Sotorasib - KRAS G12C- A regulatory failure
Two recent papers -- by Fojo and Olivier -- tell the full story
Just out this week are two paired papers on Sotorasib!
One appears in the Lancet Oncology from Timothee Olivier, Sruthi Ragunathan and me, and the other from Tito Fojo, Meredith LaRose and Susan Bates. Here are the key points for the busy drug developer.
First, by way of background, from our paper
“Sotorasib was approved on the basis of a single-arm, uncontrolled phase 1–2 study (CodeBreaK 100) that reported an objective response rate (ORR) of 37% and a 11·1 month median duration of response. By contrast, other molecular therapies for NSCLC have higher ORRs; osimertinib targets EGFR-mutated NSCLC and has an ORR of 90% and alectinib targets ALK-mutated NSCLC and has reported an ORR of 82·9%.
The CodeBreaK 200 trial was initially designed as a randomised, post-marketing requirement trial to convert accelerated approval of sotorasib into regular approval. On Oct 5, 2023, the FDA held an Oncology Drug Advisory Committee meeting to discuss whether CodeBreaK 200 could be considered “an adequate and well-controlled trial”. The meeting concluded that, on the basis of systemic biases in study conduct, the results of CodeBreaK 200 could not be reliably interpreted.”
From our paper, we highlight problems with Codebreak 200
The control group is substandard. Other regimens HAD beaten docetaxel when it was chosen as control
Crossover was used inappropriately. Crossover was permitted even though the fundamental efficacy of the drug is unknown. This masks a potential negative survival result. And hides harms.
The sponsor reduced the sample size— meaning the trial was underpowered to assess overall survival
Quality of Life was only assessed for 12 weeks— while sotorasib was given on average 20 weeks— meaning lots of harms from prolonged use might be missed
There is lots of drop out early— this could represent informative censoring— meaning that the PFS results observed might not be valid
Additionally we perform a sensitivity analysis for the missing people and show significant results can vanish easily.
Fojo makes different points, and one which I think is quite smart
They write, “Discovering which tumor will respond to sotorasib should be the focus of clinical trials for this agent, not ever larger comparisons against standard of care therapies.”
What he means is that:
The entire post market agenda for sotorasib is to eeak out a PFS benefit against some chemo control arm, but isn’t the real question why 2/3s of patients with kRAS g12C don’t respond?
Surely there is a biological answer to this question, but neither FDA nor the company is interested in figuring it out. That could only mean loss of market share. Fojo writes this brilliantly. He writes, WE NEED…
Here are the papers
