The Kirsten rat sarcoma viral oncogene (KRAS) encodes the protein KRAS and represents the most altered protein across solid tumors underscoring the need for successful targeting (1). KRAS which is anchored as a membrane-bound protein in all human cells, acts as a central node for multiple signaling pathways important in normal cellular function (2). KRAS cycles between its inactive and active forms, facilitated by the exchange of KRAS-bound guanine diphosphate (GDP) and guanine triphosphate (GTP). Guanine exchange factors (GEFs) and GTPase activating proteins (GAPs) orchestrate this exchange and are therefore critical in maintaining KRAS activity (Figure 1). This well-controlled cycle was long considered ‘undruggable’ due to two main factors: the high affinity of KRAS for GTP and the absence of appropriate binding pockets suitable for the development of small molecule inhibitors (3, 4). The last 10 years have heralded a new era in personalized medicine with the discovery of the covalent KRAS^G12C inhibitors and a plethora of new investigational KRAS targeting agents are now entering the clinic. Innate and acquired resistance mechanisms to KRAS inhibitors have already been documented and our understanding of the pleiotropic effects of oncogenic KRAS on the tumor microenvironment has evolved. This has opened up new avenues in the treatment of KRAS mutated tumors with combination-based approaches in addition to emerging vaccines and adoptive cell therapy (ACT) strategies. Although oncology has seen the evolution of agnostic biomarkers to facilitate precise approaches, it remains to be seen whether mutations in the KRAS gene will represent universal biomarkers or whether inhibitor effects will be cell lineage dependent. In particular, pancreatic cancer has been a global challenge from a therapeutic perspective. Herein, we review approaches to KRAS targeting in pancreatic cancer to date and suggest some challenges for the future.

KRAS is located on the short arm of chromosome 12 and is a member of the RAS oncogene family along with 2 other isoforms, neuroblastoma rat sarcoma viral oncogene (NRAS) and Harvey rat sarcoma viral oncogene (HRAS) (chromosomes 1 and 11 respectively) (5). KRAS is composed of 2 major domains, a catalytic domain called the G domain, and a hypervariable region at the C-terminal (6). The G domain consists of 3 regions: switch I, switch II and the P loop. Notably, as KRAS cycles between active and inactive states, the switch I and II pockets alter their conformation, providing a therapeutic vulnerability.

In the resting state, KRAS binds to GDP, which does not activate downstream signaling and is therefore considered “OFF.” GEFs limit KRAS-GDP affinity and act to replace GDP for GTP when activated by extracellular signals in the form of growth factors, e.g., epidermal growth factor (EGF), cytokines, and other molecules (Figure 1). Son of sevenless 1 and 2 (SOS1/SOS2), growth factor receptor-bound protein (GRB2) and RAS protein-specific guanine nucleotide-releasing factor 2 (RASGRF2) are common GEFs that enable the exchange (7). Src homology-2 domain containing protein tyrosine phosphatase (SHP2), which is considered a scaffolding protein, facilitates the GRB2/SOS1 complex at the cell membrane through direct binding to GRB2, promoting KRAS activation (8).

When KRAS is bound to GTP, it is considered “ON,” activating downstream signaling cascades. To convert back to the “OFF” state, KRAS-GTP must undergo hydrolysis. Intrinsic hydrolysis of GTPases is relatively ineffective and requires the assistance of GTPase-activating proteins (GAPs) such as neurofibrin-1 (NF1) or p120-RasGAP protein (encoded by RASA1) (9).

When GTP-bound, subsequent interaction with RAS-binding domains (RBDs) of effector proteins permits downstream signaling (10). The three major RAS effectors include rapidly accelerated fibrosarcoma (RAF) proteins, ral guanine nucleotide dissociation stimulator (RALGDS) and the phosphoinositide 3-kinases (PI3Ks) (Figure 1). Perhaps the most notable interaction is the activation of the mitogen activated protein kinase pathway (MAPK) RAF–MEK–ERK pathway. Activated KRAS ignites phosphorylation of RAF and subsequently extracellular signal regulated kinase (ERK) 1 and 2 following RAF dimerization (11). Phosphorylation of ERK results in ERK translocation to the nucleus where further transcriptional factors are phosphorylated and cell cycle progression through G0/G1 mitogenic signals ensues (12). Notably, the degree of activation of RAF kinases is likely variable which may influence prognostic outcomes of different mutated alleles (13).

Cellular growth and proliferation also result from signaling through the PI3K/protein kinase B(AKT) pathway. Activated PI3K phosphorylates phosphatidylinositol(PIP)2 to PIP3 promoting AKT phosphorylation which in turn stimulates many downstream pathways and regulates cell proliferation, cell death and metabolic processes, including cellular responses to insulin (14). Interactions with mTOR target proteins via feedback phosphorylation promote cell proliferation but also activate Bcl-XL/Bcl-2 to facilitate apoptosis (15, 16). The third major effector of RAS is the Ral guanine nucleotide exchange factor pathway. Ral proteins (RalA and RalB) are GTPases also cycling between GDP and GTP bound states. This effector of RAS complements RAS/RAF and RAS/PI3K signaling and is known to stimulate the Jun-N terminal kinase pathway promoting cell migration and proliferation (17).

There are two predominant precursor lesions to invasive PDAC, namely pancreatic intraepithelial neoplasia (PanIN) (85–90%) and intraductal papillary mucinous neoplasms (IPMNs) (10–15%). KRAS mutations are evident even in low-grade PanIN highlighting its principal role early in oncogenesis but also underscoring the need for the acquisition of co-mutations for invasive disease to develop. Indeed a stepwise model for progression has long been accepted (35, 36), although alternate models propose large-scale chromothriptic events which permit rapid disease progression, bypassing traditional models (37). Increasing knowledge of the malignant potential of cystic lesions, which are readily apparent on imaging, has revealed recurrent genetic drivers in KRAS, GNAS and RNF43, with additional alterations seen in invasive disease (38–40). With the successive inactivation of tumor suppressors in PDAC, cell cycle progression increases, as measured by RNA sequencing or Ki-67 (41). Transcriptional subtypes from primaries and metastases document major classifiers as basal-like and classical, with hybrid phenotypes evident which suggests tumoral plasticity. Although the incidence of the basal-like subtype is more apparent in metastases and associates with major imbalances in mutant KRAS, it is not yet clear what drives the progression to this phenotype. The influence of the tumor microenvironment (TME) and epigenetic factors are likely to be important. A full understanding of the progression of precursor lesions to invasive disease will underpin where KRAS can be targeted and what combination strategies may be needed.

PDAC harbors a unique tumoral niche characterized by a dense fibrotic stromal reaction and an immunosuppressive TME. Infiltration of immunosuppressive cells, encouraging immune evasion, is thought to occur early, highlighting the cell-extrinsic impacts of KRAS mutations (41, 42). In fact, it has been well documented in the genetically engineered KPC (Kras ^{LSL-G12D/+Trp53LSL-R172H/+} Pdx-Cre) and KC mouse (Kras^LSL-G12D/+ Pdx-Cre) models, that KRAS mutations in PanIN lesions associate with inflammatory environments with overexpression of COX2 (42), and early infiltration of T regulatory cells (Tregs), Tumor associated macrophages (TAMs) and Myeloid derived suppressor cells (MDSCs). This indicates a shift toward early immune evasion (43) and KRAS mutations induce multiple cytokines to promote and maintain this tumor promoting environment in invasive disease. RAS mutations induce the secretion of IL-6 in different cell types including myeloid cells, resulting in JAK1 activation and phosphorylation of the STAT3 pathway, culminating in a pro-inflammatory TME and shifting homeostasis toward tumorigenesis (44–46). Additionally, granulocyte macrophage colony stimulating factor (GM-CSF) secretion in response to mutated KRAS, can occur early and promotes MDSC infiltration, restricting anti-tumor immunity (47, 48). Moreover, continued secretion by mesenchymal stem cells within the stroma encourages tumor invasion and metastasis formation (49, 50) (Figure 1). Pivotal studies have demonstrated the upregulation of chemokines including CXCL-8 (IL-8) (51) and CXCR2 (52, 53) in response to RAS signaling, together with transcription factors such as NF-κB, promoting inflammation (54). Immune escape is further reinforced through the upregulation of PD-L1 and the conversion of CD4+ cells to Tregs (55). The latter is enhanced following ERK activation and secretion of IL-10 and TGF-B1, illustrating the importance of downstream pathways in shaping the TME (55). Nevertheless, the presence of a distinct chemokine signature (CCL4, CCL5, CXCL9, CXCL-10) can associate with T cell infiltration (56), and represents a biomarker of response to immune checkpoint inhibitor combinations in certain PDAC genotypes (57). This highlights the potential importance of profiling the TME for combination approaches.

The characteristic desmoplasia of PDAC is inherently reliant on cancer-associated fibroblasts (CAFs) with varying CAF subtypes identified (58). It is now known that targeting stroma may accelerate PDAC progression. The origin of CAFs remains unclear with a minority produced from the pancreatic stellate cell (59). Immunosuppressive and immune-enhancing CAFs exist (60), however oncogenic KRAS promotes non-cell autonomous programming of CAFs to stimulate cytokines, which regulate MDSCs and other immune cells (61). Understanding the pleiotropic effects of KRAS mutations in molding the TME early in oncogenesis will be critical in trial design of combination approaches in PDAC. Notably, the unselected use of immune checkpoint inhibitors has been unsuccessful to date (62).

Cancer cell metabolism is critical to survival and in PDAC, KRAS mutations shape a unique metabolic environment which drives aerobic glycolysis (63). The adaptive switch to produce glucose, glutamine and fatty acids facilitates a supply for growth and proliferation. Notably, RAS-dependent metabolic adaption is likely tumor-specific, owing to the prevalence of specific KRAS mutations (64, 65). Mutations in KRAS are known to upregulate the GLUT1 transporter together with induction of enzymatic activity to increase glycolysis (66). Glutaminolysis also increases in the presence of oncogenic RAS signaling, providing glutamate from glutamine, fuelling tricarboxylic acid cycle (TCA) and providing a pool of amino acids (67, 68). Nutrient scavenging pathways such as autophagy are known to be upregulated in PDAC (69), and continue to feed the TCA cycle sustaining growth (70). Targeting autophagy has been a huge therapeutic strategy in the last number of years with multiple trials ongoing with hydroxychloroquine. Studies have shown that targeting RAS effector pathways such as ERK signaling may increase reliance on autophagy, necessitating combination approaches (71). Macropinocytosis is an additional metabolic process where cells can engulf extracellular material and is thought essential for PDAC survival (72, 73). The transcription factor MYC is important in this pathway, and notably, recent work has established mechanistic differences in macropinocytosis according to the mutant allele. KRAS^G12R exhibits impaired activation of the effector p110alpha PI3K, exposing potential therapeutic vulnerabilities (64). The crosstalk between KRAS, the TME and metabolic pathways will be critical in advancing combination therapeutic strategies, especially given the recent disappointments in targeting metabolism in the AVENGER trial (74).

Peptide vaccines are increasingly being developed to elicit an immune response in PDAC and may provide greatest benefit in the adjuvant setting. ELI-002 consists of amphiphile-modified KRAS mutant peptides (G12D and G12R) together with amphiphile-modified Toll-like receptor (TLR) 9 agonistic CPG-7909 DNA. The novel approach through its binding to endogenous albumin encourages transit from subcutaneous tissues into lymph glands rather than the bloodstream. In AMPLIFY-201, patients were treated who had ctDNA detected or elevation in serum biomarkers following completion of surgery/adjuvant therapy. Twenty-five patients were included and 84% demonstrated mutated KRAS-specific T cell responses. Reduction in ctDNA was seen in 77% with 33% showing clearance of ctDNA (138). AMPLIFY-7P will further investigate a related compound, ELI-002 7P in a phase I/II study.

TG01 is a vaccine containing seven synthetic RAS peptides that target KRAS. In phase I/II trial, 32 patients with resected stage I or II PDAC were treated with TG01/GM-CSF in combination with adjuvant gemcitabine. Over 90% developed an immune response as defined by a delayed-type hypersensitivity response and/or a positive T cell proliferation assay. Median OS was 33.1 months (95% CI 16.8, 45.8) (139). A phase II trial is underway (NCT 05638698). Another long peptide pooled mutant-KRAS vaccine is being investigated in a phase I trial in combination with nivolumab and ipilimumab in patients with resected MMR-proficient colorectal cancer and pancreatic cancer (NCT04117087).

The KISIMA vaccine platform includes 3 components: a cell-penetrating peptide for intracellular antigen delivery, a multi-antigenic cargo tailored to the tumor type and a TLR peptide agonist to enhance the immune response. A phase I trial is underway in which patients are treated with a ‘prime-boost’ vaccine, consisting of the protein vaccine described (ATP150 or ATP152) in combination with a viral vector VSV-GP154 and an immune checkpoint inhibitor (KISIMA-02). Initial testing will be done in locally advanced/metastatic KRAS^G12D or KRAS^G12V mutated PDAC, progressing to randomized parallel cohorts of resected PDAC and a control arm (NCT05846516). DC3/8 is a dendritic cell (DC) vaccine loaded with KRAS mutated peptide which is currently being investigated in resected PDAC patients. An accrual target of 29 patients will receive autologous DCs with mutant KRAS peptides according to the patient’s tumor mutation and HLA subtype (NCT03592888). The mRNA vaccine mRNA-5671/V941 is enclosed in a lipid nanoparticle which targets KRAS G12D, G12V, G13D, and G12C. The mRNA is absorbed by APCs and translated into peptides to be presented. This has elicited strong T-cell responses in mouse models (140, 141). A phase I in-human clinical trial of V941 alone or with pembrolizumab has completed accrual, and results are awaited (NCT03948763).

Although KRAS mutations are prevalent across tumor types, it’s clear that there is heterogeneity with regard to responses. Understanding both innate and acquired resistance mechanisms to KRAS inhibitors is key to identifying optimal combination strategies and overcoming the modest duration of response we have seen with existing direct KRAS inhibitors. Adaptation to G12C inhibitors has been described as happening in a G12C-dependent or independent manner and can be rapid (142). Dependant adaptation occurs post treatment with G12C inhibitors, when some cancer cells are sequestered in an inactive state and newly formed G12C cancer cells are amplified and resume proliferation (143). Independent adaptation happens through a variety of mechanisms (89, 143). KRAS cells may acquire additional KRAS mutations on codons 12, 13, and 61. They may also acquire mutations on the switch II binding pocket, e.g., KRAS R68S, H95D/Q/R, Y96C/D (89, 143). These mutations may lead to noncovalent binding inhibition at the pocket site. Interestingly, some pocket site mutations which appeared after adagrasib treatment, conferred marked resistance to adagrasib however remained sensitive to sotorasib (143). Therefore, resistance to some inactive-state inhibitors can be overcome by a functionally distinct KRAS^G12C inhibitor (89). In addition, the tricomplex inhibitors may overcome resistance from mutations in the switch II pocket.

Secondary bypass alterations in the RAS/MAPK/PI3K signaling pathway in patients who have progressed on sotorasib/adagrasib have also been documented (143, 144). These included treatment-emergent amplifications, mutations or fusions of genes encoding receptor tyrosine kinases MET, EGFR, RET, ALK, and FGFR3 leading to investigation of combination strategies. However, the specific receptor tyrosine kinase driving the rebound in signaling can vary widely among tumors suggesting it might be advantageous to look at downstream blockade. Other bypass mechanisms include mutations or fusions of downstream effector kinases: BRAF, CRAF, and MEK1; loss of function mutations in NF1 and PTEN; NRAS and PI3K mutations. Cell state transitions are an additional mechanism of resistance and in NSCLC cell lines epithelial to mesenchymal transition was responsible for both intrinsic and acquired resistance (145). The contribution of the TME and stromal compartment in resistance is yet to be established.

The diversity of resistance mechanisms that have emerged in response to KRAS G12C inhibition highlights the challenges in targeting RAS. As data on pan-RAS inhibitors emerge, these will no doubt expand, underscoring the need for rapid adaptive combination approaches in the trial setting.

Targeting the RAS pathway remains the biggest challenge in precision oncology. The future in PDAC will likely see selection of first line combinations based on KRAS allelic status.

Vaccine strategies herald huge promise, especially in early-stage disease. The modest duration of response and emergence of resistance mechanisms to the KRAS^G12C inhibitors points to challenges that we are likely to see across the direct KRAS inhibitors. Overcoming resistance and defining optimal combination strategies is likely to be crucial to improving patient outcomes. This requires an in-depth knowledge of not just the mutational profile but of metabolic needs and distinct TMEs of each patient. The plethora of new agents under investigation has the potential to revolutionize the treatment paradigm for patients with KRAS-mutated pancreatic cancer.

AL: Writing – review & editing, Writing – original draft. MO’R: Writing – review & editing, Writing – original draft. RM: Writing – review & editing, Writing – original draft. GO’K: Writing – review & editing, Writing – original draft.