The endogenous activity of the MAPK pathway begins with the binding of an extracellular mitogen to a receptor tyrosine kinase (RTK). This leads to a cascade of events which ultimately triggers transcriptional changes that result in increased cellular proliferation and survival. Common mitogen-cell surface receptor pairs which initiate this pathway include epidermal growth factor (EGF) binding to EGF receptor (EGFR) and fibroblast growth factor (FGF) binding to FGF receptor (FGFR). Binding of mitogen to RTK triggers activation of tyrosine kinase activity of the RTK cytoplasmic domain. Docking proteins then bind to phosphotyrosine residues on the activated receptor. These cell-surface signals are transduced through KRAS. KRAS exists in the cellular context in two forms, an “off” state bound to guanosine diphosphate (GDP) and an “on” state bound to guanosine triphosphate (GTP). The protein contains intrinsic GTPase activity to turn itself off, and a variety of associated proteins can shift the balance between GDP-bound (GAPs, or GTPase activating proteins) and GTP-bound (GEFs, or guanine exchange factors) KRAS. In the presence of activating signals, GEFs replace GDP with GTP, allowing KRAS to turn “on.” A downstream kinase cascade, involving serial phosphorylation of downstream kinases, such as RAF, ERK and MEK, ultimately triggers transcription factor activation, leading to increased cellular proliferation, migration and survival (1–3). In the normal cellular context, the MAPK pathway is regulated by intrinsic feedback loops and negative regulators, such as Sprouty, RAS-association domain family (RASSF) proteins, and Son of Sevenless homolog 1 (SOS-1) which help to prevent aberrant activation and signaling (4).

There are extensive repositories documenting tumor mutational frequency in KRAS and other genes, including the Catalogue of Somatic Mutations in Cancer (COSMIC), The Cancer Genome Atlas (TCGA), and The International Cancer Genome Consortium (ICGC) (5). While the specific frequencies reported varies by source, activating mutations in KRAS are present in approximately 25-30% of NSCLC (6, 7), 80-95% of pancreatic ductal adenocarcinoma (PDAC) (8, 9) and 40-50% of CRC (9, 10) ( Figure 1 ). The most common KRAS activating mutations cluster around the nucleotide-binding pocket, and involve codons 12, 13 and 61. These mutations alter the homeostatic balance between active, GTP bound KRAS and inactive, GDP-bound KRAS, either by reducing GTP hydrolysis or by increasing the rate of GTP loading. This leads to constitutive activation of the MAPK pathway (10).

Among activating KRAS mutations, KRAS G12 mutations are particularly common. KRAS G12 mutations are present in 25% of NSCLC, 78% of PDAC, and 30% of CRC ( Figure 1 ). KRAS G12 mutations are missense mutations which result in the substitution of the amino acid glycine for another amino acid at codon 12, stabilizing KRAS in the active, GTP-bound state, though with slightly different features for each mutation (11). In KRAS G12C mutations, cysteine is substituted for glycine (12). KRAS G12C mutations are present in approximately 12% of NSCLC, 1.5% of PDAC, and 3% of CRC ( Figure 1 ) (7, 9, 10).

KRAS G12C mutations are present in approximately 3% of CRC cases with relatively little variation in mutation rates across geographic areas. Patient cohorts of 4,632 at MD Anderson Cancer Center in Houston, TX; 2,457 in Japan; 4,897 in Brazil; and 2,559 from a Nordic cohort with CRC who underwent genomic profiling, demonstrated KRAS G12C mutational rates of 2.9%, 2.8%, 3.4%, and 3%, respectively (13–15). Within the Nordic cohort, the frequency of KRAS G12C mutations reported at eight sites in Norway, Sweden, Finland and Denmark varied from 2%-4% (15).

KRAS G12C-mutant CRC appears to occur approximately equally in men and women, and has a 60% left-sided predisposition, compared with approximately 43% in unselected CRC in the US (13, 14, 16, 17). Although some studies have reported a difference in stage at presentation and metastatic sites in KRAS G12C-mutated CRC, the preponderance of data does not support a consistent significant difference (13–15, 18, 19).

Although there is some variability in studies, most large retrospective studies have demonstrated an association between KRAS G12C mutations and worse progression-free survival (PFS) and overall survival (OS) when compared with other KRAS-mutated CRC (13, 14, 19, 20). KRAS G12V mutations may also convey a poor prognosis (19, 20). One study, which included 187 patients with KRAS G12C mutations and de novo metastatic disease, demonstrated a median PFS of 6.4 months, 3.9 months, and 3.0 months after first, second and third lines of therapy, with an OS of 25.1 months (14).

The most common mutations in CRC occur in APC, TP53, KRAS and PIK3CA ( 21, 22 ). Rates of APC and TP53 mutations do not vary consistently based on the presence or type of KRAS mutation (14, 19). PIK3CA mutations are less common in patients with KRAS G12C mutations than in patients with KRAS mutations at alternative sites (14, 19). Mutations in other genes involved in the MAPK pathway, such as BRAF and NRAS, are much less likely to be mutated in CRC cases with KRAS mutations including G12C mutations (14, 17, 19). This is consistent with molecular data from other cancers, such as NSCLC, in which KRAS mutations are almost always exclusive of other strong driver mutations, such as EGFR (23, 24).

Alterations in the microsatellite instability (MSI) pathway are of particular clinical interest as they facilitate sensitivity to immunotherapy (25, 26). MSI is caused by a deficiency of mismatch repair (dMMR) and is found in approximately 10-15% of CRC. Rates of dMMR are very low in KRAS G12C-mutant CRC (1.6% vs 2.9% in other KRAS mutants and 4.5% in all cases) (19).

KRAS G12C mutated CRC displays lower rates of CIMP pathway activation (hypermethylation) compared with KRAS WT CRC (14). Overall, these data point to KRAS G12C CRC as being a biologically distinct class of CRC with different therapeutic opportunities.

Given the high frequency of KRAS mutations across multiple tumor types, including approximately 45% of CRC, mutant KRAS has been a pharmacologic target for over three decades (27). Direct KRAS targeting has been extremely challenging, as its affinity for GTP and mostly featureless surface precluded traditional approaches of small-molecule medicinal chemistry (28). This led to a variety of secondary approaches to RAS targeting, either by attempting to target other pathway “nodes” or the translation and processing of KRAS on its way to the cell surface. The enzyme farnesyltransferase, for example, catalyzes the addition of a farnesyl lipid group to a carboxyl terminal cysteine of KRAS, enabling localization to the cell membrane (29). Farnesyltransferase inhibitors were consequently an early class of pan-RAS inhibitors which, despite many preclinical successes, failed to show success in the treatment of KRAS-mutant cancers, possibly due to alternative prenylation of KRAS (30, 31).

In the context of repeated failures to directly target KRAS, it is notable that other inhibitors of the MAPK pathway have proven central in defining treatment paradigms for CRC. The EGFR inhibitors cetuximab and panitumumab proved effective in RAS WT CRC (32, 33) and the BRAF inhibitor encorafenib demonstrated efficacy in the treatment of BRAF V600E mutant CRC (34). Confining EGFR inhibitors, which act “upstream” of KRAS, to only KRAS WT disease makes conceptual sense, but limited the usefulness of these drugs.

The first success in directly targeting KRAS grew out of the unique biochemistry of the codon 12 glycine-to-cysteine (G12C) mutation. Cysteine thiols are uniquely nucleophilic amongst amino acid side chains. This is classically seen in formation of disulfide bonds and made the cysteine residue in mutant KRAS G12C an attractive pharmacologic target (12). This property enabled disulfide-fragment-based techniques to screen hundreds of potential small-molecule targets of G12C-mutant KRAS in a mutant-specific manner, sparing WT KRAS. Building on the fact that G12C was targetable, two clinical compounds have now obtained approval: sotorasib and adagrasib—with a half-dozen or more molecules close behind.

Sotorasib was the first KRAS G12C inhibitor approved in the United States. It demonstrated efficacy and acceptable toxicity in patients with previously treated KRAS G12C mutant advanced NSCLC. In CodeBreaK100, a single-arm phase 2 trial of 126 patients, Sotorasib demonstrated an objective response rate (ORR) of 37% with a median duration of response of 11.1 months. PFS was 6.8 months and OS was 12.5 months (35). A subsequent randomized controlled trial of 345 patients demonstrated superior PFS with sotorasib than with docetaxel (5.6 vs 4.5 months; HR 0.66 with p=0.0017) and fewer grade 3 or worse adverse events (AEs) (36). Common treatment-related AEs seen with sotorasib include diarrhea, nausea, vomiting, fatigue and liver function test elevation (35, 36).

Adagrasib was evaluated in KRYSTAL-1, a phase 2 study of 116 patients with previously treated KRAS G12C mutant NSCLC. Adagrasib demonstrated similar outcomes to sotorasib in this cohort, with an ORR of 43%, median duration of response of 8.5 months, disease control rate of 80%, PFS of 6.5 months and OS of 12.6 months. Common AEs were similar (37). These two proof-of-principle trials for direct KRAS inhibition were landmarks that led to the exploration of new KRAS-targeted drugs and KRAS G12C strategies in other histologies.

Sotorasib was evaluated in a phase 1-2 trial of 38 patients with previously treated metastatic KRAS G12C-mutant PDAC, with a median of two prior lines of prior therapy. The ORR was 21%; PFS was 4.0 months and OS was 6.9 months (38)

These promising but time-limited responses have prompted further investigation into the mechanisms of KRAS G12C inhibitor resistance, and drug combinations which could help increase response rates or prolong the duration of response (39).

Early clinical studies of KRAS-targeted therapies in G12C-mutant colorectal cancer showed less activity than what is reported in NSCLC. A single-arm study of sotorasib monotherapy in previously treated KRAS G12C-mutant CRC, for instance, demonstrated an overall response rate of 9.7% across 62 evaluable patients (40). More recently, adagrasib alone was tested in a cohort of 43 CRC patients in the KRYSTAL-1 trial. The investigator-assessed objective response rate was 19%, with a median PFS of 5.6 months (4.1-8.3, 95%CI) and median OS of 19.8 months (12.5-23.0, 95%CI) (41). Safety data was comparable to the results seen in NSCLC, with the most common adverse events being diarrhea, nausea, vomiting, and fatigue (all occurring in >40% of patients).

The mechanisms underlying these discordant responses to G12C targeting in CRC and NSCLC are not clearly understood. Why do a smaller percentage of patients respond? One hypothesis is that CRC tumors undergo a significant rebound in ERK phosphorylation soon after starting KRAS inhibition, suggesting rapid development of treatment-related adaptive signaling resistance (42). KRAS-mutant CRC has also been shown to have higher levels of “upstream” receptor tyrosine kinase phosphorylation as compared to NSCLC, especially in EGFR. EGFR-mediated resistance to other targeted therapies has been seen in CRC, including with BRAF inhibition, making this a natural place to look (43).

This observation has prompted studies combining small molecule KRAS inhibitors with anti-EGFR monoclonal antibodies in CRC. CodeBreaK-101, for example, included a cohort of 40 metastatic CRC patients treated with a combination of sotorasib and panitumumab, noting an ORR of 30% (16.6-46.5, 95%CI) (44). Adagrasib was similarly evaluated with or without concurrent cetuximab in 28 heavily pretreated G12C-mutant CRC patients. Without cetuximab, ORR was 19%; with cetuximab, the ORR was 46%. Notably, the standard paradigm in CRC has been that KRAS-mutant tumors do not have responses to anti-EGFR mAb therapy. Similar results have been presented for compounds earlier in development, including GDC-6036, a G12C inhibitor being developed by Genentech, which showed an even higher ORR of 66% when combined with cetuximab (45). Overall these results suggest that this combination may receive approval concurrently, or before, monotherapy treatment for KRAS G12C CRC. Still, despite the improved responses relative to KRAS inhibition alone, these results suggest a similar discrepancy in efficacy between CRC and NSCLC patients receiving identical combinations. These trials are summarized in Table 1 .

These somewhat blunted responses in CRC have not deterred continued development of new G12C inhibitors, either as monotherapy or in novel combinations for this indication. Using a search term of “G12C” for registered clinical trials yielded 76 unique entries (last updated April 27, 2023), of which 39 were trials that included patients with G12C-mutant CRC. Thirty-three studies were excluded from further review because they included only lung cancer patients, two were excluded because they were diagnostic/non-interventional studies, and one each was excluded for not being specific to G12C-mutant tumors or having KRAS mutations as an exclusion criterion. Within these exclusions, there are 12 distinct G12C inhibitors currently under investigation ( Table 2 ).

Essentially all are KRAS-GDP covalent inhibitors. Subtle differences in pharmacology and pharmacokinetics between these compounds might translate into efficacy benefits or differing potential in combination therapy regimens. Novartis’ JDQ443, for example, binds without involving the H95 residue (46). As a result of its H95-independent mechanism, it appears to maintain activity even among tumors with G12C/H95 double KRAS mutations, which might help mitigate some acquired resistance (47). Such second-site mutations in H95 appear to occasionally mediate resistance to other KRAS G12C inhibitors, such as adagrasib (48).

Janssen’s JNJ-74699157 similarly binds near the switch II pocket through an interaction with a differing cysteine residue, suggesting it may be able to evade these resistance mechanisms as well. In a phase 1 study of 10 patients across multiple tumor types, however, zero patients had an objective response and six experienced dose-limiting muscle toxicity in the form of CPK elevations (49). In contrast, the KontRASt-01 trial of JDQ443 is ongoing, with no alarming safety signals publicly reported to date (50).

Most recently, Eli Lilly’s LY3537982 presented a preliminary read-out of its phase 1 LOXO-RAS-20001 study at the 2023 AACR Annual Meeting (51). The ORR specifically among patients with CRC (n=20) receiving LY3537982 as monotherapy was 10%, compared to 42% among those with pancreatic (n=12) or 38% in NSCLC not previously treated with a KRAS G12C inhibitor (n=8). In one cohort of 11 CRC patients, the sponsors investigated the novel agent in combination with cetuximab, demonstrating an ORR of 45%. Liver enzyme elevations were more common in this cohort. This trial was notable for including several patients who had stopped prior G12C inhibitors for toxicity. Whether that toxicity would have abated on its own or LY3537982 has a more tolerable safety profile remains to be determined in larger studies.

Apart from these investigations into novel monotherapies, there is considerable interest in combining KRAS G12C inhibitors with other targets as a way to either potentiate their clinical effect or counteract the development of resistance. While the addition of anti-EGFR antibodies was one of the earliest studied combinations, concurrently targeting BRAF, MEK, and other downstream steps in the RAS/MAPK signaling pathway holds theoretical promise as well (52). In vitro and mouse models of lung cancer, for instance, have shown that adding the mTOR inhibitor everolimus to a KRAS G12C inhibitor reduces tumor cell viability (53). Dozens of other preclinical investigations have supported combination therapies.

One prominent combination is with checkpoint inhibitors, based on the hypothesis that KRAS inhibition induces pro-inflammatory changes in the tumor microenvironment of KRAS mutated tumors. Early mouse studies combining sotorasib and anti-PD-1 therapy illustrated increased CD8+ T-cell infiltration into the tumor microenvironment, an effect not seen in subjects treated with checkpoint inhibition alone (54). This finding correlated with improved in vivo efficacy, with 9/10 mice achieving complete responses, compared to 1/10 in the monotherapy arms. Such responses in preclinical studies are typically not seen with chemotherapy-immunotherapy combinations.

In an attempt to replicate that success in humans, the KRYSTAL-7 trial tested the combination of adagrasib and pembrolizumab as a first-line therapy in patients with advanced KRAS G12C-mutant NSCLC. The preliminary results presented at the 2022 ESMO Immuno-Oncology Annual Congress featured an objective response rate of 49% (35-63, 95%CI) across 53 patients, with rates at least directionally varying by PD-L1 tumor proportion score (TPS): 59% for TPS≤50, 48% for TPS 1-49, and 30% for TPS<1 (55). Sotorasib has similarly reported results for a combination arm with pembrolizumab, but this combination led to a notably high rate of liver toxicity. The main current question is whether this liver toxicity will be a class effect or if it will be specific only to some KRAS G12C inhibitors. Trials are likely to answer this question in the coming months and years.

Several ongoing trials include combinations of existing RAS inhibitors with SHP-2 inhibitors, including Novartis’s TNO 155, BridgeBio’s BBP-398, and Revolution Medicine’s RMC-4630. A combination of BBP-398 and sotorasib had a synergistic inhibitory effect on MAPK signaling and cell viability in an in vitro model and was granted fast-track designation (56). SHP-2 is a non-receptor type protein tyrosine phosphatase encoded by the gene PTPN11. Because SHP-2 acts downstream of most receptor tyrosine kinases and is essential for RAS activation and signaling rebound, there is a preclinical rationale to support combination therapy. Monotherapy trials of SHP-2 inhibition have shown modest efficacy in KRAS-mutated tumors, especially KRAS G12C (57), but most compounds are now in development in combination with other MAP/ERK pathway inhibitors, such as the KRAS G12C inhibitors. Sotorasib combinations have started to read out (58), and further data will clarify whether this combination has value in CRC or other cancers.

There are novel molecules targeting other companion proteins of RAS for which combination therapy hypotheses are natural. One is BI-1701963, a pan-RAS inhibitor targeting the catalytic site of SOS-1. SOS-1 is a GEF that binds to GDP-bound RAS proteins and is critical for tumorigenesis; in fact, the absence of SOS-1 has been to shown to have a deleterious effect on cancer cells containing KRAS mutations (59, 60). In a phase 1 dose-escalation study of 31 patients treated with BI-1701963 monotherapy, none achieved clinical responses and only seven had at least stable disease (61).