Association between patient-reported outcomes and therapeutic outcomes in patients with breast cancer: A pooled individual-participant data analysis.

Abstract

530 Background: Patient-reported outcomes (PROs) are captured in validated tools to provide the patients’ perspective and voice on their physical, social, emotional, functional, and cognitive abilities. Pre-treatment PROs have shown prognostic importance in other cancer types, however, the prognostic value of PROs in breast cancer has been minimally explored. Methods: In a pooled analysis of contemporary clinical trial IPD from patients with breast cancer, cox-proportional hazard analysis and binary logistic regression was used to assess the association between potential predictors with overall survival (OS) and adverse event (grade ≥ 3) outcomes, respectively. PROs were recorded using the EORTC QLQ-C30 version 3.0 questionnaire in the pooled cohort. Statistical significance was set at a threshold of P<0.05 and was determined via the likelihood ratio test. Model fit and linearity of variable associations were assessed via the Akaike information criterion (AIC). All analyses were stratified by age, performance status, treatment arm, and study. The primary assessed outcome was OS, with grade ≥ 3 adverse events assessed as a secondary outcome. A forward inclusion process (starting with the variable of lowest AIC, PRO domains were retained in the model if they decreased the AIC by 2 or more on addition, and remained statistically significant) was used to evaluate the value of multiple PRO domains on prognostic performances. Results: Within data available, the EORTC QLQ-C30 version 3.0 questionnaire was used in a pooled cohort of 8,544 patients across 8 clinicals trials. In the pooled cohort, the association between PROs and outcomes was best described by a linear association. Patient-reported physical functioning, appetite loss, and pain were significantly associated with OS. On forward-inclusion, only physical functioning remained within the OS prognostic model. Except for patient-reported financial difficulties, all PRO domains were significantly associated with grade ≥ 3 adverse events. On forward-inclusion, physical functioning, pain, and constipation all remained statistically significant. Conclusions: Within large high-quality IPD, pre-treatment PROs demonstrated significant prognostic relationships with therapeutic outcomes in patients with breast cancer initiating contemporary anticancer treatments. Patient-reported physical functioning was found to be the most prognostic PRO domain for OS, while patient-reported physical functioning, pain, and constipation were retained in a multivariable model prognostic of grade ≥ 3 adverse events.