DETECTION AND ISOLATION OF CIRCULATING TUMOR CELLS FROM WHOLE BLOOD USING A HIGH-THROUGHPUT MICROCHIP SYSTEM

Circulating tumor cells (CTCs) have been proved to possess great value and potential in detection, diagnosis, and prognosis of non-haematologic cancers. Their unique characteristics in providing both phenotypic as well as genotypic information make them highly valuable in liquid biopsy assays. At the same time, though numerous studies and research have been done, identification and enumeration of CTCs is still technically challenging due to their rarity and heterogeneity. The primary goal of the thesis is to develop a CTC detection and isolation system with ultra-high sensitivity and purity, while keeping it fast and scalable. We proposed a microfluidic system that integrates positive immunomagnetic capturing, high-throughput parallel flow and size filtration. In this thesis, two generations of the system have been developed to achieve the goal, and are approved to be able to effectively detect and isolate CTCs from hundreds of breast cancer blood samples in real clinical applications.

The first-generation system is based on a sandwich-structured microfluidic chamber, which has a micro-aperture chip as the core to detect and isolate immunomagnetically targeted CTCs. The system achieves high detection yield (>95%) and purity (>99.9998% depletion of leukocytes) by streamlining the workflow and using unprocessed whole blood (without centrifuging), as well as utilizing an advanced surface coating approach to passivate the microchip surface. We first demonstrate experiments for determining the optimal detection parameters. Then we characterize the system by isolating deterministically spiked 1, 10, and 100 single MCF-7 breast cancer cells into tubes of whole blood, and show that >95% of cells were captured. A detection yield of 100% from single cell spiking experiments (n = 6) demonstrates excellent detection capability and repeatability of the system. We finally demonstrate the use of the system for CTC detection in the context of a phase II clinical trial of early-stage triple-negative breast cancer (TNBC) patients. As a part of the trial, 182 blood samples were collected from 124 early-stage TNBC patients at high-risk of relapse. We detected CTCs in 36.3% of patients who had already completed chemotherapy and surgery for curative intent and were thus nominally expected to have very few to zero CTCs. Moreover, increasing CTC count from the same patients shows good correlation with their clinical course. The ability to detect CTCs’ presence using this first-generation system illustrates its important clinical utility.

The second-generation system applies a similar detection strategy but employs an upgraded microchip and device, as well as a further streamlined process flow to achieve an even higher detection efficiency, especially for capturing the target cells with low surface marker expression level. We first did modeling and simulation of the new system to find the optimal magnet configuration and verify the detection sensitivity improvement on the first-generation system. Then we characterized the new system by detecting spiked JEG-3 and JAR cells in both cell culture medium and human blood. The result demonstrates that the detection yield increased by ~20% using the second-generation system under the same experiment condition. Next, we applied the system to a phase I clinical trial for CTC detection from metastatic triple-negative breast cancer (mTNBC) patient blood samples. CTCs of mTNBC are known to with in the low marker expression phenotype, which requires ultra-high detection sensitivity. Our system captured CTCs from 48 out of 102 (47%) blood samples, the positivity rate agrees with the conclusions from other studies and presents the reliability to the system. Finally, we explored a novel 4-marker panel for CTC detection from mTNBC patient blood samples. We conducted paired comparisons using the 4-marker panel versus a single marker for detection. The 4-marker panel yielded more CTCs in 5/8 complete paired assessments, and less CTCs in 1/8. The association missed the significance level only slightly (p = 0.08), and the result strongly illustrates the potential for using the panel to cover the mTNBC cells’ heterogeneity for enhanced CTC detection. Furthermore, the presence of CTCs from blood samples correlates well with the patient’s disease progression.

Finally, we demonstrated downstream analysis ability of the CTCs detected by the second-generation system. Captured CTCs can be readily released from our system without any loss or damage to a secondary microchip device to be further isolated as single cells, and picked up individually for downstream analysis like DNA/RNA sequencing or single-cell cultivation. Directions for future work is also discussed. We envision this versatile and efficient system to be highly beneficial in a broad range of clinical and research applications regarding CTCs.