Cross validation of the prostate cancer radiotherapy late toxicity (PCRT) questionnaire with the expanded prostate cancer index composite (EPIC) instrument
Rodrigues George; Bauman Glenn; Venkatesan Varagur; Ahmad Belal; Lock Michael; Sexton Tracy; DSouza David; Stitt Larry; Eid Somaya;
Department of Radiation Oncology, London Regional Cancer Program, London Health Sciences, London, Ontario, Canada
INTRODUCTION: A 29-item prostate cancer radiotherapy (PCRT) questionnaire with genitourinary (GU), gastrointestinal (GI), and sexual (S) domains has been previously validated for the assessment of late toxicity health-related quality of life (HRQoL) effects. The study objective was to cross-validate the PCRT domains versus the expanded prostate cancer index composite (EPIC) questionnaire urinary (U), bowel (B), hormonal (H), and S subscales. METHODS AND MATERIALS: A single-institution cross-sectional PCRT patient cohort was surveyed. Descriptive and intra- and inter-class correlation coefficient statistics for the various EPIC and PCRT HRQoL domain scores were generated. Univariable and multivariable Cox and logistic regressions were performed depending on the HRQoL endpoint being assessed. RESULTS: A total of 189/276 patients (68%) completed questionnaires with EPIC and PCRT missing data rates of 9% and 4%, respectively. Mean age was 75.8 years (SD 5.5) and the mean time of questionnaire completion after radiotherapy was 852 days (range 212-1454 days). Mean EPIC urinary (85.1 SD 12.9), bowel (84.1 SD 15.8), sexual (21.8 SD 20.7), and hormonal (85.3 SD 13.7) as well as PCRT genitourinary (66.1 SD 15.3), gastrointestinal (83.6 SD 14.3), and sexual (39.4 SD 21.6) domain scores were calculated. Intraclass correlation coefficients comparing corresponding EPIC/PCRT domains ranged from 0.50-0.88. Interclass correlation coefficients for non-corresponding EPIC/PCRT domains ranged from 0.16-0.43 and 0.23-0.30, respectively. EPIC B/U, PCRT GI/GU and PCRT S required arcsin square root transformation and EPIC S/H domains required dichotomous transformations prior to univariable/multivariable analyses. Multivariable analysis demonstrated novel associations between predictive variables and HRQoL domains including between the PTV-bladder overlap volume and PCRT GU score. CONCLUSIONS: The PCRT is a compact, valid, and HRQoL instrument with very high questionnaire compliance rates and similar statistical properties to the EPIC instrument. However, dichotomization of the PRCT S data was not required which suggests some potential statistical advantage to the PCRT.