| MET — KRAS | Druggability and tractability | A wins | gpt-5.6-luna | 82% | MET is modestly more tractable for the specified small-molecule modality because it has multiple approved oral kinase inhibitors and direct clinical efficacy in biomarker-selected NSCLC, providing stronger mature precedent for target access, pharmacology, and delivery to lung tumors. KRAS also has clear small-molecule tractability: the switch-II pocket and mutation-created chemistry support selective covalent binding, and KRAS G12C inhibitors are clinically validated. However, the evidence is strongest for G12C rather than the broader KRAS G12C/G12D target definition, while alternative KRAS pockets and non-G12C alleles remain less validated. MET's advantage is therefore narrow, not clear; neither dossier provides direct pocket-quality metrics or delivery data in the specified mucus-producing glandular cells. | - Direct head-to-head structural pocket-quality metrics for MET and KRAS in the specified lung mucus-producing glandular-cell context.
- Direct delivery and intracellular target-engagement data in lung mucus-producing glandular cells or the specified NSCLC PDX context.
- Clinical or in vivo small-molecule precedent for KRAS G12D comparable to the established KRAS G12C precedent.
| |
| EGFR — KRAS | Druggability and tractability | B wins | gpt-5.6-luna | 78% | Both targets have strong small-molecule precedent. EGFR has multiple approved TKI generations and available tool compounds, supporting broad kinase-pocket tractability (C1-EV_TRACTABILITY_0001; C1-CAT-ONCO-BIO-115). However, the dossier also documents on-target wild-type EGFR dermatologic and gastrointestinal toxicity, indicating a narrower selectivity window (C1-EV_SAFETY_0001). KRAS has a chemically defined Switch-II pocket, a mutation-created reactive cysteine for covalent binding in KRAS G12C, and clinically validated mutant-selective inhibitors that largely spare wild-type KRAS (C2-CAT-ONCO-BIO-032; C2-CAT-ONCO-BIO-032b; C2-CAT-ONCO-BIO-091b; C2-EV_TRACTABILITY_0001). This gives KRAS the stronger evidence for allele-selective small-molecule tractability, despite the allele-specific limitation and the lack of direct delivery data in the specified lung glandular-cell context. The advantage is therefore narrow rather than clear. | - Direct comparative evidence of small-molecule exposure and target engagement in lung mucus-producing glandular cells or lung PDX tissue, including delivery, tissue distribution, and intracellular free-drug concentrations.
- Comparative experimental structures and quantitative pocket-quality metrics for the relevant EGFR mutants and KRAS alleles, especially KRAS G12D versus KRAS G12C.
- Direct comparative evidence of selectivity and therapeutic window in the specified lung cell type.
| |
| EGFR — ALK | Evidence robustness (meta-evidence) | B wins | gpt-5.6-luna | 93% | ALK has the more robust meta-evidence package. Its dossier identifies distinct source families and study designs, including a primary discovery study, independent tumor-genome profiling, large prevalence cohorts, tissue cohorts, perturbation studies, randomized clinical trials, regulatory sources, and explicitly reported contradictory or limiting findings. The contradictions are not hidden: ALK dependence is shown to vary across models, fusion variants, resistance mechanisms, and diagnostic assays (C2-EV_MECHANISM_0006, C2-EV_MECHANISM_0007, C2-EV_BIOMARKER_0004, C2-EV_BIOMARKER_0006). EGFR contains strong biological claims but is dominated by numerous catalogue-style entries with limited provenance, replication-status, cohort-overlap, or negative-result metadata. Under this criterion, ALK therefore has superior source structure and contradiction transparency, despite some documented dependence among individual ALK items. | - For EGFR, item-level source provenance, independent replication mapping, sample-overlap flags, and systematically identified negative or null studies would most change the comparison.
- For both candidates, a formal publication-bias assessment and complete study-level cohort-overlap table are not supplied.
| |
| EGFR — KRAS | Translational model validity | A wins | gpt-5.6-luna | 80% | FIRST has the stronger directly relevant model evidence: the dossier identifies genetically engineered lung models carrying EGFR activating alterations and reports treatment-associated tumor-cell reduction in an EGFR-L858R lung adenocarcinoma model (C1-CAT-ONCO-BIO-117, C1-CAT-ONCO-BIO-117b). SECOND describes appropriate KRAS lung GEMM classes, but its concrete efficacy example is a subcutaneous MiaPaCa-2 xenograft, not NSCLC (C2-CAT-ONCO-BIO-117b). The advantage is narrow because neither dossier supplies strong, explicit evidence in newly diagnosed patient-derived human systems or demonstrates that the reported models specifically represent newly diagnosed disease. | - Direct efficacy results in newly diagnosed NSCLC patient-derived organoids, primary patient cells, ex vivo lung tumor tissue, or humanised/PDX models explicitly representing newly diagnosed disease for both targets.
- For KRAS, a concrete NSCLC-specific in vivo efficacy result is missing; the supplied example in C2-CAT-ONCO-BIO-117b is a MiaPaCa-2 xenograft rather than an NSCLC model.
| |
| ALK — EGFR | Human genetics and causal evidence | A wins | gpt-5.6-luna | 78% | FIRST has more directly documented and independently corroborated human NSCLC genetic evidence: the EML4-ALK fusion is assigned unambiguously to ALK, shown to transform human NSCLC-derived systems, and identified again in comprehensive tumor-genome profiling (C1-EV_GENETICS_0001, C1-EV_GENETICS_0002, C1-EV_GENETICS_0003). SECOND also has strong causal somatic-driver evidence for EGFR Exon19del and L858R (C2-EV_GENETICS_0001), but the dossier provides substantially less study-level detail on replication and does not document the requested population-genetic dimensions. The advantage is narrow because both dossiers primarily contain somatic driver evidence rather than GWAS, QTL, rare-variant, or ancestry-diverse human genetic analyses. | - Neither dossier supplies GWAS or fine-mapping, variant-to-gene or L2G scores, molecular-QTL colocalisation, rare-variant burden analyses, ancestry-diverse replication, or a detailed allelic-series analysis in NSCLC.
- For EGFR, the dossier does not provide study-level details supporting the claimed replication of Exon19del and L858R causal effects.
| |
| MET — KRAS | Translational model validity | A wins | gpt-5.6-luna | 84% | FIRST has a peer-reviewed pharmacologic perturbation study using six MET-amplified lung-cancer patient-derived xenograft models with antitumor responses and mechanistically linked MET signaling, providing direct translational model evidence relevant to NSCLC (C1-EV_PERTURBATION_0002). SECOND provides general catalogue descriptions of KRAS GEMMs and xenografts, but no concrete efficacy result in a clinically relevant NSCLC model; the specific xenograft result cited is in a non-NSCLC model (C2-CAT-ONCO-BIO-117b). The advantage is narrow because neither dossier supplies direct model evidence specifically matched to newly diagnosed disease or mucus-producing glandular cells. | - Direct comparative efficacy data in newly diagnosed NSCLC-relevant human organoids, primary patient cells, humanised or genetic in vivo models—especially models representing mucus-producing glandular cells—for both MET and KRAS.
- For KRAS, a concrete efficacy study in NSCLC patient-derived or genetically engineered models rather than the supplied general catalogue descriptions or a non-NSCLC xenograft.
| |
| KRAS — EGFR | Human natural experiments | Abstained | gpt-5.6-luna | 93% | Neither dossier supplies a qualifying human natural experiment demonstrating both NSCLC efficacy and tolerability for inhibition. KRAS germline disease evidence concerns rare RASopathy-associated variants and does not establish inhibition-direction phenotypes; lifetime genetic exposure therefore does not match the therapeutic direction (C1-CAT-ONCO-BIO-056). EGFR has human germline loss-of-function disease evidence, which is directionally relevant to inhibition, but it indicates severe developmental toxicity rather than a tolerated NSCLC intervention and does not provide a matched human efficacy-and-tolerability natural experiment (C2-CAT-ONCO-BIO-005; C2-CAT-ONCO-BIO-016). The dossiers mention approved pharmacology for both targets, but do not provide the required human exposure-cohort evidence linking efficacy and tolerability; EGFR toxicity is explicitly documented (C2-EV_SAFETY_0001). | - A human NSCLC-focused natural-experiment dataset for each target showing aligned loss-of-function or pharmacological inhibition, with both efficacy and tolerability; specifically, protective-allele or biobank LoF evidence, or a well-characterized long-term exposure cohort with clinical outcomes.
| |
| EGFR — ALK | Human natural experiments | Abstained | gpt-5.6-luna | 96% | The dossiers do not provide the rubric-specified human natural-experiment evidence establishing both efficacy and tolerability for inhibition. EGFR has rare human loss-of-function disease phenotypes, but these indicate substantial consequences of reduced EGFR function and do not establish antitumor efficacy or a tolerable therapeutic inhibition window (C1-CAT-ONCO-BIO-005). Reported EGFR germline NSCLC predisposition is gain-of-function or susceptibility evidence, not a direct human experiment of EGFR inhibition (C1-CAT-ONCO-BIO-016, C1-CAT-ONCO-BIO-056). ALK human germline evidence concerns inherited neuroblastic-tumor susceptibility rather than NSCLC or ALK inhibition (C2-EV_GENETICS_0004), and the germline NSCLC cohort did not establish ALK as an NSCLC predisposition gene (C2-EV_GENETICS_0007). The ALK clinical-treatment items are pharmacological efficacy and safety evidence, but are not the specified long-term drug-exposure natural-experiment evidence. Therefore the criterion cannot be adjudicated without stretching the admissible evidence or ignoring the genetic-direction guardrail. | - A human Mendelian, protective-allele, biobank loss-of-function, or long-term drug-exposure cohort for either target that directly supports NSCLC efficacy of inhibition and tolerability in the same directional context.
| |
| MET — EGFR | Cell and tissue relevance | Abstained | gpt-5.6-luna | 94% | The dossiers provide broad tumor-cell or epithelial expression evidence for both targets, but neither provides cell-resolved evidence specifically in mucus-producing glandular cells. MET has stronger human IHC/pathology evidence in NSCLC tumor cells, while EGFR has mainly general catalogue-level statements about malignant epithelial expression; neither establishes target dysregulation or localisation in the specified critical cell type while controlling for composition and disease stage. Because the criterion is cell-specific and absence of such evidence cannot be treated as a low ranking, the appropriate verdict is ABSTAIN. | - Direct single-cell or spatial transcriptomic, proteomic, or cell-resolved human pathology evidence specifically in lung mucus-producing glandular cells, with control for cell-composition shifts and disease stage, showing target localisation and dysregulation.
| |
| EGFR — ALK | Disease biology and mechanism coherence | B wins | gpt-5.6-luna | 80% | Both targets have biologically coherent oncogenic mechanisms. EGFR has strong evidence that activating mutations are causal drivers and produce ligand-independent kinase signaling into MAPK and PI3K pathways (C1-EV_GENETICS_0001, C1-EV_MECHANISM_0001). ALK is favored narrowly because the dossier provides more convergent, high-quality experimental linkage from a defined human fusion to transformation, in vivo tumor formation, pathway dependence, and selective genetic knockdown effects in ALK-fusion-positive versus negative NSCLC models (C2-EV_GENETICS_0002, C2-EV_MECHANISM_0003, C2-EV_MECHANISM_0004, C2-EV_PERTURBATION_0001, C2-EV_TISSUE_0003). This conclusion is tempered by evidence that ALK dependence varies by cell line, fusion variant, tumor subclone, and genotype (C2-EV_MECHANISM_0006, C2-EV_MECHANISM_0007, C2-EV_TISSUE_0004, C2-EV_PERTURBATION_0006). The decision concerns mechanistic coherence, not druggability or clinical utility. | - Direct, independent perturbation studies of EGFR and ALK inhibition in the specified mucus-producing glandular lung-cell context, ideally using matched patient-derived xenograft models.
- For EGFR, a clearly identified primary study directly linking selective EGFR perturbation to genotype-matched NSCLC tumor-cell dependence is missing from the structured high-trust evidence items.
| |
| ALK — EGFR | Druggability and tractability | A wins | gpt-5.6-luna | 86% | Both targets have strong small-molecule precedent. ALK has especially direct regulatory and clinical evidence for oral kinase inhibition in NSCLC, including systemic and intracranial activity and an established companion-diagnostic infrastructure (C1-EV_TRACTABILITY_0001 through C1-EV_TRACTABILITY_0006). EGFR also has multiple approved TKI generations and documented addressable kinase-site changes plus preclinical tool-compound precedent (C2-EV_TRACTABILITY_0001, C2-CAT-ONCO-BIO-032, C2-CAT-ONCO-BIO-010). ALK is ranked narrowly higher because its dossier supplies more direct, high-quality, lung-specific clinical validation of the specified modality. The comparison remains narrow because neither dossier provides pocket-quality metrics or delivery/target-engagement data specifically in the stipulated mucus-producing glandular cells, and both face acquired-resistance limitations. | - Head-to-head structural and pocket-quality comparison of ALK versus EGFR in the specified mucus-producing glandular lung-cell context.
- Direct biodistribution, intracellular target-engagement, and protein-turnover data in lung mucus-producing glandular cells for each target.
- Comparative selectivity and off-target profiling at clinically relevant exposures.
| |
| KRAS — EGFR | Clinical validation | B wins | gpt-5.6-luna | 78% | SECOND (EGFR) has the stronger indication-relevant clinical validation on the supplied record, but only narrowly. The dossier documents multiple approved generations of small-molecule EGFR TKIs and specifically identifies approval and clinical use for activating EGFR alterations in NSCLC, including exon-19 deletions and L858R (C2-EV_TRACTABILITY_0001, C2-CAT-ONCO-BIO-091b, C2-CAT-ONCO-BIO-046, C2-CAT-ONCO-BIO-088). KRAS also has genuine clinical validation: mutant-selective inhibitors are approved for KRAS-G12C NSCLC, but the supplied evidence describes this as a narrower, mainly previously treated population (C1-EV_TRACTABILITY_0001, C1-CAT-ONCO-BIO-091b). EGFR therefore has greater demonstrated clinical breadth and maturity in NSCLC. However, the dossiers do not provide sufficiently detailed comparative information on dose, exposure, target engagement, trial design, endpoint selection, or population-level outcomes, so the margin is narrow rather than clear. The documented EGFR on-target dermatologic and gastrointestinal toxicity is a countervailing molecule-quality and therapeutic-window consideration, but it does not outweigh the broader clinical validation record (C2-EV_SAFETY_0001). | - Directly comparable indication-matched clinical trial data reporting molecule, dose, achieved exposure, target engagement, trial design, endpoint hierarchy, and biomarker-selected population for KRAS versus EGFR in NSCLC.
| |
| KRAS — MET | Novelty, crowding, and strategic optionality | A wins | gpt-5.6-luna | 68% | KRAS has broader evidenced strategic avenues: mutation-specific targeting is clinically established for G12C, while the dossier describes active expansion toward G12D, G12V, pan-KRAS/pan-RAS strategies, combinations, resistance management, and relevance across multiple tumor types. MET also has platform leverage through selective TKIs and an ADC, but its demonstrated optionality is concentrated in NSCLC biomarker segments and faces explicit crowding from multiple approved or clinically active MET inhibitors plus a non-TKI competitor. Thus FIRST has a narrow optionality edge. Crowding is not netted into biological confidence: MET's competition supports biological validation but reduces strategic differentiation, while a direct KRAS crowding comparison is not supplied. | - A directly comparable competitor pipeline by phase for KRAS versus MET, including active programs in additional indications
- A patent landscape and freedom-to-operate comparison for both targets
- Validated indication-expansion data for KRAS beyond NSCLC and for MET beyond NSCLC
| |
| MET — EGFR | Disease biology and mechanism coherence | A wins | gpt-5.6-luna | 91% | MET has a more experimentally triangulated mechanism: recurrent human METex14 driver alterations are linked to impaired receptor down-regulation and sustained MAPK signaling, and genetic and pharmacologic perturbation selectively affects MET-dependent NSCLC models, including six MET-amplified patient-derived xenografts (C1-EV_GENETICS_0002; C1-EV_GENETICS_0003; C1-EV_MECHANISM_0002; C1-EV_PERTURBATION_0001; C1-EV_PERTURBATION_0002; C1-EV_PERTURBATION_0003). EGFR also has strong mechanistic plausibility and causal activating-mutation evidence, but much of the supplied support is catalogue-level synthesis rather than direct primary perturbation evidence, and the dossier explicitly describes context-dependent dependence (C2-EV_GENETICS_0001; C2-EV_MECHANISM_0001; C2-CAT-ONCO-BIO-031; C2-CAT-ONCO-BIO-070). MET's limitations indicate biomarker-restricted rather than incoherent biology, so they do not outweigh its stronger direct experimental support for inhibition altering NSCLC pathophysiology. | - Directly comparable, primary mechanistic perturbation studies of MET versus EGFR inhibition in the specified mucus-producing glandular NSCLC cell context, including matched patient-derived xenograft models.
| |
| MET — KRAS | Combination potential | Abstained | gpt-5.6-luna | 98% | The dossiers provide target biology, monotherapy validation, and some combination or resistance evidence, but the combination data are directed mainly to other targeted pathways rather than chemotherapy. Neither candidate has direct evidence covering the criterion’s required chemotherapy-specific pathway complementarity, resistance mechanisms, preclinical combination efficacy, and overlapping-toxicity/dose feasibility. Under the rubric and guardrail, there is therefore insufficient evidence to compare them. | - Direct MET-plus-chemotherapy and KRAS-plus-chemotherapy preclinical efficacy studies in relevant NSCLC models, ideally including patient-derived xenografts.
- Evidence identifying chemotherapy resistance mechanisms addressed by either MET or KRAS inhibition in NSCLC.
- Comparative overlapping-toxicity and dose-feasibility assessment for combining each target-directed agent with chemotherapy.
- Pathway-complementarity data specifically linking MET or KRAS inhibition to the chemotherapy backbone.
| – |
| ALK — MET | Clinical validation | A wins | gpt-5.6-luna | 97% | ALK has stronger indication-relevant clinical validation. It is supported by randomized phase 3 evidence in previously untreated ALK-positive advanced NSCLC directly against platinum-pemetrexed chemotherapy, with improvements in progression-free survival, response, symptoms and quality of life, plus multiple later randomized first-line trials and durable intracranial/systemic benefit. The modality is an approved oral small-molecule inhibitor selected by an approved biomarker test. METex14 has meaningful validation through selective-inhibitor phase 2 responses and traditional approvals, including treatment-naive patients, but the dossier provides no randomized first-line comparison against chemotherapy; its strongest evidence is single-arm and biomarker-defined. Negative MET trials largely concern nonselective IHC selection or broad pathway inhibition and therefore do not invalidate selective METex14 targeting, but they reinforce the importance of population and biomarker selection. The dossiers do not provide detailed dose or exposure data sufficient for a finer molecule-quality comparison. | - A randomized, first-line METex14-skipping NSCLC trial directly comparing a selective MET inhibitor with platinum-based chemotherapy, including prespecified endpoint results, dose/exposure adequacy, and pharmacodynamic target-engagement data in the indication-relevant population.
- Clinical validation specifically in mucus-producing glandular NSCLC cells or a matching patient-derived xenograft context.
| |
| ALK — KRAS | Patient stratification | A wins | gpt-5.6-luna | 96% | ALK supports the clearer patient-selection hypothesis. ALK rearrangement is explicitly defined as a predictive biomarker, selected using FDA-authorized assays, and tested prospectively in randomized NSCLC trials against chemotherapy. This directly satisfies the requirement for an operationalizable genomic stratification hypothesis. KRAS has strong evidence for recurrent mutation-defined subsets and established small-molecule tractability, but the dossier provides less direct prospective clinical selection evidence for KRAS in NSCLC and emphasizes allele-, histology-, co-mutation-, and context-dependence. ALK's operational limitations—nonproductive FISH signals, heterogeneity, and histologic variability—warrant assay refinement but do not outweigh its established prospective testing infrastructure. | - For KRAS, a prospective NSCLC trial or regulatory companion-diagnostic evidence showing that a prespecified KRAS allele, especially G12C, prospectively selects patients for benefit from KRAS-directed therapy would most change the answer.
| |
| MET — ALK | Time to decisive experiment | Abstained | gpt-5.6-luna | 98% | The dossiers provide biological and clinical tractability evidence, and MET includes pharmacologic testing in six MET-amplified lung-cancer PDX models (C1-EV_PERTURBATION_0002), while ALK includes in vivo validation in patient-derived NSCLC material (C2-EV_PERTURBATION_0001; C2-EV_PERTURBATION_0002). However, neither dossier supplies the operational evidence required to compare time or cost to a decisive in vivo proof-of-concept experiment: assay/reagent readiness for the specified cell context, model availability and throughput, internal capacity, or historical cycle times. Ranking either candidate would treat missing operational evidence as evidence of faster execution. | - Organization-specific assay and reagent availability for the stated lung mucus-producing glandular-cell PDX experiment
- Validated readiness, throughput, and capacity of the required patient-derived xenograft models for each target
- Historical cycle times and costs from target selection through in vivo proof-of-concept
- Organization-specific internal capability, staffing, and operational constraints
| – |
| MET — ALK | Modality–direction–exposure fit | Tie | gpt-5.6-luna | 75% | Both targets have strong, directionally appropriate evidence for inhibition by an oral small molecule, including established clinical modality precedent and pharmacodynamic or target-engagement readouts. MET has direct pharmacologic activity in MET-dependent NSCLC patient-derived xenografts and validated pMET suppression assays, while ALK has established oral, companion-diagnostic-supported inhibitors and longitudinal ctDNA evidence consistent with on-target inhibition. The dossiers do not provide direct exposure, biodistribution, or target-engagement data in the specified mucus-producing glandular cell context. Biomarker heterogeneity and nonproductive or uneven rearrangement/expression can reduce effective fit for both, but neither candidate is supported as clearly superior on modality-direction-exposure fit. | - Direct PK/PD modelling, target-occupancy or exposure-response data, and delivery/biodistribution measurements in mucus-producing glandular NSCLC cells or the specified patient-derived xenograft context.
| |
| EGFR — KRAS | Perturbational evidence | B wins | gpt-5.6-luna | 68% | KRAS has the clearer dossier-level linkage between direct chemical perturbation, KRAS-mutant NSCLC-relevant models, and stated use of PDX models for target validation (C2-CAT-ONCO-BIO-010), with direct KRAS inhibitor tools identified (C2-CAT-ONCO-BIO-115). EGFR also has a stated body of preclinical PDX validation (C1-CAT-ONCO-BIO-010), but the supplied evidence is less specific about PDX-level phenotype reversal and reproducibility. The margin is narrow because neither dossier provides sufficiently detailed, independent, study-level PDX outcomes or orthogonal perturbations to establish a strong reproducibility advantage. | - Direct, study-level PDX results for both candidates showing tumor regression or another disease-relevant phenotype reversal, with independent PDX models, perturbation details, biomarker-matched cohorts, and orthogonal genetic confirmation.
| |
| EGFR — KRAS | Human genetics and causal evidence | B wins | gpt-5.6-luna | 84% | KRAS has stronger and more independently replicated human tumor-genetic evidence in NSCLC: recurrent activating alleles with an allelic series, a primary lung adenocarcinoma cohort, an independent prospective metastatic sequencing cohort, and a large additional NSCLC cohort showing substantial prevalence and subtype distribution. The mutual exclusivity with EGFR supports distinct driver-defined subsets. EGFR also has evidence for causal activating mutations, but the dossier provides less detailed cohort-level replication and lacks direct variant-to-gene, QTL-colocalisation, GWAS/fine-mapping, and ancestry-diversity evidence. The margin is narrow because the strongest evidence for both candidates is somatic driver genetics rather than the full human-genetics evidence types specified in the question. | - Direct GWAS/fine-mapping or L2G variant-to-gene assignment, disease-relevant molecular-QTL colocalisation, and ancestry-diverse replication for both EGFR and KRAS in NSCLC; the dossiers rely mainly on somatic tumor-sequencing evidence.
| |
| EGFR — ALK | Novelty, crowding, and strategic optionality | A wins | gpt-5.6-luna | 72% | FIRST has the stronger, though narrow, strategic-optionality case because the dossier documents EGFR relevance across multiple cancer indications and multiple addressable modalities, including small-molecule and extracellular targeting approaches (C1-CAT-ONCO-BIO-046; C1-CAT-ONCO-BIO-023; C1-CAT-ONCO-BIO-088). ALK has clear expansion beyond metastatic NSCLC into adjuvant resected disease and strong small-molecule/CNS platform validation (C2-EV_TRACTABILITY_0003; C2-EV_TRACTABILITY_0005), but its current segment is explicitly highly crowded with several preferred first-line inhibitors and continuing fourth-generation competition (C2-EV_COMPETITIVE_0001; C2-EV_COMPETITIVE_0005). That crowding increases biological confidence in ALK while reducing its strategic room. EGFR is also an established, competitively developed target, so the advantage is not clear; the dossiers lack a comparable EGFR competitor-pipeline and patent analysis. No net strategic benefit is assigned solely from biological validation. | - A directly comparable, phase-resolved competitor pipeline and patent-landscape analysis for EGFR and ALK, including freedom-to-operate, exclusivity duration, and expansion programs in indications beyond NSCLC
- Evidence specifically assessing platform fit and indication expansion for the stated mucus-producing glandular-cell and chemotherapy-backbone context
| |
| ALK — KRAS | Druggability and tractability | A wins | gpt-5.6-luna | 96% | ALK is more tractable for the specified small-molecule inhibition modality and lung setting. It has multiple approved oral small-molecule inhibitors, regulatory validation, established companion-diagnostic selection, demonstrated systemic and intracranial activity, and precedent extending into adjuvant NSCLC treatment (C1-EV_TRACTABILITY_0001 through C1-EV_TRACTABILITY_0006). KRAS also has genuine small-molecule tractability, including approved mutant-selective covalent inhibitors and a structurally defined switch-II-pocket strategy, but the strongest evidence is allele-specific to KRAS-G12C and broader KRAS-G12D/pan-KRAS tractability remains less established (C2-EV_TRACTABILITY_0001; C2-CAT-ONCO-BIO-032; C2-CAT-ONCO-BIO-115). Resistance and bypass mechanisms reduce durability for both targets but do not outweigh ALK's broader and more mature modality, delivery, biomarker, and clinical precedent. | - Direct comparative structural pocket-quality metrics, binding affinities, and experimentally resolved structures for ALK versus the specific KRAS allele relevant to the lung mucus-producing glandular-cell context.
- Direct delivery and target-engagement data in patient-derived NSCLC xenografts involving mucus-producing glandular cells for both targets.
- Evidence establishing small-molecule tractability of KRAS-G12D or other non-G12C KRAS alleles at a level comparable to the established G12C precedent.
| |
| EGFR — ALK | Safety and essentiality | B wins | gpt-5.6-luna | 91% | ALK has the more favourable anticipated on-target safety profile. The dossier supports relatively restricted adult expression and viable, fertile homozygous ALK-knockout mice, although with reproductive and neuroendocrine phenotypes (C2-EV_SAFETY_0001, C2-EV_SAFETY_0004). In contrast, EGFR is broadly important in normal epithelia, including lung, skin and gastrointestinal tissues; constitutive knockout causes severe epithelial, developmental and respiratory phenotypes (C1-CAT-ONCO-BIO-098, C1-CAT-ONCO-BIO-102, C1-CAT-ONCO-BIO-114). Direct on-target EGFR inhibition is also associated with dermatologic and gastrointestinal toxicity (C1-EV_SAFETY_0001). ALK inhibitors have important clinical toxicities, including hepatotoxicity, pneumonitis, cardiac, visual, metabolic and neurologic effects (C2-EV_SAFETY_0005, C2-EV_SAFETY_0006, C2-EV_SAFETY_0007), but these are less compelling evidence of broad target essentiality and may include molecule- or modality-specific components. Therefore, despite residual ALK safety concerns, ALK is favoured for anticipated on-target safety. | - Direct comparative human loss-of-function and PheWAS data for EGFR versus ALK.
- Quantitative, matched tissue-expression breadth and pan-essentiality scores for both targets, including the specified lung mucus-producing glandular cells.
- Direct human genetic evidence defining the adult normal-tissue consequences of ALK loss.
| - Favourable on-target safety
- Wild-type ALK appears to have limited essentiality for survival of adult normal tissues, although loss of ALK produces specific reproductive and neuroendocrine phenotypes in mice.
- Welche Rolle spielt EGFR in gesunden Zellen/Geweben?
- Welcher Phänotyp tritt auf, wenn EGFR deletiert wird?
- Welche Gewebe oder Organsysteme sind bei der Maus nach dem Knockout des EGFR-Gens betroffen?
- On-target dermatologic and GI toxicity
- 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.
|
| EGFR — KRAS | Disease biology and mechanism coherence | A wins | gpt-5.6-luna | 72% | Both candidates have strong mechanistic plausibility and evidence that activating alterations can drive NSCLC biology. EGFR has a somewhat more direct and coherent inhibition-linked chain in the supplied dossier: recurrent activating NSCLC mutations are identified as causal, they produce ligand-independent kinase activity with PI3K/MAPK signaling, and functional validation is described through transformation, pathway activation, genetic perturbation, and EGFR-mutant lung tumor models showing response to inhibition. KRAS has comparably strong human-genomic support and a coherent constitutive RAS-MAPK mechanism, but the supplied evidence emphasizes allele-, histology-, and context-dependent dependencies and does not provide an equally specific high-trust perturbation item for the stated cell type and model. This is a narrow advantage for FIRST, not evidence that KRAS lacks a coherent mechanism. | - A head-to-head, allele-matched genetic inhibition and rescue study in mucus-producing glandular NSCLC cells, followed by validation in the specified patient-derived xenograft context, would most change the answer.
| |
| ALK — MET | Druggability and tractability | A wins | gpt-5.6-luna | 78% | Both targets have strong small-molecule tractability in lung cancer, with approved oral kinase inhibitors and clinical activity. ALK is favored narrowly because the dossier provides broader and more mature small-molecule precedent across multiple approved inhibitors, randomized NSCLC trials, companion diagnostics, systemic efficacy, and CNS-penetrant activity, supporting delivery and practical deployment in lung cancer (C1-EV_TRACTABILITY_0001 through C1-EV_TRACTABILITY_0006). MET also has direct regulatory and clinical validation through capmatinib and tepotinib (C2-EV_TRACTABILITY_0001, C2-EV_TRACTABILITY_0002, C2-EV_TRACTABILITY_0004), but its tractability is more dependent on the specific alteration and amplification level, with weaker activity in low-level MET amplification (C2-EV_TRACTABILITY_0007). The dossiers do not provide the requested comparative structural, pocket-quality, selectivity, protein-turnover, or direct lung-delivery measurements, so the margin is narrow rather than clear. | - Comparative experimental structures, binding-pocket quality metrics, and selectivity profiles for ALK versus MET small-molecule inhibitors.
- Direct lung-tissue exposure and target-engagement measurements in the specified mucus-producing glandular cell context.
- Comparative protein-turnover and degradation data relevant to small-molecule inhibition.
| |
| ALK — KRAS | Clinical validation | A wins | gpt-5.6-luna | 98% | ALK has substantially stronger indication-relevant clinical validation. Multiple randomized phase 3 studies in biomarker-selected NSCLC populations demonstrated improved progression-free survival, response, symptom or intracranial outcomes versus chemotherapy or active ALK therapy, and several ALK inhibitors have regulatory approval in NSCLC (C1-EV_CLINICAL_0001, C1-EV_CLINICAL_0002, C1-EV_CLINICAL_0003, C1-EV_CLINICAL_0004, C1-EV_CLINICAL_0005). The PROFILE 1014 evidence directly matches untreated advanced disease and chemotherapy comparison, while the other trials provide replicated clinical proof-of-concept and modality validation. The lack of an unadjusted overall-survival benefit in PROFILE 1014 is a limitation, but extensive crossover makes it weak evidence against ALK target validity (C1-EV_CLINICAL_0006). KRAS has evidence that mutant-selective small-molecule inhibition is clinically tractable and that G12C inhibitors are approved, but the supplied dossier does not provide corresponding NSCLC trial designs, endpoints, dose or exposure data, or target-engagement-linked clinical outcomes; the approval claim alone is insufficient for a comparable adjudication (C2-EV_TRACTABILITY_0001, C2-CAT-ONCO-BIO-091b). Dose and exposure details are also incompletely supplied for ALK, but this limitation does not offset the large difference in indication-specific randomized clinical evidence. | - For KRAS: indication-specific NSCLC trial reports with randomized or otherwise controlled design, prespecified endpoints, enrolled KRAS-mutant population, administered dose, achieved exposure, target-engagement or pharmacodynamic data, and coded reasons for discontinuation or failure.
- For both candidates: directly comparable dose-exposure and target-engagement data for the molecules underlying the cited clinical outcomes, particularly in the specified newly diagnosed, chemotherapy-backbone context.
| |
| KRAS — ALK | Cell and tissue relevance | B wins | gpt-5.6-luna | 79% | ALK has stronger direct cell/tissue-level evidence: ALK rearrangements and ALK protein were detected in human NSCLC tumor tissue by FISH and immunohistochemistry, with localization to pulmonary adenocarcinoma epithelial tumor cells (C2-EV_TISSUE_0002, C2-EV_TISSUE_0006), and the fusion was detected in NSCLC tumor cells (C2-EV_TISSUE_0001). However, these data do not specifically identify mucus-producing glandular cells, and ALK can be subclonal, morphologically heterogeneous, or nonproductive by FISH (C2-EV_TISSUE_0004, C2-EV_TISSUE_0005, C2-EV_BIOMARKER_0006). KRAS has strong human NSCLC mutation evidence but lacks comparable direct localization or expression evidence in the specified cell type; the supplied KRAS tissue evidence instead describes broad/near-ubiquitous expression and notes that increased expression is not a general feature (C1-CAT-ONCO-BIO-047, C1-CAT-ONCO-BIO-058, C1-CAT-ONCO-BIO-097). Thus ALK is favored narrowly, while the criterion-specific cell-type evidence remains incomplete. | - Direct single-cell or spatial transcriptomic, proteomic, or immunohistochemical comparison of KRAS and ALK specifically in human NSCLC mucus-producing glandular cells, with adjustment for cell-composition shifts and disease stage.
- Direct evidence that either target is dysregulated specifically within mucus-producing glandular cells rather than merely present in bulk tumor or other epithelial tumor compartments.
| |
| MET — EGFR | Evidence robustness (meta-evidence) | A wins | gpt-5.6-luna | 86% | MET has explicit provenance-like study descriptions, multiple identifiable source families, documented cohort-dependence flags, independent-looking genomic cohorts, perturbational evidence across cell lines and PDX models, and accessible negative clinical and perturbational findings. The dossier also exposes contradictions rather than presenting only supportive results. EGFR contains extensive catalogue-style claims, but little study-level metadata and no documented sample-overlap, replication, or negative-result structure. Because this is a meta-evidence judgment rather than an evidence-count judgment, MET is favored narrowly; the EGFR dossier is substantially under-characterized rather than proven non-robust. | - For EGFR, study-level provenance metadata identifying source families, cohorts, sample overlap, independent replication, and availability of negative results
- For EGFR, primary experimental or clinical reports supporting the catalogue-answer claims, including replication across independent cohorts or laboratories
- For EGFR, a documented set of contradictory or negative perturbational and clinical findings comparable to the negative-result availability documented for MET
| |
| ALK — MET | Translational model validity | B wins | gpt-5.6-luna | 78% | MET has the stronger directly documented translational-model evidence: selective MET inhibition produced antitumor responses in six MET-amplified lung-cancer PDX models and in six patient-derived EGFR-mutant/MET-amplified models tested in vivo (C2-EV_PERTURBATION_0002, C2-EV_PERTURBATION_0003). Its clinical activity in treatment-naive METex14 NSCLC further supports relevance to newly diagnosed disease (C2-EV_SAFETY_0002, C2-EV_CLINICAL_0001). ALK has relevant genetically engineered mouse-model and patient-derived evidence, but the dossier provides less explicit detail on the patient-derived in vivo model and does not establish comparable newly diagnosed model provenance (C1-EV_MECHANISM_0005, C1-EV_TISSUE_0003). The margin is narrow because both targets have clinically relevant, biomarker-restricted translational evidence and both show context-dependent resistance or heterogeneity. | - Directly matched, newly diagnosed patient-derived organoid or PDX comparisons for ALK versus MET, with explicit treatment-naive model provenance and standardized efficacy endpoints.
| |
| ALK — MET | Novelty, crowding, and strategic optionality | B wins | gpt-5.6-luna | 79% | MET offers slightly greater documented strategic optionality because the dossier supports multiple actionable biomarker contexts and two distinct modalities: genotype-selected small-molecule inhibition and a c-Met antibody-drug conjugate for high protein overexpression (C2-EV_TRACTABILITY_0005, C2-EV_TRACTABILITY_0006). This creates more platform and patient-segment expansion routes than the ALK dossier, whose strongest expansion evidence is extension of the same small-molecule approach into adjuvant NSCLC (C1-EV_TRACTABILITY_0005). However, METex14 is already crowded with multiple approved or clinically active inhibitors and competition now spans an ADC modality (C2-EV_COMPETITIVE_0005, C2-EV_COMPETITIVE_0006); amplification is also a less reliable differentiation route (C2-EV_TRACTABILITY_0007, C2-EV_COMPETITIVE_0004). ALK is likewise heavily crowded with several preferred first-line TKIs and emerging fourth-generation agents (C1-EV_COMPETITIVE_0001, C1-EV_COMPETITIVE_0005), although it has evidence for adjuvant expansion. Thus MET has a narrow optionality advantage, while both targets face substantial crowding that reduces strategic value even as it supports biological validation. | - Patent-landscape and freedom-to-operate analyses for ALK and MET; direct evidence of expansion into additional non-NSCLC indications; comparative platform or pipeline data beyond NSCLC.
| |
| MET — ALK | Integrated therapeutic hypothesis | B wins | gpt-5.6-luna | 90% | ALK represents the stronger integrated target–disease–modality hypothesis. Its oncogenic fusion mechanism is precisely defined, perturbational evidence supports selective dependence in ALK-rearranged NSCLC, and randomized evidence directly demonstrates superiority of ALK inhibition over platinum-pemetrexed chemotherapy in untreated disease (C2-EV_BIOMARKER_0002, C2-EV_CLINICAL_0001). Multiple randomized trials further support strong systemic and intracranial activity, including durable first-line benefit with lorlatinib (C2-EV_CLINICAL_0002, C2-EV_CLINICAL_0004, C2-EV_TRACTABILITY_0003, C2-EV_TRACTABILITY_0004). MET is also a well-validated, tractable driver with treatment-naive responses and PDX support (C1-EV_SAFETY_0002, C1-EV_CLINICAL_0001, C1-EV_PERTURBATION_0002), but its evidence is more biomarker- and alteration-dependent, especially for amplification (C1-EV_TRACTABILITY_0007), and its clinical differentiation is weakened by substantial class competition and the absence of demonstrated efficacy advantage among MET inhibitors (C1-EV_COMPETITIVE_0005, C1-EV_COMPETITIVE_0006). ALK also faces meaningful competition and resistance, but the direct chemotherapy comparison and stronger intracranial and long-term first-line evidence provide the more coherent differentiated-medicine hypothesis. | - The Q1-Q19 criterion scores and criterion-weighted composite/posterior are not supplied, so the required Q20 coherence check against those prior judgments cannot be performed.
- Direct evidence in the specified mucus-producing glandular-cell context and in the stated newly diagnosed NSCLC patient-derived xenograft context is limited or not clearly identified for either candidate.
- Comparative evidence showing differentiated efficacy of a MET- or ALK-directed medicine against current standard therapy specifically in the supplied newly diagnosed PDX context would most affect the comparison.
| - 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 validated as a predictive therapeutic biomarker in advanced NSCLC by randomized phase 3 evidence showing superior efficacy of ALK inhibition over platinum-pemetrexed chemotherapy.
- Lorlatinib provides durable long-term clinical benefit in treatment-naive advanced ALK-positive NSCLC, with sustained systemic and intracranial disease control.
- First-line ALK inhibition produces strong systemic and intracranial clinical efficacy, supporting high tractability of ALK in NSCLC.
- Alectinib demonstrates that selective, CNS-penetrant ALK inhibition is an effective modality in treatment-naive advanced NSCLC.
- Direct ALK knockdown selectively impairs viability and clonogenic growth of EML4-ALK-positive NSCLC cells.
- 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 exon 14 skipping is clinically validated as a predictive biomarker in metastatic NSCLC through durable responses to selective MET inhibitors.
- Broad MET amplification is less reliably tractable than MET exon 14 skipping and requires biomarker refinement, particularly regarding amplification level.
- ALK-positive NSCLC is a highly competitive targeted-therapy segment with three preferred first-line ALK TKIs—alectinib, brigatinib, and lorlatinib—and no direct randomized head-to-head trials establishing superiority among them.
- Competition is extending toward fourth-generation ALK inhibitors designed to address compound resistance mutations that can emerge after lorlatinib.
- 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.
- ALK positivity does not ensure durable dependence on ALK signaling because resistance can arise through both on-target and off-target mechanisms.
- 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.
- MET amplification has weaker and threshold-dependent clinical validation than MET exon 14 skipping in NSCLC.
- 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.
|
| EGFR — MET | Perturbational evidence | B wins | gpt-5.6-luna | 93% | MET has specific, high-trust perturbational evidence across orthogonal modalities: MET knockdown selectively impaired growth and survival in MET-amplified NSCLC models, pharmacologic MET inhibition produced antitumor responses in six MET-amplified PDX models, and a separate patient-derived model series showed responses to MET inhibition alone in EGFR-mutant, MET-amplified disease. The evidence is biomarker-restricted rather than universal, and CRISPR knockout in A549 plus nonselective tivantinib activity provide countervailing findings, but these do not directly negate the PDX evidence. EGFR has catalogue-level statements that tools were used in PDX studies, but the dossier does not provide comparably specific, reproducible EGFR PDX perturbation results or orthogonal intervention data. Therefore MET has stronger reproducible evidence for reversing disease-relevant phenotypes in PDX. | - A directly described EGFR perturbation study in patient-derived xenografts showing disease-relevant phenotype reversal, ideally with orthogonal genetic and pharmacologic perturbations and biomarker-matched models.
- For MET, direct confirmation that the reported patient-derived xenograft models represent the specified mucus-producing glandular-cell context.
| |
| ALK — KRAS | Reversibility and therapeutic window | A wins | gpt-5.6-luna | 78% | FIRST has the better-supported practical reversibility profile: ALK is treated with an oral, once-daily small-molecule inhibitor and its regulatory evidence explicitly describes treatment interruption as a management option for serious adverse effects (C1-EV_TRACTABILITY_0002; C1-EV_SAFETY_0005). SECOND has a potentially favorable target window because mutant-selective KRAS-G12C inhibitors spare wild-type KRAS (C2-EV_SAFETY_0001), but the intervention is explicitly irreversible and covalent, which limits reversal after an adverse effect emerges (C2-CAT-ONCO-BIO-032; C2-CAT-ONCO-BIO-032b). The comparison is narrow because direct half-life, exposure-response separation, and post-interruption recovery data are not supplied for either candidate, and ALK-associated serious toxicities remain documented. | - Direct comparative human pharmacokinetic half-life, exposure-response separation, washout kinetics, and recovery of target modulation after treatment interruption for the ALK and KRAS small molecules.
| - ALK is clinically tractable with an established oral small-molecule kinase-inhibitor modality in metastatic NSCLC.
- 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.
- Weist KRAS Mutationen auf, die neue medikamentös adressierbare Bindungsstellen schaffen?
- Verändern Mutationen in KRAS bestehende medikamentös adressierbare Stellen?
- 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.
|
| EGFR — KRAS | Human natural experiments | Abstained | gpt-5.6-luna | 98% | Neither dossier contains the criterion-specific human natural experiment required: evidence in humans indicating both efficacy and tolerability of inhibition in NSCLC. The EGFR germline loss-of-function and KRAS germline gain-of-function findings describe inherited disease phenotypes, not therapeutic inhibition efficacy in NSCLC, and do not establish tolerability of a time-limited pharmacological intervention. The supplied safety and pharmacological evidence is not a qualifying human natural experiment and cannot substitute for the missing evidence. | - For both candidates, a human NSCLC natural experiment showing that genetically or pharmacologically reduced target activity produces both antitumor efficacy and acceptable tolerability, with genetic direction matching therapeutic inhibition, is missing.
- EGFR dossier evidence concerns rare germline loss-of-function disease or germline predisposition rather than inhibition efficacy in NSCLC (C1-CAT-ONCO-BIO-005, C1-CAT-ONCO-BIO-016, C1-CAT-ONCO-BIO-056).
- KRAS dossier evidence concerns germline gain-of-function RASopathies rather than KRAS-inhibition efficacy and tolerability in NSCLC (C2-CAT-ONCO-BIO-005, C2-CAT-ONCO-BIO-016, C2-CAT-ONCO-BIO-056).
- Human long-term exposure cohorts or protective-allele studies linking target inhibition to NSCLC outcomes and tolerability are absent for both candidates.
| – |
| EGFR — MET | Clinical validation | B wins | gpt-5.6-luna | 97% | MET has substantially stronger indication-relevant clinical validation. The dossier provides genotype-selected phase 2 evidence with durable responses in MET exon 14-skipping metastatic NSCLC, traditional FDA approvals for capmatinib and tepotinib, prospective biomarker selection, and on-treatment METex14 ctDNA depletion associated with response (C2-EV_CLINICAL_0001, C2-EV_CLINICAL_0002, C2-EV_CLINICAL_0003, C2-EV_BIOMARKER_0001, C2-EV_BIOMARKER_0002). It also provides a defined approved oral regimen for tepotinib (C2-EV_COMPETITIVE_0002). Negative results for nonspecific MET immunohistochemistry, low-level amplification, and broad MET-pathway inhibition do not invalidate METex14 because they concern different biomarker populations (C2-EV_CLINICAL_0006, C2-EV_BIOMARKER_0004, C2-EV_MECHANISM_0006, C2-EV_MECHANISM_0007). EGFR is described as clinically established and approved in EGFR-mutant NSCLC, but the supplied EGFR dossier does not provide comparable indication-specific trial datasets, endpoint results, dose/exposure information, or target-engagement data; its clinical claims are largely catalogue-level summaries (C1-CAT-ONCO-BIO-091b, C1-CAT-ONCO-BIO-046). | - For EGFR, indication-specific completed-trial reports with prespecified endpoints, comparator arms, enrolled biomarker-defined populations, dose and exposure data, pharmacodynamic target engagement, and coded trial-stoppage reasons.
- For both targets, direct clinical validation in the specified mucus-producing glandular-cell, newly diagnosed, chemotherapy-backbone population is not provided.
| |
| MET — EGFR | Combination potential | Abstained | gpt-5.6-luna | 97% | Neither dossier provides direct evidence for combination with chemotherapy. MET has combination evidence with EGFR inhibition and a documented toxicity burden, but not chemotherapy combination studies or chemotherapy-overlap assessment (C1-EV_MECHANISM_0005, C1-EV_SAFETY_0004). EGFR has evidence of bypass and resistance biology and on-target toxicity, but no chemotherapy combination efficacy or overlapping-toxicity assessment (C2-CAT-ONCO-BIO-004, C2-EV_SAFETY_0001). Because the criterion specifically concerns combination potential with chemotherapy, the available evidence is insufficient to compare the candidates. | - Direct MET-plus-chemotherapy and EGFR-plus-chemotherapy combination studies in NSCLC, including pathway complementarity, resistance-mechanism evaluation, efficacy in relevant in vivo or PDX models, and formal assessment of overlapping toxicities with chemotherapy.
| – |
| ALK — EGFR | Translational model validity | A wins | gpt-5.6-luna | 88% | ALK has more direct and target-specific translational evidence in mechanistically relevant lung cancer models, including genetically engineered mouse tumors and patient-derived NSCLC material, with concordant human NSCLC clinical validation. EGFR has catalogue summaries describing relevant GEMM, xenograft, and PDX model classes, but these are less specific and do not provide comparably direct primary-study efficacy evidence. The ALK evidence is not uniform across variants or models, but that heterogeneity does not outweigh its stronger documented model validity and translational linkage. | - Directly comparable, primary-study efficacy results in patient-derived or genetically engineered NSCLC models for both targets, specifically using newly diagnosed, mucus-producing glandular-cell disease contexts.
- For EGFR, a clearly documented primary PDX, organoid, primary patient-cell, or genetically engineered model study with quantitative efficacy results and study-level methods.
| |
| MET — KRAS | Modality–direction–exposure fit | A wins | gpt-5.6-luna | 78% | Both targets have evidence of small-molecule tractability. MET has the stronger modality–direction–exposure fit in this dossier because selective oral MET small molecules have direct clinical precedent, MET inhibition has produced responses in MET-dependent lung-cancer PDX models, and validated pMET:total-MET assays support proximal target-engagement measurement (C1-EV_TRACTABILITY_0002, C1-EV_TRACTABILITY_0003, C1-EV_PERTURBATION_0002, C1-EV_BIOMARKER_0005). KRAS also has clinically validated mutant-selective small-molecule tractability and a mechanistically suitable inhibition strategy (C2-EV_TRACTABILITY_0001, C2-EV_MECHANISM_0001), but the dossier supplies less direct evidence linking a specific KRAS small molecule to PK/PD depth, duration, lung exposure, or PDX efficacy in the stated context. The margin is narrow because neither dossier provides direct biodistribution or exposure data in the specified mucus-producing glandular cell population, and MET evidence includes heterogeneity and alteration-dependent activity. | - Head-to-head PK/PD and biodistribution data in lung mucus-producing glandular cells, including intratumoral concentration, target-occupancy or pathway-suppression depth, duration of inhibition, and exposure relative to the required modulation for MET versus the relevant KRAS allele.
| |
| ALK — KRAS | Biomarker and target engagement | A wins | gpt-5.6-luna | 99% | FIRST has direct regulatory and clinical evidence that ALK positivity, identified by approved assays, selects patients for effective ALK inhibition: the FDA label establishes the predictive biomarker and assay requirement (C1-EV_BIOMARKER_0001), PROFILE 1014 shows superior efficacy over chemotherapy in centrally confirmed ALK-positive NSCLC (C1-EV_BIOMARKER_0002; C1-EV_CLINICAL_0001), and FDA-authorized FISH, IHC, and sequencing-based companion diagnostics are documented (C1-EV_TRACTABILITY_0006). Longitudinal plasma ALK mutation kinetics provide a clinical pharmacodynamic signal consistent with target inhibition (C1-EV_BIOMARKER_0005). KRAS has evidence of clinically validated small-molecule tractability and catalogue-level statements about KRAS-G12C treatment, but the dossier does not provide comparably strong, specific evidence for a validated NSCLC patient-selection assay, clinical target-engagement/PD readout, and linked early efficacy signal. The ALK limitations—nonproductive FISH results and resistance mechanisms—reduce universality but do not negate the substantially stronger overall biomarker strategy. | - For KRAS, a regulatory-validated NSCLC biomarker assay linked to patient selection, direct clinical pharmacodynamic or target-engagement measurements, and randomized or otherwise clearly documented early efficacy evidence in the specified NSCLC context.
| |
| EGFR — MET | Patient stratification | B wins | gpt-5.6-luna | 91% | MET supports the clearer prospectively testable selection hypothesis. MET exon 14 skipping is a recurrent genomic driver with diverse splice-site alterations, and patients were prospectively selected using tissue RNA or plasma cfDNA/NGS in clinical studies, with regulatory labeling requiring an FDA-approved test. This directly demonstrates an operational genomic biomarker-to-treatment hypothesis. EGFR has strong evidence for a molecularly defined NSCLC subgroup and clinical feasibility of mutation-specific targeting, but the supplied evidence is less explicit about prospective trial selection and operational validation. MET amplification and MET protein expression are weaker or heterogeneous selection strategies, but these limitations do not outweigh the unusually direct prospective evidence for METex14. | - For EGFR, prospective biomarker-selected clinical-trial evidence explicitly demonstrating operational testing of Exon-19 deletion or L858R selection in the stated treatment context would most affect the comparison.
- For MET amplification, prospective evidence defining a reproducible gene-copy threshold and assay workflow is needed; the dossier supports METex14 more clearly than amplification.
| |
| KRAS — MET | Patient stratification | B wins | gpt-5.6-luna | 93% | MET supports the clearer operational patient-selection hypothesis: MET exon 14 skipping was prospectively detected using tissue RNA or plasma cfDNA/NGS and was associated with responses to selective MET inhibitors in clinical studies, with an FDA-approved test specifying the eligible population (C2-EV_BIOMARKER_0001, C2-EV_CLINICAL_0001, C2-EV_CLINICAL_0003). Serial METex14 ctDNA depletion also provides an operational pharmacodynamic measure (C2-EV_BIOMARKER_0002). KRAS has strong evidence for recurrent genotype-defined NSCLC subsets and prospective genomic profiling (C1-EV_GENETICS_0002, C1-EV_GENETICS_0003), but the dossier does not provide comparably direct prospective interventional evidence tying a prespecified KRAS subgroup-selection strategy to treatment outcomes. MET's hypothesis is not universal across all MET assays: MET IHC alone failed as a selector and amplification thresholds are less reliable (C2-EV_BIOMARKER_0004, C2-EV_TISSUE_0004, C2-EV_BIOMARKER_0006), but the METex14 genomic subtype remains clearly actionable. | - A prospective interventional NSCLC trial directly demonstrating treatment benefit in a KRAS-defined subgroup, with a prespecified genomic selection assay and outcome comparison, would most narrow the difference.
| |
| EGFR — ALK | Combination potential | Abstained | gpt-5.6-luna | 97% | The dossiers provide substantial target biology and resistance-mechanism evidence, but they do not provide direct EGFR-plus-chemotherapy or ALK-plus-chemotherapy combination studies, nor a chemotherapy-specific overlapping-toxicity or dose-feasibility assessment. The reported target-specific toxicities indicate that safety cannot be inferred, and the guardrail requires evidence that the combination can be dosed safely. Therefore, the criterion cannot be adjudicated between the candidates. | - Direct chemotherapy-combination preclinical studies for EGFR and ALK in NSCLC, ideally in relevant in vivo or PDX models.
- Evidence that EGFR- or ALK-directed inhibition complements chemotherapy resistance mechanisms rather than producing redundant activity.
- A chemotherapy-specific overlapping-toxicity and dose-feasibility assessment for each target, including whether the combination can be administered safely together.
| |
| KRAS — MET | Competitive differentiation | B wins | gpt-5.6-luna | 84% | MET offers the stronger dossier-supported case for clinically meaningful differentiation over chemotherapy in newly diagnosed disease. METex14-selected treatment has documented activity in treatment-naive NSCLC, and the dossier explicitly states that chemotherapy and immunotherapy have not shown a preferential response in this biomarker-defined population, supporting differentiation of biomarker-selected MET inhibition from non-targeted treatment (C2-EV_COMPETITIVE_0001, C2-EV_COMPETITIVE_0003, C2-EV_CLINICAL_0001, C2-EV_CLINICAL_0002). However, the advantage is narrow because METex14 already has multiple approved or clinically active inhibitors and additional non-TKI competition (C2-EV_COMPETITIVE_0005, C2-EV_COMPETITIVE_0006). KRAS has strong small-molecule tractability, but the dossier does not provide a comparable current-treatment or chemotherapy-differentiation analysis for newly diagnosed NSCLC; that absence is treated as missing evidence rather than negative evidence. This verdict addresses competitive differentiation, not overall scientific merit or market attractiveness. | - For KRAS: direct comparative evidence in newly diagnosed KRAS-altered NSCLC showing differentiation from chemotherapy, plus a current competitor-pipeline and treatment-landscape analysis for the relevant KRAS allele.
- For MET: prospective first-line comparative evidence against chemotherapy-based standard treatment in METex14 NSCLC.
| |
| EGFR — KRAS | Biomarker and target engagement | A wins | gpt-5.6-luna | 82% | EGFR offers the more comprehensive clinical biomarker strategy in NSCLC: activating EGFR alterations are described as predictive biomarkers with established clinical testing and regulatory treatment precedent, and pEGFR is supplied as a proximal target-engagement readout (C1-CAT-ONCO-BIO-046, C1-CAT-ONCO-BIO-091b, C1-CAT-ONCO-BIO-099). KRAS also has a credible strategy for KRAS G12C, including mutation-specific clinical selection, approved mutant-selective inhibitors, and KRAS-GTP/pERK pharmacodynamic readouts (C2-CAT-ONCO-BIO-091b, C2-CAT-ONCO-BIO-099, C2-EV_TRACTABILITY_0001). However, that clinical biomarker precedent is narrower and allele-specific, while the candidate is presented as G12C/G12D. The advantage for EGFR is therefore narrow rather than clear, because both dossiers contain clinically grounded selection, engagement, and early efficacy-marker evidence, but the supplied evidence does not provide a direct prospective head-to-head validation of the complete biomarker-to-efficacy chain. | - Prospective, indication-matched clinical studies directly linking the specific genomic assay, measured target engagement, pharmacodynamic change, and early efficacy endpoint in newly diagnosed NSCLC under the stated chemotherapy backbone.
- For KRAS, clinical biomarker and target-engagement validation for KRAS G12D in NSCLC, since the supplied clinical precedent is specific to KRAS G12C.
| |
| ALK — MET | Modality–direction–exposure fit | A wins | gpt-5.6-luna | 78% | Both targets have evidence supporting inhibition by oral small molecules, including clinical precedent and biomarker-linked target engagement. ALK has the stronger modality–exposure fit because multiple approved oral ALK inhibitors show systemic and intracranial activity in biomarker-selected NSCLC, with longitudinal molecular kinetics consistent with on-target inhibition (C1-EV_TRACTABILITY_0002, C1-EV_TRACTABILITY_0003, C1-EV_TRACTABILITY_0004, C1-EV_BIOMARKER_0005). MET is also clinically tractable, with approved oral inhibitors and pharmacodynamic evidence from METex14 ctDNA depletion and pMET suppression (C2-EV_TRACTABILITY_0002, C2-EV_BIOMARKER_0002, C2-EV_BIOMARKER_0005), but its fit is more dependent on the precise alteration and amplification level, with weaker activity in low-level amplification and substantial biomarker heterogeneity (C2-EV_TRACTABILITY_0007, C2-EV_BIOMARKER_0006). The advantage for ALK is narrow because neither dossier supplies direct PK/PD modelling, biodistribution, or exposure-duration evidence in the specified mucus-producing glandular-cell PDX context. | - Direct comparative PK/PD or exposure data in mucus-producing glandular tumor cells or the specified patient-derived xenograft context
- Modality-specific biodistribution data demonstrating tumor-cell exposure and duration for either candidate
- Direct target-turnover measurements in the specified tissue and cell type
| |
| ALK — KRAS | Novelty, crowding, and strategic optionality | B wins | gpt-5.6-luna | 68% | KRAS has broader documented strategic optionality through multiple disease indications and an expanding allelic/platform scope, including G12C clinical tractability and development directions spanning G12D, G12V, pan-KRAS, and combination strategies (C2-CAT-ONCO-BIO-046, C2-CAT-ONCO-BIO-088, C2-EV_TRACTABILITY_0001). ALK has demonstrated expansion within NSCLC into adjuvant treatment (C1-EV_TRACTABILITY_0005), but its strategic value is constrained by explicit crowding from multiple preferred first-line TKIs and emerging fourth-generation competitors (C1-EV_COMPETITIVE_0001, C1-EV_COMPETITIVE_0005). That crowding also supports biological confidence, but reduces differentiation and strategic value. The margin is narrow because KRAS-specific NSCLC competitor density, phase-by-phase pipeline evidence, and patent position are not supplied; absence of those data cannot be treated as evidence of lower crowding. | - A directly comparable competitor pipeline by phase for KRAS in NSCLC, including G12C, G12D, G12V, and pan-KRAS programs.
- Patent-landscape and freedom-to-operate evidence for both targets.
- Comparative evidence of platform leverage and clinically demonstrated expansion beyond the lead indication for each target.
| |
| MET — EGFR | Patient stratification | A wins | gpt-5.6-luna | 96% | MET supports the clearer prospectively testable patient-selection hypothesis. MET exon 14 skipping was prospectively assayed using tissue RNA or plasma cfDNA/NGS, and selective MET inhibitors produced responses in the defined subgroup; capmatinib labeling explicitly requires an FDA-approved test. The perturbational evidence also supports restriction to molecularly defined subsets rather than broad MET expression. EGFR has strong evidence that exon 19 deletions and L858R are causal drivers, but the supplied dossier does not provide comparably direct prospective trial-level evidence linking a specified genomic assay-defined EGFR subgroup to patient selection and treatment response. Its catalogue evidence is largely general or non-validated for alternative germline polymorphism-based stratification. | - For EGFR, a prospective clinical validation dossier directly demonstrating that a specified genomic assay-defined EGFR-mutant NSCLC subgroup can be operationally enrolled and predicts response to a defined small-molecule treatment.
| |
| ALK — EGFR | Reversibility and therapeutic window | A wins | gpt-5.6-luna | 72% | FIRST has the stronger therapeutic-window evidence: ALK knockout mice were viable and fertile with limited adult essentiality, despite specific phenotypes, whereas EGFR loss causes severe epithelial developmental defects and EGFR inhibition produces on-target rash and diarrhoea. ALK is also explicitly supported by an established oral small-molecule modality, which is generally compatible with titration and interruption. However, ALK has important clinical toxicities, and the dossier lacks direct half-life, washout, and target-occupancy comparisons. EGFR includes both reversible and irreversible small-molecule inhibitors, making its reversibility modality-dependent. The advantage for FIRST is therefore narrow. | - Direct comparative clinical pharmacokinetic and pharmacodynamic evidence for half-life, target-occupancy washout, dose interruption, and reversibility across the specific ALK and EGFR small molecules; the EGFR dossier includes both reversible and irreversible inhibitors without identifying the intervention being compared.
| - Wild-type ALK appears to have limited essentiality for survival of adult normal tissues, although loss of ALK produces specific reproductive and neuroendocrine phenotypes in mice.
- ALK is clinically tractable with an established oral small-molecule kinase-inhibitor modality in metastatic NSCLC.
- Several approved ALK TKIs
- On-target dermatologic and GI toxicity
- Welcher Phänotyp tritt auf, wenn EGFR deletiert wird?
- Welche Toolverbindungen für EGFR sind verfügbar?
- 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 — EGFR | Disease biology and mechanism coherence | Tie | gpt-5.6-luna | 87% | Both candidates have comparable, coherent mechanistic evidence for NSCLC biology: recurrent activating human tumor mutations, a defined causal signaling mechanism, and experimentally supported dependence on the altered target. KRAS mutations constitutively activate RAS-MAPK signaling and genetic suppression can impair tumor-cell fitness (C1-EV_GENETICS_0001; C1-EV_GENETICS_0002; C1-EV_MECHANISM_0001; C1-CAT-ONCO-BIO-070; C1-CAT-ONCO-BIO-098). EGFR exon-19 deletions and L858R cause ligand-independent kinase activation, feed MAPK and PI3K pathways, and have functional and in vivo validation (C2-EV_GENETICS_0001; C2-EV_MECHANISM_0001; C2-CAT-ONCO-BIO-083; C2-CAT-ONCO-BIO-117b). The dossiers do not provide target-specific evidence in the stated mucus-producing glandular-cell PDX context that would distinguish them. Druggability evidence was not used. | - Direct, head-to-head inhibition studies in mucus-producing glandular NSCLC cells and matched patient-derived xenografts comparing KRAS versus EGFR pathway dependence and pathological response.
| |