No abstract

Less than one-third of contemporary RCTs with statistically significant results meet ESMO thresholds for meaningful clinical benefit, and this represents only 15% of all published trials. Investigators, funding agencies, regulatory agencies, and industry should adopt more stringent thresholds for meaningful benefit in the design of future RCTs.

No abstract

6609 Background: ASCO and ESMO have developed frameworks to evaluate the benefit of cancer therapies. Here, we apply the frameworks to a cohort of contemporary randomized controlled trials (RCTs) to explore agreement and to evaluate the relationship between treatment benefit and cost. Methods: Characteristic and outcome data from RCTs evaluating systemic therapies in non-small cell lung cancer (NSCLC), breast cancer, colorectal cancer (CRC), and pancreatic cancer published and cited in PubMed between 2011-2015 were abstracted. Trial endpoints were evaluated using ASCO and ESMO frameworks. Cohen’s kappa statistic was calculated to determine agreement between the two frameworks, using the median ASCO score as a benefit threshold. Differences in monthly drug cost between RCT experimental and control arms were derived from 2016 average wholesale prices. Analyses included Pearson chi-square tests, Fisher’s Exact tests, independent samples t-tests, and Pearson correlation to assess the association between continuous variables. Results: Fifty percent (136/271) of published RCTs favoured the experimental arm; scoring rubrics were applicable to 109 RCTs (39% NSCLC, 33% breast, 23% CRC, 5% pancreas). ASCO scores ranged from 2 to 72; median score was 25. Thirty seven percent (40/109) of RCTs met benefit thresholds using the ESMO framework. Agreement between frameworks was fair at best (κ = 0.28, p = 0.002). When stratified by treatment intent (19 curative, 90 palliative RCTs), agreement remained poor (κ = 0.23, p = 0.115; κ = 0.34, p < 0.001). Major differences leading to limited agreement includes the relative weights each framework places on HR, endpoints, and toxicity/QOL analysis. Smaller RCT sample size was the only trial characteristic associated with higher ASCO scores (p = 0.015). Among the 100 RCTs for whom drug costing data were available, there was no association between ASCO benefit score and monthly drug costs (r = -0.12, p = 0.22); those meeting ESMO thresholds had a lower mean drug cost than those who did not (p = 0.046). Conclusions: There is only fair correlation between ASCO and ESMO clinical benefit frameworks. Drug costs are not associated with ESMO/ASCO measures of magnitude of clinical benefit.

No abstract