The largest transcriptomic resource for radiotherapy-treated high-risk prostate cancer: paving the way for companion diagnostic biomarkers (2021)

Type of publication:Conference abstract

Author(s):Thiruthaneeswaran N.; Bibby B.; Pereira R.; More E.; Hoskin P.; Bristow R.; Choudhury A.; West C.; Wylie J.; *Denley H.; Henry A.

Citation:Journal of Medical Imaging and Radiation Oncology; Sep 2021; vol. 65 ; p. 252

Abstract:Purpose: The Cancer Genome Atlas (TCGA) is a valuable resource for developing and validating gene signatures for personalising treatments. TCGA samples came from patients who received heterogeneous treatments-dominated by surgery. Improving the biological precision of radiotherapy is hampered by the lack of well annotated cohorts that reflect patient populations relevant for radiation oncologists. We aimed to generate transcriptomic data from needle core biopsies for a large multicentre cohort of high-risk prostate cancer patients and use the data to validate published gene signatures. Methods and materials: A total of 478 NCCN classified high-risk patients treated from 2008-2016 were identified: 244 patients received intensity modulated radiotherapy (IMRT) to the prostate only (BEDalpha/beta 1.5-3Gy of 120-180 Gy) and 234 patients received IMRT to the prostate and a high dose rate (HDR) brachytherapy boost (BEDalpha/beta 1.5-3Gy 159-265 Gy). Androgen deprivation was given to all patients for 3-36 months. Biochemical failure was defined as prostate-specific antigen (PSA) rise of >=2 ng/ml above nadir post-radiotherapy. The primary clinical end-point was 7-year biochemical relapse-free survival (bRFS). Gene expression data were generated from diagnostic needle core biopsies using Affymetrix Clariom S arrays. Two (28-gene and 32 gene) published hypoxia gene signatures and a tumour radiosensitivity index (RSI) were tested for prognostic significance [1-3]. Result(s): The median follow-up for the entire cohort was 6.3 years. Both the 28 gene (p = 0.021) and 32-gene (p = 0.033) hypoxia signatures were prognostic for 7-year bRFS. Non-prostate hypoxia signatures were not prognostic. The bRFS for radioresistant (RSI-R) vs radiosensitive (RSI-S) was prognostic in the IMRT EBRT only cohort (p = 0.01) and not in the HDR boost cohort (p = 0.8). Conclusion(s): We generated the largest high-risk prostate radiotherapy cohort with full gene expression data and showed its value in validating published gene signatures. The RSI signature should be explored further to select patients with high-risk prostate cancer who should benefit from dose escalation with a HDR brachytherapy boost. This resource will be a valuable asset for future research generating and validating signatures for personalising radiotherapy in men with prostate cancer.

Link to full-text [no password required]