KRAS: 40 Years Undruggable, Then Drugged. Then What?

The target

KRAS is a molecular switch. In healthy cells, it cycles between an active state (bound to GTP) and an inactive state (bound to GDP). When a growth signal arrives, KRAS flips to GTP-on and tells the cell to divide. When the signal stops, KRAS flips back to GDP-off and the cell stops dividing.

In cancer, KRAS mutations lock the switch in the "on" position. The cell divides continuously, regardless of signals. The most common of these mutations - G12C, G12D, G12V - change a single amino acid at position 12, jamming the switch.

KRAS G12C is found in roughly 13% of non-small cell lung cancers, 3-5% of colorectal cancers, and 1-2% of pancreatic cancers. These are not rare cancers. Across all tumor types, KRAS mutations appear in roughly one in every five to six cancer patients.

Why it was undruggable

Most small-molecule drugs work by fitting into a pocket on the target protein - like a key in a lock. KRAS has no obvious pocket. Its surface is smooth and featureless. The protein is small (21 kDa), tightly folded, and binds its natural substrate (GTP) with picomolar affinity - meaning any drug would need to compete with a molecule the protein holds a trillion times tighter than most drug-target interactions.

For decades, researchers tried to block KRAS from the outside, target proteins downstream of KRAS, or prevent KRAS from reaching the cell membrane. None of these strategies produced an approved drug. The National Cancer Institute launched a dedicated RAS Initiative in 2013, reflecting both the importance of the target and the frustration of the field.

How a UCSF chemist cracked the pocket

In 2013, Kevan Shokat's lab at UCSF published a paper in Nature that changed the field. They discovered something no one had seen in three decades of studying KRAS: a pocket.

The pocket - called the switch II pocket - does not exist in normal KRAS. It only forms in the G12C mutant, and only when the protein is in its inactive (GDP-bound) state. The G12C mutation replaces glycine with cysteine at position 12. Cysteine has a sulfur atom - a reactive handle that a drug can grab onto permanently through covalent binding.

Shokat's group designed compounds that slip into this pocket and form an irreversible bond with the mutant cysteine. Once bound, the drug locks KRAS in its inactive state. The switch cannot flip back on. The cancer signal stops.

The elegance of this approach is its selectivity. Normal KRAS (with glycine at position 12) does not have the cysteine, so the drug cannot bind to it. The drug hits only the mutant protein in tumor cells and leaves the normal protein in healthy cells untouched.

The lesson: discovering the target does not guarantee the best drug. Amgen and Mirati read the same 2013 paper, targeted the same pocket with the same warhead chemistry, but designed their own scaffolds - and avoided the toxicity that killed the discoverer's compound.

Because the two surviving drugs bind the pocket differently, they have complementary resistance profiles. Mutations at H95 that resist adagrasib have no effect on sotorasib. Conversely, G13D and A59S mutations that resist sotorasib remain sensitive to adagrasib. Two drugs that look similar on paper may turn out to be sequential treatment options.

Lumakras generates roughly $300 million a year. For context, Keytruda - the PD-1 inhibitor that Lumakras patients typically receive before or alongside - generated $29.5B in 2024. Lumakras is roughly 1% of Keytruda's revenue, treating a subset of the same patient population.

Several factors explain the gap between scientific achievement and commercial performance:

• Small addressable population. KRAS G12C appears in ~13% of NSCLC patients. Of those, most receive Lumakras only after progressing on prior therapy. The eligible population at any given time is limited.
• The confirmatory trial stumbled. CodeBreaK 200 compared sotorasib to docetaxel in second-line NSCLC. It met its primary endpoint (progression-free survival), but showed no improvement in overall survival. An FDA advisory committee voted 10-2 in October 2023 that the PFS results could not be reliably interpreted, citing potential bias from differential dropout rates. The FDA issued a Complete Response Letter, rejecting full approval. Lumakras remains on accelerated approval, with a new confirmatory study required by February 2028.
• Competition from checkpoint inhibitors. Keytruda-based combinations dominate first-line NSCLC treatment. By the time patients progress to second-line where Lumakras is used, many have limited options and limited time - narrowing the commercial window.

Why the science still matters

The commercial disappointment of first-generation KRAS G12C inhibitors does not diminish what they proved: KRAS is druggable. The pocket exists. The covalent chemistry works. The selectivity is real. What did not work was the clinical context - second-line monotherapy in a market dominated by immunotherapy combinations.

Who benefits most: the biomarker picture

Biomarker data is revealing which patients gain the most from KRAS G12C inhibitors - and which don't. Patients with STK11 co-mutations - who respond poorly to immunotherapy - show response rates as high as 64% on KRAS G12C inhibitors (KRYSTAL-1 subgroup analysis). These are exactly the patients with the fewest other options. Conversely, KEAP1 co-mutations predict poor response (14-29%), emerging as the key negative biomarker (Journal of Translational Medicine, 2025).

The median duration of response for those who do respond is 10-12 months - far longer than the 5-6 month median PFS suggests, because the median is dragged down by non-responders.

Next-gen G12C: divarasib (Roche)

Divarasib (Roche/Genentech) is a next-generation G12C inhibitor with 5-20x higher potency than sotorasib. In a Phase I study of 60+ NSCLC patients, it showed a 56% response rate and 13.8 months median PFS - nearly triple the first-generation numbers. Its Phase III trial (KRASCENDO-1), a head-to-head comparison against sotorasib and adagrasib, reads out in 2026.

Beyond G12C: zoldonrasib and daraxonrasib (Revolution Medicines)

Revolution Medicines is leading on the mutations that first-generation drugs cannot touch. Zoldonrasib (RMC-9805) targets KRAS G12D - the mutation most common in pancreatic cancer, which has zero approved targeted therapies. It received FDA Breakthrough Therapy Designation in January 2026 after showing a 61% response rate in KRAS G12D-mutated NSCLC. Breakthrough Therapy Designation means the FDA saw data compelling enough to accelerate review; roughly 60-70% of drugs that receive it eventually get approved, compared to ~10-15% of all drugs entering trials.

Daraxonrasib (RMC-6236) is a multi-selective RAS inhibitor designed to work across multiple KRAS mutations - potentially addressing the full 30% of cancers with RAS alterations rather than the ~2% with G12C. In April 2026, the Phase 3 RASolute 302 trial in previously treated metastatic pancreatic cancer reported a median overall survival of 13.2 months versus 6.7 months on investigator's-choice chemotherapy (HR 0.40, p<0.0001) - roughly a doubling of survival in a disease where every prior targeted therapy has failed, and across a population that included RAS G12D, G12V, G12R, and RAS wild-type tumors.

Kevan Shokat - the UCSF chemist whose 2013 paper opened the field - is an academic co-founder of Revolution Medicines and a member of its scientific advisory board. After ARS-3248 failed at his first venture (Araxes/Wellspring), his second company is now leading the next generation.

After note (May 2026): The Phase 1/2 results that justified the Phase 3 design were published in the New England Journal of Medicine, and on May 1, 2026 the FDA issued a "safe to proceed" letter authorizing an expanded access protocol for previously treated metastatic PDAC - signed just two days after the request was received.

Pan-KRAS degraders: destroy the protein entirely

A third approach is emerging: pan-KRAS degraders that destroy the KRAS protein rather than just blocking it. Multiple programs are entering clinical trials, aiming to work regardless of which KRAS mutation is driving the tumor. The shift from one mutation, one drug, one setting to broad RAS inhibition across cancer types is the real story now.

The bottom line

Forty years of failure. One hidden pocket. Two approved drugs - and a pipeline that suggests they were just the beginning. Lumakras stalled commercially, but Krazati's brain penetration advantage, BMS's $5.8 billion bet, Roche's next-gen data tripling PFS, Revolution Medicines' breakthrough into G12D, and daraxonrasib doubling overall survival in second-line pancreatic cancer all signal that the market hasn't given up on KRAS. The target is cracked open. The question is whether next-generation drugs can turn a proven mechanism into a blockbuster.