Predicting survival with health-related quality of life scores in locally advanced breast cancer: Results from an international randomized controlled trial

Abstract

618 Background: Whilst evidence suggests that baseline health related quality of life (HRQOL) scores predict survival in metastatic cancer patients, few trials have examined prognostic HRQOL factors in non-metastatic cancer patients. In this international phase III randomized controlled trial (RCT), we address these issues and retrospectively explore if baseline HRQOL scores predict survival in locally advanced breast cancer (LABC) patients. Methods: 435 LABC patients were enrolled over a period of three years randomized to receive in arm A: F (500 mg/m2 IV d1, 8 q 28 d x 6), E (60 mg/m2 IV d1, 8) and C (75 mg/m2 p.o. d1–14) versus E (120 mg/m2 IV d 1), C (830 mg/m2 IV d 1) and G-CSF (5 ug/kg/d s.c. d 2–13) q 14 d x 6 in arm B. HRQOL was assessed using the EORTC QLQ-C30 with baseline assessments being conducted before randomization. The Cox proportional hazards regression model was used for both univariate and multivariate analyses of survival. In addition, a bootstrap re-sampling technique was used to assess the stability of the outcomes. In this prognostic factor analysis, 359 (82.5%) of the 435 patients had a HRQOL assessment available at baseline and were then evaluated. Age, TNM stage, ER status as well as pre-selected HRQOL variables from the EORTC QLQ-C30 were analyzed for prognostic factor analyses in predicting overall survival (OS). Results: Diagnosis of inflammatory breast cancer (T4d, any N, M0) significantly predicted poor survival with a hazard ratio (HR) 1.375 (95% CI 1.027 –1.840, p=0.0323). However, no HRQOL scores at baseline showed up as prognostic indicators of OS, either in the univariate or multivariate analysis. A confirmatory analysis using a model-averaging technique, based on repeated forward Cox PH model-building on bootstrap generated data-sets were applied and confirmed the previous results. Conclusions: This is the first HRQOL prognostic factor analysis (PFA) study in LABC, and overall one of the largest HRQOL PFA using a robust methodology carried out in a homogenous cancer population. Our results indicate that HRQOL scores have no prognostic value in LABC patients. No significant financial relationships to disclose.