| MET — ALK | Cell and tissue relevance | Abstained | gpt-5.6-luna | 97% | Both dossiers provide human tumor-tissue evidence for localization in NSCLC malignant cells or adenocarcinoma compartments (C1-EV_TISSUE_0001; C2-EV_TISSUE_0001; C2-EV_TISSUE_0002), and both have evidence of tumor-associated genomic dysregulation. However, neither dossier directly evaluates the specified mucus-producing glandular cells using single-cell, spatial, proteomic, or cell-specific pathology methods. The supplied evidence therefore does not establish or compare disease-relevant expression, localization, and dysregulation in the critical cell type. Under the rubric and guardrail, a ranking would require unsupported extrapolation from broad tumor-cell or histology-level findings. | - Direct single-cell or spatial transcriptomic, proteomic, or immunohistochemical evidence measuring MET and ALK expression/localisation and dysregulation specifically in human NSCLC mucus-producing glandular cells, with adjustment for cell-composition shifts and disease stage.
- Evidence linking either MET or ALK dysregulation specifically to the mucus-producing glandular-cell compartment rather than to NSCLC tumor cells broadly.
| – |
| EGFR — MET | Resistance and escape liability | B wins | gpt-5.6-luna | 84% | Under the reversed direction, SECOND (MET) is less vulnerable on the supplied evidence, but only narrowly. EGFR has multiple documented escape routes: on-target resistance mutations including T790M and C797S, MET or other bypass signaling, alternative downstream pathway activation, intratumoral heterogeneity, and therapeutic-selection-driven clonal evolution (C1-CAT-ONCO-BIO-004, C1-CAT-ONCO-BIO-013, C1-CAT-ONCO-BIO-063, C1-CAT-ONCO-BIO-066, C1-CAT-ONCO-BIO-067, C1-CAT-ONCO-BIO-078). For MET, the supplied resistance evidence identifies pre-existing RAS–MAPK alterations as a mechanism that can cause primary resistance or reduce response durability (C2-EV_GENETICS_0005), but does not document a comparably broad set of acquired-resistance and clonal-escape mechanisms. This should not be interpreted as evidence that MET lacks such mechanisms; the limited MET evidence makes the margin narrow. Biomarker strategies would mitigate risk for both targets: EGFR mutation-specific selection and serial genomic monitoring could identify resistance states, while MET exon-14 selection and amplification-level refinement could reduce biological and assay-related misclassification. Combination mitigation is directly supported for EGFR escape caused by MET/HGF bypass, where combined EGFR and MET inhibition overcame resistance in vivo (C2-EV_MECHANISM_0005). The dossiers do not provide comparable direct evidence that combination therapy mitigates MET's own identified RAS–MAPK resistance risk. | - A matched, longitudinal clinical comparison of acquired resistance, clonal evolution, pathway bypass, and resistance duration for biomarker-selected EGFR- versus MET-directed therapy in NSCLC; the MET dossier particularly lacks comparable evidence on acquired resistance mechanisms.
| |
| MET — KRAS | Disease biology and mechanism coherence | A wins | gpt-5.6-luna | 93% | MET has a coherent causal chain from recurrent human NSCLC driver alterations to a defined molecular mechanism: METex14 removes the CBL-binding region, impairs receptor down-regulation, sustains MET signaling, and activates downstream MAPK pathways (C1-EV_GENETICS_0002, C1-EV_MECHANISM_0002, C1-EV_MECHANISM_0003). This is reinforced by selective genetic dependency and pharmacologic inhibition in MET-amplified NSCLC models, including patient-derived xenografts (C1-EV_PERTURBATION_0001, C1-EV_PERTURBATION_0002). KRAS has strong human-genetic plausibility and a coherent RAS-MAPK model (C2-EV_GENETICS_0002, C2-EV_GENETICS_0003, C2-EV_MECHANISM_0001), but the supplied structured evidence contains substantially less direct, primary, NSCLC-specific perturbational validation of inhibition altering disease biology. MET evidence is not universal: dependency is biomarker-restricted and heterogeneous, which limits generalization but does not overturn the stronger mechanistic support for the defined MET-altered subset. This judgment concerns biological coherence and experimental support, not druggability. | - For KRAS: primary, disease-specific mechanistic perturbation studies in NSCLC using genetic knockdown, knockout, or allele correction, ideally in patient-derived models and with pathway readouts.
- For both targets: direct pathology and perturbation evidence specifically in the stated mucus-producing glandular lung-cell context.
| |
| ALK — EGFR | Competitive differentiation | A wins | gpt-5.6-luna | 82% | FIRST has the stronger dossier-level evidence for clinically meaningful differentiation over chemotherapy: randomized evidence in untreated, biomarker-selected NSCLC showed superior progression-free survival and response versus platinum-pemetrexed chemotherapy for ALK inhibition (C1-EV_CLINICAL_0001). The dossier also documents clinically meaningful differentiation within the ALK competitive landscape, particularly long-term systemic and intracranial control with lorlatinib, while acknowledging that the segment is crowded and that fourth-generation competitors are emerging (C1-EV_COMPETITIVE_0001, C1-EV_COMPETITIVE_0002, C1-EV_COMPETITIVE_0005). This is therefore a narrow evidence-based advantage, not a claim of superior market attractiveness. EGFR has evidence of biological tractability and established small-molecule inhibition, but the supplied dossier does not provide comparable direct clinical evidence of differentiation over chemotherapy or a sufficiently specific competitor-landscape analysis. | - For EGFR: randomized first-line clinical evidence directly comparing an EGFR-directed small molecule with chemotherapy in newly diagnosed EGFR-mutant NSCLC.
- For EGFR: a current treatment-landscape and competitor-pipeline analysis establishing differentiation versus existing EGFR therapies and chemotherapy.
- For both targets: a direct, contemporary head-to-head comparison against the relevant standard and preferred targeted competitors in the newly diagnosed setting.
| |
| MET — EGFR | Resistance and escape liability | A wins | gpt-5.6-luna | 78% | Applying the reversed direction, MET (FIRST) is less vulnerable on the supplied evidence, but only narrowly. MET has a documented primary-resistance liability from pretreatment RAS–MAPK alterations (C1-EV_GENETICS_0005). However, EGFR has substantially more direct evidence of biological escape: MET/HGF bypass and other compensatory signalling (C2-CAT-ONCO-BIO-004; C2-CAT-ONCO-BIO-013), acquired EGFR resistance mutations such as T790M and C797S (C2-CAT-ONCO-BIO-063; C2-CAT-ONCO-BIO-078), and clonal evolution involving additional EGFR, MET, or amplification events (C2-CAT-ONCO-BIO-066). Biomarker strategies can mitigate risk for both targets: METex14 or high-level amplification selection should reduce heterogeneity and primary nondependence, while EGFR mutation-specific selection identifies the sensitive population. These strategies do not eliminate acquired escape. Combination strategies are biologically indicated for EGFR bypass, particularly targeting MET/HGF or downstream signalling, but the dossier does not provide a definitive clinical combination result for this criterion. For MET, a combination strategy to overcome RAS–MAPK-mediated resistance is not directly demonstrated in the supplied evidence. | - Direct, matched clinical resistance and clonal-evolution data under selective MET inhibition versus EGFR inhibition, including the frequency of acquired bypass alterations and outcomes of resistance-directed combinations.
| |
| ALK — EGFR | Time to decisive experiment | Abstained | gpt-5.6-luna | 94% | Both dossiers support availability of small-molecule tools and some model-related infrastructure: ALK has approved TKIs and companion-diagnostic infrastructure, while EGFR has described tool compounds and published preclinical work including PDX models. However, neither dossier provides organisation-specific capacity, historical cycle times, direct cost evidence, or readiness of the exact specified PDX and mucus-producing glandular-cell system. The available evidence therefore does not support a defensible comparison of which target would reach a high-information go/no-go decision faster and at lower cost. | - Direct comparative assay and reagent availability for the specified mucus-producing glandular-cell PDX experiment.
- Readiness, reproducibility, and expected execution time of the exact patient-derived xenograft models for each target.
- Organisation-specific internal capability, capacity, vendor access, and historical cycle times.
- Direct cost or resource estimates and predefined go/no-go decision criteria for the in vivo proof-of-concept studies.
| |
| KRAS — ALK | Biomarker and target engagement | B wins | gpt-5.6-luna | 96% | ALK offers the more credible end-to-end clinical biomarker strategy. ALK rearrangement is explicitly established as a regulatory predictive biomarker with FDA-approved testing, including FISH, IHC, and companion-diagnostic infrastructure (C2-EV_BIOMARKER_0001; C2-EV_TRACTABILITY_0006). Biomarker-selected randomized evidence demonstrates an early efficacy signal versus chemotherapy (C2-EV_BIOMARKER_0002; C2-EV_CLINICAL_0001), while longitudinal plasma ALK mutation kinetics provide clinical pharmacodynamic evidence consistent with target engagement (C2-EV_BIOMARKER_0005). Durable systemic and intracranial efficacy further supports the strategy (C2-EV_CLINICAL_0004). KRAS has strong genetic and therapeutic precedent, including approved KRAS-G12C inhibitors and mutation-based selection (C1-CAT-ONCO-BIO-091b; C1-CAT-ONCO-BIO-115), but the supplied dossier does not document an equivalently complete clinical assay, target-engagement, pharmacodynamic, and biomarker-linked early efficacy chain. ALK evidence has important limitations: ALK positivity can be nonproductive by FISH alone and resistance or variant heterogeneity can weaken dependence (C2-EV_BIOMARKER_0004; C2-EV_BIOMARKER_0006; C2-EV_MECHANISM_0006; C2-EV_MECHANISM_0007), but these caveats do not outweigh its substantially stronger clinical validation. | - For KRAS, a directly documented FDA-authorized clinical assay strategy, clinically measured target-engagement pharmacodynamics, and biomarker-linked early NSCLC efficacy data comparable in specificity to the ALK evidence.
| |
| EGFR — MET | Modality–direction–exposure fit | B wins | gpt-5.6-luna | 82% | SECOND has the stronger modality-direction fit. MET exon 14 skipping produces impaired receptor ubiquitination and delayed down-regulation, providing a documented rationale for sustained inhibition by a small molecule (C2-EV_MECHANISM_0002). Direct pMET:total-MET assays support measurement of target engagement and duration of pathway suppression (C2-EV_BIOMARKER_0005), while oral genotype-selected small-molecule inhibition is clinically established (C2-EV_TRACTABILITY_0002) and has shown activity in MET-amplified patient-derived xenografts (C2-EV_PERTURBATION_0002). FIRST also has strong small-molecule precedent through multiple approved EGFR TKIs (C1-EV_TRACTABILITY_0001), but the dossier provides less direct evidence tying inhibition depth or duration to the specified tissue and cell type. The margin is narrow because neither dossier supplies direct biodistribution or PK/PD exposure data in mucus-producing glandular cells, and MET expression/alteration-to-protein concordance is heterogeneous (C2-EV_TISSUE_0006; C2-EV_BIOMARKER_0006). | - Direct comparative PK/PD and biodistribution data for the proposed small molecules in lung mucus-producing glandular cells, including exposure duration, target occupancy or suppression, and confirmation in the specified patient-derived xenograft context.
| |
| EGFR — MET | Time to decisive experiment | Abstained | gpt-5.6-luna | 97% | The dossiers document that EGFR tool compounds are available and that EGFR validation has used PDX and other preclinical models (C1-CAT-ONCO-BIO-115; C1-CAT-ONCO-BIO-010), while MET has validated pharmacodynamic assays and MET-amplified PDX evidence (C2-EV_BIOMARKER_0005; C2-EV_PERTURBATION_0002). However, Q17 asks which target can reach a high-information in vivo go/no-go decision faster and at lower cost. Neither dossier provides organisation-specific readiness, internal capacity, historical cycle times, or comparative cost, and neither establishes readiness in the specified mucus-producing glandular-cell PDX context. These omissions prevent a defensible comparison; understudied operational evidence should not be converted into a ranking. | - Organisation-specific assay and reagent readiness for the specified mucus-producing glandular-cell NSCLC PDX context
- Internal capability, staffing, capacity, and access to the required patient-derived xenograft models for each target
- Historical cycle times from study initiation to interpretable in vivo proof-of-concept go/no-go data for each target
- Direct cost or resource estimates for the EGFR and MET in vivo studies
| – |
| KRAS — EGFR | Combination potential | Abstained | gpt-5.6-luna | 97% | The dossiers provide mechanistic pathway information and opposing general safety signals, but they do not contain direct combination-preclinical evidence with chemotherapy or a documented overlapping-toxicity assessment for either target. KRAS mutant-selective inhibition is described as sparing wild-type KRAS (C1-EV_SAFETY_0001), whereas wild-type EGFR inhibition is associated with dermatologic and gastrointestinal toxicity (C2-EV_SAFETY_0001), but these facts alone do not establish that either target can be dosed safely and effectively with chemotherapy. Under the guardrail, the available evidence is insufficient to rank combination potential. | - Direct KRAS-plus-chemotherapy and EGFR-plus-chemotherapy combination studies in NSCLC, preferably in the stated patient-derived xenograft context, including efficacy, dose scheduling, pharmacodynamic interaction, and formal overlapping-toxicity/tolerability assessment.
| - On-target dermatologic and GI toxicity
- Mutant-selective inhibitors spare wild type
|
| MET — KRAS | Competitive differentiation | A wins | gpt-5.6-luna | 68% | FIRST has the stronger dossier-supported differentiation rationale over chemotherapy: regulatory review explicitly identifies chemotherapy as a non-targeted background option for METex14-positive NSCLC and supports biomarker-selected MET treatment as the differentiation opportunity (C1-EV_COMPETITIVE_0003). METex14 also has substantial activity in treatment-naive patients, although the cited evidence is not a direct newly diagnosed chemotherapy comparison (C1-EV_SAFETY_0002; C1-EV_CLINICAL_0001). This advantage is narrow because MET already faces multiple approved or clinically active competitors and a non-TKI competitor (C1-EV_COMPETITIVE_0005; C1-EV_COMPETITIVE_0006). KRAS has clinically validated small-molecule tractability, but the supplied evidence does not establish greater clinical benefit over chemotherapy in newly diagnosed NSCLC (C2-EV_TRACTABILITY_0001). This verdict addresses scientific and clinical differentiation evidence, not market attractiveness. | - Direct randomized evidence in newly diagnosed NSCLC comparing MET- or KRAS-directed therapy with chemotherapy, including clinically meaningful endpoints.
- For KRAS, a target-specific unmet-need analysis and competitor-pipeline evidence directly addressing differentiation over chemotherapy in the newly diagnosed setting.
| |
| EGFR — KRAS | Reversibility and therapeutic window | B wins | gpt-5.6-luna | 78% | KRAS offers stronger evidence for a therapeutic window because mutant-selective G12C inhibitors can spare wild-type KRAS, whereas wild-type EGFR inhibition is associated with rash and diarrhea that narrow the window. However, KRAS intervention is covalent and therefore less readily reversible at the target than the reversible EGFR TKIs documented in the dossier. EGFR also includes irreversible small molecules, so the overall comparison is modality- and compound-dependent. The combined advantage therefore favors KRAS only narrowly, not clearly. | - Direct comparative exposure-response data, pharmacokinetic half-life and target-residence data, and clinical evidence on adverse-event reversibility after treatment interruption for representative EGFR and KRAS small-molecule inhibitors.
| |
| KRAS — ALK | Reversibility and therapeutic window | Abstained | gpt-5.6-luna | 94% | The dossiers contain relevant but non-comparable evidence. KRAS has explicit evidence for irreversible covalent binding of G12C inhibitors, while mutant selectivity may improve the therapeutic window; ALK has evidence suggesting limited normal-tissue essentiality but also clinically important toxicities. However, the ALK dossier does not establish reversibility, half-life, washout, or durability, and neither dossier provides a direct comparative exposure-response assessment. Ranking ALK higher would improperly treat missing evidence as evidence of reversibility, so the appropriate verdict is ABSTAIN. | - Direct comparative pharmacokinetic and exposure-response data for the candidate interventions, including modality half-life, washout after interruption, and recovery of target activity.
- Direct evidence establishing whether the ALK small-molecule interventions are reversible or have durable target engagement in vivo.
- Human dose-phenotype relationships or clinical interruption/rechallenge data comparing KRAS and ALK inhibition.
| - Weist KRAS Mutationen auf, die neue medikamentös adressierbare Bindungsstellen schaffen?
- Mutant-selective inhibitors spare wild type
- Favourable on-target safety
- ALK inhibition has a generally manageable clinical safety profile but carries clinically important risks, including hepatotoxicity, interstitial lung disease/pneumonitis, QT prolongation, bradycardia, severe visual loss, and embryo-fetal toxicity.
- Alectinib has clinically important hepatotoxicity and other organ toxicities that complicate the safety case for ALK inhibition.
- Mutant-selective inhibitors spare wild type
- ALK inhibition has a generally manageable clinical safety profile but carries clinically important risks, including hepatotoxicity, interstitial lung disease/pneumonitis, QT prolongation, bradycardia, severe visual loss, and embryo-fetal toxicity.
- Alectinib has clinically important hepatotoxicity and other organ toxicities that complicate the safety case for ALK inhibition.
- First-line lorlatinib produces a high burden of metabolic, neurologic and other adverse events, with more grade 3–4 toxicity than crizotinib.
|
| KRAS — MET | Biomarker and target engagement | B wins | gpt-5.6-luna | 98% | MET offers the more complete clinical biomarker chain. MET exon 14 skipping has prospective clinical validation for patient selection, serial METex14 ctDNA depletion is an on-treatment pharmacodynamic biomarker associated with radiographic response, and a validated pMET:total-MET assay provides a direct target-engagement readout. These biomarkers are connected to durable clinical responses and FDA-approved testing and treatment. KRAS has clinically validated tractability and mutation-based selection for KRAS G12C NSCLC, but the supplied dossier does not provide an equally explicit clinically validated target-engagement/PD assay linked to early efficacy; the cited KRAS-GTP and pERK measures are presented as biological readouts rather than established clinical assays. MET IHC and lower-level amplification have important limitations, but the METex14 strategy is specifically supported despite those caveats. | - For KRAS, a prospectively validated clinical target-engagement and pharmacodynamic assay in NSCLC that links on-treatment KRAS pathway suppression to early efficacy, rather than primarily documenting mutation-based treatment selection.
| |
| KRAS — ALK | Patient stratification | B wins | gpt-5.6-luna | 96% | ALK supports the clearer prospectively testable patient-selection hypothesis. ALK rearrangement is a defined molecular subtype with validated testing, regulatory companion-diagnostic infrastructure, and randomized prospective evidence showing greater benefit from ALK inhibition than chemotherapy in ALK-selected NSCLC (C2-EV_BIOMARKER_0001, C2-EV_BIOMARKER_0002, C2-EV_CLINICAL_0001, C2-EV_TRACTABILITY_0006). KRAS is also a strong genomic stratifier, with recurrent mutation-defined NSCLC subsets and prospective sequencing evidence (C1-EV_GENETICS_0002, C1-EV_GENETICS_0003, C1-EV_GENETICS_0004), but the dossier does not provide equivalently direct prospective clinical evidence operationalizing KRAS selection in the stated context. ALK heterogeneity, nonproductive FISH results, and variant-specific sensitivity require assay and subtype refinement but do not remove the established feasibility of prospective stratification (C2-EV_BIOMARKER_0006, C2-EV_TISSUE_0004, C2-EV_MECHANISM_0007). | - A prospective, biomarker-selected KRAS NSCLC trial using a prespecified genomic assay and defined KRAS allelic/co-mutation strata, directly comparable with the established prospective ALK-selection evidence.
| |
| ALK — EGFR | Patient stratification | A wins | gpt-5.6-luna | 96% | ALK supports the clearer prospectively testable selection hypothesis. The ALK-rearranged subgroup is explicitly operationalized through FDA-authorized assays and companion-diagnostic infrastructure, and its predictive value is demonstrated in randomized biomarker-selected NSCLC trials against chemotherapy. EGFR has a biologically defined subgroup based on activating mutations and the dossier describes mutation-directed testing and treatment selection, but the supporting evidence is predominantly catalogue-level and lacks comparable direct prospective clinical-validation or regulatory diagnostic evidence. ALK heterogeneity, subclonality, and occasional nonproductive FISH results qualify implementation but do not negate the established operational selection framework. | - A prospective, biomarker-defined EGFR NSCLC trial or regulatory companion-diagnostic record directly demonstrating operational patient selection, comparable to the ALK-selected randomized trials and authorized assays.
| |
| KRAS — EGFR | Reversibility and therapeutic window | B wins | gpt-5.6-luna | 73% | EGFR offers documented reversible small-molecule TKIs, providing greater ability to titrate or interrupt modulation if toxicity emerges (C2-CAT-ONCO-BIO-115). Although its therapeutic window is narrowed by on-target wild-type EGFR toxicity such as rash and diarrhoea (C2-EV_SAFETY_0001), this reversibility is important under the criterion's combined guardrail. KRAS has a favorable mutant-selective safety rationale because wild-type KRAS is spared (C1-EV_SAFETY_0001), but the clinically validated KRAS-G12C approach is covalent and irreversible (C1-CAT-ONCO-BIO-032), reducing reversibility if adverse effects occur. The overall advantage for EGFR is therefore narrow rather than clear. | - Direct comparative exposure-response data, pharmacokinetic half-life and target-engagement durability, and clinical dose-interruption/dechallenge outcomes for matched KRAS- and EGFR-directed small molecules.
| |
| MET — KRAS | Integrated therapeutic hypothesis | A wins | gpt-5.6-luna | 91% | MET provides the more coherent target–disease–modality hypothesis for this context: recurrent NSCLC driver biology, a defined mechanism of constitutive activation, selective perturbational and PDX evidence, validated pharmacodynamic biomarkers, oral small-molecule tractability, and direct clinical responses in treatment-naive METex14 NSCLC. KRAS has strong driver genetics and established small-molecule tractability, but the supplied KRAS evidence is dominated by broad catalogue material and lacks comparable direct clinical-validation, newly diagnosed, or NSCLC PDX evidence for the proposed setting. MET is not an unqualified choice: activity is alteration-dependent, resistance and toxicity are documented, and the class is competitively crowded. Those limitations reduce differentiation but do not offset the substantially stronger integrated evidence base. | - The Q1–Q19 scores and criterion-weighted composite required by the Q20 guardrail were not supplied
- Direct head-to-head evidence in newly diagnosed, mucus-producing glandular-cell NSCLC PDX models comparing MET inhibition with KRAS inhibition on a chemotherapy backbone
- Clinical validation of a KRAS small-molecule strategy specifically in newly diagnosed NSCLC, beyond the supplied general tractability evidence
- Evidence that either target produces differentiated efficacy over existing targeted competitors in the newly diagnosed setting
| - TCGA-integrated human tumor sequencing identified recurrent somatic MET exon 14 skipping as a likely driver alteration in lung adenocarcinoma.
- MET exon 14 skipping activates MET by removing the juxtamembrane CBL-binding region, impairing receptor ubiquitination and down-regulation.
- Pharmacologic MET inhibition suppresses MET-dependent lung cancer models in vitro and in vivo, with sensitivity linked to MET signaling complexes.
- MET exon 14–altered NSCLC shows clinically meaningful dependence on MET signaling, supporting MET as an essential therapeutic driver in this molecular subset.
- MET is clinically tractable in NSCLC through genotype-selected, oral small-molecule kinase inhibition of MET exon 14–skipping tumors.
- MET exon 14 skipping is a clinically validated predictive biomarker for selective MET inhibition in NSCLC.
- MET exon 14 skipping is clinically validated as a predictive biomarker in metastatic NSCLC through durable responses to selective MET inhibitors.
- Tepotinib has traditional FDA approval for metastatic NSCLC harboring MET exon 14 skipping alterations, with efficacy confirmed in a 313-patient dataset.
- KRAS is a recurrent, activating somatic driver in human lung adenocarcinoma, supporting KRAS inhibition as a genetically defined therapeutic strategy in NSCLC.
- Constitutive RAS-MAPK activation
- Clinically validated tractability
- Co-occurring RAS–MAPK pathway alterations are common in MET exon 14-altered NSCLC and can limit response to MET-directed therapy.
- MET immunohistochemical positivity did not identify a clinically responsive NSCLC population in the phase III METLung trial.
- Adding the MET inhibitor tivantinib to erlotinib failed to improve overall survival in the phase III ATTENTION trial.
- MET perturbational dependency in NSCLC is restricted to a molecularly defined subset rather than being general across lung cancer models.
- Clinical targeting of MET has substantial but monitorable toxicity, including edema, interstitial lung disease, hepatotoxicity, pancreatic toxicity, and embryo-fetal toxicity.
- MET amplification has weaker and threshold-dependent clinical validation than MET exon 14 skipping in NSCLC.
- The METex14 NSCLC market is already competitive, with multiple approved or clinically active MET inhibitors rather than a single differentiated standard.
- MET-targeted competition now includes a non-TKI modality with a broader MET-overexpression biomarker strategy.
- Gibt es kompensatorische Mechanismen, die die Auswirkungen der Hemmung von KRAS abmildern können?
- Ist KRAS üblicherweise essenziell?
- Ist KRAS für das Überleben oder die Proliferation von Krebszellen im Vergleich zu normalen Zellen essenziell?
- Ist KRAS von Resistenzmutationen betroffen?
|
| MET — EGFR | Integrated therapeutic hypothesis | A wins | gpt-5.6-luna | 84% | MET provides the more coherent supplied target–disease–modality chain: recurrent activating genomic alterations, mechanistic activation, selective perturbational and PDX responses, validated pharmacodynamic and predictive biomarkers, and direct clinical efficacy including treatment-naive patients. The conclusion is only narrow because METex14 therapy is already crowded and differentiation is not directly demonstrated; toxicity, alteration-dependent activity, and resistance further limit the hypothesis. EGFR has strong catalogue-level assertions and excellent small-molecule tractability, but the dossier supplies less directly adjudicable, high-trust evidence linking the stated context to a differentiated newly diagnosed medicine. | - Direct evidence in the specified mucus-producing glandular-cell context for either candidate.
- Head-to-head or otherwise directly comparative evidence showing differentiation versus existing targeted medicines in newly diagnosed disease.
- For EGFR, a directly documented, high-trust clinical and PDX evidence chain specific to the stated context rather than catalogue-level summaries.
| - TCGA-integrated human tumor sequencing identified recurrent somatic MET exon 14 skipping as a likely driver alteration in lung adenocarcinoma.
- MET exon 14 skipping activates MET by removing the juxtamembrane CBL-binding region, impairing receptor ubiquitination and down-regulation.
- Pharmacologic MET inhibition suppresses MET-dependent lung cancer models in vitro and in vivo, with sensitivity linked to MET signaling complexes.
- MET exon 14–altered NSCLC shows clinically meaningful dependence on MET signaling, supporting MET as an essential therapeutic driver in this molecular subset.
- MET is clinically tractable in NSCLC through genotype-selected, oral small-molecule kinase inhibition of MET exon 14–skipping tumors.
- MET exon 14 skipping is a clinically validated predictive biomarker for selective MET inhibition in NSCLC.
- MET exon 14 skipping is clinically validated as a predictive biomarker in metastatic NSCLC through durable responses to selective MET inhibitors.
- Tepotinib has traditional FDA approval for metastatic NSCLC harboring MET exon 14 skipping alterations, with efficacy confirmed in a 313-patient dataset.
- The METex14 NSCLC market is already competitive, with multiple approved or clinically active MET inhibitors rather than a single differentiated standard.
- MET-targeted competition now includes a non-TKI modality with a broader MET-overexpression biomarker strategy.
- Clinical targeting of MET has substantial but monitorable toxicity, including edema, interstitial lung disease, hepatotoxicity, pancreatic toxicity, and embryo-fetal toxicity.
- On-target dermatologic and GI toxicity
|
| KRAS — MET | Cell and tissue relevance | B wins | gpt-5.6-luna | 78% | SECOND has the stronger directly disease-relevant tissue evidence: human NSCLC pathology shows MET protein in neoplastic tumor cells, including adenocarcinoma and metastatic tissue, with altered prevalence in matched nodal metastases (C2-EV_TISSUE_0001, C2-EV_TISSUE_0002). However, this evidence is not specific to mucus-producing glandular cells, and MET IHC is substantially discordant with actionable genomic alterations and heterogeneous across tumors (C2-EV_TISSUE_0004, C2-EV_TISSUE_0006). FIRST provides broad or near-ubiquitous KRAS expression claims and strong genomic driver evidence, but does not supply cell-resolved expression, localization, or dysregulation evidence in mucus-producing glandular cells (C1-CAT-ONCO-BIO-047, C1-CAT-ONCO-BIO-058, C1-CAT-ONCO-BIO-097). Therefore SECOND is favored narrowly, not clearly. | - Single-cell or spatial transcriptomic, proteomic, or cell-resolved immunohistochemical evidence specifically quantifying KRAS and MET expression, localization, and dysregulation in mucus-producing glandular cells of human NSCLC, with adjustment for cell-composition shifts and disease stage.
| |
| KRAS — ALK | Integrated therapeutic hypothesis | B wins | gpt-5.6-luna | 86% | ALK provides the more coherent integrated target–disease–modality hypothesis for a differentiated medicine in newly diagnosed NSCLC. The dossier links a causal EML4–ALK fusion to constitutive kinase signaling, selective perturbational dependence, a validated biomarker, oral small-molecule tractability, and multiple randomized first-line clinical studies showing superiority to chemotherapy with systemic and intracranial benefit (C2-EV_GENETICS_0001, C2-EV_MECHANISM_0002, C2-EV_PERTURBATION_0001, C2-EV_BIOMARKER_0002, C2-EV_CLINICAL_0001, C2-EV_CLINICAL_0004). Its main limitation is competition and heterogeneous or bypassable dependence. KRAS has strong genetic-driver evidence and clinically validated mutant-selective small-molecule tractability, especially for G12C (C1-EV_GENETICS_0001, C1-EV_GENETICS_0002, C1-EV_TRACTABILITY_0001), but the dossier provides less direct evidence for the stated newly diagnosed setting, chemotherapy backbone, and exact cell/model context, while documenting substantial allelic, contextual, and resistance heterogeneity. The conclusion is therefore based on stronger coherence and translational validation for ALK, not on absence of KRAS activity in understudied contexts. | - Direct comparative in vivo evidence in the stated patient-derived xenograft model using mucus-producing glandular NSCLC cells, in newly diagnosed disease, with chemotherapy backbone, including a head-to-head assessment of KRAS versus ALK target engagement and tumor response.
- For KRAS, randomized first-line clinical evidence in newly diagnosed KRAS-mutant NSCLC comparable to the extensive first-line ALK evidence is missing.
- For ALK, direct evidence that the specified mucus-producing glandular-cell context is representative of the ALK-positive models is missing.
| |
| MET — EGFR | Clinical validation | A wins | gpt-5.6-luna | 98% | MET has direct, indication-relevant clinical validation in biomarker-defined NSCLC: two selective small-molecule MET inhibitors have regulatory approval for MET exon 14 skipping disease, with prospective molecular selection, substantial response rates, durable responses, and pharmacodynamic ctDNA evidence associated with radiographic response (C1-EV_CLINICAL_0001, C1-EV_CLINICAL_0002, C1-EV_CLINICAL_0003, C1-EV_BIOMARKER_0001, C1-EV_BIOMARKER_0002). The dossier also documents small-molecule tractability and a second approved agent (C1-EV_TRACTABILITY_0002, C1-EV_TRACTABILITY_0004). EGFR has strong mechanistic and genetic support and a generic claim of multiple approved TKIs (C2-EV_TRACTABILITY_0001), but the supplied dossier does not provide comparable indication-specific clinical trial datasets, dose/exposure details, endpoint results, or regulatory evidence tied to the candidate in the stated context. MET's negative studies concern broad or poorly selected MET populations and do not negate the biomarker-defined METex14 validation, consistent with the guardrail (C1-EV_MECHANISM_0006, C1-EV_MECHANISM_0007, C1-EV_BIOMARKER_0004, C1-EV_CLINICAL_0006). | - For EGFR, indication-specific completed clinical trial and regulatory evidence with prospectively defined NSCLC population, molecule identity and quality, dose and exposure, comparator or trial design, endpoints, response durability, target-engagement or proof-of-mechanism data, and coded stoppage reasons.
- For a direct modality comparison, matched evidence under the stated lung NSCLC context, including dose/exposure and endpoint-level results for EGFR-directed molecules.
| |
| EGFR — KRAS | Modality–direction–exposure fit | Abstained | gpt-5.6-luna | 97% | Both candidates have evidence of small-molecule modality precedent: approved or clinically validated EGFR TKIs for EGFR [C1-EV_TRACTABILITY_0001] and mutant-selective covalent KRAS inhibitors [C2-EV_TRACTABILITY_0001]. However, the criterion requires fit between the biological modulation requirement and tissue exposure, and the dossiers do not provide candidate-specific PK/PD, delivery, biodistribution, or target-turnover data for the stated lung, mucus-producing glandular-cell, PDX context. The available evidence therefore establishes tractability but not comparative modality–direction–exposure fit. | - Candidate-specific PK/PD data linking small-molecule exposure to depth and duration of EGFR or KRAS inhibition in lung mucus-producing glandular cells, including target engagement and downstream modulation in the stated PDX model.
- Delivery, biodistribution, and measured tumor-cell exposure in the lung PDX context.
- Target-turnover or recovery data establishing the required dosing interval and duration of inhibition for either target.
| – |
| MET — KRAS | Clinical validation | A wins | gpt-5.6-luna | 98% | MET has direct, indication-relevant clinical validation in NSCLC: prospective biomarker-selected phase 2 data show durable responses with capmatinib and tepotinib, matched molecular response supports target engagement, and both agents have FDA approval for MET exon 14-skipping metastatic NSCLC (C1-EV_CLINICAL_0001, C1-EV_CLINICAL_0002, C1-EV_CLINICAL_0003, C1-EV_BIOMARKER_0001, C1-EV_BIOMARKER_0002). The evidence also includes a second selective molecule and clinically characterized toxicity, strengthening interpretation across molecules and exposure contexts (C1-EV_SAFETY_0002, C1-EV_SAFETY_0005). Negative MET trials mainly evaluated broad or poorly predictive populations and do not invalidate the biomarker-defined result (C1-EV_MECHANISM_0006, C1-EV_MECHANISM_0007, C1-EV_BIOMARKER_0004). KRAS has evidence of clinical tractability and a catalogue claim of approved G12C inhibitors in NSCLC, but the dossier does not provide comparable indication-specific trial results, endpoint data, dose/exposure details, target-engagement data, or trial-design evidence. Therefore MET has substantially stronger clinical validation for the stated NSCLC context. | - For KRAS, indication-specific NSCLC clinical validation with completed-trial design, dose and exposure, prospectively selected population, target-engagement or pharmacodynamic data, and clinical endpoints such as response rate and duration of response.
- For KRAS, direct dossier evidence describing the specific approved NSCLC indication, trial population, endpoint results, and coded trial-stoppage or failure reasons.
| |
| ALK — EGFR | Biomarker and target engagement | A wins | gpt-5.6-luna | 98% | ALK has explicit regulatory validation as a predictive biomarker, FDA-authorized testing and companion-diagnostic infrastructure, biomarker-selected randomized evidence against platinum-pemetrexed chemotherapy in untreated advanced NSCLC, and longitudinal plasma mutation kinetics consistent with target engagement and pharmacodynamic monitoring. EGFR has biologically plausible and clinically relevant mutation claims, but the supplied dossier does not provide comparably specific validated-assay, prospective pharmacodynamic, or biomarker-selected efficacy evidence. ALK therefore offers the more credible clinical biomarker strategy for the stated criterion. | - For EGFR, a directly documented FDA-authorized clinical assay for the relevant activating NSCLC mutations, prospective pharmacodynamic or target-engagement measurements, and a biomarker-selected randomized comparison demonstrating an early efficacy signal are not supplied with comparable specificity.
| |
| KRAS — MET | Safety and essentiality | A wins | gpt-5.6-luna | 72% | KRAS has broad normal-tissue expression and important physiological essentiality, including developmental lethality after complete knockout, so its on-target inhibition risk is not negligible. However, the supplied safety evidence indicates that mutant-selective small-molecule inhibition can spare wild-type KRAS, providing a mechanism for a wider safety margin. MET dependence is restricted to biomarker-defined subsets, but clinical MET-inhibitor class evidence documents substantial toxicity, including edema, interstitial lung disease/pneumonitis, hepatotoxicity, pancreatic toxicity, and embryo-fetal toxicity. Because the MET toxicity evidence is clinically direct but not fully separable into on-target versus molecule/modality-specific effects, the advantage for KRAS is narrow rather than clear. | - Direct human KRAS and MET loss-of-function phenotypes, comparable pan-essentiality scores, and systematic normal-tissue expression breadth data for both targets.
- Target-specific attribution separating MET on-target toxicity from molecule-specific or modality-specific toxicity.
| |
| MET — KRAS | Resistance and escape liability | B wins | gpt-5.6-luna | 66% | SECOND (KRAS) is judged less vulnerable, but only narrowly. MET has direct evidence that co-occurring RAS–MAPK alterations can cause primary resistance or reduce the magnitude and duration of response to MET-directed therapy, and that MET/HGF signaling can bypass EGFR inhibition; these risks are mitigated by genomic biomarker selection and combination strategies such as combined EGFR plus MET inhibition (C1-EV_GENETICS_0005, C1-EV_MECHANISM_0004, C1-EV_MECHANISM_0005). KRAS also has substantial escape liability through feedback-mediated RTK–SHP2–SOS–RAS reactivation, alternative survival signaling, and acquired on-target mutations or KRAS allele changes (C2-CAT-ONCO-BIO-004, C2-CAT-ONCO-BIO-013, C2-CAT-ONCO-BIO-063). However, KRAS is a more central downstream signaling hub and mutant-selective inhibition provides a biomarker-defined intervention, whereas the MET dossier contains stronger direct evidence of pathway bypass and baseline RAS–MAPK-mediated resistance. KRAS resistance risk could be mitigated by allele-specific genomic biomarker selection and combinations targeting RTK/SHP2/SOS1 or downstream MAPK feedback, but the supplied evidence does not establish a large margin over MET. | - Head-to-head, clinically matched NSCLC resistance datasets comparing MET inhibition with allele-specific KRAS inhibition, including time to resistance, resistance frequency, and clonal evolution under comparable treatment settings.
| |
| EGFR — MET | Competitive differentiation | B wins | gpt-5.6-luna | 73% | MET offers the stronger, though narrow, differentiation case because the dossier explicitly identifies chemotherapy as a non-targeted background option and supports biomarker-selected METex14 therapy with substantial responses in treatment-naive patients, including intracranial activity (C2-EV_COMPETITIVE_0003, C2-EV_CLINICAL_0001, C2-EV_CLINICAL_0002). However, the same dossier documents a crowded METex14 competitive field and additional non-TKI competition, limiting distinctiveness between MET agents (C2-EV_COMPETITIVE_0005, C2-EV_COMPETITIVE_0006). The EGFR dossier establishes target tractability and clinical relevance but does not supply comparably direct evidence of differentiation over chemotherapy in newly diagnosed NSCLC or a similarly explicit competitor-pipeline analysis. This is a scientific differentiation judgment, not a market-attractiveness assessment. | - Direct comparative evidence in newly diagnosed NSCLC showing incremental benefit over chemotherapy for each target, preferably in the specified PDX/in-vivo context and with a defined competitor landscape for EGFR.
| |
| MET — EGFR | Competitive differentiation | A wins | gpt-5.6-luna | 78% | MET offers the stronger, though narrow, evidence-supported differentiation case. The dossier explicitly identifies chemotherapy and immunotherapy as non-targeted background options in METex14-positive NSCLC and reports substantial activity in treatment-naive patients, supporting a biomarker-selected scientific differentiation over chemotherapy rather than merely broad pathway inhibition (C1-EV_COMPETITIVE_0003, C1-EV_COMPETITIVE_0001). However, this advantage is weakened by an already crowded MET landscape with multiple approved or clinically active inhibitors and a non-TKI competitor (C1-EV_COMPETITIVE_0002, C1-EV_COMPETITIVE_0005, C1-EV_COMPETITIVE_0006). EGFR has strong tractability and multiple approved TKI generations, establishing competition, but the supplied dossier does not provide a comparable criterion-specific analysis of differentiation over chemotherapy in newly diagnosed NSCLC (C2-EV_TRACTABILITY_0001). This is a narrow verdict based on greater direct competitive-differentiation evidence for MET, not a conclusion about market attractiveness or overall scientific merit. | - Direct randomized or comparative evidence in newly diagnosed NSCLC showing MET or EGFR benefit over chemotherapy.
- For EGFR, a criterion-specific current treatment-landscape, unmet-need, and competitor-pipeline analysis addressing differentiation over chemotherapy.
- Head-to-head evidence demonstrating differentiation between MET- or EGFR-directed agents and the existing targeted-treatment standard.
| |
| KRAS — MET | Evidence robustness (meta-evidence) | B wins | gpt-5.6-luna | 93% | MET has the more robust meta-evidence package. Its dossier explicitly identifies multiple shared cohorts and prevents treating dependent reports as independent, while also documenting contradictory findings and negative clinical, biomarker, tissue, and perturbation results (C2-EV_MECHANISM_0006, C2-EV_MECHANISM_0007, C2-EV_BIOMARKER_0004, C2-EV_PERTURBATION_0004, C2-EV_PERTURBATION_0005, C2-EV_PERTURBATION_0006). It also spans distinct source families including genomic cohorts, mechanistic studies, randomized trials, phase 2 trials, regulatory records, and patient-derived models. The KRAS dossier contains supporting human-genomic and mechanistic claims but does not supply comparable provenance metadata, explicit sample-overlap analysis, replication status, or systematic negative-result availability. This is a metadata robustness advantage for MET, not a judgment based on the larger number of MET items. | - For KRAS, explicit provenance metadata, source-family structure, cohort-overlap flags, replication status, and systematically reported negative-result availability comparable to the MET dossier
- For both candidates, a complete publication-level audit identifying all shared cohorts, unpublished or null studies, and independent replication datasets
| |
| EGFR — ALK | Human genetics and causal evidence | B wins | gpt-5.6-luna | 88% | SECOND has stronger supplied human genetic evidence for a causal role in NSCLC. The EML4–ALK fusion is directly identified in human NSCLC, shown to transform cells, and independently confirmed as a recurrent ALK rearrangement in comprehensive lung adenocarcinoma profiling, with a clear direction of effect and high-confidence gene assignment (C2-EV_GENETICS_0001, C2-EV_GENETICS_0002, C2-EV_GENETICS_0003). EGFR has strong supplied evidence that specific activating somatic mutations are causal in NSCLC (C1-EV_GENETICS_0001), plus catalogue-level claims about rare germline susceptibility and SNP associations, but the dossier does not provide comparable independent human genetic study detail or replication. The ALK evidence is not fully comprehensive under the requested GWAS/QTL/rare-variant dimensions, but it is substantially more direct, specific, and independently corroborated for NSCLC causality. | - For both candidates, no GWAS fine-mapping, L2G scores, disease-relevant molecular-QTL colocalisation, rare-variant burden analysis, quantitative allelic series, or ancestry-diverse replication is supplied
- For EGFR, independent human NSCLC genetic studies with variant-to-gene assignment, effect sizes, replication, and ancestry diversity are not supplied
- For ALK, independent ancestry-diverse replication and quantitative human genetic analyses beyond recurrent somatic fusion detection are not supplied
| |
| MET — ALK | Resistance and escape liability | A wins | gpt-5.6-luna | 78% | MET is judged less vulnerable, but only narrowly. MET has documented escape through pre-existing RAS–MAPK alterations and bypass signaling, so it is not resistant to biological escape. However, ALK has stronger and more direct evidence of acquired resistance liability: repeat-biopsy studies identify both on-target and non-ALK-dependent resistance, sequential inhibitors select compound ALK mutations in clinical samples, and CRISPR studies identify EGFR/MAPK, PI3K/AKT/mTOR, and KEAP1-mediated bypass routes. Combination and biomarker strategies can mitigate these risks: for MET, genotype- and amplification-level selection should avoid weak MET dependence, and combined EGFR/MET inhibition can overcome some EGFR-context bypass mechanisms; for ALK, resistance-mutation profiling can guide lorlatinib or subsequent-generation inhibitors, while EGFR blockade can resensitize selected MIG6/ERRFI1-loss models. These mitigation strategies do not eliminate the greater documented acquired-resistance burden for ALK. | - A matched, head-to-head longitudinal clinical comparison of acquired resistance frequency, time to resistance, and resistance mechanisms in biomarker-selected METex14 NSCLC versus ALK-fusion NSCLC under comparable targeted-treatment sequences.
| |
| MET — EGFR | Biomarker and target engagement | A wins | gpt-5.6-luna | 98% | MET offers the more credible clinical biomarker strategy. MET exon 14 skipping was prospectively assessed using tissue RNA or plasma cfDNA/NGS and was associated with clinical responses, while serial METex14 ctDNA depletion provided an on-treatment pharmacodynamic signal associated with radiographic response (C1-EV_BIOMARKER_0001, C1-EV_BIOMARKER_0002). A validated pMET:total-MET assay provides a direct biochemical target-engagement readout (C1-EV_BIOMARKER_0005), and FDA approvals explicitly tie METex14 testing to treatment selection (C1-EV_CLINICAL_0002, C1-EV_CLINICAL_0003). The strategy has recognized limitations, including failure of nonspecific MET IHC selection and assay/tumor heterogeneity (C1-EV_BIOMARKER_0004, C1-EV_BIOMARKER_0006), but these limitations reinforce the need for genotype-defined selection rather than eliminate the validated strategy. The EGFR dossier contains evidence of biological and therapeutic tractability, including approved small-molecule TKIs (C2-EV_TRACTABILITY_0001), but does not supply comparable direct evidence for a specific clinical biomarker assay, target-engagement pharmacodynamic assay, serial biomarker response, or early efficacy signal in NSCLC. Therefore, MET is clearly better supported for the full biomarker-and-target-engagement criterion. | - For EGFR, prospective clinical evidence linking a defined clinical assay to patient selection, target-engagement pharmacodynamics, serial biomarker response, and an early efficacy signal in NSCLC; specifically, a validated clinical EGFR biomarker/PD assay dataset and corresponding regulatory precedent.
| |
| EGFR — MET | Integrated therapeutic hypothesis | B wins | gpt-5.6-luna | 68% | SECOND is narrowly stronger because MET has a more directly integrated, biomarker-defined therapeutic hypothesis in NSCLC: MET exon 14 skipping is a recurrent genomic driver, has mechanistic support, produces selective responses in MET-dependent PDX models, and is prospectively linked to clinical response and regulatory-approved small-molecule therapy (C2-EV_GENETICS_0002, C2-EV_MECHANISM_0002, C2-EV_PERTURBATION_0002, C2-EV_BIOMARKER_0001, C2-EV_CLINICAL_0001, C2-EV_CLINICAL_0003). Treatment-naive activity further aligns with the stated patient segment (C2-EV_SAFETY_0002). FIRST also has exceptionally strong causal and tractability evidence for EGFR-mutant NSCLC, including replicated driver mutations, approved small-molecule inhibitors, and preclinical validation (C1-EV_GENETICS_0001, C1-EV_TRACTABILITY_0001, C1-CAT-ONCO-BIO-010). However, the MET hypothesis is more explicitly biomarker-treatment-linked and supported by patient-derived models in the supplied dossier. The margin remains narrow because MET dependence is restricted to molecular subsets, MET amplification is threshold-dependent, assay heterogeneity is substantial, and the class is already crowded (C2-EV_PERTURBATION_0004, C2-EV_TISSUE_0004, C2-EV_CLINICAL_0006, C2-EV_COMPETITIVE_0005, C2-EV_COMPETITIVE_0006). EGFR likewise has on-target toxicity and no supplied evidence specific to the stated cell type or chemotherapy-backbone setting (C1-EV_SAFETY_0001). | - Direct, head-to-head in vivo PDX comparison of EGFR versus MET inhibition in newly diagnosed NSCLC with chemotherapy backbone, specifically in mucus-producing glandular cells.
- Evidence linking either target's activity to the stated mucus-producing glandular-cell context and demonstrating differentiation over existing standard targeted medicines.
- The Q1–Q19 criterion-weighted composite and posterior required by the Q20 guardrail are not supplied.
| |
| KRAS — EGFR | Perturbational evidence | Abstained | gpt-5.6-luna | 93% | Both dossiers mention published or available tool-compound studies involving PDX or xenograft models, but neither supplies sufficiently specific, outcome-level PDX evidence demonstrating reproducible reversal of disease-relevant phenotypes. The KRAS xenograft result is not clearly identified as patient-derived, and the EGFR dossier likewise provides general PDX claims without model-specific results. Therefore the evidence is insufficient for a criterion-specific comparison. | - Candidate-specific, outcome-level PDX data showing that KRAS or EGFR perturbation reverses disease-relevant phenotypes, with perturbation details, magnitude and durability of response, replication across independent PDX models, and preferably orthogonal genetic/chemical validation or rescue experiments.
| |
| ALK — EGFR | Clinical validation | A wins | gpt-5.6-luna | 98% | ALK has extensive, indication-relevant clinical validation in NSCLC: multiple randomized phase 3 trials in molecularly selected patients, including treatment-naive populations, demonstrate improved progression-free survival and intracranial control versus active comparators, with regulatory approvals and an established small-molecule modality. The dossier also supplies a clear predictive biomarker and a defined clinical dosing example for lorlatinib. EGFR has evidence of approved small-molecule tractability and a validated driver hypothesis, but the supplied dossier does not provide comparable completed randomized clinical-trial results, endpoint data, dose/exposure information, or coded trial outcomes. The ALK evidence is therefore substantially stronger despite the limitation that several ALK trials concern advanced rather than the exact contextual disease setting and despite the lack of an unadjusted overall-survival benefit in PROFILE 1014. | - For EGFR, randomized indication-relevant clinical trial records with explicit comparator, dose, exposure, population selection, endpoint results, and completed-trial status are not supplied; such evidence would most affect the comparison.
- For both candidates, directly comparable dose/exposure and pharmacodynamic target-engagement data from the cited pivotal trials are incompletely supplied.
| - ALK is clinically validated as a predictive therapeutic biomarker in advanced NSCLC by randomized phase 3 evidence showing superior efficacy of ALK inhibition over platinum-pemetrexed chemotherapy.
- Second-generation ALK inhibition provides superior systemic and intracranial disease control compared with first-generation therapy in untreated advanced ALK-positive NSCLC.
- Brigatinib demonstrated superior first-line efficacy, including intracranial activity, versus crizotinib in ALK-positive advanced NSCLC.
- Lorlatinib provides durable long-term clinical benefit in treatment-naive advanced ALK-positive NSCLC, with sustained systemic and intracranial disease control.
- Ensartinib provides regulatory-confirmed clinical validation of ALK as a treatment target in previously untreated locally advanced or metastatic NSCLC.
- In untreated advanced ALK-positive NSCLC, biomarker-selected patients derived substantially greater benefit from ALK inhibition with crizotinib than from platinum-pemetrexed chemotherapy.
- ALK is clinically tractable with an established oral small-molecule kinase-inhibitor modality in metastatic NSCLC.
- First-line ALK inhibition produces strong systemic and intracranial clinical efficacy, supporting high tractability of ALK in NSCLC.
- The pivotal PROFILE 1014 trial did not demonstrate a statistically significant overall-survival benefit for first-line crizotinib in the unadjusted analysis.
- ALK inhibition has a generally manageable clinical safety profile but carries clinically important risks, including hepatotoxicity, interstitial lung disease/pneumonitis, QT prolongation, bradycardia, severe visual loss, and embryo-fetal toxicity.
- Alectinib has clinically important hepatotoxicity and other organ toxicities that complicate the safety case for ALK inhibition.
- First-line lorlatinib produces a high burden of metabolic, neurologic and other adverse events, with more grade 3–4 toxicity than crizotinib.
|
| KRAS — ALK | Resistance and escape liability | B wins | gpt-5.6-luna | 68% | Applying the reversed direction, ALK (SECOND) is judged less vulnerable, but only narrowly. Both targets have substantial escape liability. KRAS inhibition is exposed to pathway reactivation through RTK–SHP2–SOS–RAS feedback, parallel survival pathways, and acquired KRAS secondary mutations or allele amplification (C1-CAT-ONCO-BIO-004; C1-CAT-ONCO-BIO-013; C1-CAT-ONCO-BIO-063). ALK also has bypass resistance through EGFR–MAPK and PI3K/AKT/mTOR, KEAP1-mediated resistance, and acquired on-target compound mutations, including clinically observed compound mutations after sequential TKI treatment (C2-EV_PERTURBATION_0002; C2-EV_PERTURBATION_0005; C2-EV_TRACTABILITY_0007; C2-EV_BIOMARKER_0004). However, ALK has stronger direct clinical and translational evidence for a durable, biomarker-defined dependency and a developed mitigation framework: fusion confirmation and longitudinal resistance genotyping, sequential next-generation inhibitors with activity against major single resistance mutations, and combination strategies such as EGFR blockade for EGFR-bypass resistance (C2-EV_COMPETITIVE_0004; C2-EV_BIOMARKER_0003; C2-EV_BIOMARKER_0005). Thus ALK is the less escape-vulnerable target on the supplied evidence, while explicitly recognizing that combination and biomarker strategies are needed to mitigate its residual risk. For KRAS, allele-specific biomarker selection and combinations targeting SHP2, SOS1, RTKs, or downstream pathways would similarly be needed, but the dossier does not establish comparable clinical mitigation of its broader feedback and bypass liability. | - A matched, head-to-head NSCLC clinical comparison of acquired-resistance frequency, time to resistance, and bypass mechanisms for KRAS inhibition versus ALK inhibition in biomarker-selected, treatment-naive patients.
| |
| MET — EGFR | Reversibility and therapeutic window | A wins | gpt-5.6-luna | 72% | MET has the stronger documented therapeutic-window evidence: an explicitly described manageable on-target profile and extensive clinical use of orally administered small-molecule inhibitors, supporting titration and interruption (C1-EV_SAFETY_0001, C1-EV_TRACTABILITY_0002, C1-EV_TRACTABILITY_0003). However, this advantage is narrow because MET inhibition also has serious and potentially fatal pulmonary, hepatic, and other toxicities (C1-EV_SAFETY_0004, C1-EV_SAFETY_0006, C1-EV_SAFETY_0007). EGFR has clear small-molecule precedent, including reversible and irreversible inhibitor classes (C2-CAT-ONCO-BIO-115, C2-EV_TRACTABILITY_0001), but its dossier documents on-target rash and diarrhea that narrow the window (C2-EV_SAFETY_0001) and does not provide comparable clinical exposure-response or interruption/reversibility data. Thus FIRST is favored narrowly, not clearly. | - Direct comparative exposure-response data and clinically characterized dose-interruption/rechallenge outcomes for MET versus EGFR inhibitors
- Human pharmacokinetic half-life and duration-of-target-inhibition data for the relevant agents
- Comparative evidence distinguishing reversible interruption from persistent pharmacodynamic effects, particularly for irreversible EGFR inhibitors
| |
| MET — EGFR | Modality–direction–exposure fit | A wins | gpt-5.6-luna | 72% | Both targets have strong small-molecule modality precedent and inhibition is directionally supported. MET has the stronger supplied fit evidence because selective oral MET TKIs are clinically established, MET inhibition has shown antitumor activity in MET-amplified lung-cancer PDX models, and a validated pMET:total-MET assay supports direct target-engagement measurement; METex14 biology also directly documents impaired receptor turnover. EGFR likewise has multiple approved small-molecule TKI generations and reported preclinical xenograft/PD use, but its dossier is dominated by general catalogue summaries and lacks target-specific turnover, quantitative PK/PD, delivery, or biodistribution evidence. The advantage is therefore narrow, not clear. | - Direct comparative PK/PD exposure data for the proposed small molecules in the specified mucus-producing glandular-cell NSCLC PDX context, including lung-tumor biodistribution, duration of target suppression, and exposure relative to the depth of inhibition required.
- EGFR-specific target-turnover data and direct PK/PD or biodistribution data in the specified tissue context.
| |
| KRAS — EGFR | Novelty, crowding, and strategic optionality | Abstained | gpt-5.6-luna | 97% | Both dossiers provide evidence of established small-molecule tractability and broad biological relevance, but they do not provide the competitor-pipeline-by-phase or patent-landscape evidence required to judge strategic optionality relative to crowding. The supplied material therefore cannot support a defensible comparison. Crowding should be reported as a biological-confidence signal and a strategic constraint, but its comparative magnitude is not documented here. | - Comparable competitor pipelines for KRAS and EGFR in NSCLC, including development phase, indications, and differentiation.
- Candidate-specific patent landscapes, including composition-of-matter, biomarker, combination, and freedom-to-operate information.
- Direct comparative evidence of indication-expansion potential and platform leverage beyond the general statements that both targets have activity across multiple tumor types and clinically validated small-molecule tractability.
| – |
| KRAS — ALK | Disease biology and mechanism coherence | B wins | gpt-5.6-luna | 88% | Both targets have biologically coherent evidence. KRAS has strong human genomic support as an activating lung-cancer driver and a coherent constitutive RAS-MAPK mechanism. ALK is better supported for this criterion by convergent causal evidence: recurrent EML4-ALK fusions transform cells, human tumor pathology localizes the alteration to malignant lung epithelial tumors, pathway studies connect ALK to downstream survival signaling, and direct ALK knockdown selectively reduces viability and clonogenic growth in ALK-positive NSCLC models. ALK also has genetically engineered in vivo mechanistic support. The ALK evidence is not uniform across fusion variants or models, so the advantage is narrow rather than clear; this is a mechanistic-coherence judgment and does not rely on ALK's separate druggability or clinical-validation evidence. | - Direct, high-quality inhibition or genetic-dependence experiments for KRAS performed specifically in mucus-producing glandular NSCLC cells and matched NSCLC PDX models.
- Direct comparative evidence in the stated mucus-producing glandular-cell context showing whether ALK or KRAS inhibition more consistently alters tumor pathophysiology in vivo.
| |
| MET — KRAS | Perturbational evidence | A wins | gpt-5.6-luna | 96% | MET has direct, peer-reviewed pharmacologic perturbation evidence in six MET-amplified lung-cancer PDX models showing antitumor responses, plus independent patient-derived model evidence of in-vivo sensitivity to MET inhibition. The MET CRISPR result in A549 cells weakens claims of universal MET essentiality but does not negate biomarker-restricted PDX activity. KRAS has evidence of tractability and general preclinical tool use, but the supplied dossier does not provide a comparably direct KRAS perturbation result in NSCLC PDX models demonstrating reversal of disease-relevant phenotypes. Therefore, MET has stronger reproducible PDX perturbational evidence. | - A KRAS-directed perturbation study in NSCLC patient-derived xenografts reporting reversal of tumor-growth or other disease-relevant phenotypes, with orthogonal genetic or pharmacologic confirmation.
| |
| EGFR — ALK | Biomarker and target engagement | B wins | gpt-5.6-luna | 94% | ALK offers the more credible end-to-end clinical biomarker strategy. ALK rearrangement is explicitly established as a regulatory predictive biomarker requiring an FDA-approved test, has randomized evidence of superior efficacy versus platinum-pemetrexed chemotherapy in biomarker-selected patients, has an established companion-diagnostic infrastructure, and has longitudinal plasma mutation kinetics consistent with pharmacodynamic target engagement (C2-EV_BIOMARKER_0001, C2-EV_BIOMARKER_0002, C2-EV_TRACTABILITY_0006, C2-EV_BIOMARKER_0005, C2-EV_CLINICAL_0001). Direct ALK knockdown selectively affecting ALK-positive models further supports mechanistic engagement and early efficacy interpretation (C2-EV_PERTURBATION_0001). EGFR has strong clinical target precedent and catalogue evidence for mutation testing and pEGFR as a biological readout, but the supplied dossier provides less direct, dossier-level evidence connecting a validated clinical assay, pharmacodynamic measurement, and early efficacy signal than the ALK evidence. ALK’s strategy is not perfect: FISH may detect nonproductive rearrangements and ALK positivity does not guarantee durable dependence (C2-EV_BIOMARKER_0004, C2-EV_BIOMARKER_0006), but these limitations do not outweigh its explicit regulatory, diagnostic, pharmacodynamic, and randomized efficacy chain. | - A direct, head-to-head comparison of validated EGFR and ALK biomarker and pharmacodynamic strategies in the specified newly diagnosed NSCLC, chemotherapy-backbone, in-vivo PDX context.
- For EGFR, a dossier-level regulatory companion-diagnostic record plus prospective clinical pharmacodynamic and early-efficacy data directly linking the selected EGFR genomic alteration to target engagement.
| |
| EGFR — KRAS | Patient stratification | B wins | gpt-5.6-luna | 82% | Both candidates support genomic patient selection in NSCLC. EGFR has a clearly defined mutation-selected subgroup, particularly Exon-19 deletions and L858R, and the dossier states that mutation-specific targeting is clinically realizable. KRAS provides the clearer prospective stratification package: an independent prospective sequencing cohort explicitly identifies KRAS and EGFR as distinct driver-defined subsets, and the dossier further defines an operational KRAS G12C subgroup with clinically validated mutant-selective inhibition. This direct prospective evidence and allele-level operationalization give KRAS a narrow advantage, while the absence of direct prospective biomarker-stratified trial feasibility data prevents a clear-margin verdict. | - Direct prospective biomarker-stratified clinical trial evidence comparing operational EGFR-mutant versus KRAS-mutant selection workflows, including assay eligibility, enrollment feasibility, and treatment outcomes.
| |
| EGFR — ALK | Integrated therapeutic hypothesis | B wins | gpt-5.6-luna | 82% | ALK represents the stronger integrated target–disease–modality hypothesis because the dossier links a causal oncogenic fusion to defined constitutive signaling, selective genetic perturbation effects, a validated predictive biomarker, and multiple randomized treatment-naive NSCLC trials showing superiority over chemotherapy or strong systemic and intracranial control (C2-EV_GENETICS_0001; C2-EV_MECHANISM_0002; C2-EV_PERTURBATION_0001; C2-EV_BIOMARKER_0002; C2-EV_CLINICAL_0001; C2-EV_CLINICAL_0002; C2-EV_TRACTABILITY_0002). The margin is narrow because EGFR also has strong causal and small-molecule tractability evidence (C1-EV_GENETICS_0001; C1-EV_MECHANISM_0001; C1-EV_TRACTABILITY_0001), while ALK has important limitations: heterogeneous pathway dependence, subclonality, acquired resistance, toxicity, and a crowded competitive landscape (C2-EV_MECHANISM_0006; C2-EV_TISSUE_0004; C2-EV_BIOMARKER_0004; C2-EV_COMPETITIVE_0001). The absent Q1–Q19 scores prevent the prescribed posterior-versus-composite coherence review. | - Scores or a criterion-weighted composite for Q1–Q19 are not supplied, so the required Q20 coherence check against that composite cannot be performed.
- Direct evidence in the specified mucus-producing glandular-cell, patient-derived xenograft, newly diagnosed, chemotherapy-backbone in-vivo proof-of-concept setting is not supplied for either candidate.
- A direct head-to-head comparison of EGFR versus ALK under the stated model and treatment context is not supplied.
| |
| KRAS — MET | Time to decisive experiment | Abstained | gpt-5.6-luna | 97% | The dossiers establish that both targets have relevant tools and prior in vivo or PDX evidence (KRAS: C1-CAT-ONCO-BIO-010 and C1-CAT-ONCO-BIO-115; MET: C2-EV_PERTURBATION_0002), but they do not provide the criterion-specific evidence needed to compare speed or cost: organisation-specific capability, model availability, capacity, historical cycle times, or procurement lead times. Existing clinical tractability and biological evidence cannot substitute for learning-efficiency evidence. Therefore neither candidate can be ranked reliably. | - Documented availability and lead times for the required patient-derived xenograft models matching mucus-producing glandular cells and newly diagnosed NSCLC
- Organisation-specific internal in vivo capability, staffing, facility capacity, and access to assay/reagent resources
- Historical cycle times and total costs from study initiation through a decisive in vivo proof-of-concept readout
- Study-readiness details for the chemotherapy-backbone combination and predefined go/no-go decision criteria
| – |
| MET — ALK | Druggability and tractability | B wins | gpt-5.6-luna | 79% | Both targets have strong small-molecule tractability in lung cancer, with approved oral kinase inhibitors and clinically validated biomarker-selected use. ALK is modestly better supported for the specified modality because the dossier provides multiple approved inhibitors, randomized evidence of systemic and intracranial activity, CNS-relevant delivery precedent, and established companion-diagnostic infrastructure. MET also has two approved oral inhibitors and durable responses in METex14-positive NSCLC, but its tractability is more dependent on the precise alteration and amplification level, with weaker evidence for low-level amplification. The margin is narrow because neither dossier supplies direct structural, pocket-quality, or cell-type-specific delivery comparisons, and both have documented acquired-resistance limitations. | - Direct experimental or predicted structures, binding-pocket quality metrics, and comparative biochemical selectivity data for MET versus ALK.
- Direct evidence of small-molecule delivery and target engagement in mucus-producing glandular cells of lung tumors or in the specified patient-derived xenograft context.
- Comparative protein-turnover data relevant to small-molecule inhibition.
| |
| EGFR — MET | Evidence robustness (meta-evidence) | B wins | gpt-5.6-luna | 95% | SECOND (MET) has the more robust meta-evidence package. Its dossier identifies study types and provenance, distinguishes independent genomic, tissue, perturbational, clinical, and biomarker evidence, explicitly flags shared cohorts, and includes multiple contradictory or negative findings, including failed phase III trials and biomarker discordance (C2-EV_MECHANISM_0006, C2-EV_MECHANISM_0007, C2-EV_BIOMARKER_0004, C2-EV_TISSUE_0004, C2-EV_TISSUE_0006). The package is therefore auditable for source dependence and selective reporting, even though several efficacy items share GEOMETRY, VISION, or LUMINOSITY datasets. EGFR has strong biological claims but is dominated by catalogue-style entries with limited provenance metadata, no explicit cohort-overlap or replication map, and little directly documented negative-result availability; the single safety contradiction (C1-EV_SAFETY_0001) does not provide comparable meta-evidence. This decision is based on evidence structure and transparency, not on the larger number of MET findings. | - For EGFR, an explicit provenance and source-family map identifying the underlying studies, cohorts, sample overlap, independent replications, and systematically available negative results is missing.
- For EGFR, independent replication and negative-result evidence directly addressing the stated lung, mucus-producing glandular-cell, PDX, and in-vivo proof-of-concept context is missing.
- For both candidates, a complete publication-bias assessment or prospectively defined evidence search is missing.
| |
| ALK — MET | Reversibility and therapeutic window | B wins | gpt-5.6-luna | 69% | Both candidates use approved oral small-molecule kinase inhibitors, supporting titration and interruption rather than an intrinsically irreversible intervention; the dossiers do not establish a meaningful reversibility advantage for either target. MET has somewhat more favorable stated on-target window evidence, including a generally manageable profile and a 28% grade ≥3 treatment-related adverse-event rate for tepotinib, although important edema, pulmonary, hepatic, pancreatic, and embryo-fetal toxicities remain. ALK has evidence of a less favorable and more heterogeneous toxicity burden, including serious hepatotoxicity, interstitial lung disease/pneumonitis, QT effects, bradycardia, neurologic/metabolic toxicity, and 72% grade 3–4 adverse events with first-line lorlatinib. Because the comparison is indirect and exposure-response, half-life, and recovery data are missing, the advantage is narrow rather than clear. | - Direct comparative exposure-response data and therapeutic-index measurements.
- Drug-specific half-life, duration-of-effect, and pharmacokinetic recovery data after interruption for the ALK and MET small molecules.
- Direct evidence distinguishing reversibility of the relevant kinase inhibitors beyond oral dosing and label-based interruption/discontinuation provisions.
- Human genetic dose-phenotype relationships for therapeutic-window assessment.
| - Manageable on-target profile
- A second selective MET inhibitor confirms both target dependence and a clinically relevant toxicity burden in MET exon 14–skipping NSCLC.
- MET is clinically tractable in NSCLC through genotype-selected, oral small-molecule kinase inhibition of MET exon 14–skipping tumors.
- ALK inhibition has a generally manageable clinical safety profile but carries clinically important risks, including hepatotoxicity, interstitial lung disease/pneumonitis, QT prolongation, bradycardia, severe visual loss, and embryo-fetal toxicity.
- First-line lorlatinib produces a high burden of metabolic, neurologic and other adverse events, with more grade 3–4 toxicity than crizotinib.
- ALK is clinically tractable with an established oral small-molecule kinase-inhibitor modality in metastatic NSCLC.
- Clinical targeting of MET has substantial but monitorable toxicity, including edema, interstitial lung disease, hepatotoxicity, pancreatic toxicity, and embryo-fetal toxicity.
- Tepotinib has clinically important, potentially fatal pulmonary toxicity in MET-altered NSCLC.
- Tepotinib safety includes hepatotoxicity, pancreatic enzyme elevations and embryo-fetal toxicity requiring monitoring or treatment modification.
- Alectinib has clinically important hepatotoxicity and other organ toxicities that complicate the safety case for ALK inhibition.
|
| ALK — KRAS | Translational model validity | A wins | gpt-5.6-luna | 94% | ALK has direct efficacy evidence in patient-derived NSCLC cells and in vivo validation, including a patient-derived ALK-rearranged model, plus randomized efficacy evidence in previously untreated advanced ALK-positive NSCLC. KRAS has strong genetic and mechanistic support, but the dossier does not provide comparable direct efficacy results in clinically relevant newly diagnosed NSCLC patient-derived or genetic models; its cited xenograft efficacy example is not NSCLC. ALK model heterogeneity and variable dependency temper, but do not overturn, the stronger translational efficacy evidence. | - Direct KRAS efficacy results in newly diagnosed NSCLC-relevant patient-derived organoids, primary patient cells, NSCLC PDXs, or genetically engineered lung models, with comparator and treatment-response data.
| |