Divarasib (GDC-6036)

Divarasib (GDC-6036) is Genentech/Roche's covalent KRAS G12C inhibitor and is widely regarded as the most potent and selective molecule in its class. In a Phase 1 single-agent study it produced a confirmed response in 53.4% of KRAS G12C NSCLC patients with a median PFS of 13.1 months - numbers that compare favorably to the approved agents sotorasib and adagrasib, whose monotherapy NSCLC response rates and PFS sit lower [1][2]. Its second strategic asset is the cetuximab combination in colorectal cancer, where divarasib plus cetuximab delivered a 62% confirmed response rate in KRAS G12C CRC, addressing the EGFR-feedback resistance that limits single-agent G12C inhibition in colon [3]. The thesis is best-in-class potency plus a credible Roche development engine, against the reality that it entered a class where two drugs are already approved.

Phase 1/2. The single-agent NSCLC and CRC dose-escalation/expansion data are mature and published in NEJM and JCO, with long-term follow-up presented in 2024-2025 [1][2]. The divarasib + cetuximab CRC combination has reported Phase 1b data [3]. As of mid-2026 the molecule is in the registrational-transition phase, with Roche's program prioritizing combinations (cetuximab in CRC; broader I/O and targeted pairings) rather than relying on monotherapy. The central commercial question is whether superior potency and selectivity translate into a registrational advantage in a market where sotorasib and adagrasib already hold approvals.

Divarasib is an oral, covalent inhibitor that binds the mutant cysteine of KRAS G12C in its inactive GDP-bound state, locking the oncoprotein and blocking downstream MAPK signaling. Its design achieves higher potency and greater selectivity than sotorasib and adagrasib, with less off-target activity - the practical consequence is a cleaner safety profile that supports combination dosing. In colorectal cancer, single-agent G12C inhibition is undermined by rapid EGFR-mediated reactivation of the RAS pathway; combining divarasib with the anti-EGFR antibody cetuximab blocks that feedback loop, which is the mechanistic rationale for the markedly higher CRC response rate in combination versus monotherapy.

The Phase 1 study dosed divarasib orally once daily from 50 to 400 mg in 137 patients (60 NSCLC, 55 CRC). NSCLC: 53.4% confirmed ORR, median PFS 13.1 months; CRC monotherapy: 29.1% confirmed ORR, median PFS 5.6 months [1]. Long-term NSCLC follow-up (data cutoff April 1, 2024) showed 48% of patients treated beyond one year with a median time on treatment of 11 months [2]. The Phase 1b cetuximab combination in KRAS G12C CRC reported a 62% confirmed ORR among 29 evaluable patients [3]. Treatment-related adverse events occurred in 93% of patients but were predominantly low-grade (Grade 3 in 11%, one Grade 4), consistent with the selectivity claim.

Our model gives this drug a 33% chance of eventually being approved. That number starts from the historical approval rate for Phase 3 drugs in this area - about 48% - then adjusts based on ten facts about the trial and its sponsor. The estimate is pulled up by the trial's light or open-label blinding, more secondary endpoints than usual, and the sponsor's strong record of getting drugs approved; it is pulled down by weak or limited earlier-phase results. The remaining facts are close to average for this stage and leave the estimate roughly where the base rate put it.

Market-timing risk dominates: sotorasib and adagrasib are approved, and a best-in-class latecomer must prove superiority head-to-head or win in a combination niche to gain share. Resistance is a class-wide problem - acquired resistance via secondary RAS-pathway mutations and bypass signaling emerges in all G12C inhibitors, limiting durability. The CRC combination, while impressive at 62% ORR, is based on small numbers and needs randomized confirmation. Finally, the G12C mutation itself is a minority KRAS allele (most common in NSCLC, less so in CRC and rare in pancreatic), which caps the addressable population relative to pan-RAS or G12D-directed approaches now advancing.

Divarasib is the strongest molecule in a validated class, and the data support the best-in-class label on both potency and tolerability. The signal that matters is not the NSCLC monotherapy numbers - impressive but in a crowded indication - but the cetuximab combination in CRC, where 62% ORR addresses a genuine unmet need and a real resistance mechanism. The honest tension is commercial: being third to market with a better drug is a harder value-creation story than being first, and Roche will need a differentiated registrational path (likely combination-led) to make divarasib more than a fast-follower. Watch the randomized CRC combination data and any move toward frontline or I/O-combination positioning. The broader KRAS narrative is also shifting toward G12D and pan-RAS(ON) agents (Revolution Medicines), which over time reframes G12C-selective drugs as one chapter rather than the whole story.