Overall survival and quality of life superiority in modern phase III oncology trials.

Abstract

11015 Background: The use of alternative endpoints, such as progression-free survival, has increased over time in phase III randomized clinical trials (RCTs). However, PFS and other alternative endpoints are often not valid surrogates for overall survival (OS) and quality of life (QOL), and may be less relevant to patients. We sought to determine the proportion of phase III oncology RCTs with OS or QOL superiority. A secondary goal was to evaluate the approach of QOL analyses, since “change-from-baseline” approaches may bias results (Bland and Altman. Trials. 2011;12:264). Methods: We performed a meta-epidemiological study of two-arm, superiority-design, interventional phase III oncology RCTs screened from ClinicalTrials.gov. RCT publications were reviewed for alternative endpoint, OS, and QOL results by at least two investigators. Alternative endpoint and OS superiority were defined for the experimental arm vs control arm according to the pre-specified statistical criteria for each RCT. QOL superiority was defined by either statistically significant or minimal clinically important differences (MCID). QOL was sub-classified as global QOL, defined by the composite summary measure obtained using the patient-reported outcome instrument, or domain QOL, referring to measures obtained from instrument subscales. Results: We included 791 RCTs published between 2002 and 2024, representing 555,580 enrolled patients. Primary RCT results were published between 2002 and 2024. Alternative primary endpoints were most common (n = 495, 63%). The primary endpoint was met in 53% of the RCTs (n = 420). Alternative endpoint superiority was shown in 55% of the RCTs (n = 434). OS was reported by 705 RCTs (89%), and OS superiority was shown in 28% of the RCTs (n = 221). Patient-reported outcomes were collected in 61% of the RCTs (n = 482), and 34% of the RCTs published global QOL results (n = 271). Most global QOL analyses were change-from-baseline (55%, n = 148). Global QOL superiority was shown in 11% of the RCTs (n = 84). In a sensitivity analysis of QOL subscale outcomes, 80 trials (10%) showed superiority in at least one QOL domain. Collectively, in 32% of the RCTs (n = 257), superiority of either OS or global QOL was demonstrated. In 6% of all RCTs (n = 48), both OS and global QOL superiority was shown. Conclusions: Phase III, superiority design oncology RCTs are commonly interpreted as “positive.” However, this is usually based on improvements in unvalidated alternative endpoints. Gains in either OS or QOL are uncommon, and exceedingly rare in combination. QOL appears both under-evaluated and under-reported. Furthermore, the majority of phase III QOL analyses, which are based on change-from-baseline comparisons, may be misleading. To increase the meaningfulness of late-phase research, future trial designs and regulatory processes should be re-focused towards OS and methodologically rigorous QOL improvements.