Discussion
To our knowledge, this study represents the largest analysis evaluating a set of five biomarkers in CTCs from sequential blood samples of patients with non-metastatic BC. In our work, we found a slightly higher detection rate in a similar amount of blood in comparison with those other studies using the CellSearch System (Veridex, LLC) before any treatment. Besides, differences in the CTC count between the baseline samples and the post-treatment samples were not observed. However, owing to the small number of patients in each treatment group, our analysis was not sufficiently powered to draw any definitive conclusion in this endpoint.
Our observations indicate that phenotyping/genotyping analysis of CTCs is highly dependent on the detection rate in three different tubes and the low number of cells captured in the non-metastatic setting. Of note, the threshold of at least one CTC has been used before as a prognostic factor in patients with non-metastatic BC, reflecting that CellSearch shows similarly low CTCs counts in this setting. Optimization of CTC assays for high-throughput processing will be required to allow a comprehensive characterization of CTCs and large-scale clinical trials that use this emerging technology.
Several studies have addressed the correlation between CTC status and clinical-pathological parameters, but reported observations are still controversial. Lang and colleagues found that CTCs were more frequently found in patients with HER2 tumors, whereas other researchers have not found any association between the CTC status and HER2 tumors. In this study, CTC cells were found more frequently in patients with HER2 tumors, whereas other classic clinical-pathological parameters studied did not show a correlation with CTC status.
Non-detectable CTCs were observed in patients who were younger than 50 years old had primary tumors with HER2 amplification and G1-G2, and the higher risk was in the G1-G2 group. As in the metastatic setting, patients with poor prognostic factors had non-detectable CTCs. Such apparently contradictory results may be explained, in part, by the acquisition of mesenchymal antigens during the epithelial-mesenchymal transition (EMT) which facilitate the process of invasion and the metastatic cascade. EMT-derived CTCs may have modulated their phenotype and acquired mesenchymal-like properties difficult to detect with the currently used detection methods. As a result, patients with non-detectable CTCs and poor prognostic factors might represent a subset of patients with partial or complete EMT phenotype instead of an unequivocal undetectable CTC population.
Intriguingly, less than 35% of our population did not change their CTC status, suggesting that CTC population follows a pattern of neutral drift dynamics. This finding diverges with those of previous data that reported a lower incidence of CTC detection rate after systemic treatment, especially in those patients who had received anti-HER2 therapies. This discrepancy may be explained, at least in part, by the fact that we analyzed 30 mL instead of 7.5 mL and because only 14.3% of our patients received trastuzumab as part of their treatment.
In concordance with the study by Fehm and colleagues, our findings showed that ER and PR expression in CTCs were not correlated with ER or PR expression in the primary tumor. Heterogeneous CTC subpopulations with different HR phenotypes coexisting in the same blood sample were observed. Remarkably, an RNA-based method is not able to evaluate individual cells and detect heterogeneity among a CTC population, whereas the IF approach provides an additional biologically relevant characterization of different CTC subpopulations. Thus, it could be speculated that distinct HR expression in CTCs in the same patients might, in part, explain differences in response to both endocrine and chemotherapy treatments, although this association needs to be further characterized.
Changes of ER/PR phenotype or persistence of CTC phenotypes other than the primary tumor phenotype was also observed in our study in the samples after treatment. As all of the patients with sequential samples received chemotherapy, it cannot be excluded that drug-induced changes and clonal selection may be influenced by the interaction of CTCs with chemotherapy.
EGFR protein was expressed in 27% of CTCs at baseline and did not correlate with clinical and pathological parameters except for HR tumors. Preclinical data have provided evidence that cross-talk between growth factor receptor (GFR) and ER pathway may mediate the development of endocrine therapy resistance in HR disease, although EGFR expression has been widely related to triple-negative BC tumors. The proposed biological mechanisms to explain how GFR signaling results in endocrine therapy resistance are conflicting. Thus, we hypothesized that EGFR CTCs might represent a potential negative biomarker of response to certain anti-cancer agents, including endocrine therapy in patients with HR BC.
Besides, less than 25% of the EGFR CTC patients became EGFR CTC after treatment, suggesting that conventional agents like chemotherapy or even trastuzumab eradicate partially EGFR CTC subpopulations. We fully acknowledge that our results should be interpreted with caution because the sample size is limited and the small numbers of events limit our conclusions.
HER2 overexpression of CTCs in patients with BC has been well characterized in recent studies. Discrepancies between HER2 status in CTCs and their corresponding primary tumors have been described in patients with early and metastatic BC. In our study, the rate for HER2-amplified CTCs detected by FISH was null at baseline, which differs from the HER2 CTC rate reported by other groups using an IF approach. The lack of HER2 CTCs may be influenced by the fact that CTC populations are heterogeneous, and analyzing such a small number of CTCs may underestimate HER2 populations. Besides, CTCs with (2+) HER2 IF staining remain unresolved and could justify partially HER2 CTCs. The optimal HER2 testing performance in CTCs has not been validated yet. Although IHC was the original method of assessment for HER2 status, IHC or IF alone cannot be recommended now for determining anti-HER2 treatment.
Previous studies have demonstrated that amplification of TOP2A in BC is not confined to those who are concomitantly HER2-amplified, suggesting that a proportion of HER2 patients exhibit TOP2A alterations. Our findings that two TOP2A-amplified CTCs come from HER2 primary tumors and that the HER2 gene is not co-amplified in these CTCs are consistent with previous observations described in primary BC tissues.
After three cycles of AT, co-amplification of HER2 and TOP2A in CTCs was observed in one patient with HER2 and TOP2A tumor. This finding is consistent with a shift in tumor genotype and possibly dependency to alternative signaling pathways in HER2 primary tumors. As all of the patients included in the sequential blood analysis received chemotherapy, it cannot be excluded that HER2 and TOP2A alterations in CTCs after treatment are influenced, at least in part, by the interaction with chemotherapy. According to other research groups, this finding opens up a window of opportunity because HER2 BCs with co-amplification of HER2 and TOP2A in CTCs may exquisitely benefit anti-HER2 agents as well as antracycline-based regimens.
HER2 CTCs were isolated in patients with either HER2 or HER2 primary tumors after systemic treatment and trastuzumab therapy when recommended. This finding suggests that in HER2-amplified BC, HER2 CTCs may have been selected by trastuzumab therapy. It is noteworthy that anti-HER2 agents are given in combination or sequentially with chemotherapy, and the precise mechanism by which HER2 CTCs persist is currently unknown.
Standard predictors for BC treatment selection are HR expression for endocrine therapy and HER2 status for anti-HER2 therapy. Among the whole set of biomarkers evaluated in CTCs, only EFGR CTCs were more frequent in luminal tumors compared with triple-negative and HER2-amplified tumors. It could be speculated that the association between EGFR CTCs and luminal BC patients is explained, in part, by an increase of cancer cells expressing EGFR involved in the paracrine loop in which epidermal growth factor produced by tumor-associated macrophages increases the invasiveness and migration of BC cells that express EGFR, although EGFR expression has been widely related to lower HR levels, higher proliferation, genomic instability, and HER2 overexpression. Although this association needs to be further characterized, luminal tumors might be more dependent than other BC subtypes on this mechanism that promotes cell migration and intravasation.