DTC-Flow: a flow cytometry-based detection platform for characterizing bone marrow disseminated tumor cells in breast cancer

Abstract

The presence of bone marrow (BM) disseminated tumor cells (DTC_bm) identifies early-stage breast cancer patients at increased risk of recurrence and poorer overall survival. However, limitations in detecting DTC_bm by standard immunohistochemical approaches have hampered clinical application. To address this gap, we developed a flow cytometry-based method, DTC-Flow, that enables the sensitive and efficient detection and molecular characterization of breast cancer DTC_bm. Our analysis identified HER2 as a sensitive marker for detecting breast cancer cells, including those lacking HER2 amplification are claudin-low. DTC-Flow using a HER2/EpCAM/CD45 marker panel enabled >90% cancer cell recovery and sensitivity of one cancer cell per million nucleated BM cells across a range of breast cancer subtypes. Molecular analyses of DTC-Flow-sorted DTC_bm from metastatic patients suggested a quiescent state and demonstrated their close genomic relationship to primary/metastatic tumors, as well as continued genetic evolution. In early-stage breast cancer patients, DTC-Flow detected DTC_bm with greater sensitivity than cytokeratin-based immunohistochemical approaches. Our data support the development of DTC-Flow as a sensitive and specific platform to identify breast cancer patients harboring DTC_bm and better understand the biology of minimal residual disease. Ultimately, this platform could enable the selection of personalized therapeutic approaches based on molecular features of DTC_bm, monitoring of DTC_bm to assess the efficacy of such therapies, and the development of novel therapeutic approaches targeting unique biological vulnerabilities of DTCs in order to eradicate these cells before they can give rise to lethal recurrent cancers.

Fig. 1. Assessment of EpCAM and HER2 expression in cell lines representing major human breast cancer subtypes.

Flow cytometric analysis of EpCAM (a, b) and HER2 (c, d) expression in human breast cancer cell lines representative of the major breast cancer subtypes. Histogram plots display cells stained with APC anti-EpCAM (a) or APC anti-HER2 (c) antibodies in red and unstained control cells in blue, along with the percentage of EpCAM-positive or HER2-positive cells for each cell line. The numbers at the top right of each histogram plot are the fold-change in median fluorescence intensity for stained cells compared with unstained cells for each cell line. Bar graphs display cell surface expression levels (median fluorescence intensity) for EpCAM (b) and HER2 (d) in an exemplar experiment for a panel of cell lines representing luminal, basal, and claudin-low breast cancer subtypes. Hatched bars in (d) indicate cell lines with known HER2 amplification and/or overexpression (HER2+). (BMC = CD45– bone marrow cells). (e) Summary data depicting the percentage of cells within each tested human breast cancer cell line that express detectable cell surface HER2 (white), EpCAM (gray), or both markers (black). Data shown are mean percentages across an average of four independent experiments per cell line.

Fig. 2. DTC-Flow assay workflow and gating strategy for analysis of bone marrow aspirates.

a Schematic depicting the DTC-Flow assay “stain/lyse/no wash” sample processing protocol. b Overall DTC-Flow gating strategy. The third dot plot shows the establishment of DTC gating cutoffs for HER2 and EpCAM. The polygonal gate shown in blue reflects the DTC gating cutoffs, defined by background staining levels in healthy donor control bone marrow aspirates (BMAs) (blue shaded area, “non-DTCs”), that were used to classify each bone marrow cell (BMC) as a DTC or not a DTC. The quadrant gates shown in black, defined using control BMAs in which HER2 and EpCAM antibodies were omitted from the staining protocol, were used to define cells as positive or negative for EpCAM and HER2 staining. c Dot plots showing background levels of HER2 and EpCAM expression (blue shaded area) in CD45– BMCs from representative control donors. Arrows indicate the “EpCAM-low” normal BMC population present in most control BMAs. d Dot plots showing HER2 and EpCAM expression levels in representative human breast cancer cell lines, relative to the gating cutoff established for CD45– cells from control BMAs. Percentages of cells from each cell line with HER2 and/or EpCAM levels above the control bone marrow gating cutoff are shown.

Fig. 3. Improved recovery of spiked tumor cells using DTC-Flow assay.

a Assessment of the percentage of spiked tumor cells successfully recovered using DTC-Flow assay, as compared to samples without pre-enrichment. Average recovery of spiked tumor cells was 93.7 ± 6.1% (mean ± standard deviation, n = 4). b Assessment of spiked tumor cell recovery at 3-log range of spiking levels, with linear regression analysis. c Flow cytometry dot plots demonstrating detection of spiked human breast cancer cells (MDA-MB-453, orange boxed area) in control bone marrow at spiking ratios of 1 tumor cell per 100,000 nucleated bone marrow cells (BMCs) (left panel) or per 1 million BMCs (right panel). d Assessment of DTC-Flow assay performance at low spiking ratios. Spiking ratios shown are tumor cells per given number of nucleated BMCs. e Comparison of spiked tumor cell recovery (1:10,000 spiking ratio) for the DTC-Flow assay as compared to the current standard DTC detection method, consisting of Lymphoprep™-based mononuclear cell isolation followed by IHC detection of cytokeratin expression (CK-IHC). (*p = 0.0221; paired two-tailed t-test, n = 3).

Fig. 4. Detection of bone marrow DTCs in breast cancer patients with ER+/PR+ metastatic/recurrent tumors.

a Flow cytometry dot plots showing the level of expression of HER2 and EpCAM in bone marrow (BM) DTCs (DTC_bm) (unshaded area, arrows) detected by DTC-Flow in two patients with ER+/PR+ metastatic breast cancer. The table below indicates DTC counts for each patient along with DTC frequencies relative to total nucleated BM cells (BMCs) prior to FACSFocus pre-enrichment. b ImageStream analysis of HER2, EpCAM, and CK expression in sorted DTC_bm from patient B compared to CD45 + BM cells and human breast cancer cell lines. Acquisition channels and fluorochromes used are indicated below the images. (BF = brightfield; scale bars = 10 µm). c Heat map depicting targeted qPCR analysis of mRNA expression in sorted DTC_bm pools (~10 cells each) from two metastatic breast cancer patients, along with sorted ~10-cell aliquots of representative human breast cancer cell lines. Sorted Jurkat T cells and normal BMC cDNA served as positive controls for BMC markers. Values shown are threshold cycles (Ct) for the indicated genes. d Plot displaying mRNA expression levels of proliferation/cell cycle marker genes in DTC_bm versus average expression in four cultured human breast cancer cell lines (HCC1954, T47D, MDA-MB-231, and Hs 578 T) (ND = not detected). Data are presented as mean ± standard deviation of three technical replicates. (*p < 0.05, **p < 0.01, ***p < 0.001 vs. breast cancer cell line average; one-way ANOVA with Bonferroni’s multiple comparisons test).

Fig. 5. Genomic profiling of bone marrow DTCs in comparison to primary or metastatic tumors.

Non-synonymous COSMIC mutations detected by whole-exome sequencing of bone marrow (BM) DTCs (DTC_bm) from patient A (~200 cell pool, a) or patient B (~10-cell pool, c), along with the corresponding primary or metastatic tumor. Predicted functional impact and pathogenic status of each mutation are per the COSMIC database. Numbers in parentheses indicate the variant allele frequency of each mutation. Mutations were not detected in germline DNA from sorted CD45 + BM cells or buffy coat cells. Mutations annotated as Pathogenic in COSMIC are in bold. Discordances between DTC and tumor mutation status are in blue. b PIK3CA sequencing demonstrating that DTC_bm from patient A exhibit the mutation c.3140A>G, which was also present in the corresponding metastatic tumor. Copy number (CN) plots generated from sWGS data from DTC_bm, primary or metastatic tumor FFPE gDNA, or CD45+ normal BM cells from patients A (d) and B (e). Insets show regions of CN gain (red arrows) and CN loss (blue arrows) shared between DTC_bm and the corresponding primary or metastatic tumors, but not CD45 + BM cell germline DNA. For patient B, repeat sequencing libraries were prepared from the same WGA products for CD45+ normal BM cells and DTC_bm.

Fig. 6. DTC detection frequency, counts, and phenotypes in PENN-SURMOUNT bone marrow aspirates screened by DTC-Flow.

a Flow cytometry dot plots showing the level of expression of HER2 and EpCAM in bone marrow aspirates (BMAs) analyzed by DTC-Flow in four early-stage PENN-SURMOUNT patients. Putative DTCs (unshaded area) are indicated by arrows. b Comparison of DTC-IHC vs. DTC-Flow results for PENN-SURMOUNT patient BMAs (n = 42). Concordant IHC and Flow results are shown in black, while discordant DTC results are shown in red. DTC-Flow identified 83% (10 of 12) of those BMAs screened as DTC-positive by IHC, while DTC-IHC only identified 37% (10 of 27) of the samples screened as DTC-positive by DTC-Flow. c Box plot showing distribution of DTC counts for DTC-positive patients as identified by either DTC-IHC (n = 10) or DTC-Flow (n = 27). DTC counts for the IHC assay are per 5 million bone marrow mononuclear cells, while Flow DTC counts are per 0.5 mL BMA. (Boxes extend from the 25th to 75th percentiles; center lines = median; whiskers = maximum and minimum values). d DTC counts in BMAs as determined by DTC-Flow (per 0.5 mL BMA). Total DTC counts are shown for PENN-SURMOUNT patient BMAs (blue dots) and healthy donor control BMAs (pink dots) across HER2; EpCAM phenotypes. Overall median counts (including DTC– samples) are indicated by lines.