725 Background: Pancreatic ductal adenocarcinoma (PDAC) is largely driven by oncogenic alterations in the KRAS , TP53 , CDKN2A/B , and SMAD4 genes. KRAS G12R accounts for nearly 20% of KRAS alterations in PDAC, is biologically unique, and offers the potential for improved response to MEK inhibitor (MEK-inh)-based therapy. As part of the MCW-Master-PREDICT (NCT05802069) observational study, we evaluated the efficacy of MEK-inh-based therapy matched to the unique molecular alterations in the tumor of each patient as recommended by the molecular tumor board. Methods: From March 2022 to December 2023, 8 patients with metastatic PDAC were treated with MEK-inh-based therapy. Molecular profiling was performed by Tempus (648 genes) in 7 patients, and FoundationOne (324 genes) in 1 patient. Description of molecular alterations, matched therapies, and efficacy of treatment are provided. Results: The median age of patients starting MEK inhibitor-based therapy was 69 years, with 62% of them being female. The molecular characteristics and matched therapies are summarized (Table). MEK-inh-based therapy was administered to 3 (37.5%) patients as a first or second-line treatment, and to 5 (62.5%) patients as a third to fifth-line treatment for metastatic PDAC. The overall survival and progression free survival (PFS) on MEK-inh-based therapy was 8.4 and 4.9 months, respectively. Four patients had PFS beyond 4 months (4.9, 5.5, 5.6 and 6.9 months). Conclusions: Individualized MEK-inh-based therapy demonstrated efficacy in late-line treatment of metastatic PDAC with KRAS G12R alterations. Using this approach earlier in the treatment course and in combination with RAS inhibitors or cytotoxic systemic therapies may further enhance outcomes. Tumor alterations and matched therapies. Gene altered Number of patients (%) Number of patients matched (%) Drug family matched KRAS G12R 8 (100) 8/8 (100) MEK inhibitor TP53 7 (87.5) 4/7 (57) VEGF/VEGR inhibitor* CDKN2A 4 (50) 4/4 (100) CDK4/6 inhibitor** SMAD4 2 (25) 2/2 (100) MEK inhibitor + (EGFR or Pan-HER inhibitor)*** *PMID: 27466356; **PMID: 33472910; ***PMID: 36127339, PMID: 24625091.

726 Background: Up to 95% of pancreatic ductal adenocarcinomas (PDAC) are driven by oncogenic alterations in the KRAS gene, with KRAS G12D (40%) and KRAS G12V (30%) mutations being the most common. However, in most cases, KRAS alterations are accompanied by alterations in other tumor suppressors such as TP53 , CDKN2A/B , and SMAD4 . Therefore, targeting KRAS alone may not be sufficient for effective treatment of PDAC . As part of the MCW-Master-PREDICT (NCT05802069) observational study, we evaluated the effectiveness of MEK inhibitor (MEK-inh)-based therapy, tailored to the unique molecular alterations of each patient’s tumor, in treating metastatic PDAC (mPDAC) with KRAS G12D or KRAS G12V alterations. Methods: From April 2022 to December 2023, 19 patients ( KRAS G12D=10 (53%), KRAS G12V=9 (47%)) were treated with MEK-inh-based therapy. Comprehensive genomic profiling was performed using Tempus xT (648 genes) in 15 patients (79%), FoundationOne (324 genes) in 4 patients (21%), and STRATA (429 genes) in 1 patient. The genomic characteristics, matched therapies, and treatment outcomes are provided (Table). Results: Of the 19 patients, 11 (58%) were female. The median age at the initiation of MEK-inh-based therapy was 67 years. Six (32%) patients received MEK-inh-based therapy as first or second-line treatment, while 13 (68%) patients were treated in the third to fifth-line. Among the 19 patients, the median overall survival (mOS) and median progression-free survival (mPFS) from the initiation of MEK-inh-based therapy were 5 and 2 months, respectively. The mOS from the time of MEK-inh-based therapy was 5.1 months for KRAS G12D, and 4.7 months for KRAS G12V (P=0.87). The mPFS from the time of MEK-inh-based therapy was 2.3 months for KRAS G12D and 1.4 months for KRAS G12V (P=0.12). Among all 19 patients, 4 (21%) patients had mPFS above 4 months ( KRAS G12V=1, KRAS G12D=3; 4.1, 4.3, 4.7, 9.9 months). Conclusions: The efficacy of MEK-inh-based therapy was limited in the late-line treatment of mPDAC for patients with KRAS G12D and KRAS G12V mutations. The potential role of MEK inhibitors in earlier lines of therapy, and their combination with KRAS inhibitors, requires further prospective evaluation. Tumor alterations and matched therapies. Altered gene KRAS G12V(N=9) KRAS G12D(N=10) Patients matched(N=19) Type of matched therapy KRAS G12D/V 9 (100%) 10 (100%) 19 (100%) MEK inhibitor TP53 7 (78%) 8 (80%) 8 (42%) VEGF/VEGR inhibitor* CDKN2A/B 8 (89%) 5 (50%) 9 (47%) CDK4/6 inhibitor** SMAD4 3 (33%) 4 (40%) 4 (21%) MEK inhibitor + (EGFR or pan-HER inhibitor) *** *PMID: 27466356; **PMID: 33472910; ***PMID: 36127339, PMID: 24625091. N= number of patients.

Background: KRAS wild-type (WT) pancreatic ductal adenocarcinoma (PDAC) represents a distinct entity with unique biology. The therapeutic impact of matched targeted therapy in these patients in a real-world setting, to date, is less established. Objectives: The aim of our study was to review our institutional database to identify the prevalence of actionable genomic alterations in patients with KRAS -WT tumors and to evaluate the therapeutic impact of matched targeted therapy in these patients. Design: We reviewed electronic medical records of patients with KRAS-WT PDAC and advanced disease ( n = 14) who underwent clinical-grade tissue ± liquid next-generation sequencing (315–648 genes for tissue) between years 2015 and 2021. Methods: Demographic and disease characteristics were summarized using descriptive parameters. Progression-free survival (PFS) and overall survival (OS) were estimated using the Kaplan–Meier method. Results: Of 236 PDAC patients, 14 had advanced/metastatic disease with KRAS-WT tumors. Median age at diagnosis was 66 years. There was a high frequency of potentially actionable genomic alterations, including three (21%) with BRAF alterations, two (14%) with fusions [ RET-PCM1 and FGFR2-POC1B ( N = 1 each)]; and one with a druggable EGFR ( EGFR E746_A755delISERD) variant; two other patients had an STK11 and a MUTYH alteration. Five patients were treated with matched targeted therapy, with three having durable benefit: (i) erlotinib for EGFR -altered tumor, followed by osimertinib/capmatinib when MET amplification emerged (first-line therapy); (ii) pralsetinib for RET fusion (fifth line); and (iii) dabrafenib/trametinib for BRAF N486_P490del (third line). Duration of time on chemotherapy-free matched targeted therapy for these patients was 17+, 11, and 18+ months, respectively. Conclusion: Sustained therapeutic benefit can be achieved in a real-world setting in a subset of patients with advanced/metastatic KRAS-WT PDAC treated with chemotherapy-free matched targeted agents. Prospective studies are warranted. Keywords case series , KRAS protein , molecular targeted therapy , oncogene fusion , pancreatic neoplasms

Background: KRAS wild-type (WT) pancreatic ductal adenocarcinoma (PDAC) represents a distinct entity with unique biology. The therapeutic impact of matched targeted therapy in these patients in a real-world setting, to date, is less established. Objectives: The aim of our study was to review our institutional database to identify the prevalence of actionable genomic alterations in patients with KRAS -WT tumors and to evaluate the therapeutic impact of matched targeted therapy in these patients. Design: We reviewed electronic medical records of patients with KRAS-WT PDAC and advanced disease ( n = 14) who underwent clinical-grade tissue ± liquid next-generation sequencing (315–648 genes for tissue) between years 2015 and 2021. Methods: Demographic and disease characteristics were summarized using descriptive parameters. Progression-free survival (PFS) and overall survival (OS) were estimated using the Kaplan–Meier method. Results: Of 236 PDAC patients, 14 had advanced/metastatic disease with KRAS-WT tumors. Median age at diagnosis was 66 years. There was a high frequency of potentially actionable genomic alterations, including three (21%) with BRAF alterations, two (14%) with fusions [ RET-PCM1 and FGFR2-POC1B ( N = 1 each)]; and one with a druggable EGFR ( EGFR E746_A755delISERD) variant; two other patients had an STK11 and a MUTYH alteration. Five patients were treated with matched targeted therapy, with three having durable benefit: (i) erlotinib for EGFR -altered tumor, followed by osimertinib/capmatinib when MET amplification emerged (first-line therapy); (ii) pralsetinib for RET fusion (fifth line); and (iii) dabrafenib/trametinib for BRAF N486_P490del (third line). Duration of time on chemotherapy-free matched targeted therapy for these patients was 17+, 11, and 18+ months, respectively. Conclusion: Sustained therapeutic benefit can be achieved in a real-world setting in a subset of patients with advanced/metastatic KRAS-WT PDAC treated with chemotherapy-free matched targeted agents. Prospective studies are warranted.

BACKGROUND & AIMS: The complex tumor microenvironment (TME) of pancreatic ductal adenocarcinoma (PDAC) has hindered the development of reliable predictive biomarkers for targeted therapy and immunomodulatory strategies. A comprehensive characterization of the TME is necessary to advance precision therapeutics in PDAC. METHODS: A transcriptomic profiling platform for TME classification based on functional gene signatures was applied to 14 publicly available PDAC datasets (n = 1657) and validated in a clinically annotated independent cohort of patients with PDAC (n = 79). Four distinct subtypes were identified using unsupervised clustering and assessed to evaluate predictive and prognostic utility. RESULTS: TME fibrotic (IE/F); fibrotic (F); and immune depleted (D). The IE

754 Background: ATM and CHEK2 mutations are linked to homologous recombination DNA repair deficiency, with the potential for improved therapeutic response to DNA damaging agents. To investigate our clinical observation of improved outcomes with irinotecan (IRI) based chemotherapy in advanced/metastatic pancreatic ductal adenocarcinoma (PDAC) patients (pts) with somatic or germline ATM/ CHEK2 mutations, we examined our institutional real world experience. Methods: Between 2015-2021, 33 pts with ATM or CHEK2 mutations treated with chemotherapy were identified, of which 16 pts had advanced/metastatic disease. Progression-free survival (PFS) was calculated and compared (Kaplan Meier, log-rank test) in several ways to assess the impact of IRI vs platinum or other regimens. The event for PFS was progression (or death), and pts without progression were censored at treatment end, or at last follow up if treatment was ongoing. Results: Among 16 pts with advanced/metastatic PDAC, 8 (50%) had ATM (5 germline, 3 somatic) and 8 (50%), CHEK2 alterations (4 germline, 4 somatic). Overall, pts received 48 lines of chemotherapy (platinum-based-(No-IRI) N=14; IRI-based-(No-Platinum) N=8; both N=4; none N=22). For best-PFS among pts analysis, median line of best-PFS therapy for IRI-Ever (N=9 pts) was 2nd line (range 1-7) and for IRI-Never (N=7 pts) was 1st (range, 1-1); median PFS was 13 vs 3 months (mo) (IRI-Ever vs IRI-Never; p=0.0076). For PFS analysis within lines of treatment of each pt, median therapy line for both IRI-containing and No-IRI lines was 2nd line (range 1-7); median PFS for IRI-containing (N=12 lines of therapy) vs No-IRI (N=36) was 12.1 vs 3 mo (p<0.001). Median PFS for best IRI-based-(No-Platinum) line (N=7) was 13 mo; for best platinum-based-(No-IRI) line (N=9) was 2.8 mo; and for best treatment containing no IRI or platinum (N=13) was 5.1 mo (P= 0.003) (median therapy line for each was 3, 2 and 1.5, respectively). Conclusions: Irinotecan is a topoisomerase 1 inhibitor that induces DNA strand breaks, potentially causing synthetic lethality in tumors with DNA damage repair deficits due to ATM or CHEK2 alterations. IRI-containing therapy out-performed other treatment types, including platinum-containing regimens, in pts with ATM or CHEK2-mutated advanced/metastatic PDAC. Prospective trials are warranted.

The combination of KRAS G12C inhibitors with EGFR inhibitors has reproducibly been shown to be beneficial. Here, we identify another benefit of this combination: it effectively inhibits both wild-type and mutant RAS. We believe that targeting both mutant and wild-type RAS helps explain why this combination of inhibitors is effective.

e16260 Background: Therapeutic inhibition of constitutive signaling mediated by mutated KRAS in PDAC remains a challenge except for modest success reported with KRAS G12C inhibition. A combinatorial strategy utilizing simultaneous MEK and autophagy inhibition holds therapeutic promise based on mechanism of action and preclinical data. We described characteristics and outcomes of patients (pts) treated with MEK-inh and HCQ at our institution. Methods: Ten KRAS-mutated advanced PDAC pts were treated with trametinib-HCQ (n = 9) or cobimetinib-HCQ (n = 1) off label due to lack of standard treatment options or toxicity concerns with cytotoxic systemic therapy. Trametinib dose was 2 mg once daily orally, Cobimetinib dose was 20 mg BID orally for 3/4 weeks cycles. HCQ was started at 200 mg BID and up-titrated weekly to 600 mg BID. Description of baseline and treatment (tx) characteristics, safety and efficacy is provided. Results: Median age at diagnosis was 61.3 years, and 7 pts were female. The number of prior lines of tx were 0/1/2/3/4 in 3/2/1/2/2 pts, respectively. KRAS mutations were: G12R/G12D/G12V/Q61H in 6/2/1/1 pts. Median overall survival was 6.6 months (m) in all pts, and 6.6/1.7 m in KRAS G12R/other KRAS (p = 0.31). Median progression-free survival was 5.7/6.2/1.5 m in all/ KRAS G12R/other KRAS (p = 0.16). Among 8 pts with evaluable response, 1 (12%) had partial response ( KRAS G12R) and 4 (50%) stable disease (3/4 KRAS G12R) as best response with disease control rate of 63%/80%/33% in total/ KRAS G12R/other KRAS (p = 0.29). Toxicity data are summarized in table 1. Conclusions: MEK-inh-HCQ demonstrated modest efficacy and manageable toxicities among KRAS G12R PDAC pts. Unlike G12D and G12V mutations in the KRAS gene, G12R is defective of conductive interactions for both PI3Ka and NF1. This ultimately results in a weakened signal being shunted through MAPK cascade and provides a unique opportunity where MEK inh can ablate signaling without the alternate pathways and WT-RAS isoforms compensating. Furthermore, as activation of PI3Ka is known to suppress autophagy, its lack of activation by KRAS G12R further sensitizes cells to HCQ. The combination therapy MEK-inh-HCQ is therefore mechanistically-rationale and warrants the further investigation of KRAS G12R as an actionable biomarker.[Table: see text]

The combination of KRAS G12C inhibitors with EGFR inhibitors has reproducibly been shown to be beneficial. Here, we reveal a new benefit of this combination: it effectively inhibits both wild-type and mutant RAS. A role for WT RAS inhibition has not previously been reported for this important combination of targeted therapies. We believe that targeting both mutant and wild-type RAS helps explain why this combination of inhibitors is effective. ### Competing Interest Statement The authors have declared no competing interest.

We report on a woman with aggressive estrogen receptor-positive, KRAS-mutated ovarian cancer who achieved a remarkable response to combination therapy with the MEK inhibitor (trametinib) and the aromatase inhibitor (letrozole), even though the disease had failed to respond to a combination of a PI3K inhibitor and different MEK inhibitor, as well as to trametinib and the estrogen modulator, tamoxifen, and to letrozole by itself. The mechanism of action for exceptional response was elucidated by in vitro experiments that demonstrated that the fact that tamoxifen can have an agonistic effect in addition to antagonist activity, whereas letrozole results only in estrogen depletion was crucial to the response achieved when letrozole was combined with an MEK inhibitor. Our current observations indicate that subtle variations in mechanisms of action of outwardly similar regimens may have a major impact on outcome and that such translational knowledge is critical for optimizing a precision medicine strategy. KEY POINTS: This report describes the remarkable response of a patient with KRAS-mutated, estrogen receptor-positive low-grade serous ovarian cancer treated with trametinib (MEK inhibitor) and letrozole (aromatase inhibitor), despite prior progression on similar agents including tamoxifen (estrogen modulator). In vitro investigation revealed that tamoxifen can have agonistic in addition to antagonistic effects, which could be the reason for the patient not responding to the combination of trametinib and tamoxifen. The current observations suggest that drugs with different mechanisms of action targeting the same receptor may have markedly different anticancer activity when used in combinations.

Abstract A number of ATP competitive inhibitors of RAF onco-protein have been successful in tumors with RAF V600 mutations but have failed for more common cases with upstream mutations. Instead, a paradoxical activation (PA) of downstream signaling is observed in response to these inhibitors when RAF is not the driver mutation. A detailed understanding of processes that potentiate PA would help in minimizing side effects, choosing appropriate mono- or combination treatments and guide future drug development. Several mechanisms of PA have been proposed in the past, such as negative co-operativity to additional drug binding and RAF dimer stabilization. However, the evidence for these mechanisms is indirect and does not explain PA for all the different types of drugs. In our work, we created a mathematical model of RAF activation solely based on canonical RAF signal regulatory processes to show how autoinhibition provides a general mechanism of PA. We also used this model to show that 14-3-3 proteins could further potentiate PA by stabilizing autoinhibited RAF. Based on the computational work, we predicted that transfecting 14-3-3 proteins would potentiate PA even with drugs which are otherwise expected to provide minimal or no PA. Our experiments with these drugs in SKMEL2 melanoma cells and SW48 colon cancer cells validated our predictions thereby providing support for our general model of PA. Citation Format: Gaurav Mendiratta, Thomas McFall, Edward Stites. Breaking the paradox breakers - RAF inhibitor mechanisms [abstract]. In: Proceedings of the Annual Meeting of the American Association for Cancer Research 2020; 2020 Apr 27-28 and Jun 22-24. Philadelphia (PA): AACR; Cancer Res 2020;80(16 Suppl):Abstract nr 5509.

Abstract Controversially, KRAS G13D colorectal cancer patients have been reported to benefit from treatment with the EGFR inhibitor cetuximab. That EGFR inhibitors could somehow reduce signaling through a constitutively active KRAS mutant has been taken as inconsistent with known mechanisms of RAS biology. Through a combination of computational modeling and experimentation, we recently demonstrated that EGFR inhibition results in a reduction of wild-type (HRAS and NRAS) signaling in KRAS G13D colon cancer cells, but not in colon cancer cells with the other common KRAS mutants, G12D and G12V. We demonstrated this in multiple colon cancer cell lines, and we developed the hypothesis originally from our mathematical model of oncogenic RAS signal regulation adapted to biochemical rate constants for the KRAS G13D mutant. Here, we develop and analyze an alternatively parameterized version of the computational model. Analysis of this alternative model similarly suggests that EGFR inhibition causes a greater reduction of wild-type RAS signal in KRAS G13D cancers. This further substantiates our findings, analogous to demonstration in an additional cell line or an additional mouse model. In addition to reinforcing our proposed mechanism, this work also further demonstrates the power of mechanism-based computational models of protein network biochemistry in personalized cancer medicine. Note: This abstract was not presented at the conference. Citation Format: Thomas McFall, Edward C. Stites. Computational analysis of the KRAS G13D colorectal cancer response to EGFR inhibition with an alternatively parameterized model [abstract]. In: Proceedings of the AACR Special Conference on Advancing Precision Medicine Drug Development: Incorporation of Real-World Data and Other Novel Strategies; Jan 9-12, 2020; San Diego, CA. Philadelphia (PA): AACR; Clin Cancer Res 2020;26(12_Suppl_1):Abstract nr 37.

Abstract Standard protocol to treat colorectal cancer (CRC) includes surgery, chemotherapy and radiotherapy as first-line treatment options. When applicable, targeting agents such as anti-EGFR therapies (i.e., cetuximab and panitumumab) have become important long-term neoadjuvant therapies that increase survival in some patients. In recent years it has become common practice to sequence patients and their tumor specimens’ DNA. This recent push in genomic medicine has led to the identification that 30-40% of CRC patient harbor an activating KRAS mutation. These mutations are thought to be an early driver in the cancer’s progression. KRAS is also considered to be the key player in conferring resistance to the anti-EGFR treatments cetuximab and panitumumab. That Ras mutations confer resistance to EGFR inhibitors is intuitive. In recent years, clinical guidelines have extended the list of contraindicated genotypes to many mutations in both the KRAS and NRAS genes. In contrast with the current guidelines for managing CRC with EGFR inhibitors, it has been shown that patients harboring a KRAS G13D mutation are sensitive to cetuximab. This has been reproducibly shown in both in vitro and in vivo model systems and is supported by retrospective analysis of initial clinical trial data. The current mutant-RAS restrictive guidelines are supported by a method that relied upon the grouping of all common Ras mutations as “equal,” which potentially washes out less common yet sensitive mutations. Whether KRAS G13D mutations are sensitive to cetuximab has remained a controversial topic for many years due to the lack of understanding of a mechanism, as it is counterintuitive to the current understanding of EGFR signaling cascade. We here identify a novel mechanism by which G13D RAS mutant is sensitive to EGFR inhibition by a nonintuitive process of reliance upon WT RAS molecules. We utilized a computational model of RAS signaling previously developed by our laboratory to explore mutant Ras signaling and thereby investigated the controversial response of KRAS G13D to anti-EGFR agents. Our computational studies of the historically reported biochemical processes that regulate Ras signals reveal a nonintuitive, mutant-specific dependency of wild-type RAS activation on EGFR. The model also reveals this dependency is determined by the interaction strength between a KRAS mutant and tumor suppressor neurofibromin. Our prospective experiments confirm this mechanism that arises from the systems-level regulation of Ras pathway signaling. Overall, our work demonstrates how systems approaches enable mechanism-based inference in genomic medicine. Citation Format: Thomas McFall, Jolene Diedrich, Stacy L. Littlechild, Laura Sisk-Hackworth, James Moresco, Meron Mengistu1, Andrey Shaw, Ed Stites. A systems biology approach to elucidate the mechanism of EGFR inhibitor sensitivity in mutant KRAS-driven colorectal cancer [abstract]. In: Proceedings of the AACR Special Conference on Targeting RAS-Driven Cancers; 2018 Dec 9-12; San Diego, CA. Philadelphia (PA): AACR; Mol Cancer Res 2020;18(5_Suppl):Abstract nr B43.

Previous analysis of Phase 3 clinical trial data for colorectal cancer patients treated with cetuximab revealed that patients harboring a KRAS mutation did not benefit from treatment. This finding set the stage for one of the first examples of cancer personalized medicine. Confusingly, patients with a Glycine to Aspartic Acid mutation at amino acid 13 of KRAS (KRASG13D) appeared to respond positively to cetuximab, suggesting this mutation is an exception to the rule that KRAS mutations confer resistance to Epidermal Growth Factor Receptor (EGFR) inhibitors. Oncologists have stated that the mechanism that explains why the KRASG13D mutation is an exception should be identified before KRASG13D colorectal cancer patients should be treated differently. We have recently elucidated this mechanism using mathematical modeling of the KRAS biochemical system coupled with experimental biology. The mechanism we revealed involves a cetuximab-mediated reduction in HRAS and NRAS signaling within KRASG13D cancer cells, owing to impaired binding of KRASG13D to the tumor suppressor, Neurofibromin (NF1).

Phase three clinical trial evidence suggests that colorectal cancers with the KRAS G13D mutation may benefit from EGFR inhibitors, like cetuximab, in contrast to the other most common KRAS mutations. A mechanism to explain why this mutation behaves differently from other KRAS mutations had long been lacking. Two recent studies have reproduced KRAS G13D specific sensitivity to cetuximab in cellular models, and both have implicated the tumor suppressor NF1 as a critical variable in determining sensitivity and resistance. One study proposes a mechanism that focuses on the inhibition of active, GTP-bound wild-type RAS, which is proposed to occur to a greater extent in KRAS G13D tumors due to the inability of KRAS G13D to bind NF1 well. The other study suggests NF1 can convert GTP-bound KRAS G13D to inactive, GDP-bound KRAS G13D. Here, we report an inability to reproduce cellular and biophysical studies that suggested NF1 has strong GTPase activity on KRAS G13D. We also report additional data that further suggests only WT RAS-GTP levels are reduced with EGFR inhibition and that KRAS G13D is impaired in binding to NF1. These new experiments further support a mechanism in which cetuximab inhibits wild-type (HRAS and NRAS) signals in KRAS G13D colorectal cancers.

Personalized cancer medicine aims to use tumor genomics to match patients with the most effective treatment option. For example, the presence of a KRAS mutation has been determined to be a contraindication to the treatment of colorectal cancer patients with anti-EGFR targeted therapy. Interestingly, the original Phase 3 clinical trials of anti-EGFR treatment for colorectal cancer included KRAS mutant patients, and analysis of that data has previously suggested that patients with the KRAS G13D mutation benefit from EGFR inhibition. However, no mechanism had been presented to explain why this specific mutation would be an exceptional responder. Clinical guidelines therefore consider KRAS G13D equivalent to the other oncogenic KRAS mutations. We have now combined computational systems biology and experimental cancer cell biology to identify a mechanism that explains why KRAS G13D, but not the other common KRAS mutations, would respond to EGFR inhibition. Oncogenic KRAS mutations are known to be insensitive to inactivation by tumor suppressor and RAS negative regulator NF1. KRAS G13D is comparatively unable to bind to NF1 while the other common KRAS mutants can bind to NF1. The model reveals that this difference in the ability to bind NF1 is critical. By binding non-productively to NF1, most KRAS mutants competitively inhibit NF1 and promote wild-type RAS activation in an EGFR independent manner. In contrast, KRAS G13D cannot promote wild-type RAS activation through the competitive inhibition of NF1, so wild-type RAS activation remains EGFR dependent. Our experiments confirm that wild-type RAS activation decreases in KRAS G13D colorectal cancer cells treated with EGFR inhibitors, but not in colorectal cancer cells that include a different KRAS mutant. Overall, this work reveals how biophysically-based mathematical models combined with experimental cell biology can elucidate biophysically based mechanisms with direct clinical impact.

Cancer treatment decisions are increasingly guided by which specific genes are mutated within each patient’s tumor. For example, agents inhibiting the epidermal growth factor receptor (EGFR) benefit many colorectal cancer (CRC) patients, with the general exception of those whose tumor includes a KRAS mutation. However, among the various KRAS mutations, that which encodes the G13D mutant protein (KRAS G13D ) behaves differently; for unknown reasons, KRAS G13D CRC patients benefit from the EGFR-blocking antibody cetuximab. Controversy surrounds this observation, because it contradicts the well-established mechanisms of EGFR signaling with regard to RAS mutations. Here, we identified a systems-level, mechanistic explanation for why KRAS G13D cancers respond to EGFR inhibition. A computational model of RAS signaling revealed that the biophysical differences between the three most common KRAS mutants were sufficient to generate different sensitivities to EGFR inhibition. Integrated computation with experimentation then revealed a nonintuitive, mutant-specific dependency of wild-type RAS activation by EGFR that is determined by the interaction strength between KRAS and the tumor suppressor neurofibromin (NF1). KRAS mutants that strongly interacted with and competitively inhibited NF1 drove wild-type RAS activation in an EGFR-independent manner, whereas KRAS G13D weakly interacted with and could not competitively inhibit NF1 and, thus, KRAS G13D cells remained dependent on EGFR for wild-type RAS activity. Overall, our work demonstrates how systems approaches enable mechanism-based inference in genomic medicine and can help identify patients for selective therapeutic strategies.

RAF kinase inhibitors can actually increase RAF kinase signaling. This process, which is commonly referred to as “paradoxical activation” (PA), is incompletely understood. RAF kinases are regulated by autoinhibitory conformational changes, and the role of these conformational changes in PA is unclear. Our mathematical investigations find that PA can result from a dynamical equilibrium between autoinhibited and non-autoinhibited forms of RAF, along with the RAF inhibitor stabilizing the non-autoinhibited form. We also investigate whether PA is influenced by 14-3-3 proteins, which can both stabilize RAF autoinhibition and RAF dimerization. Using both computational and experimental methods we demonstrate that 14-3-3 proteins potentiate PA. Third generation RAF inhibitors normally display minimal to no PA. Our mathematical modeling led us to hypothesize that increased 14-3-3 expression should also amplify PA for these agents. Subsequent experiments support our hypothesis and show that 14-3-3 overexpression increases PA in these third generation RAF inhibitors, effectively “breaking” these “paradox breakers” and pan-RAF inhibitors. We have therefore created and experimentally validated a robust mechanism for PA based solely on equilibrium dynamics of canonical interactions in RAF signaling.