NEW YORK – The US Food and Drug Administration has concerns about systemic bias in data from a Phase III trial that Amgen is hoping will support the full approval of Lumakras (sotorasib) as a treatment for advanced KRAS-mutant non-small cell lung cancer.
The FDA granted Lumakras accelerated approval in this setting in 2021 based on data from a single-arm Phase II study demonstrating a 36 percent response rate to Lumakras. Amgen is conducting the Phase III CodeBreaK 200 trial comparing Lumakras monotherapy against docetaxel in patients with previously treated, locally advanced, unresectable or metastatic NSCLC with KRAS G12C mutations, in the hopes of confirming the KRAS inhibitor's benefit and garnering full approval in this setting.
On Tuesday, the agency expressed concerns about the CodeBreaK 200 trial in a briefing document shared with the Oncologic Drugs Advisory Committee, which is meeting later this week to discuss the merits of that confirmatory study.
The regulator questioned the reliability of the trial's primary endpoint of progression-free survival and expressed concerns about how investigators assessed progressive disease favoring Lumakras, the rate of patient crossover between the Lumakras and docetaxel arms, and the higher patient dropout rate in the docetaxel arm. "The trial results are confounded by multiple sources of systemic bias, raising concerns about whether CodeBreaK 200 can be considered an adequate and well-controlled trial," the agency said in the briefing document.
In its initial review of top-line data from CodeBreaK 200, the FDA found around a five-week progression-free survival benefit for Lumakras. Given that patients' NSCLC progression is tracked via imaging every six weeks, the agency interpreted that this suggests very limited benefit for Lumakras — as little as five days — and questioned the clinical significance of the trial.
The agency further noted that 23 patients in the docetaxel arm were randomized but never treated compared to only two patients in the Lumakras arm, another indication that there was systemic bias in the study. In the docetaxel arm, there were 20 patients who withdrew consent within five weeks after randomization compared to none of the patients withdrawing consent in the Lumakras arm. The regulator noted that this imbalance in patient dropout is a potential source of bias in the trial due to the loss of patient information.
"If patients on the control arm who withdrew consent have better prognoses than those patients who remained on trial, the treatment effect of sotorasib compared to docetaxel would have been overestimated," the FDA wrote.
The FDA also assessed whether there was investigator bias in assessing progression in either treatment arm. The FDA's reviewers pointed out that more investigators deemed patients to be progressing earlier on the docetaxel arm than the Lumakras arm, suggesting that they were either more likely to take patients off docetaxel early or keep patients on Lumakras longer.
These early progression calls may introduce bias when assessments are performed by blinded independent central review (BICR), the FDA said. If patients who progressed on docetaxel began a new treatment prior to the BICR assessment, that analysis of the benefit in the docetaxel arm is confounded, since the last assessment prior to patients going on the new treatment is used. The FDA said this leads to an incomplete progression-free survival assessment by BICR, which is "particularly concerning" since this is the primary endpoint of CodeBreaK 200.
The FDA also found that patients who crossed over early from the docetaxel to the Lumakras arm survived longer after having progressed than those who crossed over to Lumakras after BICR assessment, 18.9 months versus 11.8 months, respectively.
"These results, together with eligibility criteria for crossover, suggest that patients who were willing and eligible to crossover before BICR confirmed progression could have been relatively healthier than those remaining in follow-up in the docetaxel arm," the FDA said. "If such is the case, then the observed treatment effect on progression-free survival, which is already small, could be an overestimation of the true treatment effect."
The FDA concluded by requesting that the advisory committee on Thursday focus on whether the results of CodeBreaK 200 can be reliably interpreted and whether it can be considered an adequate and well-controlled trial. The agency is expected to decide whether to grant Lumakras full approval in advanced KRAS-mutant NSCLC by Dec. 24.