Molecular characterization of circulating tumor cells from patients with metastatic breast cancer reflects evolutionary changes in gene expression under the pressure of systemic therapy

Abstract

Resistance to systemic therapy is a major problem in metastatic breast cancer (MBC) that can be explained by initial tumor heterogeneity as well as by evolutionary changes during therapy and tumor progression. Circulating tumor cells (CTCs) detected in a liquid biopsy can be sampled and characterized repeatedly during therapy in order to monitor treatment response and disease progression.Our aim was to investigate how CTC derived gene expression of treatment predictive markers (ESR1/HER2) and other cancer associated markers changed in patient blood samples during six months of first-line systemic treatment for MBC. CTCs from 36 patients were enriched using CellSearch (Janssen Diagnostics) and AdnaTest (QIAGEN) before gene expression analysis was performed with a customized gene panel (TATAA Biocenter).Our results show that antibodies against HER2 and EGFR were valuable to isolate CTCs unidentified by CellSearch and possibly lacking EpCAM expression. Evaluation of patients with clinically different breast cancer subgroups demonstrated that gene expression of treatment predictive markers changed over time. This change was especially prominent for HER2 expression.In conclusion, we found that changed gene expression during first-line systemic therapy for MBC could be a possible explanation for treatment resistance. Characterization of CTCs at several time-points during therapy could be informative for treatment selection.

Keywords: Estrogen receptor (ER); circulating tumor cells; gene expression; human epidermal growth factor receptor 2 (HER2); metastatic breast cancer.

Figure 1. Study design

Flow chart of all included patients at the different sample taking time-points in the study. Samples described as “missed sample” were only lost at the subsequent time-point and these patients had not left the study.

Figure 2. Three examples of patients with HER2 positive breast cancer

After initiation of HER2 targeted therapy, none of the patients experienced PD during the study. (A) Patient 6 had no detectable CTCs at any time-point but had EpCAM expression at BL and at 4 months. Possibly, the lack of KRT19 expression lead to that CTCs were not detected by CellSearch and AdnaTest despite initial capture. HER2 expression was also found at 4 months. (B, C) Patients 21 and 26 both showed a response to therapy with regard to CTC number and cancer related gene expression in FU samples. ALDH1 expression continued to vary in patient 26, but this did not seem to affect prognosis. Note that the time course (X-axis) and the scale for CTC number (left Y-axis) varies between patients.

Figure 3. Three examples of patients with HR positive breast cancer

(A) Patient 3´s treatment was changed from endocrine therapy to second-line chemotherapy after progression at 2 months. Before start of therapy the patient had detectable ESR1+ and HER2+ CTCs that vanished after treatment initiation. The patient responded well to chemotherapy with an overall survival of 28 months. (B) Patient 12 had very few detactable CTCs at BL with the CellSearch method and was CTC negative during therapy. However, at 6 months there was an increase in KRT19, IGF1R and ALDH1 expression and the patient was later diagnosed with clinical progression at 10 months. It is possible that this increase in markers at 6 months was an early sign of treatment resistance. (C) Patient 23 initially responded to endocrine therapy but the CTC number increased at 6 months and the simultaneous increase in HER2 expression suggested a phenotype shift in the metastasis that could possibly have been therapeutically targeted. Note that the time course (X-axis) and the scale for CTC number (left Y-axis) varies between patients.

Figure 4. Three examples of patients with TNBC treated with chemotherapy

(A) Patient 8 had detectable CTCs at BL with both ESR1 and HER2 expression despite being clinically diagnosed with TNBC. After an initial response to therapy, the number of HER2+ CTCs started to increase at 3 months and the patient was diagnosed with PD after 4 months of treatment. Interestingly, EGFR expression was also detectable at 6 months indicating a phenotype change and a possible targetable marker in addition to HER2. (B) Patient 16 had a rapid disease progression but the increased HER2 expression at 1 month could possibly have been targetable. (C) Patient 33 also had a rapid progression and despite a decrease in both CTC number and in markers such as HER2 and EGFR after initiation of chemotherapy, the patient quickly experienced disease progression and death. The only marker visible in the blood at 3 months (two days before the patient died) was expression of the stem cell marker ALDH1. However, ALDH1 expression was not correlated to prognosis in this cohort. Note that the time course (X-axis) and the scale for CTC number (left Y-axis) varies between patients.

Figure 5. Mean-centered heatmap showing the expression of 32 genes detectable in at least one of the BL samples

Three distinct patient clusters denoted 1, 2 and 3 were found with all cell line cells (positive controls) in cluster 1 and healthy donor blood (negative controls) in clusters 2 and 3. Patient ID marked with red were CTC positive by either CellSearch or AdnaTest. Genes were clustered in four groups (a-d) where group d included the most commonly used markers for CTC positivity. Genes with very low expression could be found in group b. The highest expression of CTCs captured with the EMT1 or EMT2 kit is shown.

Figure 6. Kaplan-Meier plots for progression-free and overall survival by heatmap patient cluster

No patients from heatmap cluster 3 experienced disease progression within 6 months (A) or died during the course of the study (B). P-values from log-rank test.

Figure 7. Association of KRT19 gene expression to CTC number and progression-free survival (PFS)

(A) Of the conventionally used CTC markers, KRT19 gene expression (red circles, right axis) showed highest correlation to CTC number measured by CellSearch (grey filled circles, left axis) over time (R_S = 0.61). The red dotted line illustrates median KRT19 expression and the gray solid line illustrates median CTC number at the respective timepoints. (B) KRT19 expression in CTC blood samples taken before start of therapy (BL) was associated with shorter PFS.