High-Throughput Label-Free Isolation of Heterogeneous Circulating Tumor Cells and CTC Clusters from Non-Small-Cell Lung Cancer Patients

Abstract

(1) Background: Circulating tumor cell (CTC) clusters are emerging as clinically significant harbingers of metastases in solid organ cancers. Prior to engaging these CTC clusters in animal models of metastases, it is imperative for technology to identify them with high sensitivity. These clusters often present heterogeneous surface markers and current methods for isolation of clusters may fall short. (2) Methods: We applied an inertial microfluidic Labyrinth device for high-throughput, biomarker-independent, size-based isolation of CTCs/CTC clusters from patients with metastatic non-small-cell lung cancer (NSCLC). (3) Results: Using Labyrinth, CTCs (PanCK+/DAPI+/CD45-) were isolated from patients (n = 25). Heterogeneous CTC populations, including CTCs expressing epithelial (EpCAM), mesenchymal (Vimentin) or both markers were detected. CTCs were isolated from 100% of patients (417 ± 1023 CTCs/mL). EpCAM- CTCs were significantly greater than EpCAM+ CTCs. Cell clusters of ≥2 CTCs were observed in 96% of patients-of which, 75% were EpCAM-. CTCs revealed identical genetic aberrations as the primary tumor for RET, ROS1, and ALK genes using fluorescence in situ hybridization (FISH) analysis. (4) Conclusions: The Labyrinth device recovered heterogeneous CTCs in 100% and CTC clusters in 96% of patients with metastatic NSCLC. The majority of recovered CTCs/clusters were EpCAM-, suggesting that these would have been missed using traditional antibody-based capture methods.

Keywords: CTC clusters; circulating tumor cells (CTCs); epithelial-to-mesenchymal transition (EMT); inertial microfluidics; non-small-cell lung cancer (NSCLC).

Figure 1

Optimization of Labyrinth for cell recovery. (A) The Labyrinth was loaded with red dye to show the device’s structure as well as the inlet and outlets. (B,C) Using the H1650 lung cancer cell line, different flow rates ranging from 2300–2500 μL/min were tested for inertial separation of cancer cells. Pre-labeled H1650 cell line and DAPI (4′,6-diamidino-2-phenylindole)-labeled white blood cells (WBCs) (1000 cells) were spiked into PBS and processed through the Labyrinth. Using a flow rate of 2500 μL/min, 82% ± 5% of H1650 cells were recovered from outlet #2 and 78% ± 18% of WBCs were removed through outlet #1. (D) Immunofluorescence staining optimization. Anti-human CD45 (cluster of differentiation 45) (green), anti-human PanCK (pan-Cytokeratin) (red), anti-human EpCAM (Epithelial cell adhesion molecule) (orange), and anti-human Vimentin (pink) antibodies were tested with lung cancer cell lines, H1975 and A549.

Figure 2

Isolation of circulating tumor cells (CTCs) from non-small-cell lung cancer (NSCLC) patients (n = 25). (A) Fluorescent microscope image of a single CTC. Cells are stained with DAPI (blue), PanCK (red) and CD45 (green). (B,C) Confocal microscopy images of some CTC clusters. (D) An individual bar plot of the number of CTCs recovered from NSCLC patient samples at baseline, using Labyrinth. The overall number of CTCs in NSCLC patient samples was 417 ± 1023 per mL, while healthy controls had 1 ± 1.7 CTCs per mL.

Figure 3

Identification of heterogeneous CTC subpopulations isolated from NSCLC patient samples (n = 23). (A) Fluorescent microscope images of different subpopulations of CTCs (CTCs, EpCAM+ CTCs, Vimentin+ CTCs, and Double+ CTCs). Cells are stained with DAPI (blue), CD45 (green), PanCK (red), EpCAM (orange), and Vimentin (pink). (B,C) The percentage of CTCs expressing both EpCAM (B) and Vimentin (C) recovered from each NSCLC patient sample (n = 23). The EpCAM+/− CTCs is shown in dark/light orange respectively and the Vimentin+/− CTCs is shown in dark/light pink respectively. An average of 31% of the captured CTCs were EpCAM+ and 69% were EpCAM− CTCs. An average of 45% of the captured CTCs were Vimenin+ and 55% were Vimentin− CTCs. The total number of CTCs/mL across all patient samples is shown on the bottom of the graph.

Figure 4

Isolation of CTC clusters from NSCLC patient samples (n = 25) by using Labyrinth. (A) Immunofluorescence staining images of the representative recovered CTC clusters in NCSLC patients. CTC clusters are stained with DAPI (blue), CD45 (green), PanCK (red), EpCAM (orange), and Vimentin (pink). (B) The percentage of CTCs in single (light pink) vs. cluster (red) forms. Across all patients, only one patient did not have CTC clusters. (C) Comparison between the total CTC numbers in single and cluster forms. Significantly higher numbers of clusters compared to the single CTCs were observed in the captured CTCs from NSCLC patients (n = 25) (p = 0.001). (D) Cell clusters of 2–8 CTCs were observed in 96% of patients. (E) Of the recovered CTC clusters, 41% displayed a mesenchymal or epithelial-to-mesenchymal transition (EMT) phenotype (97 CTC clusters/mL) (p = 0.007). Wilcoxon test analysis was used for comparing single vs. clusters CTCs and EpCAM+ vs. Vimentin+ CTCs. Analyses were conducted using GraphPad Prism. (F) Comparison of Kaplan–Meier progression-free survival (PFS) graph in patient samples (n = 24) with a higher number of clusters than single CTCs (n = 10) (red) and in patients who had a higher number of single CTCs (n = 14) (black). A higher number of CTC clusters than the single CTCs was correlated with worse PFS (p = 0.05). Log-rank (Mantel–Cox) tests were used to analyze the Kaplan–Meier PFS.

Figure 5

Genomic analysis of NSCLC patient samples using fluorescence in situ hybridization (FISH) analysis. (A–C) Recovered CTCs from selected patient samples with different mutations (ROS1, ALK, and RET) were evaluated. (A) A patient with ROS1 rearrangement showed 5′ deletion in some of the cells. (B) A patient with aberration in ALK showed ALK fusion. (C) A patient with aberration in RET showed 3′ deletion. Arrows indicate specific aberration in each gene. (D) Box plot of the total number of CTCs recovered from patients with different mutations (EGFR (n = 15), ROS1 (n = 6), and ALK (n = 4)). Analyses were conducted using GraphPad Prism.

Figure 6

CTC enumeration of one NSCLC patient over the treatment regime. In total, 10 different follow-up samples from patient P 14 were collected and the CTC numbers were evaluated over different treatments. According to the available clinical information, the CTC numbers tracked the patient’s outcome over the course of treatment.