RAS signaling in carcinogenesis, cancer therapy and resistance mechanisms

Abstract

Variants in the RAS family (HRAS, NRAS and KRAS) are among the most common mutations found in cancer. About 19% patients with cancer harbor RAS mutations, which are typically associated with poor clinical outcomes. Over the past four decades, KRAS has long been considered an undruggable target due to the absence of suitable small-molecule binding sites within its mutant isoforms. However, recent advancements in drug design have made RAS-targeting therapies viable, particularly with the approval of direct KRAS^G12C inhibitors, such as sotorasib and adagrasib, for treating non-small cell lung cancer (NSCLC) with KRAS^G12C mutations. Other KRAS-mutant inhibitors targeting KRAS^G12D are currently being developed for use in the clinic, particularly for treating highly refractory malignancies like pancreatic cancer. Herein, we provide an overview of RAS signaling, further detailing the roles of the RAS signaling pathway in carcinogenesis. This includes a summary of RAS mutations in human cancers and an emphasis on therapeutic approaches, as well as de novo, acquired, and adaptive resistance in various malignancies.

Introduction to RAS: Its structure and mutations in cancer

RAS was firstly identified as a virus-encoded gene by Jennifer Harvey and her colleagues in 1964 [1]. It was regarded as an oncogene, being one of the most frequently mutated genes in human cancer [2, 3]. Mutationally activated RAS is present in approximately one in five human cancers.

The RAS gene family includes three members: HRAS, NRAS, and KRAS [4]. Indeed, RAS proteins are implicated in a variety of biological responses, including cell proliferation, migration, growth arrest, senescence, differentiation, apoptosis, and survival [2]. Cancer cells with RAS mutations display more aggressive phenotypes [5]. As a result, patients with RAS mutations are more likely to experience poor prognosis and shorter survival compared to those with wild-type (WT) RAS [6, 7].

Alterations in components of the RAS signaling pathway, especially the RAS proteins that serve as central mediators, have significant consequences in various cancers, particularly in NSCLC, colorectal cancer (CRC), and pancreatic ductal adenocarcinoma (PDAC) [8]. Over the past four decades, significant efforts from both academia and industry have been directed toward developing drugs targeting RAS proteins for cancer therapy [9]. This decades-long difficulties for drug design are due to several factors: (1) Inhibitors must selectively target the dynamic conformational changes that RAS undergoes as it cycles between the GTP-bound (RAS (ON)) state and the GDP-bound (RAS (OFF)) state, each characterized by distinct structural features; (2) RAS proteins exhibit a strong affinity for GTP, compounded by the high intracellular concentrations of GTP; (3) RAS proteins lack deep small-molecule binding pockets, making them challenging pharmacological targets; (4) On-target toxicity may arise from inhibition of WT KRAS or simultaneous targeting of downstream pathways, including RAF/MEK/ERK and PI3K/AKT/mTOR; (5) Downstream pathway inhibitors can lead to a paradoxical increase in RAS signaling due to the release of negative feedback; (6) The sequestration of the KRAS-GTP complex by effector proteins [10,11,12,13]. The groundbreaking discovery of compounds and the subsequent development of covalent allosteric inhibitors, which irreversibly bind to cysteine 12 and occupy a cryptic induced pocket in the switch II region of GDP-bound KRAS, effectively trap the oncoprotein in its inactive conformation, allowing for effective inhibition of KRAS^G12C [14]. In 2021, the clinically available KRAS^G12C inhibitors sotorasib (AMG 510) [15] and adagrasib (MRTX849) [16] were approved for a specific subset of patients with NSCLC. As of 2024, sotorasib has demonstrated promising clinical activity and tolerable safety in PDAC, while adagrasib has shown similar positive outcomes in CRC [17, 18].

RAS structures: insights into hotspot mutations

As previously noted, the RAS family comprises three genes—KRAS, NRAS, and HRAS—that encode isoforms with highly conserved sequences and structural homology. Each isoform includes a G domain (residues 1–166) and a C-terminal hypervariable region (HVR) (residues 166–188/189), both essential for RAS function (Fig. 1A). The G domain, which contains the switch I (residues 30–40), switch II (residues 60–76), and P-loop (residues 10–17) regions, is crucial for binding downstream effectors, thus facilitating signal transduction [19]. Notably, the three isoforms primarily vary in their HVR, which determines their unique cellular localization and distinct activities [20].

In all three RAS isoforms—KRAS, HRAS, and NRAS—the primary mutational hotspots are located at amino acid residues G12, G13, and Q61. These mutations compromise RAS’s intrinsic GTPase activity and enhance GEF-mediated nucleotide exchange, resulting in a persistently active, GTP-bound state that promotes oncogenic signaling [21,22,23]. Thus, we focused our analysis on the three-dimensional structures of KRAS, HRAS, and NRAS proteins, with specific attention to these prevalent mutational hotspots (Fig. 1B).

Fig. 1

Structure of RAS. (A) The RAS protein structure includes the effector lobe (residues 1–86), the allosteric lobe (residues 87–165), and the HVR (residues 167–188/189). Within the effector lobe, the switch I (residues 30–40) and switch II (residues 60–76) regions are essential for binding downstream effectors and interacting with GEFs or GAPs. The HVR domain facilitates membrane attachment, playing a critical role in defining RAS’s cellular localization. (B) KRAS, HRAS, and NRAS are shown in surface representation, highlighting key mutational hotspots. The position of residue G12 is displayed in white, G13 in green, and Q61 in orange. HVR: hypervariable region

RAS mutation frequencies in human cancers

RAS isoform mutations exhibit selectivity across various cancers. Specifically, KRAS mutations are most commonly found in solid tumors, particularly in PDAC, CRC, lung adenocarcinoma (LUAD), uterine corpus endometrial carcinoma (UCEC), stomach adenocarcinoma (STAD), and testicular germ cell tumors (TGCT). HRAS mutations are primarily observed in pheochromocytoma and paraganglioma (PCPG), thymoma (THYM), head and neck squamous cell carcinoma (HNSC), bladder urothelial carcinoma (BLCA), thyroid carcinoma (THCA), skin cutaneous melanoma (SKCM), and UCEC. In contrast, NRAS mutations are predominantly found in SKCM and hematological malignancies, such as acute myeloid leukemia (AML) (data available through the cBioPortal for Cancer Genomics) (Fig. 2A-C). Therefore, it is essential for researchers to conduct comprehensive studies on various human tumor types, focusing on the differences in RAS mutations from developmental or evolutionary perspectives. In general, some RAS gene mutations result in the production of oncoproteins that drive cancer, while others may be benign [24]. The most notable mutational hotspots in all three RAS isoforms (HRAS, KRAS, and NRAS) are found at three amino acid residues: G12, G13, and Q61. Despite the identification of numerous mutation sites and varying frequencies across different cancer types, KRAS mutations predominantly (about 80%) show a preference for the G12 hotspot. In contrast, approximately 60% of NRAS mutations occur at the Q61 site. The mutational frequency of the three hotspot residues in HRAS is relatively similar, accounting for about 20–30% of mutations [25] (Fig. 2D).

Fig. 2

The frequency of RAS mutations varies across different cancer types and is notably concentrated at the G12, G13, and Q61 residues in the exons of RAS oncogenes. (A) KRAS mutations are most prevalent in pancreatic ductal adenocarcinoma, followed by colorectal cancer and lung adenocarcinoma. (B) HRAS mutations are primarily observed in pheochromocytoma and paraganglioma, thymoma, and head and neck squamous cell carcinoma. (C) NRAS mutations are mainly found in skin cutaneous melanoma, acute myeloid leukemia, and thyroid carcinoma. (D) The prevalence of G12, G13, and Q61 mutations in the exons of KRAS, HRAS, and NRAS isoforms is highlighted. The data shown in graphs (A), (B), and (C) are sourced from the cBioPortal TCGA (available via the cBioPortal for Cancer Genomics), while the data in graph (D) were derived from recent studies utilizing the COSMIC or cBioPortal databases [3, 26, 27]

RAS mutant subtypes in human cancers

RAS alterations have been identified as oncogenic drivers in several cancer types, including PDAC, CRC, LUAD, and melanoma [28,29,30,31]. The distribution of RAS mutations varies among different cancer types, with KRAS^G12X mutations accounting for 91% of KRAS mutations in PDAC, 85% in LUAD, and 68% in CRC (Fig. 3A). Mutations at G12 in KRAS are the most common, followed by alterations at G13. Both KRAS^G12X and KRAS^G13X mutations disrupt the cycle between the GTP-bound active state and the GDP-bound inactive state, favoring the GTP-bound active state [32,33,34]. The KRAS mutant subtypes are primarily classified as KRAS^G12D, KRAS^G12V, KRAS^G12C, KRAS^G12R, KRAS^G12A, and KRAS^G13D mutations, along with KRAS wild-type amplification [35, 36] (Fig. 3B). In contrast, the NRAS mutant subtypes include NRAS^Q61R, NRAS^Q61K, NRAS^Q61L, and NRAS^Q61H alterations (Fig. 3B). The KRAS^G12C mutation is the most common mutant subtype in LUAD, while KRAS^G12D is the most prevalent allele in PDAC, where KRAS^G12C is rarely observed (Fig. 3B). Additionally, NRAS^Q61R is the most common mutant subtype found in melanoma (Fig. 3C). Indeed, the codons and frequencies of RAS mutations vary by tissue type.

Fig. 3

RAS mutant subtypes in pancreatic ductal adenocarcinoma, colorectal cancer, lung adenocarcinoma, and melanoma. (A) The most common RAS mutant subtypes found in PDAC, CRC, LUAD, and melanoma include: PDAC: KRAS^G12D; CRC: KRAS^G12V; LUAD: KRAS^G12C; Melanoma: NRAS^Q61R; (B) The prevalence and types of KRAS mutations in codons 12 and 13 across pancreatic cancer, colorectal cancer, and lung adenocarcinoma show a significant frequency of KRAS^G12D and KRAS^G12V mutations, with KRAS^G12C being less common. (C) The frequency and types of NRAS mutations in codon 61 in melanoma predominantly include NRAS^Q61R, along with other alterations such as NRAS^Q61K, NRAS^Q61L, and NRAS^Q61H. PDAC: pancreatic ductal adenocarcinoma; CRC: colorectal cancer; LUAD: lung adenocarcinoma

Together, the analysis of RAS protein data provides significant insights into the functional diversity within the RAS family, which includes the main types: HRAS, KRAS, and NRAS. Each of these types consists of subtypes that exhibit distinct mutations and biochemical properties, influencing their roles in cellular signaling and oncogenesis. Understanding the differences between these family members and their subtypes can aid in developing more targeted therapeutic approaches, potentially improving the efficacy of treatments designed to inhibit RAS-driven signaling pathways.

Activation and signaling cascade

The RAS–RAF–MEK–ERK (MAPK) signaling pathway is typically activated by various factors, including cytokines, cytokine receptors, hormones, protein kinases, transcription factors, and others [37]. The canonical pathway consists of a RTK linked to the RAS–RAF–MEK–ERK cascade, in which the RAS family (KRAS, NRAS, and HRAS) acts as GDP–GTP-regulated binary on-off switches (switch 1 and switch 2) during signal transduction [38]. Both switch regions undergo conformational changes, with the switch 2 state being particularly critical for the eventual development of RAS inhibitors (Fig. 4).

Fig. 4

An overview of the various RAS signal transduction pathways and therapeutic approaches for RAS-mutant tumors. Activation of RTKs promotes the exchange of GDP for GTP in RAS, thereby activating RAS. GTP-bound RAS binds to and activates the effector RAF, which initiates the MAPK signaling cascades. Targeting RTKs can reduce the activation of RAS populations. Inhibition of SOS or SHP2 decreases the GDP–GTP exchange rate, leading to a reduction in the GTP-bound RAS population. Another effector, p110, activates the PI3K signaling cascades. Both the MAPK and PI3K signaling cascades can be inhibited at each kinase tier. ERK: Extracellular signal-regulated kinase; FAK: Focal adhesion kinase; GEFs: Guanine nucleotide exchange factors; MEK: Mitogen-activated protein kinase kinase; PDK1: 3-Phosphoinositide Dependent Protein Kinase-1; PI3K: Phosphatidylinositol 3-kinase; RAF: Rapidly accelerated fibrosarcoma; RalGDS: Ral guanine nucleotide dissociation stimulator; RAS: Rat sarcoma virus; RTK: Receptor tyrosine kinase; SH3: SRC homology 3 domain; SOS: Son of sevenless

This switch is regulated by GEFs, which stimulate the conversion of the inactive GDP-bound form to the active GTP-bound form, and GAPs, which facilitate the conversion back to the inactive GDP-bound form [38, 39]. GEFs and GAPs are typically multidomain proteins that are regulated by extracellular signals [40]. RAS activation must be tightly regulated, as aberrant activation of RAS is linked to numerous human cancers [41]. The RAS-GEF family includes RAS-GRF, RAS-GRP, and SOS. The RAS-GRF protein is responsible for Ca²⁺ influx and calmodulin-dependent activation of RAS, primarily expressed in the central nervous system (CNS). In contrast, RAS-GRP is predominantly expressed in hematopoietic cells and stimulates RAS proteins downstream of non-receptor tyrosine kinases [35]. SOS is a widely distributed RAS-GEF, and its activation of RAS is critical for various biological processes, including cell growth [42, 43]. SOS1 and SOS2 (SOS1/2), activated by RTKs and cytokine receptors, bind to the SH3 domains of the adapter protein GRB2 through C-terminal proline-rich motifs [44]. GRB2 can simultaneously bind to the SOS1-activating non-receptor protein tyrosine phosphatase SHP2 via its SH2 domain, enabling precise cooperation with the RAS-GEF SOS to activate RAS. Subsequently, active RAS recruits and interacts with downstream effector proteins, particularly RAF, which phosphorylates and activates MEK1 and/or MEK2. This activation leads to the phosphorylation of various cytosolic and nuclear proteins, including transcription factors. Additionally, active RAS-GTP can interact with the downstream effector PI3K, thereby transducing signals to regulate biological processes [45, 46]. Therefore, the RAS–RAF–MEK–ERK and RAS–PI3K–AKT–mTORC pathways serve as fundamental signaling pathways of RAS [25, 46]. Other important RAS effectors include RalGDS, PLCε, and the Rho guanine nucleotide exchange factor TIAM1, among others (Fig. 4). All of these effectors are associated with cell proliferation, differentiation, cell cycle regulation, metabolic changes, and cell survival [47].

In conclusion, the activation of the RAS signaling cascade is a pivotal process in regulating cell proliferation, differentiation, and survival. Upon activation, RAS proteins trigger a complex network of downstream pathways, including the MAPK and PI3K pathways, which play crucial roles in cellular responses to external stimuli. Dysregulation of RAS signaling, particularly through mutations in RAS genes, is a common driver of oncogenesis, leading to uncontrolled cell growth and tumor development. A deeper understanding of RAS activation mechanisms and its signaling cascades provides important opportunities for developing targeted therapies aimed at inhibiting aberrant RAS activity, especially in cancers driven by RAS mutations.

Role in carcinogenesis

Metabolic programming is crucial for RAS-induced cell proliferation and carcinogenesis. RAS signaling enhances nutrient flux in cancer by participating in central carbon metabolism and increasing glucose uptake and glycolysis, thereby providing a competitive advantage to cancer cells [48]. It also promotes multiple branching biosynthetic pathways and regulates overall mitochondrial function by inducing mitophagy, which can delay tumor progression associated with damaged mitochondria [35]. In oncogenic KRAS-induced cancer growth, glycolytic ATP generation is essential for survival under hypoxic conditions, while aerobic glycolysis is likely important for providing glycolytic intermediates necessary for nucleotide and phospholipid synthesis. Additionally, KRAS promotes the glutamine-fueled tricarboxylic acid (TCA) cycle, leading to the production of ATP, ROS, NADPH, amino acids, nucleotides, and lipids. This process is crucial for RAS-induced tumorigenicity, as it supplies substrates to the TCA cycle from amino acids and other sources, such as fatty acid oxidation [48]. Additionally, RAS signaling contributes to oncogenesis and tumor progression by inducing fatty acid oxidation, which mediates pro-tumorigenic M2 macrophage polarization [49].

Upstream of the RAS signaling pathways are primarily composed of cell surface receptors, such as the EGFR and human ERBB2, which receive external signals and transmit these signals through KRAS. This biological process primarily promotes cell proliferation and migration [50, 51].

As mentioned earlier, KRAS functions as a switch for GDP-GTP regulation, controlling the cytoplasmic signaling network and various normal cellular processes. Two splice variants of KRAS, KRAS4A and KRAS4B, have been identified, both of which are essential for tumor initiation and likely have specific roles in the tumor microenvironment. For example, KRAS4B is typically expressed at higher levels and is found in both stem and progenitor cells, while the expression of KRAS4A increases tumor cell adaptation to stressors, such as hypoxia [52]. However, recent studies have shown that KRAS4A is widely expressed, and tumors can adapt to express KRAS4A through splicing during times of stress. These findings prompt a renewed focus on the role of KRAS4A in tumorigenesis and shift the perspective on KRAS inhibition, as KRAS4A now requires careful consideration [26, 53]. It is reported that KRAS4A and KRAS4B differ only in their C-terminal membrane-targeting region [53]. The unique membrane-anchoring mechanisms of KRAS4A and KRAS4B suggest variations in their dynamics of association with the cell membrane. Recent studies have uncovered isoform-specific interactions between KRAS4A and the RAS effectors Sin1 and hexokinase I [54, 55]. These isoform-specific interactions are likely attributed to the distinct localization of KRAS4A and KRAS4B in separate membrane environments, mediated by their unique HVRs [56]. Researchers also demonstrated the contrasting activation patterns of downstream signaling pathways between the two KRAS isoforms, attributable to their divergent HVRs [56]. Furthermore, the presence of hotspot oncogenic mutations at positions 12, 13, and 61 in both KRAS4A and KRAS4B poses a significant challenge for targeted therapies due to variations in their structures and functions [56]. Therefore, future studies should focus on delineate the distinct signaling properties of KRAS4A and KRAS4B to develop novel therapeutic strategies that effectively target both splice variants.

The downstream signaling pathways mediated by KRAS have been discussed previously. In the RAS-RAF-MEK-ERK pathway, KRAS-GTP is typically activated by various extracellular stimuli, including growth factors, hormones, cytokines, and environmental stresses. Following RAS activation, the serine/threonine kinases of RAF are recruited to the cell membrane, where their C-terminal catalytic domain binds to MEK1/2 and phosphorylates multiple serine residues on these two proteins. MEK1/2 are dual-specificity kinases that phosphorylate both tyrosine and threonine/serine residues, leading to the activation of ERK1/2. The activation of the RAS-RAF-MEK-ERK cascade plays a crucial role in promoting cancer cell proliferation, survival, migration, and angiogenesis [57]. In another pathway, the PI3K-AKT-mTOR pathway, RTKs, cytokine receptors, integrins, and GPCRs activate KRAS-GTP, which then binds to and activates PI3K. Activation of PI3K stimulates the phosphorylation of its phospholipid substrate, PIP2, to produce PIP3. This lipid interacts with AKT, promoting its phosphorylation and activation by PDK1. The activation of AKT subsequently activates mTOR, thereby regulating cell growth, survival, and metabolism [58, 59] (Fig. 4).

Clinical implications of RAS mutations in different cancer types

As mentioned above, the widespread prevalence of activating RAS mutations across various malignancies has been recognized. Among these, KRAS is the most frequently altered, followed by NRAS and HRAS. For instance, KRAS mutations occur in approximately 59.24% of pancreatic cancers, 36.2% of colorectal cancers, and 26.68% of LUAD [12] (Table 1). In clinical settings, RAS mutational status is associated with various clinicopathological characteristics, prognosis, and treatment efficacy (Table 1).

Table 1 Clinicopathological features, prognosis and treatment efficacy of patients with RAS mutations

NSCLC

In LUAD, activating missense KRAS mutations are typically mutually exclusive with other clinically recognized driver mutations, such as those in EGFR and ALK [60, 61]. An early study of LUAD patients indicated that the presence of KRAS point mutations in codon 12 serves as an unfavorable prognostic factor [62]. One study found that patients with LUAD harboring the KRAS^G12C mutation were more frequently associated with invasive mucinous adenocarcinoma and solid predominant tumors. These patients also had increased lymphovascular invasion, higher programmed death-ligand 1 (PD-L1) expression, and exhibited a potentially aggressive phenotype correlated with early and locoregional recurrence [63]. Regarding prognostic value, it was also shown that KRAS^G12C is an independent prognostic factor in stage I tumors and part-solid lesions [63]. Similarly, a recent observational study indicated that detectable KRAS^G12C is considered a marker of poor prognosis in lung cancer [64]. Although KRAS^G12D-mutant lung adenocarcinoma also exhibits similar clinical features to KRAS^G12C, such as a higher prevalence in males, former or current smokers, radiologically solid tumors, and invasive mucinous adenocarcinoma [65]; In terms of prognosis, KRAS^non−G12D mutations appear to be worse prognostic factors, particularly in stage I tumors. In contrast, KRAS^G12D mutations do not seem to be associated with clinical outcomes in resected stage I-III LUAD [65]. However, some studies have reported conflicting results. Evidence suggests that a significant number of KRAS-mutant lung cancers occur in never smokers, and there is a higher frequency of KRAS^G12C mutations in women [66]. Pooled analyses of early-stage resected NSCLC suggest that KRAS mutation status is not a significant prognostic factor [67]. Mucinous adenocarcinomas with KRAS mutations were also found to be more frequently located in the lower lung lobes, exhibiting a lower frequency of nuclear atypia and a reduced proportion of geminin-positive cells [68]. Notably, KRAS mutations have been considered indicators of resistance to therapy with EGFR tyrosine kinase inhibitors (TKIs), such as erlotinib, as well as to conventional chemotherapy in NSCLC [69,70,71,72]. Dissenting reports showed the negative impact of KRAS mutations on the response to EGFR-TKIs. Therefore, the current evidence is not enough to use the KRAS mutation statues to recommend the selection of patients for anti-EGFR treatment in NSCLC. Additionally, KRAS mutations have been observed in the progression of other EGFR-TKIs, albeit with a low prevalence of approximately 1% [73]. However, a subgroup analysis of OS indicated that immune checkpoint inhibitors (ICIs), such as nivolumab, were favored among patients with RAS mutation-positive status [74]. Furthermore, several studies have reported similar findings, particularly among patients with co-existing PD-L1 expression of 50% or higher [75, 76]. However, these findings require further validation before they can be incorporated into routine patient management.

PDAC

In PDAC, two studies have reported that the KRAS^G12V mutation (not KRAS^G12D) detected in plasma and serum is associated with poor survival, partly due to a high circulating proportion of regulatory T cells (Tregs) [77, 78]. However, more studies have shown that KRAS mutations, particularly the G12D mutation subtype in circulating tumor DNA (ctDNA), are independent predictors of poor prognosis and could also serve as early biomarkers of treatment response [79,80,81,82]. However, some studies present conflicting results. One study suggested that KRAS mutation status is more predictive than prognostic in advanced pancreatic cancer, suggesting that KRAS mutation status may be more useful for predicting how a patient will respond to treatment rather than determining their overall prognosis or survival outcome [83], other studies have suggested that KRAS WT status provides a significant advantage in OS for patients with PDAC treated with gemcitabine/nimotuzumab or gemcitabine/erlotinib, compared to those with KRAS mutations [84, 85]. Therefore, before a definitive conclusion can be reached regarding the impact of KRAS mutations on prognosis in PDAC, further research and additional investigations are required.

CRC

In CRC, RAS mutations are associated with more aggressive biological behaviors compared to their WT counterparts. These include a higher prevalence in mucinous tumor types, an increased tendency for lung metastases, and a preference for primary tumors to occur on the right side [86]. As a result, a study observed that KRAS mutations were independently associated with tumor location, and patients harboring KRAS or NRAS mutations in CRC demonstrated shorter OS [87]. Additionally, the metastatic potential of CRC varies with the presence of RAS mutations; these mutations are more frequently found in lung and brain metastases, whereas RAS WT CRC shows a significantly higher cumulative incidence of liver metastases [88, 89]. It is important to note that in patients with colorectal liver metastases, RAS mutations are independently associated with worse recurrence-free survival (RFS) and OS following repeat hepatectomy (RH) [90,91,92]. Notably, KRAS WT status predicts survival and is associated with an early radiological response to anti-EGFR therapies, such as cetuximab and panitumumab [93,94,95]. Reportedly, in a colorectal cancer mouse model, KRAS mutations are associated with suppressed Th1/cytotoxic immunity. Specifically, KRAS^G12D–mediated repression of IRF2 contributes to the resistance of colorectal cancer to anti–PD-1 therapy [96, 97]. Therefore, whether RAS mutational status should be considered prior to initiating ICI treatment requires further investigation.

Other malignancies

RAS mutational status also correlates with clinicopathological features in various other cancer types, including melanoma, thyroid cancer, cholangiocarcinoma, myeloid leukemia, and women’s cancers. In melanoma, NRAS mutations are associated with the presence of mitoses, lower tumor-infiltrating lymphocyte (TIL) grade, locations in the extremities (such as brisk lesions) rather than on the scalp or neck, advanced American Joint Committee on Cancer (AJCC) stages, and poorer melanoma-specific survival [98,99,100]. A recent study reported that NRAS-mutant cutaneous melanoma is associated with a worse prognosis compared to WT melanoma when treated with ICIs. It also showed an increased recurrence in both primary and relapsed cases, although OS was similar between the subgroups [101]. In thyroid cancer, RAS mutations are indicative of aggressive biological behavior, including poor differentiation or undifferentiated characteristics, and are associated with a poorer prognosis [102]. In cholangiocarcinoma, KRAS and NRAS mutations are associated with a pattern indicative of a more immune-inflamed microenvironment, characterized by higher M1 macrophage activation and increased interferon-γ expression compared to WT tumors [34]. Additionally, RAS mutations mediate resistance to FGFR inhibitors in FGFR2 fusion-positive cholangiocarcinoma [103]. KRAS mutations are also linked to aggressive behavior in intrahepatic cholangiocarcinoma (ICC), including the development of extrahepatic metastasis. These mutations affect the responsiveness of tumor cells to interferon immune signals and are associated with poor prognosis following surgical resection [104]. In chronic myelomonocytic leukemia (CMML), mutations in the RAS pathway are associated with a high risk of disease progression and resistance following treatment with hypomethylating agents (HMAs), the current standard of care for this condition [105]. In female cancers like ovarian cancer, activating KRAS mutations are frequently found in low-grade ovarian carcinomas, those at less advanced clinical stages, and in the mucinous histological subtype [106]. These findings indicate that patients with RAS mutations exhibit unique clinicopathological characteristics and have varying treatment responses depending on the specific tumor type and targeted therapies employed.

RAS proteins with alterations at codons 12, 13, or 61 result in a locked state of the enzyme in its GTP-bound, activated form, which is considered oncogenic [107]. However, recent studies indicate that each RAS mutation exhibits functional differences [108]. For example, the mutational status of KRAS is widely recognized as a predictor of resistance to therapy with EGFR antibodies, such as cetuximab [109,110,111]. However, retrospective analyses show that patients with KRAS codon 13 mutations, unlike those with codon 12 mutations, may benefit from cetuximab therapy [112]. Furthermore, in the colonic epithelium, the expression of KRAS^G12D, but not NRAS^G12D, stimulated hyperproliferation [113]. The expression of NRAS^Q61R in melanocytes induced the development of melanomas, whereas the expression of NRAS^G12D in these cells did not promote melanoma formation [108]. Therefore, both the specific isoform and the codon mutation should be taken into account when designing strategies to target RAS-driven cancers.

In summary, the clinical implications of RAS mutations can vary depending on the specific mutation and tumor type, which partly explains the conflicting results observed in different studies. Further research examining various mutant types is needed to evaluate the true clinical significance of RAS mutations in tumors.

RAS suppression strategies

Targeting RAS directly

For patients with RAS-mutant cancers, directly inhibiting RAS is a desirable treatment approach. The timeline of RAS inhibitor discovery is illustrated in Fig. 5. We also emphasize the recent development of various RAS inhibitors in this field, including mutant-specific RAS (KRAS^G12C) switch-II covalent inhibitors, therapies targeting KRAS^G12D, pan/multi-RAS/KRAS inhibitors, and immune therapies (Fig. 6).

Fig. 5

The timeline from the discovery of RAS to the development of KRAS inhibitors is as follows. 1964 to 1978: The ras oncogenes were identified; 1982: The nucleotide sequences of the HRAS and KRAS oncogenes were published; 1983: NRAS was identified; 1995 to 2013: RAS was historically considered “undruggable.”; 2021: The first KRAS^G12C inhibitor received approval from the Food and Drug Administration (FDA). Recently, numerous preclinical and clinical studies have focused on RAS inhibitors and their associated resistance mechanisms

Fig. 6

Therapeutic approaches to target RAS in cancer. This includes various strategies to target RAS mutations, particularly: KRAS^{G12C Inhibitors}: Data from clinical trials demonstrate their efficacy; KRAS^G12D Inhibitors: Targeting this specific mutation to provide therapeutic benefit; KRAS^G12V Inhibitors: Developing agents to inhibit this variant effectively; Multi-KRAS Inhibitors: Agents designed to target multiple KRAS mutations simultaneously; Pan/Multi-RAS Inhibitors: Broader inhibitors that target various RAS isoforms and mutations; Immune Therapies: Approaches that harness the immune system to target RAS-driven tumors. These diverse therapeutic strategies aim to improve outcomes for patients with RAS-mutant cancers

From the RAS discovery to the clinical development of KRAS^G12C inhibitors

In mouse development, KRAS is essential, whereas HRAS and NRAS are not required [114, 115]. As shown in Fig. 5, research on the direct inhibition of KRAS mutations can be traced back to the period from 1964 to 1978, when retroviral isolates were observed and subsequently identified to carry ras oncogenes [1, 116,117,118]. In the spring of 1982, the laboratories of Robert Weinberg, Michael Wigler, and Mariano Barbacid reported the molecular cloning of a human transforming gene from bladder carcinoma cell lines [119,120,121]. By the autumn of 1982, the nucleotide sequences of the HRAS and KRAS oncogenes were published, marking a shift in the field toward the recently isolated human oncogenes. NRAS was subsequently identified in 1983 [4]. Then, in 1995, one of the earliest examples of rational drug design based on ras oncogene research emerged with the development of peptidomimetic inhibitors of mammalian farnesyltransferase (FT). These inhibitors were found to block RAS-induced tumors in mice [122, 123]. However, targeting RAS farnesylation is not ideal, as many other proteins are also farnesylated. Nevertheless, tipifarnib, a farnesyltransferase inhibitor (FTI), has shown encouraging clinical activity for hematological malignancies, particularly in relapsed or refractory peripheral T-cell lymphoma (PTCL), and is currently being evaluated in clinical trials (NCT02464228) [124]. In solid tumors, such as NSCLC, tipifarnib has been identified as one of the most effective drugs in preventing relapse to targeted therapies, including EGFR-TKIs (such as erlotinib or osimertinib), the KRAS^G12C inhibitor (sotorasib), ALK-EML4 inhibitors (such as lorlatinib), and BRAF^V600E inhibitors (such as dabrafenib) [125]. These findings pave the way for the combination of FTIs and targeted therapies.

Small molecule inhibitors targeting RAS have faced challenges due to the lack of an adequate binding pocket, leading to several decades of effort in drug discovery. In 2013, the laboratory of K. Shokat made a significant breakthrough in targeting KRAS [126]. They screened for KRAS^G12C-specific small molecules that irreversibly bind to the cysteine at the mutation site and selectively target the KRAS^G12C-GDP inactive state [126]. Therefore, inhibitors of this type do not affect RAS signaling in non-malignant cells, theoretically resulting in a low risk of on-target, off-tumor toxicities [126,127,128]. In 2019, sotorasib became the first clinical KRAS^G12C inhibitor to enter trials for advanced solid tumors [129]. As illustrated in Fig. 3B, KRAS^G12C mutations primarily occur in NSCLC and CRC. Following the success of clinical trials, such as the CodeBreaK100 study and the phase 1/1b KRYSTAL-1 study in NSCLC and CRC, the U.S. FDA granted accelerated approval for sotorasib in May 2021 and for adagrasib in December 2022 [16, 130,131,132]. Recent results from the CodeBreak 200 study, a randomized, open-label, phase 3 trial, demonstrated a modest progression-free survival (PFS) benefit of sotorasib compared to docetaxel (5.6 months versus 4.5 months, p = 0.0017), meeting its primary endpoint. However, OS, which was a key secondary endpoint, did not show improvement for patients with previously treated metastatic NSCLC [15]. The U.S. FDA review raised concerns about potential biases, necessitating a new confirmatory phase 3 study to secure full regulatory approval [12]. Although KRAS^G12C mutations are infrequent (approximately 1–2% of cases), recent data showed that advanced pancreatic cancer treated with sotorasib demonstrated a PFS of 4.0 months, an OS of 6.9 months, and an objective response rate (ORR) of 21%. In contrast, adagrasib suggested slightly higher response rates, with a PFS of 5.4 months, an OS of 8.0 months, and an ORR of 33% [18, 133]. The differences in response rates may be attributed to potential biases, such as the limited number of patients enrolled in the studies. Notably, in these settings, patients were heavily pretreated, with a median of 2–3 prior lines of therapy. Based on these data, both sotorasib and adagrasib have been included as approved agents in the National Comprehensive Cancer Network (NCCN) guidelines for PDAC.

Recent data on the newer KRAS^G12C selective inhibitor divarasib, which was designed for high potency and selectivity in solid tumors, demonstrated confirmed responses with a PFS of 13.1 months in NSCLC and 5.6 months in CRC [134]. Additionally, a 36% partial response rate (PCR) was observed in PDAC with the use of divarasib [134]. Increased drug potency observed with single-agent divarasib and adagrasib compared to sotorasib is reflected in slightly higher objective response rates (ORRs) and longer PFS, despite the small number of patients enrolled in the studies. Recent data on novel G12C inhibitors in NSCLC, CRC, and PDAC are also promising. For instance, Opnurasib demonstrated an ORR of 42% and a disease control rate (DCR) of 93% in NSCLC (n = 24; NCT04699188). IBI351 showed an ORR of 46.6% and a DCR of 90.5% in NSCLC (n = 116; NCT05005234) and an ORR of 47.5% and DCR of 85% in CRC (n = 40; NCT05005234). Additionally, Olomorasib achieved an ORR of 42% and a DCR of 92% (n = 24; NCT04956640), while Glecarisib exhibited an ORR of 42% and a DCR of 93.5% in PDAC (n = 31; NCT05002270). Notably, Glecarisib also demonstrated significant anti-tumor activity in CRC, with an ORR of 33.3% and a DCR of 90.9% (n = 33; NCT05002270) (Fig. 6).

Preliminary data from Phase I/II trials indicate that garsorasib, which has high oral bioavailability and CNS penetration, demonstrates promising antitumor activity in NSCLC patients with brain metastases [135]. Notably, 33–42% of patients with KRAS^G12C-mutated NSCLC are initially diagnosed with CNS metastases [136, 137]. Adagrasib is the only KRAS^G12C inhibitor with reported activity data in untreated CNS metastases, demonstrating a CNS ORR of 42% and a PFS of 5.4 months (n = 19). The CNS failure rate was 37% (7 out of 19 patients), with only two patients experiencing CNS progression in the KRYSTAL-1 trial [138]. Therefore, Adagrasib may currently be the first choice for patients with KRAS^G12C-mutant NSCLC. However, Phase 3 studies will be necessary to provide further guidance for clinical practice.

Next batter up! From targeting KRAS^G12D and pan-RAS inhibitors to emerging therapeutics

As mentioned earlier, KRAS^G12C mutations represent only a subset of KRAS mutations, primarily found in LUAD. To effectively target KRAS mutations, it is essential to develop strategies against other prevalent specific mutations, such as KRAS^G12D. With the emergence of allele-specific KRAS^G12C inhibitors, there is a growing focus on KRAS^G12D inhibitors, pan/multi-RAS/KRAS inhibitors, and novel immunotherapies, including KRAS peptides and vaccines, which are currently being tested in patients and entering clinical trials [139,140,141,142] (Fig. 6).

Targeting KRAS^G12C is achievable by designing a reactive warhead that forms an irreversible covalent bond with the mutant cysteine-12 residue [143, 144]. Due to the absence of reactive cysteines in the active site of KRAS^G12D mutations, alternative approaches are being developed for these inhibitors. Since KRAS transitions between a GTP-bound ON state and a GDP-bound OFF state, developing inhibitors for specific mutations requires evaluating which state to target. MRTX1133 is the first noncovalent, potent, and selective inhibitor for KRAS^G12D in its OFF state. It binds to the switch II pocket, inhibiting nucleotide exchange and preventing protein-protein interactions with the effector RAF [145]. Although MRTX1133 does not form a covalent bond, it demonstrates significant anti-cancer properties and is set to enter clinical trials in June 2024 (NCT05737706). In contrast, RMC-9805, a selective and orally bioavailable KRAS^G12D (ON) inhibitor, first establishes a noncovalent bond between KRAS^G12D and cyclophilin A, which subsequently allows a “cool” nonreactive covalent warhead to slowly bind to the mutant aspartate. RMC-9805 has also entered clinical trials in September 2023 (NCT06040541). Additionally, the KRAS^G12D degrader ASP3082 is currently in Phase 1 clinical trials (NCT05382559). This degrader works by binding KRAS^G12D to an E3 ligase, leading to the degradation of the protein. Another KRAS^G12D inhibitor, HRS-4642, forms a salt bridge with KRAS’s Asp12 [146]. Although HRS-4642 exhibits similar binding affinity for both GDP-bound and GTP-bound KRAS^G12D, crystallographic studies reveal the structural basis of inhibitor binding, which induces changes in the switch II pocket of KRAS^G12D [146]. Recent data from the Phase 1 clinical trials of HRS-4642 in China (NCT05533463) demonstrate encouraging efficacy in NSCLC, showing a 10% ORR and a 90% DCR (n = 10) [12]. Further studies are necessary to identify the factors that predict responses to KRAS^G12D inhibitors and to determine which combination therapies are likely to be effective for different cancer types. Additionally, with the advancement of allele-specific KRAS inhibitors, it is essential to conduct head-to-head comparisons between KRAS alleles to better characterize the allele-specific effects on tumor biology [33].

Selectivity for KRAS was achieved through direct and/or indirect constraints imposed by the evolutionary divergence among RAS isoforms in three residues within the G domain [147]. Therefore, developing pan-RAS/KRAS inhibitors that preferentially target the inactive state of RAS/KRAS is crucial to prevent reactivation through nucleotide exchange. These pan-RAS inhibitors can address mutations across all RAS isoforms, thereby potentially benefiting the largest patient population. RMC-6236 is a potent, orally bioavailable multi-RAS (ON) inhibitor, selective for the active RAS (ON) form of both wild-type and mutant variants of the canonical RAS isoforms (HRAS, NRAS, and KRAS). It is currently undergoing Phase 1 clinical trials (NCT05379985). Recent reports indicate encouraging clinical activity signals in NSCLC with an ORR of 38% and a DCR of 85% (n = 40), as well as in PDAC with an ORR of 20% and a DCR of 87% (n = 46) [12]. However, pan-RAS inhibitors may carry a higher risk of toxicity, as they inhibit signaling through wild-type KRAS, NRAS, and HRAS isoforms. Consequently, it is rational to develop a new class of pan-KRAS inhibitors (also referred to as pan-KRAS-selective inhibitors) that target most wild-type and mutant KRAS isoforms while sparing NRAS and HRAS. The first pan-KRAS-selective inhibitor, BI-2865, along with its close analogue BI-2493, selectively binds to KRAS through an interaction with His 95, one of the four amino acids in the switch II binding pocket that vary among isoforms [148]. BI-3,706,674 is a pan-KRAS OFF state inhibitor that is currently undergoing Phase 1 clinical trials, although no published data are available yet (NCT06056024). Therapeutic nucleic acid-based approaches, including small interfering RNAs (siRNA), also hold promise for developing drugs targeting KRAS. One such clinical drug candidate, AZD4785, is a potent 2′-4′ constrained ethyl-modified antisense oligonucleotide inhibitor that selectively targets KRAS. It has the ability to target al.l mutant isoforms of KRAS, offering significant therapeutic potential across various tumor types [149].

Multiple immunotherapeutic strategies for targeting RAS are emerging. Earlier data indicated that mutant RAS peptide vaccines can induce host T cell responses, with potentially improved survival observed in a small single-arm study [150]. In 2016, a case was reported involving a CRC patient who received cytotoxic T cells targeting mutant KRAS^G12D, resulting in significant tumor regression [151]. Recently, researchers reported a case of a patient with progressive metastatic pancreatic cancer who achieved objective tumor regression after receiving T-cell receptor (TCR) gene therapy targeting the KRAS^G12D driver mutation [152]. More recently, data from the Phase 1 AMPLIFY-201 trial in CRC and pancreatic cancer demonstrated that the lymph-node-targeted mutant KRAS-specific amphiphile vaccine (ELI-002 2P) was safe and induced significant T cell responses. Specifically, 84% of patients exhibited mutant KRAS-specific T cell responses, with 21 out of 25 patients showing responses (59% of whom had both CD4 + and CD8 + T cells) [142]. Additionally, the median RFS was reported to be 16.33 months [142]. Seven amphiphile-modified KRAS and NRAS peptides—G12D, G12R, G12V, G12A, G12C, G12S, and G13D (Amph-Peptides 7P)—are currently being investigated in a Phase 1/2 study, verified in July 2024 (NCT05726864). Similarly, several clinical trials are underway in Phase 1, including a KRAS peptide vaccine (NCT05013216) and an anti-KRAS^G12D mTCR (NCT03745326).

As we know, after Phase 3 studies, drugs may advance to clinical use and inform therapeutic decisions. Accordingly, Table 2 outlines ongoing Phase 3 studies based on lines of therapy. Currently, only KRAS^G12C inhibitors are in Phase 3 trials, which could potentially alter first-line therapy for NSCLC and second-line therapy for CRC in the future.

Finally, we summarize KRAS^G12C inhibitors in Phase 1/2 clinical trials in Table 3, along with other inhibitors, including KRAS^G12D inhibitors, pan-RAS/RAS wild-type inhibitors, pan-RAS inhibitors, and immune therapies currently in clinical trials, as outlined in Table 4.

Table 2 KRAS^G12C inhibitors in phase 3 clinical trials based on lines of therapy

Table 3 KRAS^G12C inhibitors in phase 1/2 clinical trials

Table 4 Other inhibitors in clinical trials

Targeting upstream and downstream proteins

As illustrated in Fig. 4, the RAS signaling pathway consists of several upstream regulators and downstream effectors. Modifying one of these critical factors can serve as an indirect approach to inhibit RAS activation.

Targeting upstream mediators

The strategies include inhibiting upstream regulators, attenuating the SOS-RAS interaction, targeting the GN binding site, and repressing SHP2, as depicted in Fig. 4.

The upstream regulators of the RAS pathway include RTKs, such as the EGFR. erlotinib and afatinib are first-line EGFR-TKIs used to treat NSCLC patients with EGFR mutations. Additionally, cetuximab and panitumumab are EGFR monoclonal antibodies approved for use in metastatic CRC [153], while necitumumab is utilized for the treatment of squamous cell lung cancer [154]. Acquired KRAS mutations are widely recognized as a common mechanism of resistance to EGFR inhibitors in CRC [155], However, a study based on retrospective analyses found that KRAS codon 13 mutations may actually be associated with responsiveness to cetuximab therapy [112]. Recently, several novel drugs, including Amivantamab, Sunvozertinib, and Poziotinib, have emerged to target EGFR exon 20 insertion mutations in NSCLC [156,157,158]. Among these, Amivantamab, a bispecific antibody that directly targets both EGFR and the MET receptor, has received FDA approval based on the results of the CHRYSALIS clinical trial [159]. It is essential to understand how intrinsic and acquired KRAS mutations influence the antitumor activity of these emerging therapeutic approaches in clinical settings. Identifying these effects will help optimize treatment strategies and improve patient outcomes.

As noted earlier, SOS is a key member of the GEFs that catalyzes the conversion of RAS-GDP to RAS-GTP. Consequently, the development of inhibitors targeting the SOS-RAS interaction is receiving growing attention. A combination of fragment screening and high-throughput screening led to the identification of a small-molecule compound, BAY-293, which effectively disrupts the KRAS-SOS1 interaction. This disruption inhibits the reloading of KRAS with GTP, resulting in significant antiproliferative activity [160]. Additionally, BI-3406 is a highly potent and selective small-molecule SOS1 inhibitor that is orally bioavailable. It prevents the KRAS–SOS1 interaction by binding to the catalytic domain of SOS1, effectively limiting cellular proliferation [161]. Notably, BI-3406 increases the sensitivity of KRAS-mutant cancers to MEK inhibitors by preventing the feedback reactivation that often occurs with MEK inhibition [161]. Consequently, a recent report indicated that combining BI-3406 with adagrasib appears to be a promising strategy for overcoming both intrinsic and acquired resistance to KRAS^G12C inhibitors [162]. The first inhibitor, BI-1,701,963, prevents the reactivation of KRAS and is currently undergoing clinical trials (NCT04111458).

The GN pocket appears to be an ideal target for drug design; however, the sub-nanomolar affinity of GTP and GDP for RAS, combined with their high intracellular concentrations, poses significant challenges for developing inhibitors that target this site. SML-8-73-1 is a GDP analogue specifically designed to interact with the GN-binding pocket of KRAS^G12C. Its prodrug derivative, SML-10-70-1, has demonstrated antiproliferative effects in both H23 and H358 cell lines (which are dependent on the G12C mutation and KRAS) as well as in A549 cells (which harbor a G12S mutation and are KRAS-independent) [128, 163]. Another small molecule KRAS agonist, KRA-533, activates KRAS by binding to the GTP/GDP-binding pocket, thereby preventing the cleavage of GTP into GDP [164]. KRA-533 effectively suppresses malignant growth by promoting apoptosis and autophagic cell death [164].

SHP2, encoded by the PTPN11 gene, is a non-receptor protein tyrosine phosphatase that plays a crucial role in signal transduction, promoting SOS1-mediated RAS-GTP loading. SHP099 is a moderately potent, selective, and orally bioavailable small-molecule inhibitor of SHP2. It effectively inhibits the proliferation of cancer cells both in vitro and in vivo by suppressing SHP2 activity [165]. SHP394 is an orally efficacious inhibitor of SHP2, designed to improve potency and enhance pharmacokinetic properties [166]. Additionally, RMC 4550, a potent and selective allosteric inhibitor of SHP2, represents a promising therapeutic strategy [167]. Recently, TNO155, a highly potent, selective, and first-in-class SHP2 inhibitor, has entered clinical development [168] (NCT03114319). RMC 4630 and JAB-3068, both SHP2 inhibitors, are currently undergoing clinical trials (NCT03634982, NCT05054725, NCT04916236, and NCT03989115 for RMC 4630; NCT03565003, NCT03518554, and NCT04721223 for JAB-3068). As the data from these trials have not yet been published, it will be worthwhile to monitor future developments!

Targeting downstream effectors

RAS mediates oncogenic transformation through the activation of downstream signaling pathways. As illustrated in Fig. 4, the primary downstream pathways include RAF/MEK/ERK and PI3K/AKT/mTOR. Inhibiting these effectors can effectively block oncogenic RAS signaling. Strategies targeting these pathways involve RAF inhibition, MEK inhibition, ERK inhibition, and PI3K-AKT-mTOR inhibition.

ARAF, BRAF, and CRAF comprise the RAF kinase family, all of which share a common upstream activator, RAS [169]. ZM336372 was the first small-molecule, ATP-competitive RAF inhibitor developed, targeting CRAF in cancer [170], However, among the first-generation RAF inhibitors, only Sorafenib progressed to clinical use and received FDA approval for treating advanced renal cell carcinoma (RCC) [171], and WT BRAF hepatocellular carcinoma [172]. Despite its multi-kinase profile, Sorafenib exhibits limited efficacy against BRAF-V600E mutations in cells [173]. This development resulted in the creation of BRAF-V600E inhibitors, with vemurafenib being the first RAF inhibitor to enter clinical trials. It received FDA approval in 2011 for treating patients with BRAF-V600E metastatic melanoma [174, 175]. Following this, dabrafenib was approved in 2013 for melanoma patients with BRAF-V600E/K mutations [176, 177]. In KRAS-mutant and RAS/RAF wild-type tumors, dabrafenib and vemurafenib activate the MAPK pathway instead of suppressing signaling [178, 179]. The underlying mechanism involves BRAF inhibitors driving RAS-dependent BRAF binding to CRAF, thereby activating MEK-ERK signaling [180]. Consequently, targeting both BRAF and CRAF appears essential. LXH-254 is an inhibitor that effectively targets this pathway and has shown efficacy against NRAS-mutant NSCLC cells [181]. However, LXH254 exhibits reduced activity against ARAF and may induce paradoxical activation of MAPK signaling, similar to the effects observed with dabrafenib [182]. Thus, the development of pan-RAF inhibitors is essential. Belvarafenib, a potent inhibitor of BRAF V600E, as well as wild-type CRAF, BRAF, and ARAF, binds to both protomers of a RAF dimer, thereby demonstrating clinical activity in BRAF- and NRAS-mutant melanoma cells without inducing paradoxical MAPK activation in RAS-mutant cells. However, mutations in the ARAF isoform have been identified as a potential driver of resistance to Belvarafenib [183, 184]. These resistance data support the hypothesis that novel therapeutic strategies should focus on combination approaches or the inhibition of all RAS isoforms to effectively disrupt compensatory mechanisms and achieve significant antineoplastic effects. Lifirafenib (BGB-283), an investigational reversible inhibitor of key RAF family kinases (BRAF- V600E, wild-type ARAF, BRAF, CRAF) and EGFR, has demonstrated clinical activity in solid tumors harboring BRAF and KRAS/NRAS mutations and is currently undergoing clinical trials [185] (NCT02610361). Preclinical studies also suggest that lifirafenib enhances the antitumor activity of MEK inhibitors in KRAS-mutant tumors [186]. Other RAF inhibitors, including PLX8394 and DAY101, are currently undergoing evaluation in clinical trials (NCT02428712 and NCT03429803, respectively).

MEK inhibitors, such as Trametinib, Cobimetinib, and Binimetinib, have been approved for the treatment of patients with advanced melanoma harboring the BRAF V600E/K mutation [187,188,189]. Binimetinib has demonstrated activity in patients with NRAS-mutated melanoma [190, 191]. However, MEK inhibitors, such as Trametinib and Selumetinib, have not improved survival in patients with KRAS-mutant advanced NSCLC [192, 193]. Similarly, Trametinib showed no survival benefit in patients with untreated metastatic pancreatic cancer, regardless of KRAS mutation status [194]. The underlying mechanism is CRAF-mediated MEK activation [195]. Thus, the concept of co-targeting MEK and CRAF has emerged; RAF/MEK inhibitor combinations have shown synergistic efficacy in KRAS-mutant tumor cells [196]. These findings provide the rationale for ongoing clinical trials of combination RAF and MEK inhibitors for KRAS-mutant malignancies.

Reactivation of ERK signaling is recognized as a common driver of resistance following BRAF and MEK-targeted therapies [197]. Therefore, ERK inhibitors represent an attractive downstream target. SCH772984 is a specific inhibitor of ERK1/2 activity and has demonstrated robust efficacy in BRAF-, KRAS-, and NRAS-mutant cancer cells [197]. BVD-523 (ulixertinib), a reversible ATP-competitive ERK1/2 inhibitor with high potency and selectivity, has exhibited dose-dependent growth inhibition and tumor regression. It also demonstrated anti-tumor activity in cases of acquired resistance to single-agent and combination BRAF/MEK-targeted therapies [198]. Notably, clinical trials evaluating BVD-523 are currently underway (NCT01781429, NCT02296242, and NCT02608229) [198]. The recent discovery of AZD0364, a potent and selective oral inhibitor of ERK1/2, has shown promising antitumor activity in both monotherapy and combination therapy in preclinical models, particularly in NSCLC [199, 200]. Future clinical trials will be necessary to evaluate its clinical activity. It is worth noting that cancer cells exhibit susceptibility to the hyperactivation of ERK pathway activity [201]. Depending on the cell type and stimulus, ERK activity can mediate various antiproliferative processes, including apoptosis, autophagy, and senescence, both in vitro and in vivo [201,202,203]. Gaining insight into these mechanisms is crucial for developing potential therapeutic strategies for cancer. Conversely, the scaffold protein SH3 and multiple ankyrin repeat domain 3 (SHANK3) acts as a RAS interactor, binding to active KRAS, including its mutant forms. SHANK3 competes with RAF, thereby limiting oncogenic KRAS downstream signaling and maintaining MAPK/ERK activity at an optimal level [204]. Recent data highlights SHANK3 depletion surpasses the threshold, leading to MAPK/ERK signaling hyperactivation and MAPK/ERK-dependent cell death in KRAS-mutant cancers [204]. Therefore, inhibiting the SHANK3-KRAS interaction also offers an alternative approach for selectively killing KRAS-mutant cancer cells by inducing excessive signaling.

As previously mentioned and shown in Fig. 4, PI3Ks consist of three classes (I–III). Among these, Class I PI3Ks are heterodimers composed of a catalytic subunit (p110) and a regulatory subunit (p85). Class I PI3Ks can be activated by GTP-bound RAS, subsequently phosphorylating PIP2 to PIP3, which allows the recruitment of AKT and activation of mTOR. Targeting isoform-specific p110 is more specific and provides a better toxicity profile. The first PI3K inhibitor, alpelisib, is a p110α-specific inhibitor and has demonstrated clinical activity in PIK3CA-activating mutant solid tumors [205]. In contrast, since p110δ and p110γ are exclusively expressed in leukocytes, inhibitors targeting these isoforms have gained approval for the treatment of hematological tumors [26]. From 2014 to 2021, four PI3K inhibitors received FDA approval. Idelalisib, the first PI3K inhibitor, was approved for relapsed B-cell malignancies in 2014. This was followed by the approval of copanlisib, a pan-class I inhibitor, in 2017; duvelisib, a dual PI3K p110δ/p110γ inhibitor, in 2018; and umbralisib, a PI3Kδ and casein kinase-1ε inhibitor, in 2021 [206].

Everolimus, an allosteric mTOR inhibitor and a derivative of rapamycin, is orally administered and has been approved by the FDA for the treatment of various solid tumors, including RCC and pancreatic neuroendocrine tumors (NETs) [207].

Thus, inhibitors targeting upstream and downstream mediators in the RAS signaling pathway can be developed in combination with RAS inhibitors to extend durable benefits for patients.

Resistance mechanisms of RAS inhibitors

Primary resistance

Patients with KRAS^G12C-mutant NSCLC have shown varying clinical outcomes with treatment using the KRAS^G12C inhibitor sotorasib. Specifically, among 172 efficacy-evaluable patients, 62 (36%) experienced early progression (PFS < 3 months), while 40 (23%) achieved long-term clinical benefit (PFS > 12 months), according to recently published data from the 2-year analysis of the CodeBreaK 100 trial [208]. As illustrated in Figs. 7A and 84% of patients with KEAP1 mutations exhibited early disease progression. Additionally, these patients were more prone to harboring ROS1 single-nucleotide variants and secondary RAS mutations [208]. Among 56 patients with KRAS^G12C-mutant NSCLC, early progression with sotorasib was observed across PD-L1 expression levels [208]. From Fig. 7A, it is evident that elevated PD-L1 expression correlates with a higher probability of early progression. However, the limited sample size necessitates further large-scale studies to validate these findings. Notably, patients with KEAP1 mutations exhibited a significant enrichment of early progression, independent of their STK11 mutation status. In contrast, other mutations did not show a significant association with early progression when compared to long-term benefits [208]. In a real-world cohort of patients with KRAS^G12C-mutant NSCLC treated with sotorasib, KEAP1 mutations were significantly associated with resistance to therapy [209]. Recent emerging data indicate that patients with NSCLC harboring KRAS^G12C mutations, along with co-occurring genomic alterations in KEAP1, SMARCA4, or CDKN2A, experience inferior clinical outcomes when treated with sotorasib or adagrasib monotherapy [13, 131, 208]. This may be attributed to the fact that patients with KEAP1, SMARCA4, or CDKN2A mutations are more likely to experience early disease progression, with PFS of 3 months or less [13]. Additionally, the ORRs in these patients were lower than those observed in patients with WT KEAP1, SMARCA4, or CDKN2A [13] (Fig. 7B). Although patients with PDAC commonly harbor the KRAS^G12D mutation, the underlying mechanisms of resistance share similarities with those seen in KRAS^G12C-mutant tumors. Recent data from patients with KRAS^G12D treated with MRTX1133 suggest that the loss of tumor suppressor genes such as PTEN, KEAP1, NF1, TP53, and RB1 may confer partial resistance to this drug [210]. However, the associations of STK11 mutations and DNA damage repair (DDR) gene alterations (including BRCA1/2, ATM, ATR, CHEK1/2, PALB2, RAD50/51/51B/51 C/51D) with ORRs were found to be positive [13] (Fig. 7B). There was a significant trend toward higher ORRs associated with mutations in DDR genes; however, no significant differences were observed between the ORRs of STK11 WT and STK11 mutant patients. Another recent report indicates that KRAS^G12C mutations co-occurring with STK11 are associated with lower response rates to IBI351, a potent covalent and irreversible inhibitor of KRAS^G12C [211] (Fig. 7C), inconsistent with the results from the study by Negrao et al.., which evaluated sotorasib and adagrasib monotherapy [13]. However, both reports did not find significant differences in ORRs between STK11 WT and STK11 mutant patients. KRAS^G12C allele-specific inhibitors are the first FDA-approved therapeutics for RAS-mutant tumors. Notably, a recent study reported that RMC-7977, a highly selective inhibitor targeting the active GTP-bound forms of KRAS, HRAS, and NRAS, with affinity for both mutant and WT variants, has demonstrated broad and significant anti-tumor activity in PDAC. However, resistance to RMC-7977 in PDAC has been observed, primarily driven by MYC alterations and the activation of the YAP-TAZ-TEAD pathways [212]. These findings, however, were inconsistent with the results presented in Fig. 7C, which focused on patients with KRAS^G12C-mutant NSCLC [211]. This disparity suggests that the mechanisms of resistance may differ between tumor types, highlighting the importance of understanding context-specific pathways in developing effective therapeutic strategies for KRAS-mutant cancers.

Importantly, STK11, SMARCA4, and KEAP1 not only impact the efficacy of KRAS inhibitors but also contribute to poor responses to ICIs due to a “cold” immune microenvironment, which lacks the necessary immune activation for ICIs to be effective. Skoulidis et al. identified STK11/LKB1 alterations as the most common genomic drivers of primary resistance to PD-1 axis inhibitors in KRAS-mutant LUAD [213]. STK11 and KEAP1 mutations are among the most frequently mutated genes in LUAD. These mutations are associated with lower ORRs to IBI351 in KRAS^G12C-mutant NSCLC and also contribute to resistance to ICIs in patients with KRAS-mutant LUAD [214]. Moreover, Marinelli et al. reported that co-occurring alterations in KEAP1, PBRM1, SMARCA4, and STK11 were associated with reduced efficacy of immunotherapy, even in patients with high tumor mutational burden (TMB), which is generally regarded as a marker predictive of enhanced response to immunotherapy [215, 216]. However, although tumors with KEAP1/TP53 double mutations often exhibit high TMB, they tend to respond less effectively to immunotherapy compared to tumors with only TP53 mutations, despite the elevated TMB [217]. This reinforces the complexity of predicting immunotherapy responses based solely on mutational profiles like tumor mutational burden (TMB). It is possible that the “quality” of coexisting mutations within a tumor (e.g., KEAP1, STK11, TP53) may be more predictive of response than the “quantity” of alterations represented by TMB. However, well-powered prospective studies are needed to ascertain whether prioritizing the specific identity of genomic alterations is more beneficial than merely focusing on the total number of nonsynonymous mutations, regardless of their nature.

Notably, despite the higher ORRs associated with mutations in DDR genes (BRCA1/2, ATM, ATR, CHEK1/2, PALB2, RAD50/51/51B/51 C/51D) as depicted in Fig. 7B, the ORR in patients with BRCA1 mutations was lower than that of BRCA1 WT patients, as shown in Fig. 7C. Furthermore, all genes illustrated in Fig. 7C showed no significant differences in ORRs between WT and mutant groups. These conflicting results warrant further validation in future investigations. Recently published data on the KRAS^G12C inhibitor Divarasib demonstrated that preexisting mutations in RAS genes contribute to primary resistance mechanisms [134]. This finding is consistent with the results observed in studies of sotorasib, as reported by Dy et al. [208]. Although the specific mechanisms of resistance mediated by co-occurring mutations remain unclear and addressing this type of resistance is challenging, these mutations may serve as valuable biomarkers for predicting responses to treatment. Additionally, they could be beneficial for patient stratification and therapy intensification in future randomized clinical trials.

Fig. 7

Primary resistance to KRAS^G12C inhibitors. (A) The primary resistance mechanisms identified include KEAP1 co-mutations, ROS1 single-nucleotide variants, secondary RAS mutations, and a high PD-L1 tumor proportion score (TPS) [208]. (B) Co-alterations in KEAP1, SMARCA4, and CDKN2A have been associated with lower objective response rates, while co-alterations involving STK11 and DNA damage response (DDR) genes have been linked to better objective responses [13]. (C) Co-alterations in TP53, STK11, KEAP1, PIK3CG, POLE, BRCA2, SMAD4, CDKN2A, DNMT3A, MYC, BRCA1, BRINP3, and PTEN have been linked to either improved or reduced objective response rates [211]

Acquired and adaptive resistance and corresponding combination therapy

Recently, multiple mechanisms have been identified that confer resistance to current KRAS inhibitors. A case report revealed that resistance is driven by the enrichment of clonal populations, KRAS-independent downstream signaling, and diverse remodeling of the tumor microenvironment [218]. Notably, Awad et al.. have reported that various genomic and histologic mechanisms also drive resistance to covalent KRAS^G12C inhibitors. For example, genomic mechanisms include acquired KRAS alterations (such as G12D/R/V/W, G13D, Q61H, R68S, H95D/Q/R, Y96C, and KRAS^G12C amplification), acquired bypass mechanisms of resistance (such as MET amplification), and activating mutations in NRAS, BRAF, MAP2K1, EGFR, and RET. Additionally, oncogenic fusions and loss-of-function mutations in tumor suppressor genes, including PTEN, have been implicated in resistance [219] (Fig. 8A). As illustrated in Fig. 8A-B, the most common form of acquired resistance involves upstream RTKs, including MET, HER2, RET, ALK, and EGFR, through amplifications, fusions, or mutations. This is followed by mutations in KRAS alleles (such as G12S, G13D, and Q61H) occurring in either the cis or trans configuration [219,220,221,222]. Although it has been reported that CRC cells exhibit higher basal RTK activation compared to NSCLC cells, and that EGFR signaling plays a key role in mediating resistance to KRAS^G12C inhibitors [223], this understanding provides a crucial mechanistic foundation for developing EGFR antibodies in combination with KRAS^G12C inhibitors. Such combinations could potentially overcome resistance to these inhibitors [220, 224]. However, Fig. 8C and D also demonstrate that acquired resistance primarily arises from alterations in upstream RTKs, with subsequent mutations in KRAS alleles (such as G12S, G13D, and Q61H) occurring in either cis or trans configurations [220, 222]. Similarly, in July 2024, Dilly et al. demonstrated that mutations in PIK3CA and KRAS, as well as amplifications of KRAS^G12C, MYC, MET, EGFR, and CDK6, emerged as mechanisms of acquired resistance to adagrasib or sotorasib in patients with KRAS^G12C-mutant PDAC [225]. Additionally, the study found that amplifications of KRAS, YAP1, MYC, and Cdk6/Abcb1a/b were associated with resistance to MRTX1133 [225]. These findings underscore the complexity of resistance mechanisms in KRAS-mutant tumors and the need for strategies that can address these diverse alterations to enhance treatment efficacy.

Fig. 8

Acquired mechanisms of resistance to KRAS^G12C inhibitors. (A-D) Frequency of acquired resistance mutations in MAPK pathway genes, as documented in references [219,220,221,222]

Additionally, researchers have identified adaptive feedback reactivation of WT RAS-MAPK signaling as a key mechanism of adaptive resistance to KRAS inhibitors (Fig. 9). This highlights the potential importance of vertical combination strategies to effectively overcome resistance [226, 227]. In the presence of mutant KRAS, feedback inhibition typically limits the activity of upstream RTKs and WT RAS isoforms. When treating the KRAS ‘off’ state, suppression of the MAPK pathway leads to the loss of this feedback inhibition, resulting in the upregulation of RTKs and a shift of RAS into an ‘on’ state, mediated by SOS and SHP2, which activates WT RAS isoforms. This rebound signaling can significantly limit the effectiveness of drug treatment (Fig. 9). For instance, recent published data have shown that co-targeting SOS1 enhances the antitumor effects of adagrasib by overcoming both intrinsic and acquired resistance [162]. The HER family inhibitors afatinib and cetuximab, along with the PI3Kα inhibitor BYL-719, demonstrated a combinatorial effect with MRTX1133 [210]. In contrast, inhibitors targeting SHP2, SOS1, mTOR, and CDK4/6 did not show this synergistic effect [210]. This suggests that specific pathways may interact more effectively with KRAS^G12D-targeted therapies, highlighting potential combination strategies for enhanced therapeutic efficacy. These findings are consistent with published reports suggesting that the inhibition of mutant KRAS can lead to feedback compensation, promoting the expression of ERBB receptors. This mechanism may contribute to acquired resistance to MRTX1133 treatment, highlighting the complexity of resistance pathways that can arise in response to targeted therapies [228]. In July 2024, investigators reported the mechanistically critical role of ERK in resistance to KRAS-ERK MAPK targeted therapies. This study highlighted how ERK signaling can mediate resistance mechanisms, suggesting that despite targeting KRAS directly, the downstream effects on ERK may allow cancer cells to adapt and survive treatment [229]. Understanding this relationship could lead to more effective combination therapies that simultaneously inhibit ERK alongside KRAS to overcome resistance and improve patient outcomes. Of note, the previously mentioned RMC-7977 could provide significant therapeutic benefits by not only targeting mutant RAS-driven signaling but also inhibiting the activity of WT isoforms that may contribute to tumor growth and adaptive feedback resistance mechanisms [212]. The efficacy of this inhibitor highlights its potential for offering more effective treatment options in RAS-mutant cancers.

Fig. 9

Adaptive mechanisms of resistance to KRAS inhibitors. Adaptive mechanisms of resistance. This figure illustrates the various ways in which cancer cells develop resistance to treatment, highlighting both specific mutations and broader adaptive responses

In addition to genetic alterations and adaptive mechanisms of resistance to KRAS inhibitors, histological mechanisms play a significant role in the process of histologic transformation in cancers, including those driven by KRAS mutations [219]. Histologic transformation refers to the change in the type of cancer cells, which can lead to more aggressive tumor behavior and resistance to therapies. Published data in 2024 indicate that adeno-to-squamous transition, the presence of mucinous histological features, and focal adhesion kinase (FAK)-YAP signaling play significant roles in driving resistance to KRAS inhibition [230,231,232]. Therefore, combination strategies that involve RAS inhibitors with MYC inhibition or targeting the Hippo signaling pathways in PDAC may soon be translated into clinical practice. These approaches aim to address the various resistance mechanisms that co-evolve alongside RAS inhibition. To overcome resistance, combination therapy strategies target both acquired genetic mutations and adaptive resistance to KRAS inhibitors. These strategies address upstream RTKs, secondary RAS mutations, WT RAS isoforms, and downstream effector pathways such as RAF–MEK and PI3K–AKT. By simultaneously inhibiting these various components, combination therapies aim to improve treatment efficacy and counteract resistance mechanisms. These data collectively support the advancement of multiple combination therapy strategies to effectively address the challenges posed by resistance to RAS inhibition. By simultaneously targeting RAS and related pathways, such as MYC and Hippo signaling, clinicians may enhance therapeutic effectiveness and improve outcomes for patients with KRAS-driven tumors. This comprehensive approach reflects a growing understanding of the intricate relationships between genetic alterations, histological changes, and signaling pathways in cancer biology. Given the multifactorial nature of this resistance, there is a clear need to explore more cost-effective combination therapies and alternative treatment strategies.

Importantly, the insights gained from these studies provide critical mechanistic evidence supporting the principles of personalized medicine. By identifying specific resistance mechanisms within individual tumors, clinicians can develop tailored treatment strategies that directly target these pathways, optimizing therapeutic outcomes for patients. This approach encourages the creation of rational, mechanism-driven combination therapies aimed at overcoming resistance and improving the efficacy of KRAS-targeted treatments. Ultimately, integrating these findings into clinical practice could significantly enhance outcomes for patients with KRAS-driven cancers, particularly in difficult cases such as PDAC. Emphasizing personalized and adaptive treatment strategies will be essential in addressing the evolving nature of cancer resistance and achieving greater therapeutic success.

Finally, it is important to note that the therapeutic vulnerabilities in KRAS-mutated and KRAS WT cancers differ significantly. For KRAS-mutant cancers, specific therapies targeting the KRAS^G12C mutation, such as Sotorasib and Adagrasib, have exploited the unique properties of the mutant protein. As stated above, inhibiting pathways that are essential for the survival of KRAS-mutant tumors, such as the MAPK/ERK and PI3K/AKT pathways, can lead to tumor cell death. Importantly, recent study suggests that argininosuccinate synthase 1 (ASS1) deficiency, driven by mutant KRAS, promotes DNA synthesis and creates a reliance on SLC7A1, highlighting dietary arginine restriction and SLC7A1 inhibition as promising therapeutic approaches in KRAS-mutant NSCLC [233]. Likewise, A recent report identified the EPHA2-PARD3 axis as a vulnerability in KRAS-mutant CRC [234]. Novel clinical approaches are leveraging the fact that mutant KRAS peptides are naturally processed and presented in tumors by the major histocompatibility complex (MHC) [235]. While cancers with WT KRAS may rely on other pathways, such as EGFR or HER2, making them susceptible to targeted therapies against these receptors. Besides, utilizing combinations of chemotherapy, targeted agents, and immunotherapies can exploit the unique signaling pathways activated in WT cancers. A single-institution cohort of 795 cases of exocrine pancreatic cancer revealed that 43.8% of KRAS WT cases exhibited evidence of an alternative driver of the MAPK pathway, including BRAF mutations, in-frame deletions, and RTK fusions. In contrast, 56.2% of cases did not show a clear MAPK driver alteration; however, 29.3% of these MAPK-negative KRAS WT cases displayed activating alterations in other oncogenic drivers, such as GNAS, MYC, PIK3CA, and CTNNB1. Additionally, the study demonstrated the potent efficacy of pan-RAF and MEK inhibition in patient-derived organoid models with BRAF in-frame deletions [236]. This indicates that identifying additional genetic alterations in WT KRAS tumors can uncover specific vulnerabilities, facilitating personalized treatment strategies. Moreover, WT KRAS cancers may be more sensitive to therapies that target the tumor microenvironment, including anti-angiogenic agents.

Understanding the therapeutic vulnerabilities of both KRAS-mutated and WT cancers is crucial for developing effective treatment strategies. Tailoring therapies to the unique characteristics of each tumor type can improve patient outcomes and enhance the efficacy of cancer treatments.

Conclusion

The RAS signaling pathway plays a pivotal role in carcinogenesis, underscoring the critical need for a comprehensive understanding of RAS biology to develop innovative therapeutic strategies. Extensive research efforts have been directed towards RAS inhibitors in both clinical and preclinical settings. Among these, the FDA-approved allele-specific KRAS^G12C inhibitors have notably transformed the treatment paradigm for RAS-driven tumors [16, 130,131,132, 237]. Despite these exciting advancements, the low clinical response rate to a monotherapy approach across all RAS-mutant cancers remains a significant challenge, compounded by the emergence of resistance mechanisms. Consequently, a diverse array of strategies targeting specific RAS-mutant subsets is under development. Understanding and addressing resistance mechanisms have become pivotal topics in current research. One promising direction involves the combination of different inhibitors to overcome resistance, offering a more robust and sustained therapeutic response [238]. Despite the notable advancements in combination therapies, the optimal therapeutic approach has yet to be definitively identified. Nonetheless, large-scale clinical trials of RAS-targeted therapies are showing promising efficacy in several highly refractory malignancies, including NSCLC, CRC, and PDAC. As we move forward, it is crucial to recognize both the potential opportunities and the challenges that lie ahead in refining and optimizing these therapies for broader clinical use.