Combined KRAS G12C and SOS1 inhibition enhances and extends the anti-tumor response in KRAS G12C -driven cancers by addressing intrinsic and acquired resistance

Alterations in KRAS are common across cancers, with KRAS G12C prevalent in NSCLC and CRC. Covalent KRAS G12C inhibitors (e.g., sotorasib, adagrasib) benefit patients but most develop resistance. KRAS cycles between GDP-bound inactive and GTP-bound active states, and approved KRAS G12C inhibitors preferentially bind the inactive GDP-bound form. Resistance often involves reactivation of RAS-MAPK signaling via WT RAS activation, secondary KRAS mutations, or bypass through upstream/downstream nodes. Combinations targeting regulators of KRAS GTP-loading such as RTKs or SHP2 improve efficacy by maintaining KRAS in the GDP-bound state and suppressing pathway feedback. The authors developed potent SOS1 inhibitors (BI-3406; clinical BI 1701963) that disrupt the SOS1–KRAS interaction and reduce GTP loading. Hypothesis: combining a KRAS G12C inhibitor with a SOS1 inhibitor will enhance depth and durability of response in KRAS G12C-mutant NSCLC and CRC, and help overcome acquired resistance compared with KRAS G12C inhibitor alone or combinations with SHP2 or EGFR inhibitors.

Prior clinical and preclinical studies demonstrate activity of sotorasib and adagrasib in KRAS G12C-mutant tumors but with frequent resistance and limited durability. Adaptive feedback reactivates RAS-MAPK signaling via RTKs, SHP2, or alternative pathways; secondary KRAS mutations (e.g., R68S, Y96C/D, H95 variants), KRAS amplification, and activation of PI3K/mTOR or YAP-TEAD have been implicated. Combinations of KRAS G12C inhibitors with SHP2 inhibitors or EGFR blockade enhance responses in preclinical NSCLC/CRC and have advanced clinically, improving response rates in CRC. SOS1, a GEF downstream of RTKs, is a feedback node modulating KRAS activation; SOS1 inhibition maintains KRAS in the GDP-bound state and has shown anti-tumor effects. Thus, SOS1 is a rational, potentially better-tolerated alternative to targeting individual RTKs or SHP2 to enhance KRAS G12C inhibitor efficacy.

- High-throughput drug combination screens: Combined sotorasib (AMG 510) or adagrasib (MRTX849) with 179 small molecules in KRAS G12C NSCLC cell line NCI-H2122; a focused subset evaluated in KRAS G12C CRC cell line SW837. Synergy quantified by Bliss-based cScore. - Panel studies: Eleven KRAS G12C-driven cell lines treated with adagrasib plus BI-3406 (SOS1i), TNO155/SHP099 (SHP2i), or cetuximab (EGFRi). Longitudinal confluence/apoptosis assays over 7 days in NCI-H2122 and SW837 using IncuCyte. - In vivo efficacy: Cell line-derived xenografts (CDX) NCI-H2122 (NSCLC) and SW837 (CRC); CRC PDX models F3008 and B8032. Dosing regimens included: adagrasib 100 mg/kg p.o. qd; BI-3406 50 mg/kg p.o. bid; TNO155 10 mg/kg p.o. bid; cetuximab 15–20 mg/kg i.p. twice weekly. Measured tumor volumes, tumor growth inhibition (TGI), and regression. - Pharmacodynamics and signaling analyses: RAS-GTP measured by G-LISA; Western blots for p-ERK, p-S6, DUSP6, Cyclin D1, cleaved PARP, total ERK/KRAS. Transcriptomics (RNA-seq) on xenografts (QuantSeq or TruSeq); differential expression with DESeq2; MAPK Pathway Activity Score (MPAS) to quantify pathway activity; GSEA for pathway enrichment. In situ assays: RNAscope for DUSP6/EGR1; multiplex IF/IHC for p-ERK and Ki-67 in tumor tissues at 4–48 h after last dose. - Resistance modeling: Generated adagrasib-resistant NCI-H358 cells by long-term exposure to 10× IC50 adagrasib; assessed sensitivity to combinations. Created KRAS G12C Ba/F3 transgenic library harboring comprehensive secondary KRAS mutations; performed colony outgrowth assays under KRAS G12C inhibitors ± BI-3406 at escalating doses. - Relapse retreatment study: SW837 xenografts treated chronically with adagrasib (5 days on/2 off) until outgrowth (>100 mm³ over nadir), then randomized to retreatment with adagrasib alone or plus BI-3406, cetuximab, or TNO155; evaluated tumor responses and transcriptomic changes in relapsed tumors. - Statistics: Parametric and nonparametric tests (ANOVA with Tukey, Student’s t-test, Wilcoxon rank-sum, Fisher exact), multiple-testing corrections (Bonferroni-Holm, Benjamini-Hochberg).

- High-throughput screens identified strong synergy between KRAS G12C inhibitors and inhibitors of upstream KRAS activation, including BI-3406 (SOS1i), SHP2 inhibitors (TNO155, SHP099), and ErbB family inhibitors (lapatinib, afatinib); also synergy with downstream MEK and PI3K inhibitors. - Across 11 KRAS G12C-driven cell lines, adagrasib plus BI-3406 or TNO155 showed the strongest synergistic anti-proliferative effects (BI-3406 in 6/11, TNO155 in 8/11 lines); cetuximab (5/11) and SHP099 (4/11) also synergized. In NCI-H2122 and SW837, combinations produced more profound and durable growth inhibition than adagrasib alone over 7 days. - In vivo efficacy: NCI-H2122 CDX: adagrasib alone TGI 83% at Day 16; combinations improved control—adagrasib+BI-3406 TGI 106% (mean TV change −18% at Day 16; p=0.0222; 5/6 regressions) and adagrasib+TNO155 TGI 104% (−13%; p=0.0189; 5/7 regressions). SW837 CDX: adagrasib alone TGI 108% (−35% at Day 42) vs adagrasib+BI-3406 TGI 119% (−67.8%) and adagrasib+cetuximab TGI 121% (−74.4%), with deeper regressions in all tumors. CRC PDXs (F3008, B8032): adagrasib monotherapy modest, with outgrowth by ~Day 20; combinations with BI-3406 or cetuximab drove sustained regressions throughout the study. - Mechanism: Adagrasib+BI-3406 further reduced RAS-GTP (notably at 7 and 24 h) vs monotherapies, consistent with trapping KRAS in the GDP-bound state. Combinations more strongly suppressed p-ERK and p-S6 at 6 h and sustained inhibition (including DUSP6) at 24 h, with increased cleaved PARP and reduced Cyclin D1 (stronger in SW837). Rebound of p-ERK/p-S6 under adagrasib monotherapy was attenuated by combinations. - Transcriptomics/MPAS: In NCI-H2122, combinations modulated more genes than monotherapy (adagrasib+BI-3406: 578; +TNO155: 547 vs adagrasib: 217 at 4 h) and achieved lower MPAS (−0.502 and −0.435 vs −0.35). In SW837, all treatments downregulated MPAS at 4 h (adagrasib −0.57; +BI-3406 −0.49; +cetuximab −0.68), but only combinations sustained MAPK suppression up to 48 h (e.g., CCND1 log2 fold −1.8; DUSP6 −2.4 vs vehicle). RNAscope confirmed greater DUSP6/EGR1 downregulation with combinations. - Proliferation markers: In NCI-H2122, adagrasib+BI-3406 significantly reduced p-ERK and Ki-67 at 4 h vs adagrasib alone (p-ERK p=0.0318; Ki-67 p=0.014). In SW837, Ki-67 was significantly reduced by both adagrasib+BI-3406 and adagrasib+cetuximab (p=0.008) at 4 and 48 h. - Resistance mitigation: In Ba/F3 KRAS G12C library bearing secondary mutations, adding BI-3406 to KRAS G12C inhibitors reduced outgrowth-positive wells in a dose-dependent manner (stronger at 600 nM than 300 nM BI-3406), indicating benefit against secondary on-target resistance mutations. Adagrasib-resistant NCI-H358 cells displayed improved sensitivity to adagrasib+BI-3406 (and +TNO155) vs adagrasib alone. - Relapse retreatment (SW837): Relapsed tumors under chronic adagrasib responded with regression to adagrasib+BI-3406 or +TNO155, while adagrasib alone failed and adagrasib+cetuximab induced stasis. MPAS in relapsed tumors was further reduced by adagrasib+BI-3406 (p=0.014) but not by adagrasib+cetuximab (p=0.29). - Resistance biology: No secondary KRAS mutations were detected in relapsed SW837 tumors; DUSP5/DUSP6 were downregulated (consistent with continued KRAS G12C inhibition). Upregulation of MRAS and RasGRP1 suggested alternative MAPK reactivation routes (e.g., MRAS–SHOC2–PP1 complex) may contribute to resistance despite modest overall MPAS rebound. Overall, SOS1 co-inhibition enhanced and prolonged MAPK suppression and anti-tumor effects, including in acquired resistance settings.

Combining a KRAS G12C inhibitor with a SOS1 inhibitor enhances anti-tumor potency and durability in KRAS G12C-mutant NSCLC and CRC by more effectively lowering RAS-GTP and suppressing downstream MAPK signaling than monotherapy. The efficacy of adagrasib+BI-3406 was comparable to other clinically explored combinations (with SHP2 or EGFR inhibitors) across in vitro and in vivo models, supporting SOS1 as a rational common upstream node that can buffer multi-RTK feedback and reduce adaptive RAS reactivation. The combination attenuated pharmacodynamic rebound (p-ERK/p-S6, MPAS) relative to monotherapy and deepened inhibition of proliferation markers (Ki-67) and MAPK-regulated transcripts (DUSP6, EGR1). Importantly, SOS1 co-inhibition delayed or overcame acquired resistance: it suppressed outgrowth of KRAS G12C-mutant cells harboring secondary KRAS mutations, improved sensitivity of adagrasib-resistant cells, and re-established tumor regression in relapsed CRC xenografts, outperforming adagrasib alone and exceeding cetuximab combinations in the relapse setting. Mechanistically, resistance in these CRC models did not appear driven by secondary KRAS mutations or large MAPK rebounds. Instead, transcriptional shifts (e.g., MRAS and RasGRP1 upregulation) suggest compensation through parallel RAS-family signaling and the MRAS–SHOC2–PP1 axis. Targeting SOS1 may counteract such adaptations by broadly limiting RAS GTP-loading. Given potential pleiotropic effects and tolerability considerations with SHP2 inhibitors, SOS1 inhibition may offer an alternative route to achieve robust KRAS pathway control. These findings provide a strong preclinical rationale for clinical evaluation of KRAS G12C plus SOS1 inhibition, including in patients who have progressed on KRAS G12C inhibitors.

Co-administration of the SOS1 inhibitor BI-3406 with the KRAS G12C inhibitor adagrasib produces deeper and more durable anti-tumor responses than adagrasib alone in KRAS G12C-driven NSCLC and CRC models, with efficacy comparable to combinations with SHP2 or EGFR inhibitors. The combination enhances suppression of RAS-GTP and MAPK signaling, delays emergence of resistance, and reinstates tumor control in models with acquired resistance to adagrasib. Transcriptomic analyses indicate complex adaptive changes, including MRAS and RasGRP1 upregulation, that may be mitigated by maintaining KRAS in the GDP-bound state via SOS1 inhibition. These data support ongoing and future clinical trials of SOS1 inhibitors with KRAS G12C inhibitors and suggest exploring combinations with next-generation “KRAS-on” inhibitors as a strategy to improve response rates and durability. Future work should delineate resistance circuitry (e.g., MRAS–SHOC2–PP1) and identify biomarkers to guide patient selection.

- Preclinical scope: Findings are based on cell lines, CDX, and PDX models; clinical translatability and tolerability of combinations require validation. - Observation windows: Some studies had limited durations; pharmacodynamic rebound and late resistance mechanisms beyond study windows may not be fully captured. - Sample sizes and repeats: Certain assays had small n (e.g., G-LISA n=2; some xenograft PD cohorts n=2–5), potentially limiting statistical power. - Model specificity: Resistance mechanisms observed (e.g., MRAS/RasGRP1 upregulation without secondary KRAS mutations) may be model- or tumor-type specific and not universally generalizable. - Comparative arms: While combinations with SHP2 and EGFR inhibitors were included, broader comparison with other vertical pathway inhibitors (MEK, PI3K, CDK4/6) was limited in vivo.