Abstract
METHODS
We pooled HRQoL data from five randomized trials including five cancer types (n = 1863). Multivariable Cox proportional hazards models were adjusted for age, sex, WHO performance status, metastatic status, and cancer type. QLQ-C30 scales were selected using stepwise backward elimination, with bootstrap validation (500 resamples) assessing stability. To avoid multicollinearity, the Summary Score and Global Health/QoL scale were evaluated in separate models, with an additional bootstrap analysis comparing their stability. Discriminative performance was assessed using Harrell's C-index.
CONCLUSION
The Summary Score, Physical Functioning, and Appetite Loss improve OS prediction beyond clinical factors. Although Global Health/QoL showed similar discrimination, the Summary Score demonstrated greater stability and may be preferable for robust risk stratification.
RESULTS
In separate multivariable models, the Summary Score (HR 0.904, 95% CI 0.869-0.939) and Global Health/QoL (HR 0.951, 95% CI 0.925-0.977) were retained as independent predictors, as well as Physical Functioning (HR 0.926, 95% CI 0.896-0.957) and Appetite Loss (HR 1.035, 95% CI 1.011-1.058). The Summary Score was retained in 96.6% of bootstrap samples versus 9.0% for Global Health/QoL. Compared with the clinical model (C = 0.603), C-indices increased to 0.637 with the Summary Score, 0.622 with Global Health/QoL, and 0.641 with Physical Functioning and Appetite Loss.
BACKGROUND
The EORTC QLQ‑C30 Summary Score (Summary Score) aggregates 13 functioning and symptom scales to measure overall health-related quality of life (HRQoL). We assessed its baseline prognostic contribution for overall survival (OS) alongside the global health status/quality of life (Global Health/QoL) and other QLQ-C30 scales.