Patient-reported outcomes (PROs) with durvalumab by PD-L1 expression in unresectable, stage III NSCLC (PACIFIC)

Abstract

Introduction: In the phase 3 PACIFIC study of Stage III NSCLC patients without progression after cCRT, PFS and OS were significantly improved with durvalumab vs. placebo, with no detrimental effect on PROs. We retrospectively investigated the impact of tumour PD-L1 expression on PROs. Methods: After ≥2 cCRT cycles, patients were randomised (2:1) to durvalumab 10 mg/kg or placebo IV q2w up to 12 mo. If available, optional pre-cCRT tumour tissue was tested for PD-L1 tumour cell (TC) expression using the VENTANA SP263 immunohistochemistry assay and scored at pre-specified (25% or unknown) and post-hoc (1%) cutoffs. PROs were assessed using EORTC QLQ-C30 and -LC13 with changes from BL analysed by a mixed model for repeated measures, HRs for time to deterioration (TTD) by a stratified Cox proportional-hazards model, and ORs for improvement rates by logistic regression. Results: Of 713 patients, 63% had known PD-L1 status. Similar to the intent-to-treat (ITT) population, most PROs remained stable over time from BL across the PD-L1 subgroups (TC ≥25%, < 25%, ≥1%, < 1%, or unknown), with no clinically meaningful (CM) differences (≥10 points) for durvalumab compared to placebo. However, similar to the ITT population, CM improvements (decreases ≥10 points) from BL to Week 48 were observed for dysphagia and alopecia across most PD-L1 subgroups for both durvalumab (mean changes 8.1 [not CM] - 20.9 and 15.5 - 26.9, respectively) and placebo (mean changes 10.4 - 19.4 and 15.8 - 31.3). Pre-specified and post hoc TTD analyses of PROs by PD-L1 subgroup were generally similar to those of the ITT population, with overlapping HR and 95% CIs. Similarly, PRO improvement rates by PD-L1 subgroup were generally similar to those of the ITT population, with overlapping OR and 95% CIs. Conclusions: There were no CM differences in PROs between treatment arms across various PD-L1 subgroups. Results were generally consistent with those in the ITT population, suggesting that PD-L1 expression did not influence PROs in this study.