Changes to primary endpoints in phase 3 cancer trials occur frequently and are often undisclosed in study manuscripts, new research shows.
Among 755 randomized controlled cancer trials registered in ClinicalTrials.gov, 145 (19%) had their primary endpoints changed after the trial started, and 102 (70%) of those changes were not noted in the study manuscripts.
The researchers also found that after a change in primary endpoint, trials were more likely to meet the primary endpoint compared with trials that did not change this endpoint.
“The lifecycle of a clinical trial can be exceedingly complex and nuanced and a primary endpoint change, in and of itself, even after the trial begins, does not necessarily represent a red flag,” lead investigator Ethan Ludmir, MD, told Medscape Medical News.
However, alterations in primary endpoints can undermine the validity, reproducibility, and clinical applicability of trials, the study authors explained.
These data do “raise questions as to the optimal direction going forward to improve transparency,” added Ludmir, with the University of Texas MD Anderson Cancer Center in Houston.
The study was published online May 17 in JAMA Network Open.
Lack of transparency in randomized controlled trials remains a “pressing concern,” despite efforts to make protocols available and require clinical trial registration. And it’s unclear how often clinical trials report changes in primary endpoints.
To assess the frequency of reported primary endpoint changes, Ludmir and colleagues used three methods to determine whether oncology clinical trials underwent such a change. The researchers tracked the history of changes in ClinicalTrials.gov, self-reported changes noted in published manuscripts, as well as changes reported within available versions of protocol documents among 755 randomized controlled cancer trials.
Overall, the team found that 19% of trials underwent primary endpoint changes during the trial, but the rates of primary endpoint changes detected using the three methods varied considerably. The highest rate of changes was detected in published protocols (28%), followed by ClinicalTrials.gov (16%), and self-reported (6%).
Across all methods, primary endpoint changes were detected at higher rates when multiple versions of a trial protocol were available vs just one or no version (32% vs 16%).
The most frequent primary endpoint changes involved reporting a primary endpoint as a secondary endpoint (34%), making a change to the primary endpoint definition (32%), and changing a secondary endpoint to a primary endpoint (30%).
Among the 120 trials with changes that were observed on ClinicalTrials.gov, 63 (52.5%) had changes that occurred after the reported primary study completion date.
Of 145 trials with primary endpoint changes, 61% (89 trials) were positive compared with only 51% (309 of 610 trials) without a primary endpoint change. In multivariable analysis, a primary endpoint change was independently associated with trial positivity (odds ratio, 1.86; P = .003).
Overall, “we detected a high rate of [primary endpoint] changes among active cancer [randomized controlled trials] that occurred after trial initiation,” and these were “markedly underreported in published articles.” the authors concluded. “The findings suggest that revision of journal policies to require trialists to publish and report [primary endpoint] changes within the article and to provide protocols with multiple versions for reference is needed.”
“Our data suggest that, as with many things, sunshine is the best sterilizer and it’s been policy among many journals to insist that the trial protocol is included concurrently with publication of trial results,” Ludmir said.
Support for the study was provided by grants from the National Cancer Institute, the Sabin Family Fellowship Foundation, the Fund for Innovation in Cancer Informatics, and others. Ludmir has no relevant disclosures.
JAMA Netw Open. Published online May 17, 2023. Full text
For more from Medscape Oncology, join us on Twitter and Facebook
Source: Read Full Article