Circulating tumor cells criteria (CyCAR) versus standard RECIST criteria for treatment response assessment in metastatic colorectal cancer patients

In colorectal cancer (CRC), metastasis is the main cause of death [1]. Distant metastasis is identified in approximately 25% of patients at initial diagnosis, and half of CRC patients will develop it [2]. During this process, circulating tumor cells (CTCs), detach from primary sites, enter the bloodstream and extravasate in distant organs. CTCs are now being studied in order to have a deeper understanding of the metastatic processes [3]. The phenotypic and genetic characterization of CTCs is especially important; as different subpopulations of CTCs can be detected in the blood of these patients [4]. These CTCs subclones can depict in real time the heterogeneity of a tumor, displaying its different abilities to elude therapies, and therefore, determining tumor response to treatment [5].

Metastatic colorectal cancer patients (mCRC) are currently subjected to a treatment regime combining chemotherapy with biological therapies. Bevacizumab, a monoclonal antibody inhibits the tyrosine kinase vascular endothelial growth factor A (VEGF-A) and blocks its transduction signal, through both VEGFR-1 and VEGFR-2. VEGF-A is a potent pro-angiogenic growth factor that stimulates proliferation, migration, and survival of endothelial cells. As it is one of the more important proteins expressed by tumor cells, VEGF is an important target of anticancer therapies. Cancer cells and tissues with high metabolic rates are characterized by hypoxia, which induces the transcription of VEGF protein [6]. Circulating VEGF binds with high affinity to VEGF receptors (VEGFR-1 and VEGFR-2) and its co-receptors neuropilin (NRP-1 and NRP-2), which are expressed on the surface of endothelial cells and play a critical role in the development of angiogenesis, by stimulating recruitment and proliferation of endothelial cells [7]. Bevacizumab is an IgG1 recombinant humanized monoclonal antibody that acts by selectively binding to circulating VEGF-A, creating a large molecule that renders it unable to bind to its cell surface receptors, reducing microvascular growth of tumor blood vessels and limiting blood supply of nutrients and oxygen to tumor tissues. In combination with intravenous 5-fluorouracil-based chemotherapy, it is indicated for first- or second-line treatment of patients with metastatic colorectal cancer. In combination with fluoropyrimidine–irinotecan- or fluoropyrimidine–oxaliplatin-based chemotherapy, it is indicated for second-line treatment of patients with metastatic colorectal cancer who have progressed on a first-line Bevacizumab containing regimen [8, 9]. First line bevacizumab has been demonstrated to improve overall survival (OS), progression-free survival (PFS) and treatment response rate in mCRC [10]. Despite these improvements, most mCRC patients will die due to disease progression [11]. With this in mind, extensive biomarker programs have now been built into numerous clinical studies with bevacizumab. However, predictive markers for bevacizumab treatment have yet to be validated [12].

In this clinical experimental work, we aimed to establish the predictive role of CTCs, and their expression of a treatment-associated marker (VEGFR), as response biomarkers to bevacizumab in mCRC patients, as well as their relationship with disease progression and death risk. Furthermore, we then compared the Response Evaluation Criteria in Solid Tumors (RECIST) version 1.1 [13] with our proposed Cytologic Criteria Assessing Response (CyCAR), based on CTC status, to determine their respective utility as predictive and prognostic assessments. Finally, we compared treatment responses of mCRC patients under FOLFOX–bevacizumab-containing chemotherapy by both criteria.

The study was performed in 77 patients with metastatic cancer of colon and rectum (59.7% and 40.3% respectively). The median age was 61 years and 51% of patients were men. Only 12 patients (15.6%) presented synchronous metastases while 65 (84.4%) presented metachronous metastases (metastases developed at least 12 months after the primary tumor) (Table 1). In this way, 59 (76.6%) patients developed liver metastasis, 8 (10.4%) lung metastasis and the remaining 10 (13%) developed metastasis in other organs as bone, lymph nodes or adrenal glands. Nevertheless, no relation was found between levels or status of CTCs and a predisposition to metastasis between organs (p > 0.05).

RECIST has been adopted as the standard method for tumor assessment and helps in clinical decision making. Deciding and then monitoring the effectiveness of individual therapies in mCRC patients is currently very challenging, as a result of the high prevalence of lymph abdominal, peritoneal, serous and pleural metastasis, which are particularly difficult to evaluate by RECIST [16]. In addition, the thresholds (of response or progression) for predicting differences in survival in treated patients probably differ according to the type of treatment and the type of cancer [17]. For example, targeted molecules such as anti-VEGF or anti-EGFR often induce only small size changes, whereas patient survival is significantly prolonged. Therefore, specific criteria for certain diseases or treatments are necessary, as the current biomarkers and imaging evaluation options to monitor and register treatment clinical responses do not yet allow for optimal management of individual patients yet [18].

The hypothesis that CTCs are a fundamental prerequisite to metastasis was first proposed in the mid-nineteenth century (1869) by Thomas Ashworth, an Australian pathologist [19]. The characterization of CTCs, derived from a ‘simple’ blood test, have the potential to serve as ‘real-time tumor biopsies’ permitting accurate, up-to-date pictures of tumor activity, without the need for invasive tissue biopsies. Furthermore, CTCs can be analyzed on a serial basis, allowing real-time identification of emerging treatment ‘resistance profiles’, and consequently, being of significant assistance to the radiological assessment of tumor responses [20].

We performed this study in 77 patients with metastatic colorectal cancer treated in a homogeneous manner. The overall response rate to treatment was 24 months and median overall survival was 23.3 months, which did not differ from that expected [21]. In our study, we cover responses of metastatic colon cancer patients to bevacizumab, assessing and comparing CTCs and standard evaluation criteria, and we found comparable results to other published studies.

In our study, we compared CyCAR to RECIST in the same mCRC patient cohort, to determine the specific importance of CyCAR. In addition, we analyzed the association between OS and PFS with different clinic-pathological characteristics such as the presence of CTCs, at follow-up times.

The correlation between K-RAS tumor status and the presence of CTCs was analyzed, finding borderline significant association (p = 0.06) between higher percentages of CTC positive patients with K-RAS wild-type tumors. Similar results were reported in a recent study with 24 metastatic colon cancer patients by Das et al. [22]. In the same way, Buim et al. [23] observed that only 9 of their 23 CTCs positive patients had K-RAS mutations in their corresponding primary tumors. Our results, together with the aforementioned studies, suggest that the dissemination of CTCs is an independent process to K-RAS status in the primary tumor.

On the other hand, we detected a direct association between the presence of CTCs and primary tumor surgery. In fact, most patients undergoing primary tumor surgical treatment were positive for the presence of CTCs. These results concur with several studies, which demonstrated that primary tumor resection can stimulate cellular proliferation of residual colorectal tumors [24]. However, other studies demonstrated that primary tumor resection improves OS in metastatic colon patients [25]. Interestingly, the absence of CTCs in mCRC has been also associated with a higher OS in several studies [26].

Additionally, besides analyzing the presence of CTCs, we also analyzed their heterogeneous VEGFR expression. VEGF is the target of bevacizumab, a humanized anti-VEGF monoclonal antibody (Bevacizumab; Avastin). This drug has been approved by the FDA in the first and second-line colorectal cancer setting, in combination with chemotherapy. Bevacizumab acts by selectively binding circulating VEGF, inhibiting the binding of VEGF to its cell surface receptor, and in this way preventing its activation. Despite this immobilization of VEGF, we found no association between VEGFR status in CTCs with the response to treatment, even though we describe a no statistically significant reduction in the cell subpopulations of CTC^CK+VEGFR+ and a decrease in the number of CTC^CK+VEGFR+ patients, before and after treatment. These results suggest that the treatment could be effective at removing VEGFR+ tumor cell populations, but also selecting VEGFR− populations. Interestingly, Simiantonaki et al. [27] demonstrated that negative VEGFR-1 expression was significantly associated with lymphogenous and haematogenous metastases. In the same way, Hanrahan et al. [28] found a significant increase of VEGFR-1 mRNAs in T3/T4 colorectal carcinomas compared to lymphogenously metastasising tumors, and Lebok et al. [29] showed that low VEGFR-1 tumor expression was associated with lower survival and correlated with an advanced disease status in breast cancer. According to these results, we expected CTCs^CK+VEGFR− patients to have a worse prognosis; however, we did not observe any difference in OS or PFS between patients with CTCs^CK+VEGFR+ and CTCs^CK+VEGFR−. In this context, Senger et al. [30] demonstrated the existence of autocrine VEGF signaling in human tumors that might reflect the importance of VEGF for sustaining the self-sufficiency or autonomy of tumor cells, especially relevant to aggressive cancers and to the biology of cancer stem cells. Autocrine VEGF signaling is also closely associated with tumor dedifferentiation and with epithelial–mesenchymal transition (EMT), which are processes involved in the genesis of cancer stem cells. In this context, it has been suggested that autocrine signaling blocking by bevacizumab is modest and that its action can be improved by combining autocrine signaling blocking therapies [31]. These mentioned analyses, could explain why we cannot find differences in the OS and PFS between positive and negative VEGFR expression in CTCs.

The principal objective of this study was to compare CyCAR with RECIST to predict bevacizumab response. In our analysis, we observed no differences between both criteria. In fact, we detected similar results when we compared the CyCAR and the RECIST response at 12 weeks. We also obtained similar results when analyzing the assessment response at 12 (71% vs. 93%) and 24 weeks (82% vs. 80%) for CyCAR and RECIST respectively.

Our results suggest that the presence of CTCs at 12 and 24 weeks can be useful predictive markers, and used as a complementary tool with RECIST, even in those patients where RECIST can be more difficult to apply.

We also analyzed the predictive role of the presence of CTC^CK+VEGFR+ to assess the treatment response in these patients. However, due to the low number of CTC^CK+VEGFR+ patients we have not been able to demonstrate that the presence of VEGFR in CTCs is a predictive factor of tumor response to FOLFOX–bevacizumab. However, it would be important to note that one the principal limitation of this study is the methodology used to isolate these CTCs, since this isolation is based on the epithelial markers expression, therefore, we could be losing an important subpopulation of CTCs, which are under EMT process. Recently, Zhang et al. [32], demonstrated that VEGF expression rate in mesenchymal CTCs was significantly higher than that of epithelial CTCs, which suggested that VEGF may be correlated with tumor malignancy “and probably with the resistance”.

Although these results suggest the value and importance of CTCs for monitoring treatments, further studies are necessary including, not only more patients, but also a deeper study about the autocrine and paracrine activity involved in the activation of VEGF and its role in the migration and proliferation of circulating tumor cells.

These results suggest that CyCAR is similar to RECIST criteria at evaluating the response in metastatic colorectal carcinoma. Although RECIST is useful for evaluating treatment efficacy in clinical trials and practice, it has some limitations. This way, we propose the use of CyCAR and RECIST combination for a better response prediction of metastatic colorectal carcinoma, in special when RECIST is limited.