The journey to target mutant KRAS in cancer treatment marks a significant milestone in oncological research, largely propelled by groundbreaking work initiated in Dr. Kevan Shokat’s lab. A pivotal discovery in 2013, documented in research identified by 01375 (referencing study [84]), pinpointed an allosteric binding site specific to the KRAS G12C-mutant protein, located behind the switch II region. This Switch II Pocket (S-IIP) presented a novel opportunity for targeted drug development as it was uniquely observed in the G12C-mutated form of KRAS. Leveraging the high reactivity of cysteine, researchers drew parallels from successful cysteine-targeting agents like ibrutinib and dacomitinib [85, 86]. These agents demonstrated the ability of active cysteine to act as a nucleophile, forming covalent bonds. Ostrem and colleagues built upon this by developing covalent inhibitors designed to bind to the cysteine residue within the S-IIP of the KRAS-GDP complex. This binding action induced conformational changes in the switch I and switch II regions, effectively blocking KRAS activation and subsequent downstream signaling, ultimately leading to apoptosis in cancer cells. Critically, these early compounds specifically targeted the GDP-bound state of KRAS G12C and required the protein’s intrinsic GTPase hydrolysis activity for efficacy. While compounds 6 and 12 showed potent in-vitro covalent inhibition, they lacked sufficient cellular activity.
This breakthrough was a watershed moment, paving the way for clinically effective agents. Initial prototypes included SML-8-73-1, ARS-853, and ARS-1620. SM-8-73-1, a guanosine-derived GDP analog, was designed to bind KRAS G12C but suffered from poor cell penetration [87, 88]. Conversely, ARS-853 and ARS-1620 targeted the S-IIP allosteric site, effectively inactivating KRAS G12C by trapping it in its GDP-bound state. ARS-853 demonstrated a 600-fold increase in potency compared to compound 12, inhibiting SOS-catalyzed nucleotide exchange and downstream MAPK signaling in KRAS G12C-mutant cells in a dose-dependent manner [89]. To address the metabolic and chemical instability of ARS-853, the quinazoline-based ARS-1620 was developed. This oral G12C inhibitor exhibited enhanced rigidity and a more favorable conformation due to its interaction with His-95 compared to ARS-853 [90]. Further clinical advancement necessitated additional ligand interactions beyond the limited size of the S-IIP. By utilizing the His-95 groove as an auxiliary binding site, AMG-510 (sotorasib) and MRTX849 (adagrasib) emerged as the first KRAS G12C inhibitors to enter clinical trials [71, 91, 92]. The successful introduction of these molecules sparked considerable excitement in the field, finally offering a therapeutic avenue for the previously considered ‘undruggable’ KRAS mutation. Currently, several other KRAS G12C inhibitors are in development, including GDC-6036, JDQ443, LY3537982, MK-1084, JAB-21822, BI-1823911, and D-1553.
Clinical Impact of KRAS G12C Inhibitors: Sotorasib and Adagrasib
The development of selective KRAS G12C inhibitors, spearheaded by sotorasib (AMG 510) and adagrasib (MRTX849), has had a rapid and profound impact on clinical practice. Both drugs are orally available small molecules that selectively and irreversibly bind to KRAS G12C in its GDP-bound state, effectively locking it in an inactive conformation (Fig. 5). Initial clinical trials explored their efficacy across various tumor types. While responses were observed in multiple cancers, monotherapy proved most effective in Non-Small Cell Lung Cancer (NSCLC).
Fig. 5
KRAS G12C protein structure illustrating the Switch-II Pocket inhibitor binding site.
The Phase I/II CodeBreaK100 trial investigated sotorasib monotherapy in patients with KRAS G12C-mutated tumors [93]. The Phase I dose-escalation phase found no dose-limiting toxicities across dose levels of 180 mg, 360 mg, 720 mg, and 960 mg, establishing 960 mg as the recommended Phase II dose. Among 129 patients across all dose levels, 56.6% experienced treatment-related adverse events (TRAEs), but only 11.6% experienced Grade 3 or 4 TRAEs. The most common Grade 3 or higher TRAEs included elevated ALT (4.7%) and AST (2.3%), anemia (4.7%), vomiting (3.9%), diarrhea (3.9%), and abdominal pain (3.1%). With a median follow-up of 11.7 months, the response rate in NSCLC patients (n=59) was 32.2%, with a disease control rate of 88.1%. In the Phase II cohort receiving 960 mg (n=34 NSCLC patients), the response rate was 35.3%. Responses were rapid, with a median time to response (TTR) of 1.4 months, and durable, with a median duration of response (DOR) of 10.9 months. The median Progression-Free Survival (PFS) for all NSCLC patients was 6.3 months. The Phase I portion also included 42 colorectal cancer patients, where the response rate was lower at 7.1%, although 73.8% achieved stable disease with a median duration of 5.4 months.
The Phase II portion of CodeBreaK100 evaluated sotorasib at 960 mg once daily in KRAS G12C-mutated cancers [94]. The NSCLC cohort comprised 126 patients, 81% of whom had prior platinum-based chemotherapy and PD-(L)1 inhibitor therapy. With a median follow-up of 15.3 months, the response rate was 37.1%, and 80.6% achieved disease control. Response characteristics mirrored Phase I, with a median TTR of 1.4 months and a median DOR of 11.1 months. Median PFS was 6.8 months, and median Overall Survival (OS) was 12.5 months (Table 2). The safety profile in Phase II was consistent with Phase I. Common TRAEs included diarrhea (31.7% all grades, 4% Grade 3), nausea (19.0% all grades, 0% Grade 3+), increased ALT (15.1% all grades, 6.3% Grade 3), increased AST (15.1% all grades, 5.6% Grade 3), and fatigue (11.1% all grades, 0% Grade 3+). TRAEs led to dose reduction and/or modification in 22.2% and discontinuation in 7.1% of patients. The US FDA granted accelerated approval to sotorasib for KRAS G12C-mutant NSCLC patients with at least one prior systemic therapy line on May 28, 2021 [95]. CodeBreaK200, the first randomized Phase III trial for sotorasib, compared sotorasib to docetaxel in 345 pre-treated metastatic KRAS G12C-mutant NSCLC patients [96]. The study met its primary endpoint of PFS, with 5.6 months for sotorasib versus 4.5 months for docetaxel (HR 0.66, p=0.002). However, the secondary endpoint of OS was not met. Median OS was 10.6 months for sotorasib and 11.3 months for docetaxel (HR 1.01, p=0.53). While OS data are still maturing, these results underscore the challenges in targeting KRAS.
Table 2 Efficacy results from single arm phase II studies of sotorasib (CodeBreaK100 [94]) and adagrasib (KRYSTAL-1 [100]) in NSCLC
Efficacy of sotorasib monotherapy in KRAS G12C-mutant colorectal cancer was less pronounced. In the Phase II CodeBreak100 colorectal cancer cohort of 62 patients, the ORR was only 9.7%, with a median PFS of 4.0 months, despite an 82.3% disease control rate [97]. In 38 pancreatic cancer patients, sotorasib achieved an ORR of 21.1% and a DCR of 84.2% [98].
Adagrasib monotherapy was evaluated in the Phase I/II KRYSTAL-1 trial. The Phase I portion explored doses ranging from 150 mg daily to 600 mg twice daily [99]. Similar to CodeBreaK100, enrollment was limited to KRAS G12C-mutated tumors. In 25 patients, mostly with NSCLC (72%), no maximally tolerated dose was identified, but dose-limiting toxicities were observed. The 600 mg twice-daily dose was selected for Phase II trials. At this dose, common TRAEs included nausea (80% all grades), diarrhea (70% all grades), vomiting (50% all grades), and fatigue (45% all grades, 15% Grade 3). In 15 evaluable NSCLC patients treated at 600 mg bid in Phase I, the ORR was 53.3%, with a median DOR of 16.4 months and a median PFS of 11.1 months.
The Phase II KRYSTAL-2 trial included 116 NSCLC patients treated with adagrasib 600 mg bid [100]. With a median follow-up of 12.9 months, the ORR was 42.9%. Median PFS was 6.5 months, and median OS was 12.6 months (Table 2). TRAEs were similar to Phase I, leading to dose reduction in 51.7% and interruption in 61.2%, but discontinuation due to TRAEs was low (6.9%).
Adagrasib has demonstrated efficacy in NSCLC patients with brain metastases. In patients with treated, stable CNS metastases, the intracranial ORR was 33.3% [100]. Case reports also indicate activity in untreated brain metastases [101]. Adagrasib concentrations in cerebrospinal fluid reached clinically relevant levels, consistent with preclinical mouse models.
Adagrasib is also under investigation in other KRAS G12C cancers. In a report of 46 colorectal cancer patients, adagrasib monotherapy showed a promising ORR of 22% and a disease control rate of 87% [102].
The advent of sotorasib and adagrasib has broadened treatment options for KRAS G12C-mutant NSCLC. Further understanding of the mutational landscape and its impact on response will refine patient selection and optimize therapeutic strategies. Ongoing research into rational combinations aims to enhance the effectiveness of these inhibitors in NSCLC and other KRAS G12C-mutated tumors. Biomarker testing has also become more crucial, requiring identification of specific KRAS mutations, not just presence.
Resistance Mechanisms to KRAS G12C Inhibitors (Fig. 6)
Fig. 6
Diagram illustrating the various mechanisms of resistance to KRAS G12C inhibitors.
Innate Resistance
In the CodeBreaK100 and KRYSTAL-1 Phase II trials, while most patients achieved disease control, objective responses were seen in less than 50%. Understanding innate resistance mechanisms is critical for improving treatment strategies. The heterogeneous response in KRAS-mutant cancers contrasts with the more uniform responses seen with other kinase inhibitors in NSCLC. This may be due to the higher prevalence of KRAS G12C mutations in smokers, leading to greater genomic heterogeneity [103]. Pre-existing tobacco-induced genomic alterations might provide alternative carcinogenic pathways independent of KRAS mutation. Biomarkers like PD-L1 and TMB, predictive in immunotherapy, were not predictive in CodeBreaK100 for KRAS G12C-mutant NSCLC [104].
Sotorasib responses in CodeBreaK100 Phase II were consistent across PD-L1 expression levels [94]. Similarly, adagrasib responses in KRYSTAL-1 Phase II were similar across PD-L1 strata. Sotorasib responses also did not vary by TMB status.
Concurrent mutations can also contribute to innate resistance. In CodeBreak100 Phase II, among patients with STK11 and KEAP1 co-mutation data, ORR to sotorasib varied significantly. ORR was highest in STK11-mutated/KEAP1-wild-type, and lowest in KEAP1-mutated/STK11-wild-type [Table 3]. KRYSTAL-1 showed similar trends for adagrasib. STK11 and KEAP1 are tumor suppressor genes with potential negative prognostic and predictive roles in KRAS-mutated patients treated with immunotherapy [73].
Table 3 ORR in KRAS G12C NSCLC by STK11 and KEAP1 mutation status from single arm phase II studies of sotorasib (CodeBreaK100 [94]) and adagrasib (KRYSTAL-1 [100])
Preclinical studies suggest primary resistance can arise from heterogeneous responses where some cells are eliminated while others synthesize new KRAS G12C, converting to a drug-resistant active form [105]. PI3K-driven MAPK activity can also promote tumorigenesis independent of KRAS deficiency, indicating potential innate resistance [106]. KRAS-mutant cancer cells can be classified as KRAS-dependent or -independent [107]. KRAS-independency is often associated with epithelial-to-mesenchymal transition (EMT) [108], another resistance mechanism. This can also explain non-KRAS driven tumorigenesis where KRAS mutation is a passenger event.
Acquired or Adaptive Resistance
Understanding acquired resistance is crucial for long-term efficacy. Mechanisms are broadly categorized as: (1) on-target resistance (KRAS alterations), (2) off-target resistance (bypass pathways), (3) TME changes, and (4) histological transformation [109].
Concurrent KRAS Alterations
KRAS G12C mutations can coexist with other KRAS alterations. Approximately 2.8% of KRAS-driven cancers exhibit concurrent KRAS mutations [110]. Double-mutant KRAS G12C and G12V cell lines are resistant to G12C inhibitors in vitro [111, 112]. While G12C signaling is blocked, alternative KRAS activation pathways remain unimpeded [110]. G12C inhibitors target the inactive KRAS-GDP conformation, so mutations favoring the active KRAS-GTP state confer resistance [105]. On-target resistance mutations include switch-II pocket mutations (KRAS Y96, R68, H95) and activating non-G12C KRAS mutations (G12D/V/R, G13D, Q61H) [109]. KRAS G12C amplifications are also seen [112]. In an adagrasib monotherapy study, 45% of patients developed acquired resistance mechanisms, with 88% involving RAS-MAPK pathway reactivation [113]. SIIP pocket mutations can vary in their impact on different G12C inhibitors; for example, H95 mutations decrease sensitivity to adagrasib but less so to sotorasib. In a sotorasib resistance study, 27 of 43 patients developed treatment-emergent alterations, including secondary KRAS alterations [114].
Vertical Signaling Pathway Alterations
Alterations in RTK and upstream pathways (GRB2, SHP2, SOS) can induce resistance [115]. Downstream MAPK cascade activation also impairs inhibitor efficacy. G12C inhibition suppresses negative regulators of MAPK pathway (DUSP, PHLDA, SPRY) [115]. G12C inhibitors can also increase MAPK signaling via RTK upregulation (EGFR, HER2, FGFR, ALK, MET) due to reactivated HRAS and NRAS signaling [116]. Upstream RTK inhibition can restore sotorasib sensitivity [116].
Resistance pathways differ by tumor origin. In NSCLC, sotorasib upregulates MEK and ERK, causing rapid resistance. In CRC, EGFR phosphorylation is more frequent, leading to downstream MAPK cascade activation [117]. This EGFR-mediated resistance is similar to BRAF inhibitor resistance in CRC [118, 119].
Studies of sotorasib and adagrasib resistance biopsies identified alterations in NRAS, BRAF, EGFR, MET, FGFR2, MYC, IDH1/2 [113, 114]. Optimal combination therapies to overcome these alterations require further investigation.
Other Mechanisms
Parallel pathways bypassing KRAS G12C inhibition include tumor microenvironment alterations, cell cycle regulator alterations, and phenotypic transformation.
Sotorasib treatment alters the tumor microenvironment, upregulating TGF-β signaling, EMT transformation, complement activation, neoangiogenesis, and coagulation [120]. Immune-related gene signatures are downregulated upon resistance [120]. Increased xenobiotic metabolism and adhesion kinase activation leading to fibrotic changes can also induce resistance [121].
In NSCLC, CDKN2A loss-of-function mutations (cell cycle regulator) occur in ~20% of resistant patients, leading to CDK4/6 RB phosphorylation. CDK4/6 inhibitors (palbociclib) combined with adagrasib can reverse this resistance in xenograft models [91]. AURKA inhibitors (alisertib) may have similar effects [105].
Histologic transformation is another off-target resistance mechanism. EMT transformation is seen in both EGFR and KRAS-driven tumors. Two adagrasib-resistant patients developed squamous cell carcinoma transformation [113], without identified genomic changes.
Non-G12C KRAS Inhibitors
While sotorasib is approved for KRAS G12C-mutant NSCLC, targeting other, more frequent KRAS mutations remains a major unmet need. The strategy used for G12C inhibitors is being adapted to target other KRAS mutations, including G12D, G12V, G12S, and G12R. These mutations lack a cysteine residue, requiring noncovalent approaches. They also lack intrinsic hydrolysis activity, likely remaining in the GTP-bound state [122]. BI-2852, a prototype in-vitro KRAS inhibitor, blocks both GDP-bound and GTP-bound KRAS states [123]. ‘ON’ inhibitors, targeting the GTP-bound state, are in development and may offer advantages over ‘OFF’ inhibitors, including faster signaling inhibition and resistance to upstream RTK amplification.
MRTX-1133, a noncovalent KRAS G12D inhibitor, binds both active and inactive states of KRAS G12D [124]. RMC-9805, a KRAS G12D ‘ON’ inhibitor, uses a tri-complex formation to block GTP-bound KRAS G12D [125]. RMC-6291 and RMC-8839 are selective KRAS G12C and G13C ‘ON’ inhibitors, respectively, using similar tri-complex strategies [126, 127]. RMC-6236, a non-selective RAS inhibitor, targets multiple ‘ON’ RAS isoforms [128] and is currently in Phase I clinical trials (NCT5379985). These novel inhibitors represent a significant advancement in directly targeting previously undruggable KRAS mutations, expanding therapeutic possibilities for a broader range of cancers.
