Zidesamtinib produced responses in 94% of ROS1 TKI-naive patients in ARROS-1

A 94% response rate among 94 evaluable patients, 15% of them complete

The 94 patients are the efficacy-evaluable population rather than the whole cohort. They had measurable disease by blinded independent central review and had started zidesamtinib by 15 June 2025, a cut-off chosen so that every one of them had at least nine months of follow-up available for duration of response.

Eighty-eight of the 94 responded and 14 had no measurable disease left, giving a 95% confidence interval of 87% to 98% around the response rate. GSK does not state who assessed response.

Responses held in 86% of patients at 12 months, with median progression-free survival not reached

Median follow-up was 15.2 months, with individual patients followed for between 1.1 and 30.5 months, at a data cut-off of 16 April 2026. Durability was reported at landmarks rather than as medians: 94% of patients were continuing to respond at nine months, 86% at 12 months, and 90% were progression-free at 12 months.

Neither median progression-free survival nor median duration of response had a value at this analysis, which on a trial whose longest-followed patient has reached 30.5 months is a statement about how few events have accrued.

All ten patients with measurable brain metastases responded, seven with complete clearance

Seventeen per cent of the evaluable population had central nervous system metastases at baseline, and ten of them had intracranial disease large enough to measure. Every one of the ten responded, with a confidence interval running from 69% to 100%, and in seven the detectable brain tumour cleared entirely. At 12 months, 78% of patients maintained an intracranial response.

GSK also reports no central nervous system progression events among patients who entered the trial without brain metastases. That is the stronger of the two intracranial claims, because it rests on that whole group rather than on ten patients, although no follow-up time is given for it.

Dose reductions in 11% of patients and discontinuation in 1%

The treatment-related events GSK reports in at least 15% of patients were peripheral oedema, weight increase, raised blood creatine phosphokinase, altered taste and raised aspartate aminotransferase. Most events were low grade.

One per cent of patients stopped for a treatment-related event. ROS1-positive disease accounts for roughly 2% of non-small cell lung cancers, an estimated 50,000 new diagnoses worldwide each year.

“Patients with ROS1-positive NSCLC may receive treatment for many years while continuing to work, care for their families and live their daily lives. They need treatments that provide lasting control of their disease, including in the brain, while limiting side effects and treatment disruption. The durable responses and low rates of treatment discontinuation observed in ARROS-1 represent encouraging progress toward addressing these long-term treatment needs.”Alexander Drilon, MD, ARROS-1 primary investigator and Chief of the Early Drug Development Service at Memorial Sloan Kettering Cancer Center

Memorial Sloan Kettering Cancer Center states that Drilon has financial interests related to Nuvalent, Inc.

The drug is approved only after a prior TKI, and ARROS-1 carries no comparator arm

The US Food and Drug Administration approved zidesamtinib in July 2026 for patients with ROS1-positive NSCLC previously treated with a TKI, and the drug is not approved anywhere in the world as a first-line therapy.

ARROS-1 is single-arm, so the 94% figure has no randomised comparison behind it. Patients had received no previous ROS1 TKI, although 27% had prior chemotherapy and 17% of those had prior immunotherapy as well, so this is a population whose targeted treatment starts with the drug rather than a treatment-naive one.