KRAS: The "Undruggable" Oncogene That Wasn't — How New Inhibitors Are Changing Lung and Colon Cancer

When the KRAS gene was discovered in 1982 as one of the first human cancer oncogenes, it seemed like a clear therapeutic target. KRAS mutations drive an estimated 25% of all human cancers — among the most common single-gene drivers in oncology — and are especially prevalent in three of the deadliest tumor types: lung adenocarcinoma (~30%), colorectal cancer (~45%), and pancreatic ductal adenocarcinoma (~95%). Yet for nearly four decades, KRAS remained stubbornly undruggable, frustrating researchers who watched every other major cancer driver — EGFR, ALK, BRAF, BCR-ABL — yield to targeted therapy while KRAS stood apart.

Then in 2013, a landmark crystallography paper described a previously unrecognized hydrophobic pocket on the surface of the KRAS G12C mutant protein — present only in the inactive, GDP-bound state — that could be targeted covalently by small molecules. This discovery unlocked a decade of drug development that has now produced two FDA-approved KRAS G12C inhibitors, a growing suite of combination strategies, and an emerging class of pan-KRAS inhibitors targeting mutations beyond G12C. This article explains why KRAS was so difficult to drug, how the new inhibitors work, what clinical data show, and what comes next.

Why KRAS Resisted Drugging for 40 Years

KRAS (Kirsten RAS) encodes a small GTPase — an enzyme that cycles between an active, GTP-bound "on" state and an inactive, GDP-bound "off" state. In normal cells, KRAS switches on in response to growth factor signals, activates downstream proliferation pathways (RAF-MEK-ERK and PI3K-AKT-mTOR), and then switches off when GTP is hydrolyzed to GDP. The switching-off is accelerated by GTPase-activating proteins (GAPs).

Point mutations in KRAS — most commonly at codon 12 (G12C, G12D, G12V) or codon 13 (G13D) — impair the GTPase activity of the protein, locking it in the constitutively active GTP-bound state regardless of external growth signals. Oncogenic KRAS thus drives continuous, growth-factor-independent proliferation.

Three characteristics made KRAS exceptionally difficult to drug with conventional approaches. First, its affinity for GTP is extraordinarily high — in the picomolar range — making competitive inhibition with small molecules essentially impossible given the millimolar GTP concentrations inside cells. Second, the protein's surface lacks obvious deep hydrophobic pockets that drug-like molecules typically bind to; structural analyses repeatedly revealed a frustratingly flat, featureless binding surface. Third, KRAS is present at very high cellular concentrations, so any inhibitor would need to engage a large fraction of KRAS molecules to suppress signaling below the threshold for cancer cell proliferation.

The Switch-II Pocket: The Key That Unlocked KRAS

The breakthrough came from recognizing that the G12C mutation creates a unique opportunity that does not exist with other KRAS mutations. The cysteine substituted at position 12 by the G12C mutation contains a sulfhydryl group — a highly reactive chemical handle that can be covalently targeted by specially designed electrophilic compounds. Covalent inhibitors that irreversibly modify a specific cysteine residue have been used successfully for decades (aspirin covalently modifies COX enzymes; ibrutinib covalently modifies BTK), but KRAS G12C was a harder problem because the active (GTP-bound) form of the protein is unfavorable for drug binding.

The 2013 structural studies by Shokat and colleagues at UCSF identified a cryptic hydrophobic pocket adjacent to the switch-II region of KRAS — visible only in the GDP-bound (inactive) conformation. This "switch-II pocket" or S-IIP does not exist in wild-type KRAS or in other KRAS mutants, and is inaccessible when KRAS is bound to GTP. KRAS G12C inhibitors work by preferentially binding to and covalently reacting with KRAS G12C in its GDP-bound inactive state, occupying the switch-II pocket and preventing the exchange of GDP for GTP that would reactivate the protein. The result is irreversible trapping of KRAS G12C in the inactive off state.

This mechanism is self-limiting in an important sense: for the inhibitor to bind, KRAS must be in the GDP-bound state, which requires the inherent GTPase activity of the mutant protein to hydrolyze GTP — however slowly. The inhibitor then reacts with the transiently exposed cysteine, locking the protein in the off state before it can re-exchange GDP for GTP. Over time, as more KRAS G12C molecules cycle through the GDP-bound state and are trapped by the inhibitor, the proportion of active KRAS G12C falls below the threshold needed to maintain oncogenic signaling.

Approved KRAS G12C Inhibitors: Clinical Data

Sotorasib (Lumakras) — Amgen

Sotorasib was the first KRAS inhibitor approved anywhere in the world, receiving FDA approval in May 2021 for KRAS G12C-mutated locally advanced or metastatic NSCLC following prior systemic therapy. The pivotal CodeBreaK 100 trial enrolled 126 patients with KRAS G12C NSCLC who had progressed on prior platinum chemotherapy and/or immunotherapy. The objective response rate was 37.1%, median duration of response was 11.1 months, and median progression-free survival was 6.8 months. Importantly, responses were durable in a subset of patients, with 24-month landmark data showing that approximately 20% of responders maintained their response at 2 years.

The CodeBreaK 200 Phase III trial confirmed the PFS benefit versus docetaxel (5.6 vs 4.5 months, HR 0.66), establishing sotorasib as the preferred standard second-line option for KRAS G12C NSCLC patients previously treated with platinum chemotherapy and immunotherapy. Sotorasib is dosed at 960 mg orally once daily and is generally well tolerated; the most common adverse events are diarrhea (33%), musculoskeletal pain (21%), nausea (20%), and hepatotoxicity (12% Grade 3+), with liver function monitoring required.

Adagrasib (Krazati) — Mirati Therapeutics/Bristol Myers Squibb

Adagrasib received FDA approval in December 2022 for KRAS G12C NSCLC. Adagrasib has a distinct chemical structure from sotorasib and pharmacokinetic properties that differ meaningfully: a long half-life of approximately 23 hours (enabling once-daily dosing at 600 mg twice daily), extensive tissue distribution including CNS penetration, and the ability to form covalent bonds with KRAS G12C at the switch-II pocket via an acrylamide warhead similar to sotorasib but with different selectivity.

The KRYSTAL-1 trial reported an ORR of 42.9% in NSCLC, median DOR of 8.5 months, and median PFS of 6.5 months. A critical differentiator is adagrasib's CNS penetration: the KRYSTAL-1 cohort included patients with active, untreated brain metastases and reported intracranial ORRs of 33%, positioning adagrasib as a preferred option for KRAS G12C NSCLC with brain involvement. Adagrasib is also approved for KRAS G12C colorectal cancer (in combination with cetuximab, discussed below).

KRAS G12C in Colorectal Cancer: Why Combinations Are Necessary

Single-agent KRAS G12C inhibitors have shown far more modest activity in colorectal cancer than in NSCLC — ORRs of only 10–20% versus ~40% in lung. The mechanistic explanation for this disparity is instructive. In colorectal cancer cells, KRAS inhibition creates a feedback loop: suppression of the RAS-RAF-MEK-ERK pathway triggers upregulation of EGFR signaling via epiregulin and amphiregulin, effectively bypassing KRAS blockade through a parallel growth pathway. This rapid adaptive resistance essentially rescues the cancer cell from KRAS inhibition within hours to days.

The solution is co-targeting EGFR. Clinical trials of adagrasib plus cetuximab (an anti-EGFR antibody) showed dramatically improved ORRs of 34% in KRAS G12C colorectal cancer — approximately double single-agent activity — leading to FDA approval of this combination in 2024. Similarly, sotorasib plus panitumumab (another anti-EGFR antibody) demonstrated an ORR of 30% and improved PFS in the CodeBreaK 300 trial, receiving accelerated approval in 2023. These combinations represent the first effective targeted therapies for RAS-mutated colorectal cancer, which had been previously intractable to targeted approaches.

Resistance to KRAS G12C Inhibitors

Despite response rates of 35–45%, nearly all patients eventually progress on KRAS G12C inhibitors, with median PFS of 6–7 months in NSCLC. Acquired resistance mechanisms are diverse and heterogeneous, often involving multiple simultaneous mechanisms in different tumor subclones — a feature that complicates sequential treatment strategies.

Resistance mechanisms identified from ctDNA and biopsy analyses at progression include: on-target KRAS mutations that impair inhibitor binding (particularly Y96D, which alters the switch-II pocket geometry); KRAS amplification that overwhelms the available drug; secondary RAS mutations in NRAS or HRAS that activate parallel RAS signaling; RAF-MEK bypass through BRAF fusions, RAF1 amplification, or MAP2K1/2 mutations; upstream receptor activation through MET amplification, EGFR overexpression, or RET fusions; and lineage switching to a histologic subtype that lacks KRAS dependency, including transformation to small cell carcinoma.

The breadth of resistance mechanisms argues for upfront combination strategies rather than sequential monotherapy. Numerous trials are investigating KRAS G12C inhibitors combined with SOS1 inhibitors (which block KRAS reactivation), MEK inhibitors, SHP2 inhibitors (which interrupt upstream RAS signaling), and immunotherapy agents.

Beyond G12C: Pan-KRAS and Other Mutation-Specific Inhibitors

KRAS G12C accounts for roughly 13% of NSCLC KRAS mutations, 3–4% of colorectal KRAS mutations, and only 1–2% of pancreatic KRAS mutations. The numerically more common mutations — KRAS G12D (dominant in pancreatic cancer), KRAS G12V, and KRAS G13D — lack the reactive cysteine that enabled the G12C drugging strategy. Targeting them requires fundamentally different approaches.

KRAS G12D inhibitors are in early clinical trials. The most advanced is MRTX1133 (Mirati), a non-covalent inhibitor that occupies the switch-II pocket of KRAS G12D in both GDP- and GTP-bound states. Phase 1 data from 2024 showed early clinical activity in pancreatic and colorectal cancer. Pan-RAS inhibitors that block all KRAS isoforms and mutations simultaneously are also in development, though selectivity over wild-type RAS — required to avoid unacceptable toxicity — remains a significant challenge. RAS ON inhibitors that target the active GTP-bound state of KRAS using novel binding modes (including allosteric pockets distinct from the switch-II pocket) are another active area.

An entirely different approach uses mutant-specific peptide vaccines and T-cell receptor (TCR) therapies targeting neoantigens created by KRAS mutations. Because the mutated KRAS protein sequence is presented as a peptide by MHC molecules on tumor cell surfaces, it creates tumor-specific neoantigens that — unlike normal KRAS — the immune system has not been tolerized against. Early phase trials of mRNA vaccines and adoptive T-cell therapies targeting KRAS neoepitopes (particularly G12D presented by HLA-C*08:02 and HLA-A*11:01) have shown proof-of-concept activity in pancreatic cancer.

The Unfinished Business: KRAS in Pancreatic Cancer

Pancreatic ductal adenocarcinoma (PDAC) carries KRAS mutations in approximately 95% of cases, overwhelmingly G12D and G12V. Despite being the longest-known and most prevalent KRAS-driven cancer, PDAC has seen no benefit from KRAS G12C inhibitors (only 1–2% of PDAC has G12C). The urgency is acute: PDAC has a 5-year survival of only 12%, and most patients receive gemcitabine/nab-paclitaxel or FOLFIRINOX with limited benefit.

The clinical development of KRAS G12D inhibitors and pan-KRAS approaches is therefore watched with special intensity in the pancreatic cancer field. Early signals from MRTX1133 and from the SOS1 inhibitor BI 1701963 in combination regimens suggest that KRAS inhibition in PDAC is achievable in the clinic, though the dense stromal microenvironment of pancreatic cancer — which limits drug penetration and supports immune exclusion — poses additional barriers beyond the molecular target itself.