In brief
- In the phase III Krascendo 1 trial, divarasib extended median overall survival to 17.7 months, against 12.0 months with sotorasib or adagrasib (HR 0.56; p=0.0005), in previously treated KRAS G12C-mutated NSCLC.
- Median progression-free survival by blinded independent central review, the primary endpoint, was 9.8 months against 5.7 months (HR 0.58; p<0.0001).
- The confirmed objective response rate was 58% with divarasib and 32% in the control arm, with a median duration of response of 11.9 and 6.9 months.
- Serious treatment-related adverse events occurred in 7.8% of patients given divarasib and 15.2% of the control arm, although patients stayed on divarasib for more than twice as long.
- The trial was open-label, and the overall survival result comes from the primary analysis, with no median follow-up given.
Study at a glance
- Trial: Krascendo 1 (NCT06497556), phase III, randomised, open-label, multicentre
- Population: 338 adults with previously treated, locally advanced or metastatic KRAS G12C-mutated NSCLC
- Treatment: divarasib once daily (n=167) against sotorasib once daily or adagrasib twice daily (n=171)
- Primary endpoint: progression-free survival by blinded independent central review, 9.8 vs 5.7 months (HR 0.58; p<0.0001)
- Overall survival: 17.7 vs 12.0 months (HR 0.56; p=0.0005)
- Response: confirmed objective response rate 58% vs 32% in 328 response-evaluable patients
- Safety: serious treatment-related adverse events 7.8% vs 15.2%; discontinuation for treatment-related adverse events 4.8% vs 11.7%
- Not reported: median follow-up, grade 3 or higher adverse event rates, intracranial outcomes, and results for sotorasib and adagrasib separately
Survival extended by about six months against sotorasib or adagrasib
Krascendo 1 is the only global trial to compare a next-generation KRAS G12C inhibitor directly with the first-generation drugs currently approved in this setting. The 338 previously treated patients were randomised to divarasib or to a control arm given sotorasib or adagrasib. The study met its primary endpoint and its key secondary endpoints.
At the primary analysis, median overall survival was 17.7 months in the 167 patients given divarasib and 12.0 months in the 171 patients of the control arm (hazard ratio 0.56; 95% CI 0.41 to 0.78; p=0.0005). Median progression-free survival, the primary endpoint, was 9.8 months against 5.7 months (hazard ratio 0.58; 95% CI 0.44 to 0.76; p<0.0001). At 12 months, 44% of patients given divarasib were alive without progression, compared with 24% in the control arm.
Roche reported that the overall and progression-free survival benefit was consistent across clinically relevant subgroups, including subgroups defined by PD-L1 expression and by KEAP1 mutation status. The subgroup hazard ratios themselves have not been made public in the company’s summary.
Responses in 58% of evaluable patients, lasting a median 11.9 months
Response was measured in 328 patients with measurable disease at baseline; six patients in the divarasib arm and four in the control arm had none and were excluded from that analysis. The confirmed objective response rate was 58% with divarasib and 32% with sotorasib or adagrasib.
Duration of response was calculated among the 94 responders on divarasib and the 54 in the control arm. The median was 11.9 months against 6.9 months.
“New life-extending treatments are greatly needed, as most patients with this type of cancer will eventually relapse after initial therapy, and there are limited effective treatments for recurrent disease.”Ferdinandos Skoulidis, Professor of Thoracic/Head and Neck Medical Oncology, The University of Texas MD Anderson Cancer Center
Half the rate of serious treatment-related events despite twice the time on drug
The safety profile matched the known effects of the KRAS G12C inhibitor class. Gastrointestinal toxicity, mostly diarrhoea, nausea and vomiting, and hepatotoxicity were the most common treatment-related adverse events, and Roche described them as mainly low grade.
Patients remained on divarasib for 11.3 months, compared with 5.0 months on sotorasib or adagrasib. Over that longer exposure, 7.8% had a serious treatment-related adverse event, against 15.2% in the control arm, and 4.8% stopped treatment because of a treatment-related adverse event, against 11.7%.
An open-label comparison with a pooled control arm
Two features of the design bear on how the results are read. The trial was open-label, so patients and investigators knew which drug was given; the primary endpoint was assessed by blinded independent central review, which limits that bias for progression-free survival. The control arm also combined two drugs given on different schedules, sotorasib once daily and adagrasib twice daily, so the result is a comparison with the class rather than with either agent alone.
The overall survival figure comes from the primary analysis, and Roche described the divarasib median as approximately 18 months at this point, so it may change with further follow-up.
Two further phase III trials move divarasib into earlier disease
KRAS G12C is found in roughly 12% to 14% of NSCLC, and patients with this mutation may have poorer outcomes than those with wild-type KRAS. Sotorasib and adagrasib are the approved first-generation inhibitors.
Krascendo 2 (NCT06793215) compares divarasib plus pembrolizumab, without chemotherapy, with chemotherapy plus pembrolizumab in untreated advanced KRAS G12C-mutated NSCLC. Krascendo 3 (NCT07541170) tests adjuvant divarasib against immunotherapy or observation in resected stage II to IIIB disease after chemoimmunotherapy. The US Food and Drug Administration granted divarasib Breakthrough Therapy Designation in 2022 and Orphan Drug Designation for KRAS G12C-mutated NSCLC in 2026. Roche said it is working with health authorities on the next steps; no regulatory filing has been named.
Sources
- Roche. Roche’s divarasib demonstrated best-in-class efficacy, advancing the standard of care in previously treated KRAS G12C-mutated lung cancer (2026-10-25). roche.com
This article was produced independently by the Lung Summit editorial team, without industry funding or input.
