An estimate of rate of deviation from NCCN guideline recommendations for central nervous system imaging in trials forming basis for drug approval in first line advanced non-small cell lung cancer (NSCLC)

Importance It is unknown whether and to what degree trials submitted to the US FDA to support drug approval adhere to NCCN guideline-recommended care in their baseline and surveillance CNS imaging protocols. Objective We sought to characterize the frequency with which the trials cited in US FDA drug approvals for first line advanced NSCLC between 2015 and 2020 deviated from NCCN guideline-recommended care for baseline and surveillance CNS imaging. Design, setting, and participants Retrospective observational analysis using publicly available data of (1) list of trials cited by the FDA in drug approvals for first line advanced NSCLC from 2015 to 2020 (2) individual trial protocols (3) published trial data and supplementary appendices (4) archived versions of the NCCN guidelines for NSCLC from 2009 to 2018 (the years during which the trials were enrolling). Main outcomes and measures Estimated percentage of trials for first line advanced NSCLC leading to FDA approval which deviated from NCCN guideline-recommended care with regards to CNS baseline and surveillance imaging. Results A total of 14 studies that had been cited in FDA drug approvals for first line advanced NSCLC met our inclusion criteria between January 1, 2015 and September 30, 2020. Of these trials, 8 (57.1%) deviated from NCCN guidelines in their baseline CNS imaging requirement. The frequency of re-assessment of CNS disease was variable amongst trials as well, with 9 (64.3%) deviating from NCCN recommendations. Conclusions and relevance The trials supporting US FDA drug approvals in first line advanced NSCLC often have CNS imaging requirements that do not adhere to NCCN guidelines. Many trials permit alternative, substandard methods and the proportion of patients undergoing each modality is uniformly not reported. Nonstandard CNS surveillance protocols are common. To best serve patients with advanced NSCLC in the US, drug approvals by the FDA must be based on trials that mirror clinical practice and have imaging requirements consistent with current US standard of care.

OP440 Comparison Of Evidence Supporting Cancer Drug Approvals And Prices In The US And Brazil

IntroductionCancer drug prices are high on the policy agenda worldwide. Previous research found no association between cancer drug benefits and prices at the time of regulatory approval. Drugs approved in the US with uncertain benefits may have spill-over effects in other settings. Our objective was to compare the evidence supporting cancer drug approvals in the US and Brazil, and to examine the association between cancer drug prices and availability of added therapeutic benefit.MethodsWe matched all novel cancer drugs approved in the US from 2010–2019 to approvals in Brazil. We extracted data on pivotal study design characteristics and outcomes in the US and Brazil, and evidence supporting price approval in Brazil, including availability of added therapeutic benefit.ResultsFrom 2010–2019, fifty-six cancer drugs with matching indications were approved in US and Brazil and had their prices authorized in Brazil by December 2020. Drug were available in Brazil following a median 522 days after US approval (IQR: 351–932). In the US, thirty-four (60.7 percent) of the drugs had pivotal randomized controlled trials (RCTs) and Twelve (21.4 percent) had overall survival benefit. By the time of Brazilian approval, forty-one (73.2 percent) drugs had pivotal RCTs and twenty-two (39.3 percent) had overall survival benefit. A total of twenty-eight (50 percent) drugs did not demonstrate added therapeutic benefit over other authorized drugs for the same indication and had a median reduction from requested to approved price of 6.1 percent (IQR: 0–27.8 percent) in Brazil. The twenty-seven (48.2 percent) drugs with added therapeutic benefit had a median price reduction of 2.0 percent (IQR: 0–9.2 percent).ConclusionsHalf of new cancer drugs approved in Brazil failed to demonstrate added therapeutic benefit. The Brazilian pricing system secured considerable price reductions, ensuring that prices for medicines with no added therapeutic benefit were not higher than existing treatments for the same approved indication. Although evidence was more mature by the time of Brazilian review, pivotal studies often lacked randomization and overall survival endpoints.

Landmark Cancer Clinical Trials and Real-World Patient Populations: Examining Race and Age Reporting

Background: Concern exists that the clinical trial populations differ from respective cancer populations in terms of their age distribution affecting the generalizability of the results, especially in underrepresented minorities. We hypothesized that the clinical trials that do not report race are likely to suffer from a higher degree of age disparity. Methods: Food and Drug Administration (FDA) drug approvals from July 2007 to June 2019 were reviewed to identify oncology approvals, and trials with age details were selected. The outcomes studied were the weighted mean difference in age between the clinical trial population and real-world population for various cancers, the prevalence of race reporting and association of age and race reporting with each other. Results: Of the 261 trials, race was reported in 223 (85.4%) of the trials, while 38 trials (14.6%) had no mention of race. Race reporting improved minimally over time: 29 (85.3%) in 2007–2010 vs. 49 (80.3%) in 2011–2014 vs. 145 (85.4%) during the period 2015–2019 (p-value = 0.41). Age discrepancy between the clinical trial population and the real-world population was higher for studies that did not report race (mean difference −8.8 years (95% CI −12.6 to −5.0 years)) vs. studies that did report it (mean difference −5.1 years, (95% CI −6.4 to −3.7 years), p-value = 0.04). Conclusion: The study demonstrates that a significant number of clinical trials leading to cancer drug approvals suffer from racial and age disparity when compared to real-world populations, and that the two factors may be interrelated. We recommend continued efforts to recruit diverse populations.