Abstract 5379: Non-invasive profiling of advanced prostate cancer via multi-parametric liquid biopsy and radiomics

Abstract

Abstract Introduction: Biomarkers of treatment response in advanced prostate cancer (PC) are urgently needed. Liquid biopsy profiling and radiomic imaging enable non-invasive, repeatable tumor profiling, and combining these modalities could yield powerful new predictive tools. We tested the feasibility of concurrent, multi-parametric molecular analysis of circulating tumor cells (CTCs) and matched plasma cell-free DNA (cfDNA), combined with radiomic analysis of CT scans from the same patients. Methods: Under IRB-approved informed consent, blood was collected from 23 patients with metastatic castrate resistant PC (mCRPC). CTCs were enumerated using the RareCyte and CellSearch platforms, and individual single CTCs and white blood cells (WBCs) were retrieved using RareCyte's integrated robotic micropipette. DNA was extracted from single cells (scDNA), as well as from matched cfDNA and buffy coats (bcDNA), then sequenced using Ampliseq HD Pan-cancer panel. Concurrent radiomic analysis was performed on bone metastatic lesions identified on CT scans from the same patients, and manually segmented regions of interest were queried for a variety of texture metrics. Results: Of 19 patients with both CellSearch and RareCyte enumeration, CTCs were detected in 12 by CellSearch (63%; median: 3/7.5 mL; range 1-343/ 7.5 mL) and in 14 by RareCyte (74%; median: 1/7.5 mL; range 1-363/7.5 mL). Using matched bcDNA as a gold standard, sensitivity and positive predictive value of germline SNP detection were 74% and 93% for scDNA, and 98% and 99% for cfDNA. A total of 48 and 18 somatic alterations were identified across the cohort in CTCs and cfDNA, respectively, including mutations in AR, TP53, CCND3, FGFR1, ALK, and ROS1. Of 14 patients whose single CTCs were recovered by RareCyte, 12 (86%) had detectable somatic mutations in scDNA, and 7 (50%) had detectable somatic mutations in matched cfDNA. While some mutations were concordant between matched scDNA and cfDNA, most were distinct. Radiomic entropy in CT scans was associated with CellSearch CTC-counts (AUC of 0.74) using the FDA-cleared prognostic threshold of <5 vs. ≥5 CTCs/7.5ml. Conclusion: We successfully piloted an integrated, multi-parametric analysis of matched single CTCs, WBCs, plasma cfDNA, and buffy coats, along with matched radiomic analysis of CT scans in patients with mCRPC. Germline variants were detected more readily in cfDNA, whereas somatic mutations in commonly altered genes were detected more frequently in CTCs. CellSearch CTC count, an FDA cleared prognostic biomarker, correlated with radiomic entropy, the first such association to our knowledge between liquid biopsy and radiographic readouts. The results demonstrate the feasibility of this approach and its potential to generate comprehensive molecular data and new biomarker profiles. Prospective validation in a large PC cohort is underway. Citation Format: Gareth J. Morrison, Jonathan Buckley, D. Gigi Ostrow, Bino Varghese, Nolan Ericson, Steven Cen, Alexander Cunha, Yi-Tsung Lu, Tad George, Jeffrey Smith, David Quinn, Vinay Duddalwar, Timothy Triche, Amir Goldkorn. Non-invasive profiling of advanced prostate cancer via multi-parametric liquid biopsy and radiomics [abstract]. In: Proceedings of the Annual Meeting of the American Association for Cancer Research 2020; 2020 Apr 27-28 and Jun 22-24. Philadelphia (PA): AACR; Cancer Res 2020;80(16 Suppl):Abstract nr 5379.