SOS1 vs SHP2 inhibitors in KRAS-mutant NSCLC

The KRAS signalling nexus and why upstream nodes matter

Mutant KRAS — most commonly G12C, G12D, or G12V in non-small cell lung cancer — is locked in a GTP-bound, constitutively active state that drives uncontrolled proliferation and survival through the RAF–MEK–ERK cascade. Direct KRAS inhibitors such as sotorasib and adagrasib trap the G12C mutant in its inactive GDP-bound conformation, but they do not eliminate the protein or its capacity to be reactivated by upstream signals, and they are structurally inapplicable to non-G12C mutants. This creates a fundamental vulnerability that adaptive resistance rapidly exploits.

The dominant resistance mechanism to direct KRAS inhibitors, MEK inhibitors, and SHP2 inhibitors alike is adaptive feedback reactivation of upstream receptor tyrosine kinases (RTKs), including EGFR, HER2/3, MET, FGFR, and AXL. When ERK output is suppressed, ERK-mediated negative feedback on SOS1, RAF, and Sprouty proteins is relieved. RTKs are transcriptionally upregulated, and autocrine or paracrine ligand production (e.g., HB-EGF, amphiregulin for EGFR) stimulates renewed GRB2–SOS1 recruitment to the membrane, reloading KRAS with GTP. Two non-redundant convergence points — SOS1 and SHP2 — sit at the fulcrum of this feedback loop and have emerged as the most clinically advanced upstream suppressor targets.

SOS1 and SHP2 are not redundant: SOS1 is the primary guanine nucleotide exchange factor (GEF) that catalyses GDP-to-GTP exchange on RAS, while SHP2 is a phosphatase that amplifies RTK signals by dephosphorylating inhibitory sites on adaptor proteins, thereby enabling efficient GRB2–SOS1 membrane recruitment. Inhibiting either node impairs RAS-GTP loading, but through distinct molecular mechanisms with different positional implications for blocking adaptive escape. According to NIH-funded structural biology, both proteins harbour druggable allosteric pockets that have now been successfully targeted in the clinic.

SOS1::KRAS inhibitors — blocking the GEF catalytic engine

SOS1 inhibitors (BI-1701963, RMC-0708) directly prevent the GDP-to-GTP exchange reaction on KRAS by binding a specific pocket on SOS1’s catalytic CDC25 domain, disrupting the productive SOS1::KRAS-GDP interaction and trapping mutant KRAS in its inactive state — an effect mechanistically analogous to direct G12C inhibitors but applicable across all KRAS mutant subtypes.

SOS1 is the primary GEF responsible for RAS activation in most cell contexts. It exists in an autoinhibited cytosolic form; RTK activation recruits SOS1 to the membrane via the GRB2 adaptor, where interaction with RAS-GDP at the allosteric RAS binding domain (RBD) relieves SOS1 autoinhibition and enables catalysis at the CDC25 domain. Even oncogenic KRAS mutants — which have intrinsically elevated nucleotide exchange rates — remain critically dependent on SOS1 for efficient cycling and maximal signalling output, particularly under the conditions of feedback reactivation that follow MAPK pathway inhibition.

“SOS1 inhibitors act as a gatekeeper downstream of RTK reactivation: by blocking the final catalytic step of KRAS-GTP loading, they intercept adaptive escape signals from any RTK that signals through GRB2–SOS1 — regardless of which receptor is driving the rebound.”

This positional logic is the key mechanistic rationale for SOS1 inhibitors as combination partners: because SOS1 sits at the convergence of multiple RTK-driven signalling streams, a single SOS1 inhibitor can theoretically neutralise adaptive escape through EGFR, MET, FGFR, and other RTKs simultaneously, without requiring a separate inhibitor for each receptor. This is mechanistically agnostic to the upstream RTK driving SOS1 recruitment.

Mutation-agnostic potential

Because SOS1 inhibitors do not target the mutant KRAS residue itself, they are mechanistically applicable to KRAS G12C, G12D, G12V, and potentially other RAS-driven cancers including NRAS-mutant melanoma. Clinical data from the BI-1701963 plus trametinib Phase 1 trial confirmed activity across G12C, G12D, G12V, and G13D NSCLC subtypes — a breadth not achievable with mutation-specific direct inhibitors.

Limitations and open questions

SOS1 has a paralog, SOS2, which may provide compensatory GEF activity in specific tissue contexts, potentially limiting efficacy in some tumour types. Additionally, GEF-independent mechanisms of RAS activation — such as those mediated by RASGRP family members — could theoretically bypass SOS1 inhibition, though evidence suggests SOS1 is dominant in KRAS-mutant cancers. SOS1 is also essential for normal RAS signalling in haematopoiesis and skin, making the therapeutic window a critical consideration. The BI-1701963 plus trametinib combination showed significant gastrointestinal toxicity, skin reactions, and liver enzyme elevations, necessitating dose reductions that may limit achievable pharmacological exposure.

SHP2 allosteric inhibitors — dimming the RTK feedback amplifier

SHP2 allosteric inhibitors (TNO155, RMC-4630) stabilise SHP2 in its autoinhibited closed conformation by binding a specific tunnel formed during the transition to the active state, preventing the dephosphorylation of adaptor proteins that is required for efficient GRB2–SOS1 membrane recruitment — thereby weakening the signal flow from multiple RTKs into KRAS before it ever reaches the GEF catalytic step.

SHP2, encoded by PTPN11, is a non-receptor protein tyrosine phosphatase that plays a paradoxical but dominant pro-oncogenic role downstream of most RTKs. In its active state, SHP2 dephosphorylates inhibitory phosphorylation sites on scaffold proteins such as GAB1, GAB2, and IRS1. This dephosphorylation is required for the stable formation of the full RTK–GRB2–SOS1 signalling complex at the membrane, which drives KRAS-GTP loading. By stabilising SHP2 in its autoinhibited state, allosteric inhibitors impair this coupling — weakening the signal amplitude from EGFR, MET, FGFR, and other RTKs before it converges on SOS1 and KRAS.

This upstream positioning makes SHP2 inhibitors particularly well-suited as combination partners for KRAS G12C inhibitors and MEK inhibitors: by dampening the RTK signal strength that drives adaptive reactivation, they directly counteract the relief of negative feedback caused by ERK pathway suppression. In effect, SHP2 inhibitors address the “upstream escape” mechanism that limits the durability of downstream blockade. As documented in research indexed by Nature, SHP2 also participates in immune checkpoint signalling (including PD-1), introducing both potential immunomodulatory benefits and on-target immune cell toxicities that must be managed in clinical development.

Key limitation: SHP2-independent RTK bypass

While SHP2 integrates signals from many RTKs, some receptors can signal to RAS via SHP2-independent pathways — for example, through direct GRB2–SOS1 recruitment or alternative phosphatase-adaptor combinations. This means SHP2 inhibitors weaken but do not fully eliminate RTK-to-RAS coupling, and escape through SHP2-independent channels remains a theoretical resistance mechanism. This is precisely why combinations with direct KRAS inhibitors or SOS1 inhibitors are being explored: the two classes are mechanistically complementary, with SHP2 inhibitors reducing signal amplitude and the downstream inhibitors blocking residual RAS activation.

Combination trial evidence: ORR, pharmacodynamics, and resistance patterns

Combination trials of SHP2 inhibitors with KRAS G12C inhibitors have generated the most mature clinical evidence, with ORRs of 30–35% in inhibitor-naive NSCLC patients — generally exceeding historical single-agent KRAS G12C inhibitor benchmarks in later lines — and pharmacodynamic confirmation that adaptive RTK feedback signalling is durably suppressed in responding patients.

TNO155 + sotorasib (CodeBreaK 101, Phase 1b)

This trial evaluated the SHP2 inhibitor TNO155 combined with the KRAS G12C inhibitor sotorasib in heavily pre-treated KRAS G12C NSCLC. In patients naive to KRAS G12C inhibitors, the combination achieved an ORR of approximately 30%, with some responders showing prolonged durations of response exceeding what is typically observed with single-agent sotorasib in later lines. In patients previously treated with a KRAS G12C inhibitor, the ORR was approximately 17% — providing direct clinical evidence that SHP2 inhibition can partially overcome acquired resistance to direct KRAS blockade. Pharmacodynamic analyses demonstrated significant suppression of pERK levels and modulation of key RTK signalling pathways including EGFR and HER3 in tumour biopsies, consistent with blunting of adaptive feedback. Early circulating tumour DNA (ctDNA) analysis suggested suppression of emerging KRAS-independent resistance clones.

RMC-4630 + sotorasib (Phase 1/2)

RMC-4630 combined with sotorasib showed an ORR of around 35% in KRAS G12C inhibitor-naive NSCLC patients — the highest reported ORR among SHP2 inhibitor doublet combinations — with activity also documented in patients who had progressed on prior KRAS G12C inhibitors. Pharmacodynamic analyses demonstrated robust pERK reduction and modulation of feedback markers. The adverse event profile was similar to the TNO155 combination: primarily gastrointestinal (diarrhoea, nausea) and haematological (neutropenia, anaemia) toxicities, manageable with dose modifications.

RMC-4630 + cobimetinib ± erlotinib (triple combination)

Recognising that MEK inhibition alone relieves additional negative feedback, this trial explored broader vertical RAS/MAPK pathway inhibition. The triple combination of RMC-4630 plus cobimetinib plus erlotinib showed approximately 30% ORR in broader KRAS-mutant NSCLC populations — including patients with non-G12C mutations — particularly in those with inflammatory gene expression signatures. This provides proof-of-concept that adding an RTK inhibitor (erlotinib targeting EGFR) to a SHP2–MEK backbone can yield incremental benefit by more comprehensively blocking parallel and feedback RTK signals, at the cost of additional toxicity management complexity.

BI-1701963 + trametinib (Phase 1)

The SOS1 inhibitor BI-1701963 combined with the MEK inhibitor trametinib was evaluated across solid tumours including KRAS-mutant NSCLC of multiple mutant subtypes. Partial responses were observed in KRAS G12C and G12V NSCLC patients, and disease control was achieved in G12D and G13D subtypes. Pharmacodynamic analyses confirmed significant pERK reduction, validating target engagement across mutant subtypes. However, the combination showed a particularly challenging toxicity profile — severe gastrointestinal toxicity, skin reactions, and liver enzyme elevations — that necessitated dose reductions and limited the ability to reach higher, potentially more efficacious exposures of BI-1701963. This toxicity overlap between SOS1 and MEK inhibition is a key differentiating challenge compared to SHP2 inhibitor combinations.

Comparing the two approaches: intercepting vs. dampening the RTK signal

SOS1 inhibitors and SHP2 inhibitors are not interchangeable — they occupy distinct positions in the RTK-to-RAS signal cascade, with different mechanistic strengths, mutation applicability profiles, combination partner preferences, and toxicity signatures that together define their respective clinical development paths in KRAS-mutant NSCLC.

The fundamental positional distinction is this: SOS1 inhibitors act as a gatekeeper immediately downstream of RTK reactivation, blocking the final catalytic step of KRAS-GTP loading regardless of which RTK is driving SOS1 recruitment. SHP2 inhibitors act as a dimmer switch upstream, reducing the amplitude of signal flowing from multiple RTKs toward GRB2–SOS1 before it reaches the GEF step. Both approaches impair RAS-GTP loading, but SOS1 inhibitors provide a more direct, complete blockade of the catalytic event, while SHP2 inhibitors provide broader upstream attenuation that may be more effective at suppressing the early phases of adaptive feedback before SOS1 is fully recruited.

On mutation breadth, both classes are mechanistically mutation-agnostic, but clinical evidence is more mature for SHP2 inhibitor combinations in G12C NSCLC (driven by the availability of approved G12C inhibitors as combination partners), while SOS1 inhibitor combinations with MEK inhibitors have demonstrated activity across multiple non-G12C subtypes — a potentially important advantage as direct inhibitors for G12D and G12V reach the clinic and require upstream partners. Research standards from WIPO-tracked patent filings and ClinicalTrials.gov registrations confirm that both target classes are among the most actively patented and clinically explored upstream RAS pathway nodes globally.

On tolerability, SHP2 inhibitor combinations with KRAS G12C inhibitors have shown a more manageable adverse event profile (primarily gastrointestinal and haematological) compared to the SOS1 inhibitor plus MEK inhibitor combination (BI-1701963 plus trametinib), which showed severe gastrointestinal toxicity, skin reactions, and hepatic enzyme elevations requiring dose reductions. Whether next-generation SOS1 inhibitors such as RMC-0708 — for which mature clinical data is not yet publicly available — will demonstrate improved tolerability remains a key question for the field.

Looking ahead, the most important unresolved questions are: whether biomarkers (specific RTK dependencies, co-mutations, gene expression signatures) can identify which patients benefit most from SOS1 inhibitor versus SHP2 inhibitor combinations; whether triple combinations (SHP2 inhibitor plus SOS1 inhibitor plus direct KRAS or MEK inhibitor) can achieve durable responses without prohibitive toxicity; and whether optimal dosing schedules — including intermittent dosing to reduce toxicity while maintaining efficacy — can expand the therapeutic window for both classes. The ASCO annual meeting and ongoing Phase 1/2 readouts will be critical milestones for resolving these questions in the coming 12–24 months.

“The triple combination of RMC-4630 plus cobimetinib plus erlotinib showed approximately 30% ORR in broader KRAS-mutant NSCLC populations — demonstrating that simultaneously blocking multiple RTK feedback nodes yields higher response rates than doublet combinations alone, at the cost of increased complexity in toxicity management.”