New lung cancer pill beats older KRAS drugs in head-to-head trial

A newer KRAS G12C pill called divarasib kept lung cancer in check for a median of 9.8 months, against 5.7 months on the two approved drugs in its class, in the phase 3 Krascendo 1 trial. People on divarasib also lived longer, with median overall survival of 17.7 months against 12.0 months. The results were presented in the Presidential Symposium at the European Society for Medical Oncology (ESMO) Congress in Madrid on 25 October 2026 and published the same day in the New England Journal of Medicine.

That is unusual. Most cancer trials test a new drug against chemotherapy or a placebo. Krascendo 1 put Roche's drug up against sotorasib (Lumykras) and adagrasib (Krazati), the targeted tablets doctors already use for this mutation, and it beat them on both of the outcomes patients care about most.

What did Krascendo 1 compare?

Krascendo 1 (NCT06497556) enrolled 338 adults with advanced non-small cell lung cancer carrying the KRAS G12C mutation. According to MD Anderson Cancer Center, whose Ferdinandos Skoulidis led the study, patients were treated at 151 centres in 26 countries. Everyone had already had between one and three earlier lines of treatment, and almost all had received an immunotherapy drug.

Patients were randomised to one of two groups:

• divarasib 400 mg taken by mouth once a day (167 patients)
• the doctor's choice of sotorasib 960 mg once a day or adagrasib 600 mg twice a day (171 patients, of whom 156 got sotorasib and 15 got adagrasib)

The trial was open-label, so patients and their doctors knew which tablet they were taking. To limit bias, the main outcome was judged by radiologists who did not know the treatment, a method called blinded independent central review. Our guide to open-label and blinded trials explains why that detail counts.

The main measure was progression-free survival, meaning time alive without the cancer growing or spreading. Overall survival was a key secondary measure. Patients in the control group could not switch to divarasib if their cancer grew.

How much better did divarasib do?

The data cut-off was 19 May 2026, with a median follow-up of 14.3 months. These are the main figures reported by Roche and MD Anderson.

The hazard ratio for progression or death was 0.58 (95% CI 0.44 to 0.76, p<0.0001). For death from any cause it was 0.56 (95% CI 0.41 to 0.78, p=0.0005). Response rates were measured in the 328 patients whose tumours could be assessed.

The benefit held across the subgroups the investigators looked at, according to OncLive's report from the meeting. That included people with low or high PD-L1 levels and people whose tumours also carried a KEAP1 mutation, which tends to make lung cancer harder to treat.

What do those hazard ratios mean in plain numbers?

Hazard ratios are hard to picture, so here is a worked example using the 12-month figure.

Imagine 100 people starting divarasib and 100 starting sotorasib or adagrasib. After a year, about 44 people in the divarasib group would be alive with their cancer still under control. In the other group the figure would be about 24.

That is a gap of 20 people in every 100. Put another way, roughly one extra person in every five treated (100 divided by 20) would still be free of progression at one year because they got divarasib. The median gains are easier still: 4.1 extra months before progression (9.8 minus 5.7) and 5.7 extra months of life at the midpoint (17.7 minus 12.0).

Response rates tell a similar story. With 58.4% against 32.3%, about 26 more tumours in every 100 shrank by a meaningful amount. Our explainer on reading trial results and absolute risk walks through these sums in more detail.

Medians describe the middle patient. Some people will do far better than these numbers and some far worse.

What side effects did people have?

Divarasib caused more stomach and bowel side effects, mostly mild ones. The older drugs caused more liver problems and more serious events overall. These are the treatment-related figures from Roche and OncLive.

There were no treatment-related deaths in either group. No one on divarasib had treatment-related heart rhythm changes (QT prolongation) or lung inflammation (pneumonitis), while a small number of patients in the control group did.

Most of the gut side effects were grade 1, the mildest level. Medscape reported severe diarrhoea in 9.6% of divarasib patients against 4.1% on the comparison drugs.

The ESMO discussant, Benjamin Besse of Gustave Roussy in Paris, said the trial still needed to show how long that diarrhoea lasts. "Is the diarrhea there for 10 months, or is the diarrhea here for just 2 days and then the patient has a normal life?" he asked, as quoted by Medscape. That question is fair. A daily tablet taken for nearly a year has to be livable.

People on divarasib also stayed on treatment more than twice as long, which gives side effects more time to show up. Even so, fewer of them stopped because of side effects.

Who could take part, and who was left out?

The entry rules shape who these results apply to. Based on the ClinicalTrials.gov record and the published paper, patients had to meet all of these conditions:

• advanced or metastatic non-small cell lung cancer with a confirmed KRAS G12C mutation
• one to three earlier treatments for advanced disease
• no earlier KRAS G12C or other RAS-targeted drug
• no untreated or active brain metastases (treated, stable ones were allowed)
• fit enough to be up and about most of the day (ECOG performance status 0 or 1)
• at least one tumour that could be measured on scans

The group was typical of this mutation. The median age was 66. Almost everyone (98% and 96%) had a history of smoking, and about 95% had adenocarcinoma. Around 29% to 30% had a past history of cancer in the brain or spinal cord.

Besse pointed out that many patients seen in clinics would not have qualified. People with active brain metastases or poorer general health are common in later-line lung cancer.

Why does beating the existing drugs carry so much weight?

Sotorasib and adagrasib were the first drugs to hit KRAS G12C, a target once thought impossible to block. Their approvals were big news. Their results against chemotherapy were more modest.

In the CodeBreaK 200 trial, published in The Lancet in 2023, sotorasib delayed progression compared with docetaxel chemotherapy. Median overall survival, though, was 10.6 months on sotorasib and 11.3 months on docetaxel (hazard ratio 1.01). Besse summed up that history at ESMO: "Despite the early excitement surrounding sotorasib and then adagrasib, their efficacy is only limited and an OS benefit against chemotherapy has not been demonstrated," he said, according to pharmaphorum.

Krascendo 1 is the first randomised trial to show one KRAS G12C drug extending life compared with another. Divarasib is designed to lock the KRAS G12C protein in its switched-off state, and laboratory work had suggested it is more potent than the first-generation drugs. Besse said the results "indicate that the higher potency of divarasib could translate into a new standard of care."

Skoulidis put it in patient terms. "For patients with KRAS G12C-mutant lung cancer, extending survival while delaying cancer progression are the outcomes that matter most," he said in the MD Anderson release.

Roche's chief medical officer, Levi Garraway, said the results "demonstrate the potential for divarasib to become a new standard of care for people with previously treated KRAS G12C-mutated lung cancer."

What are the weak points of the trial?

No trial is perfect, and this one has three limits worth weighing before calling it settled.

First, it was open-label. Central reviewers were blinded, which protects the main result, but patients knew what they were taking, and that can colour how side effects are reported. Second, only 15 people received adagrasib. That is too few to say how divarasib compares with adagrasib specifically, as MD Anderson itself acknowledged. Third, nobody in the control group could cross over to divarasib after progression. Besse criticised this. Without crossover, the survival gap partly reflects whether people ever got the newer drug at all.

Follow-up is also fairly short at 14.3 months. Longer data will show whether the survival curves stay apart.

Where does this leave patients in the UK?

KRAS G12C is found in about 12% to 14% of non-small cell lung cancers, according to Roche and MD Anderson. In the UK that is a meaningful group of people, but their NHS options have just narrowed.

On 9 July 2026, NICE published final draft guidance saying sotorasib should not be routinely funded on the NHS in England after its time in the Cancer Drugs Fund. People already taking it through that route can carry on. An appeal hearing was listed for 18 August 2026, and the NICE page still showed the appraisal as in progress when we checked. NICE's appraisal of adagrasib was terminated, as pharmaphorum reported.

Tom Newsom-Davis, chair of the British Thoracic Oncology Group, told pharmaphorum that the decision would leave "no options for patients with KRAS G12C mutations to receive targeted therapy" in that setting.

Divarasib is not licensed anywhere yet. The US Food and Drug Administration has given it Breakthrough Therapy designation, which speeds up review, but no filing dates have been announced. A UK route would run through the MHRA for a licence and then NICE for NHS funding. Our guide to how new medicines get approved by the FDA, MHRA and NICE sets out each step.

What other divarasib trials are running?

Roche is testing the drug earlier in the disease. Krascendo 2 (NCT06793215) is comparing divarasib plus the immunotherapy pembrolizumab with standard first-line treatment. Krascendo 3 (NCT07541170) is looking at divarasib after surgery for earlier-stage lung cancer.

If you or someone you care for has KRAS G12C lung cancer, a clinical trial may be the main way to get divarasib for now. Our guide on how to find and join a clinical trial covers searching registries and what to ask the study team.

Krascendo 1 also follows other ESMO 2026 lung and RAS news on this site. We covered tarlatamab with durvalumab in small cell lung cancer and daraxonrasib subgroup results in pancreatic cancer, another drug aimed at the RAS family.

What should you ask your lung cancer team?

If your cancer carries KRAS G12C, these questions can help at your next appointment. Has my tumour been tested for KRAS and other mutations, and do the results include KEAP1? Would I meet the entry rules for a divarasib trial, including the rules on brain metastases? If my current treatment stops working, what is the next option on the NHS, and is any targeted drug still available to me? How would side effects such as diarrhoea be managed if I joined a trial?

This article is for information only and is not medical advice. Talk to your oncologist before making any decision about your treatment.