(341d) Workflow for the Un-Biased Isolation and Genomic Characterization of Single Circulating Tumor Cells

Despite tremendous improvements in disease management, cancer remains the second leading cause of death in the United States today with over 600,000 cancer deaths projected to occur in 2023. Of these mortalities, greater than 90% are expected to be linked to metastasis – the process by which cancer spreads from one area of the body to another. Patients with metastatic cancers often show an extreme decrease in five-year survival rate relative to localized disease. Pancreatic cancer, considered to be the third deadliest cancer with an estimated 49,830 deaths per year, has a localized five-year survival rate of 41.6%, but an abysmal 3% survival rate in metastatic patients. A contributing factor to this discrepancy is that 52% of pancreatic cancers are not diagnosed until after metastases has already occurred. There is therefore a critically unmet need in the diagnosis of pre-metastatic pancreatic cancer patients.

Pancreatic cancer is traditionally diagnosed using tissue biopsies isolated from the primary tumor site. However, pancreatic tissue biopsies are often invasive and difficult to obtain due to the location of the pancreas. Liquid biopsies have emerged as a promising, minimally invasive alternative for cancer diagnosis. These biopsies analyze bodily fluids – most commonly blood – for various biomarkers, including circulating tumor cells (CTCs), circulating tumor DNA (ctDNA), and cancer-derived extracellular vesicles (EVs). Of these biomarkers, CTCs have shown strong potential for the breadth of information that can be obtained from an individual’s cancer. CTCs represent a viable population of cancerous cells, allowing for a number of downstream, multi-omic analysis approaches. However, CTCs are a heterogenous population of cells, making their bulk analysis difficult. In this study, we demonstrate a workflow for the unbiased isolation and genomic characterization of single CTCs from pancreatic cancer patients. The workflow is first optimized using a combination of conventional pancreatic cancer cell lines as well as patient-derived CTC cell lines immortalized by the Nagrath lab. As a proof of concept, the workflow is then validated using a blood sample collected from a metastatic pancreatic patient.

The workflow starts with sample preprocessing, including the removal of erythrocytes through Ficoll-Plaque density centrifugation and 5x dilution in phosphate buffered saline pH 7.4 (PBS). The sample is then processed through a microfluidic CTC enrichment device developed in the Nagrath lab called the Labyrinth. The Labyrinth uses inertial forces generated by Dean’s vortices in curved microfluidic channels to separate blood cells from the larger CTCs based on cell size. Unlike other methods of CTC isolation that isolate cells based on the presence of specific membrane antigens, the Labyrinth is an antigen-agnostic method, allowing for a more heterogenous population of CTCs to be enriched. Following Labyrinth enrichment, the samples are fixed in 2% paraformaldehyde (PFA) and stored at 4°C until further processing.

To differentiate our CTCs from background blood cell contamination, the samples are immunofluorescently stained using a combination of CTC identifying markers (pan-cytokeratin (pCK), epithelial cell adhesion molecule (EpCAM), CD133, epidermal growth factor receptor (EGFR)) and a white blood cell identifying marker (CD45). CTCs are defined as CD45- and any combination of pCK/EpCAM/EGFR/CD133+. To isolate individual CTCs for genomic analysis, the samples are then processed using the DEPArray system (Menarini Silicon Biosystems). The DEPArray utilizes the inherent membrane potential of cells and a grid of dielectrophoretic (DEP) cages to individually route and partition single cells. The DEPArray contains a built-in fluorescent microscope that allows users to individually select and isolate cell populations of interest based on fluorescent staining.

Following single cell isolation, the individual cells are processed for whole genome amplification (WGA) using the Ampli1 kit (Menarini Silicon Biosystems), library prep, quality control, and low-pass DNA sequencing. The resulting single-cell sequencing data is then processed with MSBiosuites, a cloud-based analysis software courtesy of Menarini Silicon Biosystems, to generate single-cell copy number variation (CNV) profiles. The Labyrinth-DEPArray pipeline was first optimized using three conventional pancreatic cancer cell lines: Panc-1, Capan-2, and BxPC-3. We also utilized two patient-derived CTC cell lines generated in the Nagrath lab. Finally, the workflow was validated end-to-end using a metastatic pancreatic patient sample.

To optimize the genomic characterization workflow, two to three cells from each of the pancreatic cell lines – both conventional and patient-derived CTC lines – were isolated using the DEPArray and processed for WGA. Quality control for each of the cells was evaluated using the Ampli1 QC kit (Menarini Silicon Biosystems). The QC kit uses polymerase chain reaction (PCR) to amplify and subsequently detect the cell DNA for the presence of four distinct regions on the genome across multiple chromosomes. The PCR product for each cell is visualized using gel electrophoresis and given a genome integrity index (GII) based on the number of bands present (0 to 4 bands, with 4 bands representing the highest quality cells). In order to perform low pass sequencing, a minimum of 2 bands must be present. In total, eight pancreatic cell line cells (median GII: 3) and 6 CTC cell line cells (median GII: 3.5) were selected for library preparation and low pass sequencing. The average concentration of genetic material following library prep was 26.24 ± 16.28 ng/µL per cell, with an average fragment size of 746 ± 233.79, well within the expected range of 100-2000 bp.

Following sequencing, CNV profiles were generated for each of the cells across the five cell lines. For the conventional cell lines, the copy number profiles were compared to previously published CNV data. The regions of copy number gains and losses were consistent across duplicate cells from the same cell lines. We also found several cancer-specific genes in regions of copy number variations consistent with previously published cell line data, highlighting the reliability of the workflow. CNV profiles for the patient-derived CTC cell lines were also generated. Of note, both cell lines showed consistent gains on parts of chromosome 8q, 10q, and 17q. Within these regions are the genes Myc (8q), a known oncogene, and FGFR2 (10q), a gene associated with cell proliferation, migration, and invasion of pancreatic ductal adenocarcinoma cells.

As a proof of concept, we validated the Labyrinth-DEPArray workflow end-to-end using a blood sample from a metastatic pancreatic cancer patient. Following Labyrinth enrichment, a portion of the sample was reserved for CTC enumeration analysis using immunofluorescent imaging. We found that the sample had a an average CTC concentration of 2.88 CTCs/mL of blood. The rest of the sample was fixed, stained, and loaded onto the DEPArray. We successfully managed to identify and isolate a CTC (EpCAM+/pCK-/CD45-) from the sample. Following WGA, all patient sample cells (CTC and WBC) with a GII of 2 or greater underwent library prep and low pass sequencing. The average concentration of genetic material following library prep for the patient sample cells was 14.43 ng/μL per cell, and the average fragment size was 507 bp. CNV profiles were generated for the identified CTC and a patient-matched WBC. The CTC profile had distinct regions of gains and losses that were not present in the WBC. Furthermore, the CTC had an arm-level copy number gain in chromosome 1q as well as an arm-level copy number loss in chromosome 5q. DDR2 (1q) has been shown to be associated with cell proliferation and survival in multiple cancers, while APC, CSF1R, and FGFR4 (5q) are known oncogenes.

From this study, we have demonstrated the reliability of the Labyrinth-DEPArray pipeline to accurately and consistently identity copy number variations from single CTCs. Unlike other traditional methods of CTC enrichment, such the FDA approved CellSearch system, the Labyrinth is a high-throughput (30 mL of blood per hour) technology capable of enriching an un-biased of cells. CNV profiles generated from conventional pancreatic cancer cell lines were shown to be consistent with previously published cell line data. The novel sequencing of two patient-derived CTC cell lines demonstrated the ability of the workflow to identify cancer-specific genes within regions of copy number gains and losses. This pipeline offers a diagnostic tool to perform longitudinal studies throughout the course of a patient’s treatment. CNVs have been highly correlated to differential gene expression, providing us with valuable, patient-specific information that is readily accessible. Additionally, cells isolated using the Labyrinth-DEPArray pipeline are more representative of a heterogenous cell population which allows for a more thorough characterization of a patient’s cancer. For the sake of this study, we limited our analysis to CNVs using low-pass sequencing. However, this workflow can be modified and adjusted to examine other genomic aberrations across various cancer types. In the future, we aim to apply this workflow to a larger patient cohort to evaluate how various CNV profiles within patients may correlate with disease progression.

Figure 1: Overview of Labyrinth-DEPArray workflow. 1) Cells are obtained from either culture or by processing blood through the Labyrinth system. 2) Cell samples are fixed and fluorescently stained to differentiate cell populations. 3) The cells are loaded onto the DEPArray system an individually partitioned for downstream analysis. 4-5) Single cells are processed for whole genome amplification and low-pass sequencing. Quality control is performed after each step. 6) Copy number variation (CNV) profiles are generated for each cell.