REPORTING FIDELITY OF PRIMARY ENDPOINTS IN FDA-APPROVED NSCLC THERAPIES: AN EXTENSION OF THE COMPARE METHOD
Author(s)
Jennifer M Hinkel, BS, MSc.
DPhil Student, University of Oxford, Oxford, United Kingdom.
DPhil Student, University of Oxford, Oxford, United Kingdom.
OBJECTIVES: Selective outcome reporting (SOR) is well documented in general medicine. The COMPare project at University of Oxford found that 58% of trial primary outcomes are silently omitted or substituted. This project examined whether oncology, in the context of FDA approvals, exhibits the same level of SOR. Prior literature on SOR has not used the regulatory record as a comparator source.
METHODS: Methods were prospectively registered on Open Science Framework. A corpus of all 58 FDA NSCLC approval actions from 2015 to 2024 was assembled resulting in 33 unique drugs, 59 pivotal trials, 7,612 endpoint instances, and 4,404 countable clinical endpoints. Using this corpus, I compared endpoint presence and specification across four sources: trial registry, protocol, primary publication, and FDA review documents, extending COMPare's registry-versus-publication design. I measured designated-primary-endpoint disclosure asymmetry and omission by endpoint type.
RESULTS: No approval (0/58; 0%, 95% CI 0-6.2%) showed silent omission or substitution of its designated primary endpoint, versus COMPare's 58% in general medicine. Designated primaries (Overall Survival, OS; Objective Response Rate, ORR; Progression Free Survival, PFS) were faithfully reported. However, overall endpoint omission was 44.4%, concentrated in patient-relevant and secondary measures. This level of omission was similar across accelerated approvals (42.4%) and traditional approvals (44.8%). A limitation was that this study did not include subsequent journal publications reporting aspects of the same clinical trial that may have reported additional data.
CONCLUSIONS: FDA oversight constrains selective outcome reporting in terms of primary-endpoint omission. This is a positive and novel finding in contrast to the COMPare study. However, lack of endpoint fidelity persists, concentrated in patient-relevant endpoints (such as Patient Reported Outcomes, PROs) that are increasingly important for regulators, payers, and health technology assessment (HTA) bodies.
METHODS: Methods were prospectively registered on Open Science Framework. A corpus of all 58 FDA NSCLC approval actions from 2015 to 2024 was assembled resulting in 33 unique drugs, 59 pivotal trials, 7,612 endpoint instances, and 4,404 countable clinical endpoints. Using this corpus, I compared endpoint presence and specification across four sources: trial registry, protocol, primary publication, and FDA review documents, extending COMPare's registry-versus-publication design. I measured designated-primary-endpoint disclosure asymmetry and omission by endpoint type.
RESULTS: No approval (0/58; 0%, 95% CI 0-6.2%) showed silent omission or substitution of its designated primary endpoint, versus COMPare's 58% in general medicine. Designated primaries (Overall Survival, OS; Objective Response Rate, ORR; Progression Free Survival, PFS) were faithfully reported. However, overall endpoint omission was 44.4%, concentrated in patient-relevant and secondary measures. This level of omission was similar across accelerated approvals (42.4%) and traditional approvals (44.8%). A limitation was that this study did not include subsequent journal publications reporting aspects of the same clinical trial that may have reported additional data.
CONCLUSIONS: FDA oversight constrains selective outcome reporting in terms of primary-endpoint omission. This is a positive and novel finding in contrast to the COMPare study. However, lack of endpoint fidelity persists, concentrated in patient-relevant endpoints (such as Patient Reported Outcomes, PROs) that are increasingly important for regulators, payers, and health technology assessment (HTA) bodies.
Conference/Value in Health Info
2026-11, ISPOR Europe 2026, Vienna, Austria
Value in Health, Volume 29, Issue 12S
Code
MSR25
Topic
Clinical Outcomes, Health Policy & Regulatory, Methodological & Statistical Research
Disease
Oncology