Microfluidic Isolation of Circulating Tumor Cell Clusters by Size and Asymmetry

Abstract

Circulating tumor cell clusters (CTC clusters) are potent initiators of metastasis and potentially useful clinical markers for patients with cancer. Although there are numerous devices developed to isolate individual circulating tumor cells from blood, these devices are ineffective at capturing CTC clusters, incapable of separating clusters from single cells and/or cause cluster damage or dissociation during processing. The only device currently able to specifically isolate CTC clusters from single CTCs and blood cells relies on the batch immobilization of clusters onto micropillars which necessitates long residence times and causes damage to clusters during release. Here, we present a two-stage continuous microfluidic chip that isolates and recovers viable CTC clusters from blood. This approach uses deterministic lateral displacement to sort clusters by capitalizing on two geometric properties: size and asymmetry. Cultured breast cancer CTC clusters containing between 2-100 + cells were recovered from whole blood using this integrated two-stage device with minimal cluster dissociation, 99% recovery of large clusters, cell viabilities over 87% and greater than five-log depletion of red blood cells. This continuous-flow cluster chip will enable further studies examining CTC clusters in research and clinical applications.

Figure 1

Two-Stage Cluster Capture Operating Principles: (a) Stage 1 and Stage 2 device inlet, outlets and fluidic paths. (b) Array of cylindrical micropillars in Stage 1 deflects large clusters from other blood cells using deterministic lateral displacement. (c) A 90 µm Stage 1 ceiling allows clusters to align along the Z-axis. (d) Array of asymmetric pillars in Stage 2 deflects small clusters. (e) A 30 µm Stage 2 ceiling forces clusters to align in the X-Y plane. (f) Photograph of device running whole blood (red) with colored PBS buffer (blue) next to U.S. quarter dollar coin. (g) Diagram showing asymmetric pillars inducing cluster rotation. (h) Symmetric single cells do not deflect while asymmetric small clusters sort by their longitudinal axes in Stage 2 and large clusters sort by size in Stage 1. Schematics not to scale.

Figure 2

Whole Blood Device Operation. (a) Photomicrograph of: mid-Stage 1 showing large cluster deflected into product stream (white arrow), (b) exit of Stage 1, two large clusters exiting into Stage 1 product (white arrows), and (c) Stage 2 showing two small clusters being deflected out of blood stream (black arrows). Red blood cells are dark. Scale bar: 200 µm. (d) Image of Blood input vs. Waste, Stage 1 and Stage 2 outputs showing red blood cell content. (e) Log (10) depletion of red blood cells and white blood cells in Stage 1 (dark) and Stage 2 (light) outputs. Experiments conducted in triplicate, error bars represent one standard deviation.

Figure 3

Image Processing and Device Performance. Photomicrographs of frames taken of CTC clusters recovered from whole blood. (a) raw unprocessed images and (b) processed images of Stage 1 (left), Stage 2 (middle) and Waste (right) outputs with outlines: large clusters (red), small clusters (yellow) and single cells (blue). Scale bar: 100 µm. (c) Fraction of total large clusters (red), small clusters (yellow) and single cells (blue) that partition to Stage 1, Stage 2 and Waste outputs for cell/clusters run in buffer. (d) Total recovery of large and small clusters into Stage 1 and Stage 2 outputs for runs in buffer. (e) Fraction of cells/clusters that partition to outputs for cells/clusters run in whole blood. (f) Total recovery of large and small clusters into Stage 1 and Stage 2 outputs for runs in whole blood. Recovery experiments conducted in quintuplicate. (g) Comparison of total number of cells/clusters spiked into whole blood before device operation (dark) vs. number collected from all output streams (light). Above experiments all conducted at 0.5 mL/hr blood processing rates. (h) Viability of CTCs in Stage 1 product (dark circles) and Stage 2 product (hollow diamonds) 1 hour after operation at different sample processing rates vs. untreated controls (represented by 0 mL/hr). Experiment conducted in duplicate, error bars represent one standard deviation and at least 200 cells were counted per condition.