomes inferred from single circulating tumor cells by array-CGH and next-generation sequencing. Cancer Res. 73, 2965–2975 (2013). 59. Dago, A. E. et al. Rapid phenotypic and genomic change in response to therapeutic pressure in prostate cancer inferred by high content analysis of single Circulating Tumor Cells. PLoS One 9, (2014). 60. Shaw, J. A. et al. Mutation analysis of cell-free DNA and single circulating tumor cells in metastatic breast cancer patients with high circulating tumor cell counts. Clin. Cancer Res. 23, 88–96 (2017). 61. Hollstein, M., Sidransky, D., Vogelstein, B. & Harris, C. C. P53 Mutations in Human Cancers. Science (80-. ). 253, 49–53 (1991). 62. Barbieri, C. E. et al. Exome sequencing identifies recurrent SPOP, FOXA1 and MED12 mutations in prostate cancer. Nat. Genet. 44, 685–9 (2012). 63. Li, J. et al. PTEN , a Putative Protein Tyrosine Phosphatase Gene Mutated in Human Brain , Breast , and Prostate Cancer and Ramon Parsons Published by : American Association for the Advancement of Science Stable URL : http://www.jstor.org/stable/2893082 JSTOR is a not-. Science (80-. ). 275, 1943–1947 (1997).  

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
https://iiif.library.ubc.ca/presentation/dsp.24.1-0354561/manifest

Comment

Related Items