Quality of life (QOL) variables influence local regional control in radiation therapy oncology group (RTOG) headsneck trials (9003 and 9111)

Abstract

The purpose of this analysis was to evaluate the influence of QOL (via FACT-H&N, version 2.0) in addition to disease and socio-demographic factors on local regional control for patients treated on RTOG headsneck cancer clinical trials. 1510 patients treated on RTOG 9003 (locally advanced disease) and RTOG 9111 (laryngeal preservation) were potentially analyzable; baseline FACT-H&N data were available for 1093 patients. A Cox model stratified by protocol assessed which QOL measures (baseline FACT-G, its subscales or the H&N subscale) had an independent prognostic impact on local regional control after accounting for tumor and socio-demographic variables (T-Stage, N-Stage, age, gender, race, KPS, marital status, # in household, education level, household income and smoking status). The QOL measures were considered as continuous variables with higher scores associated with better QOL. The mean FACT-G and H&N subscale scores were 86.4 and 21.2 with a standard error of measurement (SEM) of 5.3 and 4.3, respectively. Overall, patients with and without FACT-H&N had similar local regional control rates (HR = 1.094;p = 0.2827); by protocol, this was the case for RTOG 9003 (HR = 0.969;p = 0.7450), but not for RTOG 9111 (HR = 1.594;p = 0.0038). On univariate analysis, FACT-G, most of its subscales, and the H&N subscale showed significantly better local regional control for higher values. On multivariate analysis, T-sN-Stage and KPS were significant for local regional control, as expected. Socio-demographic variables found to have independent prognostic value were smoking status (HR = 1.413;p = 0.0007) and race (HR = 1.355;p = 0.0025). Moreover, FACT-G was prognostically significant when added to the fitted model (HR = 0.992;p = 0.0038). Of the subscales, functional (HR = 0.968;p < 0.0001) and emotional (HR = 1.028;p = 0.0359) well-being were found to be significant. Unlike the functional well-being scale, lower emotional well-being scores were unexpectedly associated with better local regional control. In conclusion, some baseline QOL and socio-demographic variables do have independent prognostic value for local regional control, which may identify patients requiring additional supportsearly interventions to improve outcome.