Self-reported quality of life as a predictor of survival in renal cell carcinoma.

595 Background: Quality of life (QOL) in cancer patients has gained increasing attention and may provide prognostic value above and beyond traditional demographic and disease parameters. We evaluate the utility of self-reported QOL to predict mortality in patients with renal cell carcinoma (RCC). Methods: The Medicare Health Outcomes Survey was linked to SEER data to identify patients who completed a QOL questionnaire after the diagnosis of RCC from 1998-2014. Mental component summary (MCS) and physical component summary (PCS) scores were classified as high (≥50) or low ( < 50) based on a population mean score of 50 points. Patients were classified into four groups: 1) high MCS, high PCS; 2) high MCS, low PCS; 3) low MCS, high PCS; and 4) low MCS, low PCS. Multivariable Cox proportional hazards regression evaluated associations between QOL and all-cause mortality (ACM). The Harrell’s concordance statistic (C-index) estimated predictive accuracy. Fine and Gray competing risks models adjusted for stage, demographics, and comorbidities evaluated RCC-specific and non-RCC-specific mortality. Results: A total of 1494 patients with a median age of 73.4 years (IQR 68.8-79.3) at survey completion were included. Median follow-up was 5.6 years (IQR 4.0-8.3). There were 747 deaths, of which 139 were due to RCC. Models showed that each additional MCS and PCS point reduced the hazard of ACM by 1.3% (95% CI 0.981-0.993, P< 0.001) and 2.2% (95% CI 0.972-0.985, P< 0.001), respectively. The C-index was 72.1%. In the competing risks model, the subdistribution hazard ratio (SHR) of RCC mortality in Groups 2, 3, and 4 were 2.71 (95% CI 1.18-6.22, P= 0.02), 4.55 (95% CI 1.57-13.18, P= 0.005), and 3.11 (95% CI 1.35-7.16, P= 0.008), respectively, compared to Group 1. The SHR for non-RCC mortality were 1.50 (95% CI 1.16-1.94, P= 0.002), 1.03 (95% CI 0.59-1.78, P= 0.9), and 1.83 (95% CI 1.41-2.38, P< 0.001), respectively, relative to Group 1. Conclusions: Self-reported QOL metrics can be used to predict ACM in RCC patients with good accuracy; lower PCS and MCS scores led to higher rates of ACM, even after accounting for differences in disease, demographics, and comorbidity. Furthermore, non-RCC mortality was associated more with low physical health rather than low mental health.