Improving circulating tumor cells enumeration and characterization to predict outcome in first line chemotherapy mCRPC patients

Abstract

Introduction: There is a critical need of new surrogate markers for improving the therapeutic selection and monitoring of metastatic prostate cancer patients. Nowadays clinical management of these patients is been driven by biochemical and clinical parameters without enough accuracy to allow a real personalized medicine. The present study was conducted to go insight the molecular profile of circulating tumor cells (CTCs) isolated from advanced metastatic castration-resistant prostate cancer (mCRPC) with the aim of identifying prognostic marker with potential utility for therapy selection and monitoring.

Materials and methods: CTCs isolation was carried out in peripheral blood samples from 29 mCRPC patients that undergo systemic chemotherapy based on taxanes (docetaxel/cabazitaxel) and 19 healthy controls using in parallel CellSearch and an alternative EpCAM-based immunoisolation followed by RT-qPCR analysis to characterize the CTC population. A panel of 17 genes related with prostate biology, hormone regulation, stem properties, tumor aggressiveness and taxanes responsiveness was analysed to identify an expression signature characterizing the CTCs.

Results: Patients with ≥ 5 CTCs/7.5ml of peripheral blood at baseline and during the treatment showed lower progression free survival (PFS) and overall survival (OS). Changes of CTCs levels during the treatment were also associated with the patient's outcome. These results confirmed previous data obtained using CellSearch in mCRPC. In addition, we found a CTC profile mainly characterized by the expression of relevant genes for the hormone dependent regulation of PCa such as AR and CYP19 together with genes strongly implicated in PCa progression and resistance development such as BIRC5, TUB1A, GDF15, RAB7 and SPINK1. Our gene-expression profiling also permitted the identification of valuable prognostic biomarkers. Thus, high levels of AR, CYP19 and GDF15 were associated with poor PFS rates while AR, GDF15 and BIRC5 were also found as reliable predictors of OS. Besides, a logistic model using KLK3 and BIRC5 showed a high specificity and sensitivity compared to CellSearch to discriminate patients with a more aggressive evolution.

Conclusions: The molecular characterization of CTCs from advanced mCRPC patients provided with a panel of specific biomarkers, including genes related to taxanes resistance, with a promising applicability as "liquid biopsy" for the management of these patients.

Keywords: circulating tumour cells (CTCs); prognostic markers; prostate cancer; taxanes resistance.

Figure 1. Kaplan Meier analysis for OS and PFS of CTCs levels changes within the treatment in mCRPC patients

5 CTCs/7.5 mL of blood was defined as the cut-off to separate the good and poor prognosis group.

Figure 2. Validation of the CTCs isolation approach in mCRPC patients

Box plots indicate median values in the group of control compared with the group of mCRPC patients for CD45 (A) GAPDH (B) and KLK3 (C) normalized to CD45. CD45, used as a marker of unspecific blood cells isolation showed no differences between both groups, while GAPDH and KLK3 demonstrated optimal accuracy for CTCs detection (**p < 0.01; ***p < 0.01 according to Mann-Whitney test). (D) ROC-curve showing the high sensitivity and specificity of KLK3 to detect the presence of CTCs in our mCRPC cohort.

Figure 3. Gene expression profiling in CTCs from mCRPC patients

Significant expression levels of genes involved in relevant signaling pathways for PCa biology: (A) hormone pathways (B) stem cell features and (C) associated with PCa progression and chemotherapy resistance. White boxes represent the gene expression levels in the group of healthy controls, grey boxes corresponding patients. (Mann-Whitney test, *p < 0.05; **p < 0.01; ***p < 0.001).

Figure 4. Kaplan Meier analysis for overall survival (OS) of validated CTC markers in mCRPC patients

Low/high expression were defined based on the 50% (KLK3 and AR) and 70% (BIRC5 and GDF15) percentile (Supplementary Table 2).

Figure 5. Prognosis value for a logistic model combining KLK3 and BIRC5 compared with CellSearch system

ROC-curve showing the sensitivity and specificity of CellSearch and the combination of KLK3 and BIRC5 to detect the presence of CTC in our mCRPC cohort.