ESMO Congress 2026

The evidence base for the KRAS G12C inhibitor class in previously treated NSCLC is already established: sotorasib (AMG 510) and adagrasib (MRTX849) were developed as direct inhibitors targeting the switch-II pocket of KRAS G12C for KRAS G12C-mutated NSCLC, with the CodeBreaK 200 phase III showing that AMG 510 did not significantly improve overall survival versus docetaxel ^[3][6]. The next-generation oral, highly potent and selective inhibitor divarasib (GDC-6036) had so far been at the phase I stage: 65 patients with advanced KRAS G12C-positive NSCLC received 50-400 mg once daily, with a confirmed objective response rate of 55.6% (95% CI 42.5-68.1), median duration of response of 18.0 months (95% CI 11.1-24.9), median PFS of 13.8 months (95% CI 9.8-25.4), and 48% of patients treated beyond 1 year with a post-1-year safety profile consistent with the overall profile ^[2].

LBA5, listed by the on-site post as one of the most anticipated presentations at ESMO26 (Madrid, Oct 23-27), is KRASCENDO 1: a phase III study of divarasib versus sotorasib/adagrasib in previously treated KRAS G12C-positive NSCLC. Per the on-site post, Roche reported positive toplines for both PFS and interim OS; the same post also explicitly notes that effect sizes and full safety data are still pending, with specific hazard ratios and median values undisclosed, awaiting the full report.

Relative to existing knowledge, this readout advances two points. First, the comparator choice: the prior pivotal phase III used docetaxel as control and did not significantly improve OS ^[3], whereas KRASCENDO 1 goes head-to-head against the same-class G12C inhibitors sotorasib/adagrasib. Second, interim OS reads positive, directly addressing the debate over survival benefit in this class ^[3]. Combined with the activity and tolerability signals from phase I (median PFS 13.8 months, median DoR 18.0 months) ^[2], if the effect sizes deliver, this would directly influence the sequencing of G12C inhibitors in the previously treated setting.

The limitations are equally clear: only toplines are available this time, with effect sizes, full safety, and subgroup details to be confirmed; the phase I was a single-arm, open-label dose-exploration study ^[2], and its numbers cannot be directly compared with phase III results; moreover, resistance after prolonged use remains a shared challenge for the class ^[3][6], and divarasib's final positioning rests on the complete LBA5 data.