In a 338 patient Phase 3 study that pitted two approved rivals against each other instead of placebo (a comparison drugmakers usually avoid), Roche's pill against the KRAS cancer gene improved both progression free and overall survival in KRAS G12C mutant lung cancer.
A rare thing happened in lung-cancer drug development this week. Roche ran a 338-patient Phase 3 trial that did not compare its experimental pill, divarasib, against placebo or chemotherapy. Instead, it pitted divarasib head-to-head against two already-approved drugs from Amgen and Bristol-Myers Squibb, Lumakras and Krazati, and reported wins on both progression-free survival and overall survival. The design is the story: head-to-head trials against approved competitors are unusual in oncology because they raise the evidentiary bar and hand rivals a direct comparison on the same playing field.
KRAS G12C is a specific oncogene mutation found in roughly 13% of non-small cell lung cancer (NSCLC), the most common form of the disease. For decades it was considered undruggable. That changed in 2021, when Amgen's Lumakras (sotorasib) became the first FDA-approved drug to block the mutation directly. Bristol-Myers Squibb's Krazati (adagrasib) followed in 2022, picked up as part of a $5.8 billion acquisition. Both were scientific breakthroughs, and both have commercially under-delivered. Lumakras sold $363 million in 2024, finally crossing a 2022 analyst projection of $347 million after years of sluggish uptake. Krazati did $205 million last year, a thin return on a $5.8 billion deal.
The trial, called Krascendo 1, randomized 338 previously treated patients with KRAS G12C-mutant NSCLC to once-daily divarasib, once-daily Lumakras, or twice-daily Krazati, according to Roche's announcement. Patients with prior KRAS G12C inhibitor exposure were excluded. Roche said divarasib improved both the primary endpoint of progression-free survival and a key secondary endpoint of overall survival with statistically significant margins against both approved drugs. The result was reported in a press release and ad-hoc announcement and covered by Fierce Biotech.
Roche's pre-existing rationale for divarasib has been greater in vitro potency and selectivity versus Lumakras and Krazati, which the company argued would translate into better clinical outcomes. The head-to-head design now tests that hypothesis directly, in a single trial rather than across separate studies.
The commercial stakes are concrete. Roche has guided to peak divarasib sales of 1 billion to 2 billion Swiss francs, roughly $1.2 billion to $2.5 billion, according to the company's investor materials. That forecast is management commentary rather than analyst consensus. But the runway is real: the KRAS G12C market exists, the patients exist, and the two incumbents have not grown the category to anything close to their early projections. Roche's chief medical officer, Levi Garraway, called the readout the establishment of divarasib as a new standard of care in the second-line KRAS G12C NSCLC setting. That framing is the company's own.
The biggest unresolved question is the actual magnitude of the win. Roche said divarasib beat both comparators on PFS and OS with statistical significance but disclosed no hazard ratios, confidence intervals, p-values, or subgroup breakdowns. Those numbers, along with full adverse-event and discontinuation rates, are expected at an upcoming medical meeting. Until they land, the clinical advantage is qualitative, and the cross-arm safety comparison is descriptive at best. The design itself adds caveats. Krascendo 1 is open-label, meaning patients and investigators knew which drug they were taking, a setup that can bias investigator-assessed endpoints.
What to watch next: the full Krascendo 1 readout at an upcoming medical conference; Roche's revised filing and launch timing for divarasib; results from Roche's parallel trials in first-line advanced/metastatic NSCLC and in the adjuvant lung-cancer setting; and the response from Amgen and BMS, each of which has next-generation KRAS programs and combination strategies. A head-to-head win does not automatically translate to commercial displacement, especially when the rivals can pivot to combinations, new indications, or follow-on molecules. Oncologists, however, may be quicker to migrate than payers.