KRAS inhibitors: resistance drivers and combinatorial strategies

1The year 1982 marked the beginning of a new era in cancer research, with the discovery of HRAS and KRAS as the first human cancer genes. Since then, the RAS gene family, comprising HRAS, NRAS, and KRAS, has been identified as one of the most important oncogene groups in cancer. Among all RAS genes, KRAS is the most frequently mutated, accounting for ~80% of all RAS mutations. Notably, RAS mutations exhibit significant tissue specificity, with the most common and lethal cancer types, such as colorectal cancer (CRC), pancreatic ductal adenocarcinoma (PDAC), and non-small cell lung cancer (NSCLC), being prominently associated with KRAS mutations (43%, 85%, and 31%, respectively) (Figure 1A). By contrast, HRAS mutations are commonly linked to skin cancer (excluding melanoma) (6%), bladder cancer (5%), and head and neck cancer (7%), while NRAS mutations are prominent in melanoma (23%), thyroid cancer (11%), and leukemia (9%)i (Figure 1B). RAS proteins are key regulators of oncogenic signaling pathways, including mitogen-activated protein kinase (MAPK) and phosphoinositide 3-kinase (PI3K) pathways. This is further underscored by the high frequency of cancer-driving mutations in RAS downstream effectors (Box 1), including oncoproteins, such as BRAF (8%), AKT1/2 (4%), MEK1/2 (2%), and PIK3CA (13%)i, which all result in constitutive protein activity promoting tumor growth and survival. The important role of RAS proteins and RAS downstream effectors as drivers of cancer has led to an intense search for effective RAS inhibitors.

2Early on, experimental studies using cell culture and mouse models confirmed mutant RAS as a critical driver of cancer development and progression, further validating it as a favorable cancer target. However, the tertiary structures of RAS proteins revealed the lack of deep hydrophobic pockets on the surface, limiting opportunities for the design of potent and selective small-molecule inhibitors. In addition, the exceptionally high picomolar affinity of RAS for GTP, together with the high cellular concentration of GTP (~500 μM) renders RAS targeting significantly more challenging compared with kinases, which bind ATP with micromolar affinity and ATP concentrations in the millimolar range. This contributed to the longstanding belief that RAS was undruggable. Consequently, early strategies aimed at targeting RAS were primarily indirect.

3The first and most promising approach involved the development of farnesyltransferase inhibitors (FTIs) to disrupt localization of RAS at the plasma membrane, which is essential for RAS to function in signal transduction. FTIs failed dramatically in clinical trials (NCT00100750, NCT00005989, and NCT00005648) due to unexpected RAS isoform differences, where KRAS and NRAS, but not HRAS, could escape the actions of FTIs through alternative prenylation. This failure marks one of the darkest hours in KRAS history. Subsequently, research efforts shifted focus toward targeting KRAS effector signaling, in particular the RAF-MEK-ERK MAPK pathway, which led to the development of multiple inhibitors targeting each level of this cascade. However, MAPK pathway inhibitors have shown limited efficacy in KRAS-driven cancers, primarily due to unexpected dynamic feedback mechanisms and significant toxicity in normal tissues.

4In this review, we summarize the groundbreaking journey from the development of the first KRASG12C inhibitor to FDA approval of KRAS-targeted therapies nearly 40 years after RAS mutations were first identified as human oncogenes. We discuss the current impact of, and advances in, KRAS inhibitors in clinical settings, address the persistent challenge of therapeutic resistance, and review ongoing clinical evaluations of combination therapies aimed at enhancing sensitivity to KRAS inhibitors or overcoming resistance.

5The first findings that began to dispel the notion that RAS was an undruggable protein involved the discoveries of Kobe 0065 and SML-8731. Kobe 0065 inhibits the binding of HRAS to RAS-binding domains in effector proteins and SML-8731 was reported to bind to the nucleotide-binding domain of KRASG12C. However, neither of these small molecules advanced to clinical evaluation.

6In 2013, Shokat, Wells, and colleagues reported the seminal findings that finally paved the way to success. They screened for an inhibitor that covalently binds to the cysteine of the glycine-12-cysteine (G12C) substitution of mutant KRASG12C. Using a small-molecule library of cysteine-reactive compounds (so-called ‘warheads’), they identified several hits, leading to the development of compound 12. The development of the trailblazer compound 12 was a pivotal moment in KRAS drug design. Structural analyses revealed that this compound bound to a previously unrecognized pocket within the switch II region, designated S-IIP. Although compound 12 exhibited low potency, it demonstrated selective inhibition of KRASG12C mutant cancer cells. These findings not only established that the undruggable KRAS was druggable, but also highlighted that this can occur in a mutant-selective manner, reducing toxicities to healthy tissue harboring wild-type (WT) KRAS.

7Additional chemical modification of compound 12 led to the development of ARS-853, which exhibited improved cellular potency in vitro. Subsequent further optimization resulted in the development of ARS-1620, with enhanced potency and oral bioavailability in vivo. ARS-1620 demonstrated potent antitumor activity in KRASG12C-mutant cell line-derived (CDX) and patient-derived (PDX) xenograft mouse models. A surprising characteristic of compound 12 and subsequent compounds was their selectivity for the inactive (OFF) GDP-bound form of KRASG12C. At that time, the dogma in the field was that all RAS mutants were created equal, simply frozen in the GTP-bound state and, consequently, expected to be insensitive to an OFF-state selective inhibitor. However, it was determined that the KRASG12C mutant dynamically undergoes GDP-GTP cycling and, thus, exists in a state that is responsive to OFF-selective inhibitors.

8In parallel, several pharma companies initiated their KRASG12C programs and designed small molecules chemically similar to compound 12 that bound to S-IIP and covalently modified GDP-bound KRASG12C. Sotorasib (AMG510) was the first G12C-targeted inhibitor to enter clinical evaluation. In 2018, the first clinical trial, CodeBreaK 100 (NCT03600883), started to study the clinical safety of sotorasib in patients with advanced KRASG12C-mutant solid tumors that had advanced on standard-of-care chemotherapy or immunotherapy. No dose-limiting toxicities were observed, and a 32.2% confirmed objective response rate (ORR) (see Glossary) was revealed. Disease control (DC) was observed in 88.1% of patients, with a median progression-free survival (PFS) of 6.3 months. In CRC, 7.1% had a confirmed partial response (PR), 73.8% experienced DC, and the PFS was 4.0 months. Complete response was observed in 0% of patients, with 8.5% of patients with NSCLC and 23.8% of patients with CRC presenting progressive disease (PD). Based on these positive outcomes, CodeBreaK 100 continued into a Phase 2 trial. Here, 80.6% of patients with NSCLC displayed DC, PFS was 6.8 months, and median overall survival (OS) was 12.5 months. Based on the favorable findings of CodeBreaK 100, sotorasib gained accelerated approval from the FDA in May 2021 for locally advanced or metastatic KRASG12C-mutant NSCLC. The outcomes of major clinical trials involving KRASG12C inhibitors are summarized in Table 1.

9To further evaluate the benefits of sotorasib, the Phase 3 CodeBreaK 200 trial (NCT04303780) compared sotorasib with docetaxel in patients with NSCLC. Docetaxel, which is a second-line standard-of-care treatment, displayed modest clinical benefit and substantial toxicity. PFS for patients treated with sotorasib was only 1.1 months longer than for docetaxel, 5.6 months versus 4.5 months, respectively. Disappointingly, sotorasib did not show significant improvement in OS, 10.6 versus 11.3 months, respectively, leading the FDA to refuse full approval in 2024, but with accelerated approval still in place. Suboptimal trial design was cited as one basis for this decision.

10A second KRASG12C-selective covalent inhibitor, adagrasib (MRTX849), also began clinical evaluation in 2018 (KRYSTAL-1) for patients with advanced and metastatic solid tumors bearing KRASG12C mutations who had been previously treated with chemotherapy and anti-PD-1/PD-L1 therapy (NCT03785249). Adagrasib presented good bioavailability in vivo and demonstrated effective tumor inhibition in KRASG12C CDX and PDX tumor models. In Phase 1/2 evaluation, adagrasib demonstrated positive results in NSCLC with an ORR of 42.9%, DC was 86%, PFS was 6.5 months, and OS was 12.6 months. Based on these results, the FDA also granted adagrasib accelerated approval for NSCLC in December 2022.

11Both sotorasib and adagrasib monotherapy exhibited significantly lower ORR in CRC compared with NSCLC (Table 1). This divergence with KRASG12C-mutant CRC showing poorer response was reminiscent of earlier findings with a poor response of BRAF-mutant CRC versus melanoma to BRAF-inhibitor therapy. Epidermal growth factor receptor (EGFR) activation was identified as a basis for the lower response in CRC, with additional EGFR inhibition together with BRAF inhibitor treatment overcoming this issue. This led to the approval of a combined EGFR-BRAF inhibitor treatment for CRC. As a result, the CodeBreaK 300 trial (NCT05198934) was initiated, which combined sotorasib with the EGFR inhibitor panitumumab in KRASG12C-mutated advanced CRC. Combining sotorasib with panitumumab improved PFS to 5.6 months compared with 2.2 months for the patient group receiving standard of care (trifluridine-tipiracil or regorafenib). The CodeBreaK 300 trial especially highlighted the benefit of combination therapies for KRASG12C inhibitors. A similar improvement in ORR (30.2%) was observed with adagrasib in combination with cetuximab. Based on these findings, the FDA granted accelerated approval of this combination for KRASG12C-mutated CRC in June 2024.

12Typically, the initial approvals in a new drug class will not be best in class and, consequently, a multitude of second-generation KRASG12C inhibitors are under clinical evaluation (Table 2 and Figure 2). Among those, divarasib, a promising KRASG12C inhibitor, recently entered clinical trials (NCT04449874). Studies demonstrated improved outcomes for patients treated with divarasib compared with sotorasib or adagrasib. Specifically, the ORR of patients treated with divarasib was 53.4% compared with 28.1–43% observed with sotorasib or adagrasib. Additionally, the PFS was extended to 13.1 months in patients with NSCLC, an improvement of 6 months compared with sotorasib or adagrasib treatment.

13D3S-001, another next-generation GDP-bound KRASG12C inhibitor, demonstrates faster target engagement compared with first-generation KRASG12C inhibitors. As a result, D3S-001 effectively depletes GTP-bound KRAS even in the presence of growth factors, such as EGF. D3S-001 has demonstrated enhanced inhibitory effects in vitro and is in Phase 1/2 clinical trials (NCT05410145).

14While most second-generation inhibitors are mechanistically similar to the original OFF-selective compound 12, two very distinct covalent G12C-selective inhibitors have also entered clinical evaluation. First, BBO-8520 is a covalent inhibitor selectively active on both GDP- and GTP-bound KRASG12C. Consistent with its ON-state activity, preclinical studies found that BBO-8520 was less susceptible to signaling activities that promote GTP-KRASG12C formation. BBO-8520 recently entered clinical evaluation for metastatic KRASG12C-mutant tumors (NCT06343402).

15Second, RMC-6291 is a KRASG12C(ON) inhibitor with a very distinct mode of action compared with S-IIP-targeted inhibitors. RMC-6291 cannot bind directly to KRASG12C but instead first forms a binary complex with the immunophilin cyclophilin A (CYPA). This RMC-6291-CYPA complex then forms a tricomplex with GTP-KRASG12C (RMC-6291 covalently binds to KRASG12C), thereby impairing effector binding and activation by KRAS. RMC-6291 is in Phase 1 clinical trials alone (NCT05462717) and in combination with the multi-RAS inhibitor RMC-6236 or cytotoxic drugs (NCT06128551 and NCT06162221) in KRAS-G12C-mutant solid tumors. Additional KRASG12C inhibitor candidates currently under clinical evaluation include JDQ443 (NCT04699188), BEBT-607 (NCT06117371), and GH35 (NCT05010694).

16The field of RAS inhibitors is evolving rapidly, with many new inhibitors in development (Figure 2), leading to an increasingly crowded therapeutic space. As a result, the further development of JDQ443 was recently announced to be discontinued.

17Targeting KRASG12C has shown significant promise, particularly in NSCLC, which accounts for ~75% of all KRASG12C casesii. However, non-NSCLC-related KRASG12C accounts for only ~5% of all KRAS mutationsii. Consequently, only a small portion of patients with non-NSCLC-related KRASG12C would benefit from KRASG12C inhibitors. Due to the high prevalence of the other KRAS mutations, development of other mutant-selective inhibitors as well as pan-KRAS and pan-RAS inhibitors is ongoing and a subset of these novel RAS inhibitors are being tested in clinical trials (Table 2). In particular, KRASG12D has emerged as a key area of interest in the development of new inhibitors, given that this mutation represents ~28% of all KRAS mutationsi and is the most prevalent mutation in both CRC and PDAC. Multiple KRASG12D inhibitors, including JAB-22000, UA022, and BPI-501836, are being studied in preclinical settings. However, MRTX1133, discovered by a structure-based drug design, has pioneered the field of noncovalent KRASG12D inhibitors.

18Similar to the FDA-approved KRASG12C inhibitors, MRTX1133 recognizes KRASG12D in its GDP-bound state. In addition, MRTX1133 can also bind GTP-bound KRAS, albeit with lower affinity and its selectivity is 700-fold higher for KRASG12D versus WT KRAS. In preclinical studies, MRTX1133 exhibited potent activity in KRASG12D-mutant pancreatic CDX and PDX mouse models. Additionally, upon MRTX1133 treatment, a temporal shift from a basal-like to classical phenotype was observed in both PDAC cell lines and PDX models, which renders the tumor cells more susceptible to chemotherapy. MRTX1133 was the first KRASG12D-selective inhibitor to enter clinical evaluation in 2023, with a Phase 1/2 study evaluating MRTX1133 in patients with mutant KRASG12D with advanced solid tumors (NCT05737706). Similar to the KRASG12C inhibitors, MRTX1133 is part of an expanding class of small-molecule KRASG12D inhibitors under clinical evaluation. Other candidates under clinical evaluation include GFH375/VS-7375 (NCT06500676), HRS-4642 (NCT05533463), INCB161734 (NCT06179160), LY3962673 (NCT06586515), TSN1611 (NCT06385925), QLC1101 (NCT06403735), and QTX3046 (NCT06428500). To date, clinical responses have only been reported with HRS-4642 on 18 patients with solid tumors. Six patients experienced target lesion shrinkage, with grade three or less toxicities.

19Using the CYPA-based tricomplex strategy, RMC-9805 is another KRASG12D covalent inhibitor but, in contrast to the other KRASG12D inhibitors, it is selective for the KRAS(ON) state. In seven out of nine PDAC PDX and CDX models and six out of nine NSCLC PDX models, an objective response was obtained following RMC-9805 treatment. Phase 1 clinical evaluation of RMC-9805 revealed a 30% ORR and an 80% DC, with gastrointestinal-related toxicities and rashes for limited periods of time (NCT06040541)iii.

20Besides the direct small-molecule KRASG12D inhibitors, two alternative approaches are being tested in clinical trials: proteolysis-targeting chimeras (PROTACs) and an RNAi-based biodegradable implant. PROTACs have recently gained research interest in oncology, because they offer the potential to degrade the target protein, thereby eliminating potential scaffold functions. In general, a broad range of PROTAC-based degraders are under evaluation for the treatment of prostate and breast cancer. Recently, a KRAS PROTAC, namely ACBI3, was demonstrated to target all major oncogenic KRAS alleles, effectively degrading KRASG12D and KRASG12V in CRC and ovarian leiomyosarcoma xenografts, respectively. Meanwhile, the KRASG12D-targeting PROTAC, ASP3082, has entered Phase 1 clinical trials either as a single treatment or in combination with cetuximab or chemotherapy (NCT05382559). Early results suggest an acceptable safety profile in PDAC, CRC, and NSCLC. Additionally, the RNAi-based biodegradable implant siG12D-LODER is under investigation in a Phase 2 clinical trial in combination with chemotherapy for the treatment of locally advanced pancreatic cancer (NCT01676259). Early results from this study demonstrated that local downmodulation of KRAS is well tolerable and safe.

21In cancer, RAS genes show a spectrum of mutations beyond G12C and G12D (Figure 1). Therefore, new inhibitors are being developed to target the broad spectrum of RAS mutations. The tool compounds BI-2865 (BI-2493 structural analog) and BI-2852 were designed by eliminating the cysteine-specific covalent warhead (targeting the G12C mutation) and applying structure-based modifications. These noncovalent GDP pan-KRAS inhibitors exhibit activity against a range of KRAS mutants, as well as against KRAS WT. Since these inhibitors selectively recognize the OFF-state of KRAS, their broad activities suggest that most KRAS mutants, such as G12C, dynamically cycle between GDP- and GTP-bound states. However, these inhibitors lack in vitro activity against KRAS mutations with impaired GTP-GDP cycling activity, such as G12R and most Q61 mutations. Nevertheless, these inhibitors resulted in significant growth suppression of KRAS-mutant cancer cell lines and significantly impaired growth of PDX tumors with various KRAS mutations, including G12C, G12D, and G12V.

22Several pan-KRAS inhibitors are undergoing clinal evaluation. Among these, BI 3706674 is a potent and orally available small molecule that is under Phase 1 evaluation in metastatic KRAS-WT amplified gastric, esophageal, and gastroesophageal junction adenocarcinomas (NCT06056024). Additional pan-KRAS inhibitors in clinical Phase 1 include PF-07934040 (NCT06447662), LY4066434 (NCT06607185), and QTX3034 (NCT06227377) (Table 2).

23A mechanism of resistance to KRAS mutant-selective inhibitors is the upregulation of receptor tyrosine kinase (RTK) activity, which activates WT KRAS and leads to signaling feedback loops involving other RAS members. The use of pan-RAS inhibitors might counteract this reactivation while also targeting a broader range of RAS mutations. Initially, the concept of pan-RAS inhibition faced significant backlash, because Kras had been identified early on as a gene crucial for mouse embryonic development. Thus, pan-RAS inhibition was initially thought to have detrimental side effects due to normal tissue toxicity. However, these concerns have not materialized as anticipated. Currently reported pan-RAS inhibitors include GFH547, RSC-1255, YL-17231, RMC-7977, and RMC-6236, with several in clinical trials (Table 2 and Figure 2). The most extensively studied and potentially most advanced are the CYPA-based tricomplex RAS(ON) multiselective inhibitors, RMC-7977 (tool compound) and RMC-6236 (clinical candidate). RMC-7977 targets both mutant RAS and RAS WT in the GTP-bound state and has demonstrated potent antitumor activity in RAS-dependent mouse cancer models, including PDAC, CRC, and NSCLC. RMC-7977 induced tumor-selective apoptosis, while the skin and colon of these mice exhibited only a few apoptotic cells.

24The related analog of RMC-7977, RMC-6236, is being tested in clinical trials as monotherapy (NCT05379985), in combination with chemotherapy or immunotherapy (NCT06162221), or in combination with the KRASG12C-selective inhibitor RMC-6291 (NCT06128551). Similar to RMC-7977, preclinical models treated with RMC-6236 tolerated all doses tested and induced strong tumor regression in both PDAC and NSCLC models. In both entities, the median tumor progression-free period was not reached. By contrast, CRC models responded to a lesser extent, with half the preclinical models showing progressive disease and the median tumor regression-free period being reached after 60 days. The results of this study indicate that CRC tumors may require a combination therapy, as has been successfully demonstrated with KRAS-G12C inhibitors and EGFR antibodies in CRC. Nevertheless, Phase 1/2 monotherapy results were reported for 127 patients with KRASG12X and other RAS-mutant PDAC. The ORR and DC observed were 29% and 91%, respectively in second-line treated patientsiv. Impressively, two patients exhibited complete responses, one with a KRASG12D-driven PDAC and the other with a KRASG12V-driven NSCLC. These two patients had no evidence of disease after 5 and 8 months of treatment, respectively. The predominant side effects observed in 87% of patients were rashes, 6% of which were grade 3 or higher. Dose modification due to treatment-related adverse events occurred in 35% of patients. However, no events resulted in treatment discontinuationiv. In support of these results, recruitment for the RASolute 302 Phase 3 clinical trial is ongoing (NCT06625320). The relatively low PFS of patients with PDAC receiving RMC-6236 either as second or third-line therapy (8.1 and 4.2 months) indicates that, similar to KRAS mutant-selective inhibitors, multi-RAS inhibitors are subject to rapid resistance development.

25With the large number of RAS inhibitors being tested in clinical trials, it has become clear that the development of resistance will be a major challenge to their long-term efficacy. Remarkably, multiple resistance mechanisms have been observed in individual patients treated with KRASG12C inhibitors, ~50% of which may be attributed to genetic alterations. Notably, multiple resistance-driving mutations have been identified within individual patients. These resistance mechanisms can be classified as primary, acquired, and adaptive resistance.

26Primary resistance is the underlying cause of non-responders and involves mechanisms that render the tumor unresponsive to KRAS inhibition. This form of resistance occurs in 9–24% of all patients. Although our understanding of mechanisms driving primary resistance to KRASG12C inhibitors is limited, it will be critical for patient stratification. In a clinicogenomic study of 424 patients with NSCLC treated with sotorasib or adagrasib, co-occurring mutations in KEAP1, SMARCA4, and CDKN2A were independently associated with earlier disease progression and poorer clinical outcome. Notably, the KRASG12C/KEAP1MUT/STK11MUT or WT subgroup exhibited significantly shorter PFS (2.8 months versus 5.3 months) as well as OS (6.3 months versus 10.7 months) compared with the KRASG12C/KEAP1WT/STK11WT subgroup. A meta-analysis of 1224 patients in 18 studies with adagrasib or sotorasib also identified KEAP1 mutation as a marker of poor response. KEAP1 is a component of an E3 ligase complex that promotes proteasome-dependent degradation of NRF2. Loss of KEAP1 activates NRF2, leading to the regulation of genes that promote an antioxidant protective response. KEAP1 loss has also been identified as a driver of resistance to EGFR inhibition in NSCLC.

27By contrast, acquired resistance is observed in most patients who initially respond to KRASG12C inhibition, but then develop resistance after ~4–6 months. This resistance can emerge due to the acquisition of new mutations that impair inhibitor binding to KRAS and, thus, overall efficacy (Figure 3). Within the KRYSTAL-1 trial, 45% of patients who initially had stable disease for a minimum of 12 weeks, developed resistances to adagrasib. Sequence analyses of circulating tumor DNA identified newly acquired KRAS mutations with KRASR68S, KRASH95D/G/N/R, and KRASY96C/D/H/N being among the most frequently observed resistance mutations. These mutations are located within the switch II binding pocket of KRAS, hindering inhibitor binding. In addition, secondary activating KRAS mutations were found at lower frequencies, within either the mutant or the WT allele, including G12A/D/R/S/V, G13D, and Q61H/L. These newly acquired activating KRAS mutations are not susceptible to KRASG12C-specific inhibitors and, thus, are a potentially additional way to restore oncogenic KRAS activity.

28Similarly, KRASG12C amplifications have been identified as significant resistance drivers to KRASG12C inhibitor therapy in both CRC and NSCLC. KRASG12C amplification leads to an increased amount of KRAS protein, which is initially GTP loaded due to the 10:1 GTP:GDP ratio within the cell. Given the slow intrinsic GTPase activity, the steady-state amount of GTP-bound KRAS is elevated, thereby reducing the efficacy of KRASG12C(OFF) inhibitors.

29Besides KRASG12C-altering mutations or amplifications, acquired resistance to KRASG12C inhibitors can also be driven by the paralog RAS isoforms. Upon KRASG12C mutation, the main MAPK signaling activity is driven by the mutant KRAS; NRAS and HRAS contributing only minorly. However, patient data indicate that, upon KRASG12C inhibition, mutations in HRAS or NRAS, for example, NRASQ61K/L/R and HRASG13V, can restore both MAPK and PI3K-AKT pathway activity. The same effect is also reached via loss-of-function mutations in the RAS-GAP neurofibromin 1 (NF1), which have been reported in patients. NF1 loss results in impaired GTPase activity, thereby activating WT RAS to promote reactivation of downstream effector signaling pathways following KRAS inhibition.

30Apart from RAS proteins, downstream effectors may also acquire resistance-driving gene mutations. Common examples of this are BRAFV600E mutations, MAP2K1 mutations, such as MAP2K1K57N/T, or activating MAP2K1 in-frame deletions. These alterations result in reactivation of the MAPK pathway independently of KRAS, NRAS, or HRAS, thereby promoting KRAS-G12C inhibitor resistance. Likewise, amplification in MYC, comprising a major downstream effector of ERK signaling, was observed in patients, counteracting the inhibitory effect of the KRASG12C inhibitors. Mutations in the PI3K-AKT pathway are also observed, such as loss-of-function mutation of the PTEN tumor suppressor, or PIK3CA gain-of-function mutations, such as PIK3CAH1047R, causing hyperactivation of PI3K signaling. Besides mutations in the RAS-RAF-MEK or PI3K-AKT pathways to restore KRAS effector signaling, mutations in cell cycle regulators further contribute to KRASG12C inhibitor resistance. Loss-of-function mutations in CDKN2A, leading to inactivation of the RB1 tumor suppressor, are found in ~20% of patients with KRAS-mutant NSCLC.

31Given that RTKs are the major activators of RAS, it is not surprising that mutations or amplifications in RTKs, such as EGFR and MET, are frequently observed as a KRASG12C inhibitor resistance mechanism. MET is a driver of RAS signaling through SOS-dependent activation of WT RAS. However, amplification of MET is also able to activate PI3K-AKT independently of KRAS, as well as other prosurvival pathways, including the JAK-STAT3 pathway. Loss of MET function through siRNA depletion of MET or by treatment with the MET inhibitor crizotinib restored sensitivity to sotorasib treatment both in vitro and in vivo.

32Additional mutations that may contribute to resistance against KRASG12C inhibitor have been identified, including activating mutations in the RET RTK as well as oncogenic fusion proteins, such as EML4-ALK, CCDC6-RET, and NRF1-BRAF. However, these mutations have not yet been validated as KRASG12C inhibitor resistance drivers.

33Next to the large number of genetic mechanisms, adaptive nongenetic mechanisms have also been recognized as important contributors to KRASG12C-directed therapy resistance (Figure 3). Adaptive resistance refers to the ability of tumor cells to circumvent therapy through feedback mechanisms or post-translational deregulation of protein levels. A frequently observed adaptive strategy involves reactivation of the MAPK pathway via the activation of upstream signals. WT NRAS and HRAS can become activated as a compensatory mechanism in response to the loss of KRASG12C activity. This compensatory activation often occurs through the over-expression of RTKs or downmodulation of phosphatases. For example, EGFR overexpression results in increased levels of GTP-bound KRASG12C and, thus, reduces the amount of inhibitable protein. However, more EGFR also activates WT RAS proteins, such as HRAS and NRAS, which subsequently drive activation of the RAF-MEK-ERK cascade. Similarly, FGFR1 overexpression leads to sustained ERK activation, allowing the cells to overcome inhibition of KRASG12C.

34Epithelial–mesenchymal transition (EMT), a phenomenon frequently associated with therapy resistance in various cancer entities, can promote resistance to KRASG12C inhibitors. Mesenchymal-like cells promote PI3K activity, despite the presence of KRASG12C inhibitors, thereby overcoming their inhibitory effect. Furthermore, mesenchymal cells show activation of FGFR or AXL to promote resistance, feeding into the HIPPO pathway. This pathway is associated with EMT, with the downstream activator YAP being a driver of EMT. A genome-wide CRISPR/Cas9 screen of KRASG12C-mutant NSCLC cells identified YAP1/TAZ-TEAD activation as an important driver during adagrasib resistance development. These results highlight the potential effect of the HIPPO pathway on EMT and resistance to KRAS inhibitors. Since the TEAD inhibitor, VT3989 is in clinical trials for solid tumors (NCT04665206), the effect of TEAD1 inhibition on adagrasib sensitivity of NSCLC cells was tested preclinically and proven to enhance adagrasib efficiency.

35Not only EMT, but also transition of lung adenocarcinoma to a squamous cell carcinoma phenotype (AST) has been linked to KRASG12C inhibitor resistance in NSCLC lacking known gene mutations. Experimental analyses in genetically engineered mouse models of lung cancer found that loss of function of the STK11 tumor suppressor gene, which encodes LKB1 serine/threonine kinase, drives AST and resistance to G12C inhibitors. Notably, co-mutation of STK11 with KRAS occurs in 15–30% of lung adenocarcinomas.

36Given the mechanisms described earlier, it is obvious that single-drug treatments using KRAS inhibitors frequently develop resistance. A key lesson learned from past experiences is that combinatorial therapies are essential for improving patient outcomes. These strategies can help delay or overcome resistance, thereby enhancing the overall efficacy and durability of treatment. The main challenge is the upfront recognition of the most promising combinatorial approaches for individual patients.

37Due to the high prevalence of primary and acquired resistance to KRASG12C inhibitors, research into potential sensitizers and drug combinations is rapidly expanding (Figure 4). Among the most extensively studied combination therapies for KRASG12C inhibitors are those involving EGFR antibodies, particularly in CRC. This strategy is informed by early RAF inhibitor studies, which were initially ineffective due to feedback-induced upregulation of RTK signaling, in particular EGFR signaling. By using vertical inhibition of both EGFR and RAF or EGFR and RAS, a more robust suppression of oncogenic signaling is achieved, potentiating the antitumoral effects of the inhibitors. Preclinical studies demonstrated that the EGFR inhibitor cetuximab sensitized KRASG12C-mutant CRC CDX and PDX models to KRASG12C inhibition, resulting in tumor shrinkage.

38Multiple clinical trials have investigated the efficacy of combining KRASG12C inhibitors with the EGFR antibodies cetuximab and panitumumab (Table 2). In the Phase 1b KRYSTAL-1 study, the benefit of adagrasib combined with cetuximab was evaluated in CRC, NSCLC, and PDAC. In patients with CRC, the ORR increased from 18.6% with adagrasib alone to 46.4% when treated with the combination. The addition of EGFR antibodies to KRASG12C inhibitors reduces both mutant and WT GTP-bound RAS. In June 2024, the FDA granted accelerated approval for this specific combination therapy for the treatment of patients with KRASG12C CRC. Interestingly, the combination of panitumumab with sotorasib is effective not only in patients with KRASG12C-mutated CRC, but also in those with NRASG12C-mutated CRC, as recently described in two patients. Both patients exhibited either stable disease or PR for up to 12 weeks of treatment, with the only toxicities being EGFR inhibitor-based rashes. Rubinson et al. also demonstrated that sotorasib inhibits all RASG12C isoforms, with a fivefold higher potency against NRASG12C than against KRASG12C or HRASG12C. However, this ability of sotorasib is not shared with the other FDA-approved KRASG12C inhibitor adagrasib, which inhibits KRASG12C, but not HRASG12C or NRASG12C. This specificity is linked to a necessary interaction between the inhibitor and the KRAS residue H95, which is not conserved in HRAS or NRAS. This also accounts for why H95 mutations have been identified in patients who relapsed on adagrasib but not on sotorasib.

39KRASG12C inhibitor monotherapy frequently results in loss of multiple ERK-driven negative feedback loops causing reactivation of ERK. Combining MEK inhibitors with KRASG12C inhibitors presents an approach that could counteract MAPK signaling pathway reactivation. Unfortunately, the combination of KRASG12C inhibition and MEK inhibition is insufficient to overcome EGFR reactivation. To address this, a triple vertical combination of MEK and EGFR inhibition together with KRASG12C inhibition was tested in the CodeBreaK 101 trial. The combination of sotorasib with the MEK inhibitor trametinib and the EGFR antibody panitumumab demonstrated a PR of 20% in patients with NSCLC and 9% in patients with CRC, with a DC of 87% and 82%, respectivelyv.

40While EGFR is among the most widely studied RTK targeted in combination with KRASG12C inhibitors, other RTKs, including VEGF, MET, or AXL, have also been investigated as cotargets. MET amplification drives resistance to EGFR antibodies and also to the KRASG12C EGFR antibody combination. Thus, the inhibition or knockdown of MET in combination with KRASG12C inhibition resensitizes cells and promotes tumor shrinkage in mice. Another RTK the inhibition of which was shown to sensitize to KRASG12C inhibitors is AXL. Early on in one of the first CRISPR screens to discover sensitizers for the KRASG12C inhibitor ARS1620, AXL was identified as a sensitizer in PDAC cells. Additionally, a YAP-AXL axis was described to promote adagrasib tolerance in NSCLC cells, with AXL inhibition sensitizing the cells toward KRASG12C inhibition.

41Given the role of multiple RTKs as drivers of resistance, instead of using a complex cocktail of inhibitors targeting specific RTKs, a more effective approach might be to target a common node between RTK signaling and RAS activation. The guanine nucleotide exchange factor SOS1 is crucial in mediating RTK activation of RAS. Thus, inhibiting SOS1 offers a potential pan-RTK-RAS therapeutic strategy. The SOS1 inhibitor BI-3406 in combination with adagrasib resulted in impaired tumor growth, even in adagrasib-resistant models, driven by MRAS and SHOC2. Similarly, the SOS1:KRAS inhibitor, BI 1701963, in combination with the KRASG12C inhibitor BI 1823911, resulted in tumor regression in NSCLC CDX and entered clinical trials in 2021 (NCT04973163). Additionally, a SOS1 PROTAC, based on the SOS1 inhibitor BI-3406, combined with sotorasib synergistically reduced lung tumor volume in mice. This approach inhibits oncogenic KRAS through the KRASG12C inhibitor, while additional SOS1 degradation prevents RTK-based activation of WT KRAS, NRAS, and HRAS and, therefore, prevents them from compensating for the loss of KRASG12C function.

42Another approach to uncouple RTKs from RAS is to target the SHP2 phosphatase. SHP2 functions downstream of RTKs to promote RAS activation through regulation of SOS1 activity. Consequently, SHP2 inhibitors can act as pan-RTK inhibitors and prevent RTK signaling-induced RAS activation. This leads to accumulation of KRAS in the GDP-bound state, allowing for a higher proportion of KRASG12C being targeted by KRASG12C(OFF) inhibitors. Combinations of SHP2 and MAPK inhibitors have already indicated synergistic effects. Additionally, various in vitro studies have proven SHP2 inhibitors to potently and sustainably suppress the RAS pathway upon KRASG12C inhibitor treatment. As a result, multiple SHP2 inhibitors are under clinical evaluation in conjunction with KRASG12C inhibitors. Initial results from the KontRASt-01 trial (NCT04699188), combining the KRASG12C inhibitor JDQ443 with the SHP2 inhibitor TNO155, were released at the 2023 IASLC World Conference on Lung Cancer. DC rates of 67% were documented for patients pretreated with a KRASG12C inhibitor, with an ORR of 33%, which lasted more than 6 months. While SHP2 inhibitors demonstrate strong antitumor effects, particularly in preclinical models, they encounter tolerability issues in clinical settings, necessitating optimized dosing schedules to minimize adverse events.

43To overcome feedback loops driven by activated WT HRAS, NRAS, and KRAS, multi-RAS inhibitors offer a further alternative to block RTK-induced RAS-MAPK signaling. Early Phase 1 clinical trials of these inhibitors reported PR in patients with PDAC and NSCLC. As a result, ongoing clinical trials are exploring the combination of the multi-RAS inhibitor, RMC-6236, with the KRASG12C inhibitor RMC-6291 (NCT06128551).

44Another approach to prevent activation of WT HRAS by RTK feedback loops circles back to the previously described application of FTIs. A recent study demonstrated the benefit of combining sotorasib with the FTI, tipifarnib, in NSCLC, PDAC, and CRC. While oncogenic KRASG12C signaling is inhibited by sotorasib, the alternative RAS isoforms can compensate for this inhibition. FTIs, which impair HRAS signaling, resulted in enhanced efficacy of the KRASG12C inhibitor. Here, the combination might also benefit from the broad activity of FTIs on all farnesylated proteins rather than only on HRAS. One of the best known farnesylated proteins also targeted by FTIs is the small GTPase RHEB, which is involved in mTOR signaling, suggesting that PIK3CA/ mTOR signaling is also involved in therapy response.

45With KRAS-MAPK signaling being a key driver of the G1–S phase cell cycle transition, targeting of cell cycle-controlling proteins presents an additional vulnerability to improve KRAS inhibition and to address cell cycle-induced drug resistance. Therefore, inhibiting cell cycle regulators, particularly those driving the G1–S phase transition, is a promising cotarget strategy. The cyclin-dependent kinases 4 and 6 (CDK4/6) have a major role in promoting the G1–S phase transition, with their inhibition stalling the cell cycle in G0/G1. The CDK4/6 inhibitor palbociclib enhanced the efficacy of KRASG12C inhibitors in cell lines and mouse models. Currently, five clinical trials are under way in which clinically approved CDK4/6 inhibitors are combined with KRASG12C inhibitors.

46Another cell cycle kinase found to improve KRASG12C inhibition is Aurora kinase A (AURKA). Combining AURKA inhibition or knockout with KRASG12C inhibition effectively reduced KRAS reactivation and exerted an antiproliferative effect on cancer cells. Furthermore, AURKA inhibition stimulated KRASG12C inhibitor-induced quiescence, resulting in numerous cells remaining in the G0 state. Inhibitors of AURKA in combination with sotorasib were also studied in a clinical trial (NCT05374538), although this trial was terminated due to sponsor decisions.

47Interestingly, while G0 cell cycle arrest is induced by inhibitors of the RAS-MAPK pathway or cell cycle inhibitors, it can also be triggered by the withdrawal of the KRASG12C inhibitor in KRASG12C-amplified resistant cells. In this context, drug removal induces a signaling shock, resulting in oncogene-induced senescence. The senescent phenotype of the cancer cells presents a novel vulnerability that could be explored using senolytic drugs.

48Immunotherapy is a rapidly advancing field of research, with numerous clinical trials investigating the benefits of anti-PD-1/PD-L1 immune checkpoint therapy in cancer. T cell activity is promoted through the inhibition of either PD-1 or PD-L1, mediating tumor cell death. Multiple PD-1/PD-L1-targeting drugs, such as pembrolizumab and nivolumab, have received FDA approval within the past few years for the treatment of NSCLC and other types of cancer. This approach has expanded into combination therapies involving immunotherapy together with KRASG12C inhibitors in NSCLC and CRC. Preclinical evaluation in a KRASG12C-mutant, immune-competent mouse CRC CDX found that sotorasib alone caused short-term tumor regression followed by tumor regrowth. However, when combined with anti-PD-L1 treatment, complete tumor regression was detected, which persisted for over 2 months after treatment ending. Currently, there are 24 clinical trials combining anti-PD-1/PD-L1 therapy with KRASG12C inhibitors, the most advanced being the KRYSTAL-7 trial, a Phase 2/3 trial, combining adagrasib with the anti-PD-1 monoclonal antibody pembrolizumab (NCT04613596). No results of the trial have been posted so far.

49Given that PD-L1 mRNA is stabilized by oncogenic RAS signaling and, subsequently, PD-L1 expression is also regulated, combining inhibitors for the two targets appears promising. To further improve this combination, several in vitro studies highlighted the benefit of combining KRASG12C inhibitors not only with anti-PD-1, but also with SHP2 inhibitors. This triple combination is thought to induce antitumor immunity and remodel the tumor microenvironment, potentiating the effects of KRASG12C inhibition, especially in PDAC. Currently, one clinical trial is focusing on these in vitro findings, combining JDQ443 with TNO155 and tislelizumab (NCT04699188). No results have been posted so far.

50Additional treatments of mutant-KRAS tumors exploiting the immune system involve neoantigens derived from KRASG12C protein fragments covalently bound to KRASG12C inhibitors. Once presented on the surface of cancer cells via MHC I complexes, these neoantigens allow bispecific T cell engagers to mediate the interaction between the neoantigen and T cells. This interaction triggers a cytotoxic T cell response, leading to the selective destruction of cancer cells. This approach has the advantage that all cancer cells that still harbor the KRASG12C mutation and are inhibited by the KRASG12C inhibitor will be targeted by the cytotoxic T cells, even if these cells have developed resistance through mechanisms such as downstream mutations or MYC amplification.

51Given that chemotherapy remains the standard of care, it is no surprise that 17 clinical trials are evaluating the combination of chemotherapy with KRASG12C inhibitors. Preclinical studies provide support for these trials, demonstrating that sotorasib combined with standard-of- care carboplatin resulted in a significantly greater reduction in tumor growth compared with single-drug treatment in a KRASG12C-mutant NSCLC CDX mouse model. Similarly, while treatment with MRTX1133 or standard-of-care gemcitabine alone blocked the metastatic growth in a KRASG12D-mutant PDAC mouse model for only 2 weeks, the combination resulted in prolonged tumor regression. Whether the observed effects are due to synergy between KRASG12C inhibitors and chemotherapeutics or to chemotherapy targeting rapidly replicating cells, including those resistant to KRASG12C inhibitors, remains an open question.

52With the successful approval of two KRASG12C-selective inhibitors, and with numerous inhibitors targeting other KRAS mutations, the perception that KRAS is undruggable can now be put to rest. While most inhibitors under clinical evaluation are mechanistically similar to the two approved mutant-selective OFF allosteric inhibitors, some also present mechanistically distinct approaches, thereby increasing the variety of effective options. Although conventional wisdom is that mutant-selective inhibitors will be superior to inhibitors active on both the WT and mutant KRAS proteins and not be limited by normal tissue toxicity, the unanticipated clinical efficacy of the multi-RAS inhibitor RMC-6236 is challenging this perception.

53Moving forward, the most significant challenge for the continued development of KRAS inhibitors is overcoming resistance, which limits the depth and duration of response. It is now generally accepted that the long-term effectiveness of KRAS inhibitors will not be achieved through monotherapy, but rather through combination therapy. While our knowledge of resistance mechanisms remains limited, emerging evidence suggests that these mechanisms will be diverse and complex, varying both within and between different patients with the same cancer type. Thus, establishing effective combinations remains a daunting challenge. Further research is needed, particularly in the fields of patient stratification, combinational approaches, pharmacology, and the understanding of resistance development, to ensure not only improvements in the quality of life of these patients, but also prolonged survival (see Outstanding questions).

54After nearly four decades of setbacks, the successful development of clinically effective KRASG12C-selective inhibitors marks a significant milestone in cancer drug discovery. This achievement signals the beginning of an exciting new era of basic and translational RAS research. Previous anti-RAS efforts have been limited by our incomplete understanding of RAS function and future efforts will surely reveal that RAS has many more secrets to unlock. One thing is certain, the future holds promising therapeutic options for patients with RAS-mutant cancers.