Association of dynamic change in patient-reported pain with survival in metastatic castrate sensitive prostate cancer: Exploratory analysis of LATITUDE study

Pain is an important dimension of quality-of-life in patients with metastatic castrate-sensitive prostate cancer (mCSPC). However, it is unclear if dynamic change in pain over time can predict for overall survival (OS) or time to radiographic progression (TTRP) in these patients. This is an exploratory analysis of LATITUDE, in which men with de novo mCSPC were randomized to receive either ADT plus abiraterone versus ADT alone. Information was collected on patient-reported worst pain score (WPS) from the Brief Pain Inventory-Short Form. We applied separate univariate joint models to determine the association of dynamic changes in WPS with OS and TTRP, respectively. For the time-to-event submodel, a Cox proportional hazard regression model was constructed. For the longitudinal submodel, a linear mixed effect model was built. The two submodels were linked through a random effect. The joint models were fitted with the use of Markov chain Monte Carlo algorithms. Overall, 1125 patients with at least 3 pain score measurements were eligible. Patients were initially stratified into two categories based on presence of baseline pain. Median OS for patients with and without any pain at baseline was 35.7 and 41.1 months and median TTRP was 31.8 and 23.1 months, respectively. On Cox multivariable regression, 1 unit higher baseline WPS was associated with inferior OS (hazard ratio [HR]: 1.05 [95%confidence interval [CI]: 1.02-1.09]; time dependent area under curve [tAUC] 0.64) and inferior TTRP (HR: 1.05 [1.01-1.08]; tAUC 0.64). On independent joint modeling, a dynamic increase in the current value of WPS by 1-unit was associated with inferior OS (HR: 1.32 [1.26-1.38]; tAUC 0.74) and TTRP (HR: 1.32 [1.26-1.38]; tAUC 0.70). The above findings highlight the potential dynamic interplay between patient-reported pain with OS and TTRP in mCSPC. Compared to baseline pain, such dynamic assessment of pain was found to have superior predictive ability as denoted by superior tAUC. Dynamic evaluation of pain score has the potential to tailor subsequent treatment based on response to initial therapy beyond its role as a very important dimension of quality-of-life.