Circulating free DNA (cfDNA) as biomarker of taxane resistance in metastatic castration-resistant prostate cancer (mCRPC)

Introduction Docetaxel (D) and cabazitaxel (C) are the standard chemotherapy for mCRPC. No biomarker predictive of resistance to D or C has been approved yet. Based on preliminary clinical data, we aimed to assess the association between ABCB1 amplification (ABCB1 amp) and primary resistance (RES) to D or C for mCRPC, using cfDNA. Methods A cohort (A) of 136 patients (pts) with at least 1 plasma sample drawn and stored within 12 months prior to starting D for mCRPC (2002-2014) and a cohort (B) of 42 pts with at least 1 plasma sample drawn and stored within 12 months prior to starting C for mCRPC (2010-2016) were identified from the Dana-Farber Cancer Institute IRB approved database. Whole genome sequencing (WGS) at 0.1x coverage, termed ultra-low pass WGS (ULP-WGS), was performed on the cfDNA extracted from the selected samples (1000μL/subject) and sequencing data were run through ichorCNA, a probabilistic model that allows to detect cases with sufficient tumor DNA content (>7%) and accurately identify copy number alterations (CNAs) including ABCB1 amp. Primary endpoint was assessing the association between ABCB1 amp and RES to D or C. Secondary endpoint was evaluating associations between any other CNAs with ≥10% prevalence and RES to D or C. RES was defined as the absence of a response, defined as PSA50 decline or radiologic response according to RECIST criteria version 1.1, within 16 weeks from treatment start. Odds ratio (OR) was used to compare odds of RES to D or C for pts with ABCB1 or exploratory CNAs and P-values were calculated by Fisher’s exact test or Monte Carlo simulation. Results Of the selected 178 pts, 66 had sufficient tumor purity: 45 pts in cohort A and 21 in cohort B. The rates of men with ≥4 prior therapy lines were 22.2% in cohort A and 71.4% in cohort B. No significant association was noted between ABCB1 amp and RES to D (P=0.7123; OR=1.600) or C (P=1.000; OR=1.0606). RES was observed in 26 pts (57.8%) of cohort A and 18 (85.7%) of cohort B. ABCB1 amp was observed in 9 pts (20%; 95% CI, 9.6-34.6) in group A and 6 of them (66%) had RES to D. ABCB1 amp rate among D-resistant men was 23.1% (95% CI, 9.0-43.7). In group B, 2 pts (9.5%; 95% CI, 1.2-30.4) had ABCB1 amp and both of them had RES to C. ABCB1 rate among C-resistant pts was 11.1% (95% CI, 1.4-34.7). No significant association was found between exploratory biomarkers and RES to D or C. Conclusion In this study, ABCB1 amp does not predict for RES to D or C for pts with mCRPC. Future studies including ABCB1 amp in a suite of putative biomarkers and larger samples may aid drawing definitive conclusions.