Health Related Quality of Life Outcomes are Correlated with Survival in Glioma Patients in the SEER-MHOS Database

Abstract

Health related quality of life (HRQOL) is a complex patient reported outcome defined as an individual’s self-reported well-being related to a health related event including physical, functional, social, and emotional well-being; research suggests that patient-reported HRQOL may be more predictive of cancer patients’ survival than clinician-rated performance status. In brain tumor patients specifically, the research in this area has been limited and inconclusive. To further investigate the relationship between HRQOL and survival in glioma patients, we evaluated the relationship between several HRQOL measures and overall survival in a large, nationwide sample. Among 972 brain tumor patients with 1500 surveys, 158 patients were identified with 1) glioma diagnosed between 1976 and 2013 2) no other cancers 3) ≥1 post-diagnosis survey. An NCI algorithm created comparable scores to combine data using SF-36 or VR-12. Subscales (including fatigue and general health) and mental and physical component summary scores were normed (mean 50, standard deviation 10) in the US general population with higher scores representing higher HRQOL. We defined a minimally important difference on survey outcomes as 0.5 standard deviations or 5 points. Covariates included difficulty with ≥1 activity of daily living (ADL), marital status, glioblastoma (vs lower grade), age, diagnosis to survey time, comorbidities, gender, and race. We performed univariate analysis for categorical measures using Fisher’s Exact tests and continuous measures using t-tests. Separate multivariable Cox proportional hazard models were used to examine each HRQOL measure and survival times controlling for covariates. In this cohort, 53% were female, 59% had trouble with ≥1 ADL, 53% were married, 27% had glioblastoma, 74% were white, and 32% had no comorbidities. Median time to survey was 79 months (IQR 17 - 160 months), and median age was 60 years old (IQR 48 - 70 years). At last follow-up, 48% had died (median survival 211 months, IQR 16-367 months). For patients alive vs deceased at last follow-up, there was a difference of >5 points on general health (38.4 vs 30.0) and physical component summary scores (35.0 vs 28.7), but not for mental component summary or fatigue scores (though differences were significant for all). On survival analysis adjusting for covariates, a 5 point decrease in general health and physical component summary scores was significantly associated with a 17% and 14% increase in hazard of death, respectively. Mental component summary and fatigue scores were not significantly associated with hazard of death. General health and physical component summary scores, even accounting for clinical factors, were significantly associated with overall survival. These specific HRQOL measures should further be studied in confirmatory longitudinal studies for impact on survival in glioma patients.