ROS1 fusion NSCLC: the TKI landscape in 2026

ROS1 fusions drive about 1-2% of non-small-cell lung cancers — a small slice in percentage terms, but a large absolute population and one of the most cleanly druggable oncogenes in thoracic oncology. The field now has four approved tyrosine kinase inhibitors, a dominant resistance mutation that defines the whole sequencing strategy, and a newest generation engineered specifically to keep working after the others fail. Here is where ROS1 actually stands in 2026.

Why ROS1 is so druggable

ROS1 is a receptor tyrosine kinase that is normally silent in adult tissue. In a subset of NSCLC, a chromosomal rearrangement fuses the ROS1 kinase domain to a partner gene (CD74, SLC34A2, EZR, and others), producing a constitutively active fusion protein that the tumor depends on. Because the ATP pocket of ROS1 is structurally close to that of ALK, the first drugs to hit it were repurposed ALK inhibitors. The dependency is strong and the off-target burden in normal tissue is low, which is why response rates here are among the highest in targeted oncology.

The four approved inhibitors

• Crizotinib (Xalkori) — Pfizer, FDA approved for ROS1 NSCLC in March 2016. The original ROS1 drug, borrowed from its ALK and MET activity. Objective responses around 70% in the PROFILE 1001 cohort, but weak CNS penetration, so brain progression is common and the drug has no activity against the key solvent-front resistance mutation.
• Entrectinib (Rozlytrek) — Genentech/Roche, FDA approved August 2019. Designed to cross the blood-brain barrier, so it controls CNS disease far better than crizotinib. Response rates up to roughly 80% in treatment-naive patients, but like crizotinib it is defeated by G2032R.
• Repotrectinib (Augtyro) — Bristol Myers Squibb, FDA approved November 2023. A compact macrocycle built to fit past the solvent-front residue. In the TRIDENT-1 trial it produced the longest median progression-free survival of the class in TKI-naive patients (about 35.7 months) and, critically, drove responses in roughly 59% of patients carrying the G2032R mutation.
• Taletrectinib (Ibtrozi) — Nuvation Bio, FDA approved June 2025. A next-generation selective ROS1 inhibitor with strong CNS activity and preclinical and clinical activity against both wild-type ROS1 and the G2032R mutant, positioning it alongside repotrectinib as a drug that survives solvent-front resistance.

G2032R: the mutation that defines the strategy

When tumors progress on crizotinib or entrectinib, secondary ROS1 kinase-domain mutations show up in roughly 30-60% of cases. The single most common is G2032R, a solvent-front substitution that swaps a small glycine for a bulky, charged arginine right at the edge of the ATP pocket. That arginine sticks out into the space where crizotinib and entrectinib need room to sit, creating a steric and electrostatic clash that collapses their binding affinity. It is the direct ROS1 analogue of ALK G1202R and TRK G595R — same solvent-front position, same failure mode.

Other resistance mutations recur but are rarer: D2033N (also solvent-front), L2026M (the gatekeeper), L2086F, and S1986F. The clinical logic of the whole field follows from this: the first-generation drugs control disease until G2032R emerges, and the newer drugs (repotrectinib, taletrectinib) exist precisely because they were shaped to keep binding when that arginine is in the way. A third wave, including zidesamtinib, is in registration aiming at even broader resistance coverage.

What this means for sequencing

The practical question for 2026 is no longer “which TKI works” but “which TKI first.” A front-line drug with deep CNS penetration and built-in G2032R coverage compresses the whole treatment ladder into a single agent for longer. Repotrectinib and taletrectinib are reshaping that decision, and resistance profiling at progression — to distinguish an on-target solvent-front mutation from a bypass-track pivot — is becoming standard of care.

Try the docking yourself

The clash is something you can see directly. The ROS1 kinase domain bound to crizotinib is deposited as 3ZBF. Dock a first-generation inhibitor against wild-type ROS1, then introduce the G2032R substitution and re-dock: the bulky arginine side chain occupies the solvent-front edge of the pocket and the predicted binding score degrades, while a macrocyclic next-generation compound holds up far better. Open Studio and pick ROS1 from the target catalog with G2032R from the mutation chips to dock against this structure. Liganx renders the wild-type and mutant receptors side by side, so the selectivity story shows up as a score gap rather than a hand-wave.

Liganx is molecular docking online: free, browser-based, and set up for exactly this kind of resistance question. If you want to try molecular docking on a ROS1 solvent-front mutant without a local install, that is the fastest path.

Primary sources

• Awad MM, et al. Acquired resistance to crizotinib from a mutation in CD74-ROS1. N Engl J Med 368, 2395-2401 (2013). doi:10.1056/NEJMoa1215530
• Drilon A, et al. Repotrectinib in ROS1 fusion-positive non-small-cell lung cancer. N Engl J Med 390, 118-131 (2024). doi:10.1056/NEJMoa2302299
• U.S. Food and Drug Administration. FDA approves taletrectinib for ROS1-positive non-small cell lung cancer. (June 11, 2025). fda.gov