Discordance of patient- and physician-reported toxicities in two prospective trials of stereotactic body radiotherapy (SBRT) for localized prostate cancer.

Abstract

245 Background: SBRT for localized prostate cancer (PCa) is the focus of several ongoing and reported high-impact trials, which often focus on physician-reported toxicity (P-Tox) when comparing regimens. Patient-reported quality of life (PR-QoL) may differ and provide a more sensitive comparative metric of treatment burden, especially with fewer provider interactions during SBRT than during protracted RT courses. We evaluated the concordance of prospective genitourinary (GU) and gastrointestinal (GI) P-Tox and PR-QOL in men receiving SBRT for PCa. Methods: Data from two concurrently-enrolled prospective trials of SBRT in high-risk (Phase I Safety Endpoint, NCT01896271) and low-intermediate risk (Phase II GI Toxicity Endpoint, NCT02353832) PCa were used. Matching standardized schedules of collected PR-QoL [Expanded Prostate Cancer Index Composite (EPIC)] and P-Tox (CTCAE v5.0) were analyzed over the first 18 months of follow up, where symptoms are most pronounced. We assessed concordance of Grade≥2 GU/GI physician reported toxicity with PR-QoL declines exceeding anchor based minimal clinically important difference (MCID) thresholds (-6 urinary and -5 bowel summary scores, respectively) for each patient at each time point. Patients without baseline PR-QoL data were excluded in full, while time points with missing PR-QoL or P-Tox were excluded individually without imputation. Concordance was evaluated by Cohen’s kappa statistic. Results: From 101 patients, there were 256 (64%) follow up observations through 18 months with both PR-QoL and P-Tox at the time point and baseline. Concordance of PR-QoL and P-Tox was low at all time points for both GU and GI toxicity domains (mean kappa 0.093; Table). MCID was more often reported by patients than Grade≥2 toxicity by physicians (38% vs 17% for GU and 44% vs 10% for GI). There was little overlap of PR-QoL and P-Tox reporting: Grade≥2 P-Tox reported in 17% of observations with MCID in PR-QoL, while MCID in PR-QoL reported in 54% of observations of Grade ≥2 P-Tox. Mean concordance was similarly low when analyzing sub-groups of trial, investigator, and an alternative 2xMCID threshold. Conclusions: P-Tox and PR-QoL differed dramatically in two prospective studies of SBRT despite toxicity primary endpoints. This may reflect subjective and varying intervention thresholds driving P-Tox reporting, rather than actual patient burden. These data strongly support use of PR-QoL rather than P-Tox for SBRT comparative study endpoints and guidelines in this rapidly evolving space. [Table: see text]