# HSEF v5.3 review: Oncogene Addiction Hypothesis

> Sources: 66 primary reports; claims: 70; verified: 70; unverified: 0. Verification is bounded attribution. Access: 9 partial-body and 57 complete abstract/status examinations. Literature cutoff: 2026-10-08; source examination UTC date: 2026-10-09. No complete-paper audit or blinded calibration claimed.

> **Adversarial review:** Generated by Sol 6.1; adversarially reviewed by Claude Opus 5.5 (Anthropic) on 2026-10-14. 6 corrections applied (see ADVERSARIAL REVIEW AND APPROVAL). Status: **Approved as revised.**

## AT A GLANCE

**Defined driver–tumor pairs show strong oncogene dependence for regression; this does not establish one-driver addiction across most cancers or eradication of every residual cell.**

| Level | Confidence | Causal role | Momentum | Stability | Divergence |
|---|---|---|---|---|---|
| Η — Strong convergent causal evidence, intervention-sensitive | Moderate | primary driver | Mechanistic only | Stable | Contested |

**Deciding experiment:** In baseline molecularly defined driver–tumor cohorts and matched native models, combine allele-specific withdrawal with direct engagement measures, matched driver restoration and longitudinal lineage-resolved residual-cell tracking. Prospectively include nonresponders; perturb candidate bypass and reservoir dependencies separately. Follow sustained control both during treatment and after predefined cessation.

**Belief versus evidence:** The preregistered scope-matched rule reports SC1 at Η: endogenous conditional-driver and common-conduit interventions, genetic withdrawal, mechanistic resistance/rescue and baseline molecularly defined clinical trials converge on tumor-maintenance dependence across named pairs. Significant primary nonresponse, host/apoptosis dependence and persistence prevent Θ or Ω. SC2 pan-cancer one-driver generalization and SC3 residual-cell eradication remain Γ as broader mechanistic claims without matching general causal validation.

**Best-supported rival:** Clonal bypass and lineage plasticity (Ζ), the highest-ranked non-nominal model. Resistance-allele reconstruction and Kapoor verified YAP/TEAD bypass with binding-defective controls distinguish persistent or reassigned dependence from failed drug action, and Corbin target-engaged survival marks the residual-compartment boundary (PMID11423618,15737014,24954535,21157039). The models are complementary at different times and compartments; ranking does not establish one universal cause.

## COMPETING MECHANISTIC MODELS

**Rank 1: Scoped dominant-oncogene maintenance dependence (Η).** A defined baseline driver can remain a rate-limiting maintenance node despite other genomic changes. Endogenous conditional KRAS, endogenous MYC inhibition, EGFR genetic/drug controls and independent molecularly defined clinical trials converge (PMID22606359,18716624,16705038,12637609,20573926). This model explains bulk regression, not universal response, one-node pan-cancer maintenance or every residual cell.

**Rank 2: Clonal bypass and lineage plasticity (Ζ).** Selection restores drug-resistant target signaling or substitutes a different maintenance route. Resistance-allele reconstruction and YAP gain/loss, binding-defective TEAD controls and orthotopic rescue discriminate true bypass from incomplete inhibition (PMID11423618,15737014,24954535). Serial biopsy association cannot identify every lineage mechanism; rtTA reactivation and functional bypass must remain distinct.

**Rank 3: Reversible drug-tolerant state (Ζ).** A reversibly altered cellular state survives treatment and can seed later resistance. Single-cell tolerance, independent barcode/kinetic route tests and GPX4 knockout with ferroptosis rescue and residual-xenograft course connect state survival to recurrence in bounded experimental settings (PMID20371346,26828195,29088702). Some sources share EGFR lines and investigator programmes; in vivo GPX4 tests remain xenografts. This is not all-patient endogenous proof or a clinical cure.

**Rank 4: Non-addicted residual reservoir (Γ).** A residual compartment can survive strong signaling inhibition while the bulk regresses. Philadelphia-positive stem/progenitor survival with target engagement and cytokine controls, plus MYC liver dormancy, support the mechanism (PMID21157039,15475948). The decisive human evidence is ex vivo and does not selectively remove a lineage-proven native reservoir with matched rescue and relapse confirmation. Survival does not prove total BCR-ABL protein dispensability.

**Rank 5: Non-oncogene stress-support dependency (Ε).** Residual tumor states can acquire requirements for unmutated stress-support functions. GPX4 loss and antioxidant rescue specifically kill persisters and block relapse in the tested residual melanoma xenograft (PMID29088702). A state-specific, genetically engineered xenograft dependency does not establish that stress support generally replaces mutated oncogenes as the dominant native cancer mechanism.

The tournament does not alter the nominal level. Compatible pathways and flexible alternatives are not automatically proved by explaining observations.

## FULL REVIEW

### Registered claim and scope

Cancers with prospectively defined driver–tumor pairs depend on continuous dominant oncogene signaling for maintenance, so selective signaling withdrawal produces tumor regression.

Use the index specific-pair maintenance scope and separate most-cancer generalization. Baseline driver/tumor identity, treatment setting and endpoints define enrollment; responders are never used to define scope. Regression and elimination of residual disease are different outcomes.

Selection: public-interest rank 40 from the editorial ranking compiled 2026-09-23; categorical prior verdicts were visible. Private fresh review; earlier documents are preserved.

### Evidence and causal interpretation

The registered hypothesis concerns maintenance in named baseline driver–tumor pairs. Sixty-six original reports were examined. The scope-matched rule permits a strong result for these pairs without implying that most cancers have one immutable dependency or that withdrawing a driver eradicates every cell. None of the scope is selected retrospectively from which tumors responded.

Conditional MYC, H-RAS, BCR-ABL, Neu and EGFR systems demonstrate regression after turning off a defined oncogenic stimulus (PMID10488335,10440378,10615128,12498714,16705038). Huettner models acute B-cell leukemia, not human CML. Many founding systems force expression; the endogenous conditional KRAS activity switch and inhibition of native MYC in KRAS-driven autochthonous tumors supply complementary evidence rather than merely more transgene papers (PMID22606359,18716624).

Politi supplies a particularly informative original-body comparison (PMID16705038). Genetic EGFR withdrawal causes rapid regression. Erlotinib also regresses the tumors while doxycycline induction continues, whereas placebo and KRAS-driven controls do not respond. Transgene RNA remains induced during treatment. Strong RNA induction and several-fold bulk protein increases constrain native-dosage extrapolation, and custom MRI disappearance does not establish eradication. BIM/PUMA knockout and constitutive AKT/MEK controls show that target inhibition requires a competent downstream apoptosis programme (PMID23532334).

Human evidence provides independent disease settings: IRIS CML, baseline EGFR-mutant NEJ002, HER2-selected breast cancer, baseline BRAF-mutant BRIM-3 and ALK-selected PROFILE trials demonstrate therapeutic consequence (PMID12637609,20573926,11248153,21639808,25470694). The denominators remain intact. For example, NEJ002 reports 73.7 percent response rather than every patient responding, and BRIM-3 reports 48 percent. Trastuzumab plus chemotherapy includes antibody-mediated host effects and does not isolate pure kinase withdrawal. EGFR discovery papers selected responders after outcome and therefore support mechanism discovery without supplying a prospective response frequency (PMID15118073,15329413).

IPASS uses clinically selected enrollment with mutation-positive and mutation-negative molecular subgroups, rather than genotype enrollment (PMID19692680). FLAURA, ALEX and CROWN compare active inhibitors; they improve duration, survival or CNS control but do not alone test driver-on versus driver-off (PMID29151359,31751012,28586279,33207094). CROWN five- and seven-year reports reuse the same 296 patients: seven-year progression-free survival is 55 versus 3 percent, a major durable treatment benefit, yet still an on-treatment endpoint with immature overall survival (PMID38819031,42217582). These outcomes cannot be counted as three independent trials or proof of cure.

Resistance can strengthen the target-dependence interpretation when it restores signaling. Gorre reconstructs resistant ABL, and Pao compares T790M constructs, phosphorylation and independent mutation detection (PMID11423618,15737014). Pao has a scant pretreatment sample and imperfect detection sensitivity, so a later detected allele is not necessarily newly created. MET/ERBB3 bypass, dual EGFR blockade, ALK compound mutations and serial biopsy lineage transformation instead identify different forms of reassigned dependence (PMID17463250,19759520,27432227,29650534,21430269). Retaining a driver mutation after lineage conversion does not prove functional dependence on it.

Kapoor directly distinguishes transgene reactivation from bypass (PMID24954535). All 28 advanced conditional pancreatic tumors regress after Kras withdrawal, but 20 recur. About half of examined relapses reactivate the transgene through rtTA amplification; others lack its expression. YAP gain substitutes for Kras in orthotopic growth, and YAP/TEAD knockdown or TEAD-binding-defective mutants suppress that route. Basal wild-type RAS remains possible. This functional specificity supports a real alternative maintenance node, without claiming every recurrence is YAP driven. BRAF-mutant colon versus melanoma feedback and dual MYC/KRAS models further establish tissue and cooperating-node boundaries (PMID22281684,18356293,18461184).

Corbin checks a difficult negative contrast in original text (PMID21157039). CML stem/progenitors survive several ABL inhibitors with cytokine support despite strong phospho-CRKL/STAT5 inhibition and sustained engagement. More than 95 percent of surviving cells remain Philadelphia positive; normal cells and cytokine titration help exclude simple carryover or only grossly supraphysiological rescue. Residual phosphorylation is not literally zero, and a short culture experiment cannot establish total BCR-ABL protein dispensability in every native stem cell. Together with earlier quiescent survival, it identifies a serious boundary for an all-cell killing claim (PMID11756187).

Sharma shows reversible tolerance in single-cell-derived PC9 cultures with EGFR suppressed, distinct reversibility after short versus expanded persistence, and KDM5A/HDAC/IGF1R controls (PMID20371346). Hata uses barcodes, timing, persister pools and spiked-clone controls to distinguish rare pre-existing T790M from a later persister-derived route; late resistant cells retain weaker apoptosis despite similar EGFR/ERK inhibition (PMID26828195). These studies share some platforms, and patient-derived response correlations do not retrospectively trace the patients original clones. Cycling-persister and APOBEC studies refine additional state and mutation routes (PMID34381210,37407818).

Independent GPX4 genetics adds a functional residual-course test (PMID29088702). Knockout selectively kills persisters after ferrostatin withdrawal; antioxidants and iron/lipid controls discriminate ferroptosis from nonspecific death. In melanoma xenografts, GPX4 loss is unmasked only after initial targeted-drug shrinkage, and relapse is prevented under continued treatment while wild-type tumors recur. Compounds lack systemic bioavailability, so this genetic proof is not a tested clinical regimen or treatment-free cure. FAK/YAP and TROP2 work further tests residual interventions with selected-model and preliminary-clinical limitations (PMID38702301,42314664).

Not every long-term survivor requires continuous medication. STIM deliberately enrolls CML patients with at least two years of deep molecular remission before cessation; many relapse and respond again, while a selected fraction maintains remission off treatment (PMID20965785). This does not estimate drug-free control in all newly diagnosed CML or prove zero residual cells. Repeated Omomyc also deserves a balanced reading: all animals recur after one cycle, whereas after eight cycles three of four have no scored tumor after four weeks off (PMID23475959). Histological sampling excludes atypical hyperplasias, and the small short cessation cohort is not lifelong sterile eradication. Osteosarcoma differentiation and liver dormancy show different MYC withdrawal fates (PMID12098700,15475948).

The 2026 RAS-on literature expands mechanistic testing rather than automatically expanding clinical proof. Foth compares genotype/resistance mechanisms and two divergent clinical cases; Wei and Orlen test immune-dependent responses in preclinical models; Scaparone identifies schedule-dependent ALK bypass under KRAS inhibition (PMID42171647,41670434,40057911,42329099). The Foth journal article replaces its earlier preprint. Consensus suggestions implying clinical complete responses from Wei are corrected to preclinical evidence. The nominal specific-pair model reaches Η, while its broad one-driver and residual-eradication extensions remain Γ.

## HYPOTHESIS STATE EVALUATION FRAMEWORK (HSEF v5.3)

<!-- consensus-consistency-2026-10-09 -->
### Consensus provenance reconciliation — 9 October 2026

Case-specific Consensus discovery records were located and linked to this final edition. This supplements the review's provenance; it leaves the original assessed claim, grades and source counts intact. Provider summaries and full-text badges do not establish examination of original primary passages by this reviewer. Source access and incorporation retain the limits stated in the original methods and any dated research update.

Recorded evidence artifacts:

- `deliverables/hsef_v53_popularity_ranks_36_40_2026-10-08/oncogene-addiction/consensus-anchors-search-log.json`
- `deliverables/hsef_v53_popularity_ranks_36_40_2026-10-08/oncogene-addiction/consensus-anchors.json`
- `deliverables/hsef_v53_popularity_ranks_36_40_2026-10-08/oncogene-addiction/consensus/broad-ax.txt`
- `deliverables/hsef_v53_popularity_ranks_36_40_2026-10-08/oncogene-addiction/consensus/RECONCILIATION.md`
- `deliverables/hsef_v53_popularity_ranks_36_40_2026-10-08/oncogene-addiction/consensus/targeted-ax.txt`

Reported usage is preserved in the linked source-specific records; query counts and original-passage comparisons are different measures.

No earlier adversarial approval is attached to this case; this provenance supplement is not a new independent or blinded adjudication. No score or confidence change is made by this reconciliation.
<!-- end-consensus-consistency -->


### Subclaim matrix and aggregation

| Subclaim | Scope | Level | Decision |
|---|---|---|---|
| SC1 Selective withdrawal of dominant oncogene signaling regresses tumors in defined driver–cancer pairs. | baseline BCR-ABL/CML, EGFR/NSCLC, HER2/breast, BRAF/melanoma, ALK/NSCLC and genotype-defined experimental pairs | Η | Η is supported for the prospectively named maintenance/regression pairs by endogenous conditional KRAS and MYC interventions, orthogonal EGFR withdrawal/drug controls, resistance alleles and several molecularly enrolled randomized trial families. This means causal regression in defined settings, not every enrolled tumor responding or every tumor cell requiring the driver. Nonresponse and context-specific bypass are major exceptions, preventing Θ and Ω. |
| SC2 Most cancers depend on a single dominant oncogene in the same way. | prespecified pan-cancer lineages and baseline driver status | Γ | Experimental dual-driver and tissue comparisons, heterogeneous KRAS cell dependency and genome-wide resistance screens show context dependence. These reports do not constitute a representative native pan-cancer maintenance map establishing a single dominant dependency in most cancers. The broad model has mechanistic plausibility, but its prespecified denominator and matching causal test are missing. |
| SC3 Driver withdrawal eliminates residual populations and gives durable treatment-free control. | baseline driver–tumor pair and residual-cell follow-up | Γ | Initial regression and prolonged progression-free survival differ from eradication. Target-engaged human stem-cell survival, reversible tolerance, late resistance, transgene-independent recurrence and responder-selected treatment cessation contradict universal residual-cell elimination. Selected osteosarcoma differentiation or four-week off-treatment tumor absence after repeated Omomyc does not establish general lifelong cure. |

Preregistered aggregation: scope-matched. The preregistered scope-matched rule reports SC1 at Η: endogenous conditional-driver and common-conduit interventions, genetic withdrawal, mechanistic resistance/rescue and baseline molecularly defined clinical trials converge on tumor-maintenance dependence across named pairs. Significant primary nonresponse, host/apoptosis dependence and persistence prevent Θ or Ω. SC2 pan-cancer one-driver generalization and SC3 residual-cell eradication remain Γ as broader mechanistic claims without matching general causal validation. Sensitivities: weakest-link Γ; primary-subclaim Η; scope-matched Η.

### Decisive-perturbation inventory and causal tally

[1] Conditional MYC in hematopoietic lineages yields tumors that regress after transgene inactivation. Forced expression is not native human dosage; the abstract does not establish elimination of every residual cell or all sporadic tumors.

[2] Conditional H-RAS withdrawal regresses INK4a-null melanoma; enforced VEGF does not sustain the tumor after withdrawal. Engineered induction and genotype constrain scope; vascular effects do not establish universal tumor dependence.

[3] Suppressing conditional BCR-ABL1 reverses acute B-cell leukemia, with repeated induction and remission cycles. This mouse acute B-cell leukemia is not human chronic myeloid leukemia; repeat reversibility is not proven sterilization.

[4] Neu withdrawal regresses mammary tumors and pulmonary metastases to undetectability, but most animals later recur with Neu-independent tumors. Undetectable burden is not absence of residual cells; body controls were unavailable for examination.

[5] Both EGFR transgene withdrawal and erlotinib during continued doxycycline regress mutant-EGFR tumors; placebo and KRAS tumors fail the drug response. Transgene RNA is strongly induced and bulk protein is several-fold elevated; custom mouse MRI response and mostly normal histology do not establish eradication or physiological dosage.

[6] Brief MYC withdrawal in osteosarcoma promotes differentiation and sustained loss of neoplastic behavior despite later reactivation. This differentiation response is lineage specific and cannot establish treatment-free cure in every baseline driver pair.

[7] Genomically complex MYC lymphomas often regress after MYC inactivation; some escape with further chromosomal changes. The engineered lymphoma lineage and heterogeneous escape preclude universal dependence or cure.

[8] MYC withdrawal in liver tumors induces differentiation and dormancy; reactivation restores neoplastic properties in genetically related cells. Persistence is compatible with bulk addiction, and the abstract alone does not resolve every residual-cell dependency or reservoir-specific intervention.

[9] p53 deficiency impairs sustained MYC-withdrawal regression; p53 or thrombospondin restoration restores the response in the studied setting. Thrombospondin alone is insufficient; genotype and vascular mechanisms modify dependence rather than prove all-cancer withdrawal sufficiency.

[10] CD4 immunity and thrombospondin contribute to sustained regression after MYC or BCR-ABL inactivation. Host-dependent sustained control differs from immediate cell-autonomous collapse; linked erratum details were not examined.

[11] Conditional oncogenic Kras withdrawal regresses pancreatic lesions and changes the associated stroma. The engineered pancreas and p53 settings cannot establish all sporadic pancreatic tumors or residual-cell elimination.

[12] Switching off an endogenous conditional KRAS fusion allele regresses pancreatic tumors. The switch remains an engineered fusion and genotype-specific test; a linked author-name correction does not add an independent causal experiment.

[13] Dual MYC and KRAS mammary tumors regress more strongly when both are removed; recurrent tumors can reactivate inducible transgenes through rtTA mutations. Reactivation is different from verified driver-independent bypass; the result contests one-node generalization rather than proving every recurrent mechanism.

[14] MYC and KRAS withdrawal effects differ between lung tumors and lymphomas; dual withdrawal exceeds MYC-only lung responses. Cooperating pathways and tissue context preclude a uniform one-oncogene rule.

[15] BIM or PUMA loss reduces apoptosis and slows HER2-withdrawal regression; PUMA loss also impairs EGFR-withdrawal lung regression. Constitutive AKT/MEK separates downstream apoptotic routes. Apoptotic competence modifies response, and these inducible models do not show that every residual cell must depend on the driver.

[16] Fate-labeled MYC lymphoma cells undergo arrest, differentiation and apoptosis after MYC withdrawal. A selected engineered lymphoma is not a representative pan-cancer dependency survey.

[17] Omomyc inhibition of endogenous MYC regresses KRAS-driven lung tumors with reversible effects on normal regenerative tissues. MYC is a downstream common conduit rather than the mutated initiating KRAS itself; later work shares the model family.

[18] Repeated four-week Omomyc on/off cycles lower KRAS-lung tumor recurrence; after eight cycles, three of four mice have no scored tumors after four weeks off. Refractory lesions suppress Omomyc expression. All animals recur after a first cycle; the final off-treatment cohort is small, histology excludes atypical hyperplasia, and four weeks is not lifelong sterile eradication. Linked erratum details remain unexamined.

[19] Lung-specific transgenic EML4-ALK tumors respond to an ALK inhibitor in the experimental mouse setting. Forced expression and transformation assays do not by themselves establish a native universal maintenance dependence.

[22] In phase 1 CML testing, 53 of 54 patients receiving at least 300 mg daily obtain complete hematologic response. Dose-cohort selection, nonrandomization and hematologic response do not establish cure or an all-dose response.

[23] IRIS randomizes 1106 CML patients and shows better cytogenetic response and progression control with imatinib than interferon plus cytarabine. Crossover complicates survival inference; drug treatment does not equal genetic removal or residual-cell eradication.

[26] In IPASS, gefitinib favors progression control in the mutation-positive subgroup and disadvantages mutation-negative patients relative to chemotherapy. Enrollment uses clinical features rather than EGFR genotype, and the molecular analysis is a tested subsample; subgroup estimates are not the entire enrolled cohort.

[27] Baseline EGFR-mutant NEJ002 enrollment yields 73.7 percent response and 10.8-month median progression-free survival with gefitinib, versus 30.7 percent and 5.4 months with chemotherapy. Responses are not universal; reported overall-survival comparison is not statistically decisive, and recurrence remains compatible with bulk addiction.

[28] FLAURA randomizes 556 baseline EGFR-mutant patients; osimertinib improves progression-free survival relative to active first-generation EGFR inhibitors. Both arms inhibit the same driver, so this comparison does not isolate inhibition versus none; initial response is not eradication.

[30] HER2-selected metastatic breast cancer shows greater response and longer survival with trastuzumab plus chemotherapy than chemotherapy alone. Antibody-mediated immunity and chemotherapy contribute; this is not pure selective kinase withdrawal, and many tumors do not respond.

[31] BRIM-3 enrolls 675 baseline BRAF-mutant melanoma patients; vemurafenib yields 48 percent response versus 5 percent with dacarbazine and improves early survival. More than half lack an objective response in this report; the baseline genotype pair cannot be narrowed to responders after outcome.

[32] PROFILE 1007 enrolls 347 ALK-positive previously treated patients and improves response and progression-free survival with crizotinib versus chemotherapy. Interim overall survival is not significantly different; a linked erratum was not audited, and the outcome is not cure.

[33] PROFILE 1014 enrolls 343 untreated ALK-positive patients and improves response and progression-free survival with crizotinib versus chemotherapy. Crossover and immature survival limit causal survival claims; linked correction details remain unexamined.

[34] ALEX enrolls 303 ALK-positive patients and shows better progression and CNS control with alectinib than crizotinib. Two active ALK agents are compared; this is not a driver-off versus driver-on test or a treatment-free endpoint.

[35] CROWN enrolls 296 ALK-positive patients and improves response and progression control with lorlatinib versus crizotinib. Both arms inhibit ALK; toxicity and ongoing treatment remain, and later reports reuse this cohort.

[38] Resistant CML samples contain ABL gatekeeper mutations or amplification; reconstructed mutant kinase resists imatinib. Drug resistance with restored signaling supports continuing target dependence, not independence after successful blockade; nine selected resistant cases do not define population frequency.

[39] T790M appears in three of six progressing EGFR-inhibitor cases; independent RFLP and matched mutant constructs support reduced drug-sensitive phosphorylation. One pretreatment sample is scant; sequencing sensitivity cannot exclude rare pre-existing cells. Cell-line viability contrasts have other background differences.

[40] MET amplification activates ERBB3-mediated bypass; combined MET and EGFR inhibition restores response in resistant cells, with amplification in four of eighteen resistant patients. Culture rescue and selected progressing biopsies do not establish universal clinical bypass frequency or durable cure.

[41] Combined irreversible EGFR inhibition and cetuximab regresses resistant EGFR-mutant mouse tumors and lowers receptor signaling and abundance more effectively than either alone. Combination activity includes antibody and abundance effects; the model does not establish efficacy in every human resistance state.

[43] BRAF-mutant colon models resist BRAF inhibition through EGFR feedback; combined inhibition restores response, and ectopic EGFR can confer resistance in melanoma cells. The same driver behaves differently by tissue; experimental combination rescue does not prove all baseline BRAF tumors share one maintenance rule.

[44] ALK resistance biopsies and functional tests identify inhibitor-dependent spectra of secondary mutations and bypass sensitivity. Progressor selection and the shared clinical programme limit population estimates and independence; response after resistance is context specific.

[45] Experimental ALK selection and serial patient specimens show sequential compound mutations during inhibitor exposure. Ba/F3 selection and selected progressing samples cannot establish that all resistance follows this genetic route.

[47] Snail expression promotes recurrent mammary tumor behavior after Neu withdrawal and relates to a mesenchymal recurrence phenotype. Forced Snail gain does not establish native necessity of Snail in every residual tumor.

[48] Kras withdrawal initially regresses Kras/Mad2 lung tumors, followed by frequent recurrence with mostly absent oncogenic Kras transgene re-expression. The present case examines the abstract; original-body credit belongs to the preceding genome-instability review. Recurrence and aneuploidy do not directly prove a persister lineage or zero remaining instability.

[49] After Kras withdrawal, pancreatic tumors regress and twenty of twenty-eight recur; a subset lacks transgene signaling. YAP gain substitutes for Kras, while YAP/TEAD loss or binding-defective controls suppress bypass growth. About half reactivate the transgene through rtTA amplification; basal wild-type RAS remains. Relapse YAP-amplification and the engineered line constrain universal bypass claims; linked erratum details are unaudited.

[51] Rare PC9 persisters survive concentrations far above drug IC50 with EGFR suppression; single-cell clones regenerate tolerance, withdrawal restores sensitivity, and KDM5A/HDAC/IGF1R controls separate reversible states from later T790M colonies. Cell-line state switching does not establish every patient reservoir; expanded persisters reverse more slowly, and pharmacologic prevention is not clinical treatment-free cure.

[54] Barcoding, early resistant clones, persister pools and spiked-clone kinetic controls distinguish pre-existing and later T790M routes. Late-route cells resist apoptosis despite similar EGFR/ERK inhibition; navitoclax restores death and tumor response. Human line-response correlations do not trace original patient clones; low-frequency pretreatment mutations have detection limits, and the later route is not the only clinical route.

[55] Reversible persisters selectively depend on GPX4; knockout, ferrostatin rescue and iron/lipid controls identify ferroptotic death. Unmasking GPX4 deletion after tumor shrinkage prevents relapse in the tested melanoma xenograft. The in vivo intervention is genetic because compounds lack systemic bioavailability; continued targeted treatment, one xenograft and absent clinical safety testing do not establish treatment-free human cure.

[56] Barcoded cycling persister lineages show distinct antioxidant and fatty-acid-oxidation requirements under EGFR inhibition, with related states in human residual samples. Culture lineage dynamics and patient transcriptional resemblance do not quantify every patient's persister fraction or prove all clinical recurrence is state driven.

[57] APOBEC3A loss reduces treatment-associated mutagenesis and delays resistance in the experimental systems. Delay does not establish a sole universal resistance mechanism or spontaneous cancer initiation by APOBEC.

[58] FAK-YAP inhibition with EGFR inhibitors reduces residual disease and improves tumor response in tested experimental systems. Combined intervention and selected models do not establish a safe or universally curative clinical regimen.

[60] TROP2 derepression accompanies EGFR-inhibitor persisters; experimental TROP2-directed combination delays relapse, with preliminary phase-2 observations. Preliminary nonrandomized observations are not controlled treatment-free cure or universal residual-cell dependence.

[63] NRAS melanoma RAS-on response and resistance depend on alterations including cyclophilin-A and MAP2K1; two clinical cases show divergent response. Two cases do not define response frequency or clinical cure. The peer-reviewed report replaces its earlier preprint without double counting.

[64] RAS-on inhibitor combinations regress tested tumors and require host T cells for deeper responses. These are preclinical results, not demonstrated regression or complete responses in NSCLC patients; host dependence limits purely cell-autonomous interpretation.

[65] Acquired EML4-ALK restores wild-type RAS signaling under KRAS-G12C inhibition; continued KRAS blockade creates a context-specific ALK sensitivity and schedule-dependent combination response. Selected models and treatment sequence do not establish population prevalence or clinical eradication.

[66] RAS-on inhibition regresses pancreatic tumors in tested models; T cells and dendritic cells are required for the deeper response, with greater complete responses after immune combination. Complete responses are preclinical and combination dependent; shared drug-class programmes are not independent clinical validation or treatment-free cure.

Tally: 51 discriminating contrasts. Null and negative contrasts are discriminators, not positive threshold crossings. Reports and replications are not independent causal lines. Positive contrasts are not independent causal lines. Follow-up efficacy reports are not retallied as new perturbations. STIM cessation, target-engaged survival and observational biopsy contrasts do not inflate positive-support tallies. Transformation discovery and genomic resistance screens are not whole-tumor native maintenance interventions.

### Cap audit, precedence and Β/Γ/Δ boundary

The index explicitly permits scope-matched specific-pair maintenance scoring. Η applies to the assembled evidence for SC1, which covers weaknesses of forced models with endogenous conditional perturbation, resistance reconstruction and molecularly enrolled human trial families. The same convergence does not upgrade SC2 or SC3. Their Γ scores follow missing full-scope thresholds, rather than applying a literal in-vitro-only cap to a mixed in vivo/clinical corpus.

Gamma hard-rule audit: Cell-line viability and transformation capacity alone do not establish native cancer-maintenance necessity. Patient genotype must be known before response; retrospective responder-selected mutation discovery is separate from molecularly enrolled trials. A drug-resistant kinase with restored signaling is not independence after successful driver blockade. Radiographic response, deep molecular remission and on-treatment progression-free survival are not sterile treatment-free eradication. The assembled endogenous, genetic, mechanistic and clinical evidence is assessed as a set rather than re-capped by its weakest assay class.

SC1 field-valid definition: “
- Endogenous genetic manipulation in autochthonous models
- Patient-derived system perturbation with matched rescue AND in vivo confirmation
- Clonal tracking demonstrating persistence of dependency through selection pressure” Positive full-scope result: yes. Η is supported for the prospectively named maintenance/regression pairs by endogenous conditional KRAS and MYC interventions, orthogonal EGFR withdrawal/drug controls, resistance alleles and several molecularly enrolled randomized trial families. This means causal regression in defined settings, not every enrolled tumor responding or every tumor cell requiring the driver. Nonresponse and context-specific bypass are major exceptions, preventing Θ and Ω.

SC2 field-valid definition: “
- Endogenous genetic manipulation in autochthonous models
- Patient-derived system perturbation with matched rescue AND in vivo confirmation
- Clonal tracking demonstrating persistence of dependency through selection pressure” Positive full-scope result: no. Experimental dual-driver and tissue comparisons, heterogeneous KRAS cell dependency and genome-wide resistance screens show context dependence. These reports do not constitute a representative native pan-cancer maintenance map establishing a single dominant dependency in most cancers. The broad model has mechanistic plausibility, but its prespecified denominator and matching causal test are missing.

SC3 field-valid definition: “
- Endogenous genetic manipulation in autochthonous models
- Patient-derived system perturbation with matched rescue AND in vivo confirmation
- Clonal tracking demonstrating persistence of dependency through selection pressure” Positive full-scope result: no. Initial regression and prolonged progression-free survival differ from eradication. Target-engaged human stem-cell survival, reversible tolerance, late resistance, transgene-independent recurrence and responder-selected treatment cessation contradict universal residual-cell elimination. Selected osteosarcoma differentiation or four-week off-treatment tumor absence after repeated Omomyc does not establish general lifelong cure.

### Upward threshold ladder

SC1 Δ: “Requires at least one meaningful causal perturbation showing the proposed mechanism contributes to initiation, maintenance, or relapse.” Satisfied: yes. 

SC1 Ε: “Requires context-specific causal validity with explicit limits. Cancer stem cells often belong here when a stem-like hierarchy exists in some tumor contexts but fails as a universal model.” Satisfied: yes. 

SC1 Ζ: “Requires:
- Independent causal confirmation across models
- Reduced viability of major alternatives
- Relevance beyond one assay class” Satisfied: yes. 

SC1 Η: “Requires:
- Strong endogenous evidence
- Therapeutic or mechanistic consequence
- Reproducibility across disease settings within the claimed scope” Satisfied: yes. 

SC1 Θ: “Requires that the mechanism explains most of the claimed cancer behavior with only minor exceptions.” Satisfied: no. To cross Θ, explain most behavior within every named baseline pair with only minor exceptions, including prospective nonresponders and residual compartments; independently verify target engagement, baseline modifiers, driver restoration and long-term course without responder-defined scope. Ω additionally needs necessity and sufficiency across those cancer contexts.

SC2 Δ: “Requires at least one meaningful causal perturbation showing the proposed mechanism contributes to initiation, maintenance, or relapse.” Satisfied: no. Test a prespecified representative pan-cancer sample with native allele-specific withdrawal and matched restoration, direct target engagement and in vivo maintenance endpoints. Establish the one-driver frequency before assigning a general causal rule.

SC3 Δ: “Requires at least one meaningful causal perturbation showing the proposed mechanism contributes to initiation, maintenance, or relapse.” Satisfied: no. After verified driver withdrawal, lineage-track all detectable residual compartments and compare treatment-free recurrence with matched restoration or selective reservoir elimination in independent native models; test prospective cessation cohorts without treating assay negativity as absolute absence.

Upper ladder: Genetic switches, native common-conduit inhibition, resistant-target reconstruction, genotype-defined treatment cohorts and recurrence/cessation tests are available upper-ladder routes. No RCT-only rule is imposed. They support Η for defined maintenance pairs, but do not remove major nonresponse and residual-cell exceptions or establish necessity and sufficiency across scope.

Independence: Clinical follow-ups are grouped by trial, not by paper: FLAURA has two reports and CROWN three. MYC/Omomyc, inducible Neu, EGFR transgene, inducible pancreatic Kras, MGH serial-biopsy programmes and persister lineages have shared components. Distinct drug classes alone are not independence; genetic specificity, non-target genotype controls, native-locus switches and molecular human endpoints exclude partly different errors. The retained corpus has independent disease settings rather than 66 independent causal confirmations. Marinkovic/Wirth GFP-fate transplantation is a separate laboratory model and is not assigned to the Felsher MYC programme merely because both use conditional MYC.

Endogenous conditional driver switch [12]: A native-locus conditional KRAS activity switch tests maintenance rather than mere transformed-cell sensitivity; its fusion and pancreas genotype remain limitations.

Native common-conduit dependence [17, 18]: Omomyc acts on endogenous MYC in KRAS-driven autochthonous tumors and measures recurrence; the two papers share the same model programme and form one line.

Orthogonal target withdrawal and engagement [5, 15]: EGFR genetic withdrawal and drug inhibition with continued induction and KRAS controls address drug nonspecificity; apoptotic-effector genetics map a different potential failure. Related inducible lines are grouped.

Target resistance reconstruction [38, 39, 41]: BCR-ABL and EGFR resistance alleles and complementary EGFR blockade distinguish loss of target inhibition from target independence; EGFR follow-up constructs share a programme.

Independent molecular clinical settings [23, 27, 30, 31, 33, 34]: Distinct CML, EGFR-lung, HER2-breast, BRAF-melanoma and ALK-lung trials address engineered-model artifact. Shared biomarker selection and drug specificity remain; active-inhibitor comparisons do not substitute for genetic withdrawal.

Functional bypass [49, 43]: YAP/TEAD withdrawal/rescue and BRAF tissue-specific EGFR feedback have different nodes, tissues and control failures; selected experimental models limit prevalence.

Dynamic persister mechanisms [51, 54, 55]: Single-cell reversible states and barcode/kinetic routes share some EGFR platforms. Independent GPX4 genetic/antioxidant controls and residual melanoma course test a distinct state-specific survival mechanism.

### Falsification and pitfalls

Residual phenotype persists despite target inhibition. Strongly inhibited CML stem/progenitor cultures survive with cytokines, and genotype-defined cohorts retain nonresponders and later persistent disease. These falsify all-cell elimination, while not disproving bulk regression (PMID21157039,21639808,26828195). Additional downgrade: False.

Dependency can change through selection. YAP/TEAD bypass after verified Kras withdrawal and lineage conversion with retained EGFR mutations contest an immutable dominant node (PMID24954535,21430269). Additional downgrade: False.

Scope defined by outcome. Responder-selected discovery and STIM deep-response enrollment have separate limitations; SC1 stays within baseline-defined cohorts including nonresponders.

Regression substituted for eradication. MRI absence, deep molecular response, prolonged on-treatment PFS and four-week off-treatment histology remain distinct.

Target resistance confused with target independence. Gatekeeper mutations, rtTA reactivation and genuine YAP bypass are separately appraised.

Cell-state plasticity ignored. Single-cell reversibility, slow reversion after expansion and two T790M routes remain explicit.

Repeated reports counted as replication. FLAURA/CROWN outcomes and related mouse programmes are grouped.

Provider summary substituted for original evidence. Only one of eight requested Consensus body checks was directly comparable; seven provider gaps earn no body credit.

### Adversarial steelman: above, then below

One rung higher Θ: long-term ALK control and convergent genetic/clinical regression could imply that the mechanism explains most behavior in the named pairs. But substantial initial nonresponse, primary tissue-specific feedback, apoptosis/host requirements and residual-cell escape are major rather than minor exceptions.

One rung lower Ζ: many genetic studies force transgene expression and selective drugs are pleiotropic, so independent causal confirmation might be all that is justified. That argument overlooks endogenous conditional KRAS, inhibition of native MYC, orthogonal drug/withdrawal controls and independently molecularly enrolled clinical consequence across several tumor settings.

Ζ understates the assembled endogenous and therapeutic evidence within SC1; Θ overstates completeness within the baseline cohorts. Retain Η with Moderate confidence and keep the broader SC2/SC3 grades separate.

### Final score, causal direction and M/S/D

The preregistered scope-matched rule reports SC1 at Η: endogenous conditional-driver and common-conduit interventions, genetic withdrawal, mechanistic resistance/rescue and baseline molecularly defined clinical trials converge on tumor-maintenance dependence across named pairs. Significant primary nonresponse, host/apoptosis dependence and persistence prevent Θ or Ω. SC2 pan-cancer one-driver generalization and SC3 residual-cell eradication remain Γ as broader mechanistic claims without matching general causal validation.

Apply the direction procedure to the scoped dominant-driver mechanism: it is not demonstrated to be produced by tumor regression, and causal dependence is present. Genetic engagement initiates tumors in defined experimental lineages, selective withdrawal substantially regresses or delays established disease, and switch/restoration or target-resistance contrasts identify a rate-limiting maintenance node. Thus primary driver is supported for the named bulk-tumor settings. This role does not assert that all baseline tumors or every residual compartment require the same node; initiating sufficiency from forced expression alone is not generalized to every sporadic cancer.

Momentum: Mechanistic only. The named 2026 RAS-on genotype/resistance and immune-combination programmes are active causal mechanistic work in preclinical models (PMID42171647,41670434,40057911), but none addresses the boundary check that would cross the next threshold: no named ongoing experiment prospectively maps nonresponse within each baseline pair or tests residual-compartment elimination with matched restoration. The review asserts no recruiting trial and no imminent outcome, which is the framework's definition of mechanistic rather than decisive work. Momentum modifies no letter.

Stability: Stable. No named pending result is shown to be likely to move the scoped Η grade imminently; new residual-disease results could change the separate eradication subclaim.

Divergence: Contested; no single implied-level arithmetic. Artifact for the split: the field's own framing of oncogene addiction as a near-universal single-node vulnerability implies a high rung, while the same literature's documented primary nonresponse, tissue-specific feedback and target-engaged residual survival imply a much lower one for the broad claim, so no single implied level is defensible (PMID20573926,21639808,22281684,21157039). These dimensions do not change level or confidence.

## PROPOSED DECISIVE EXPERIMENTS

In baseline molecularly defined driver–tumor cohorts and matched native models, combine allele-specific withdrawal with direct engagement measures, matched driver restoration and longitudinal lineage-resolved residual-cell tracking. Prospectively include nonresponders; perturb candidate bypass and reservoir dependencies separately. Follow sustained control both during treatment and after predefined cessation.

Readout: Bulk regression, apoptosis, driver signaling, residual clone fate, lineage transition, relapse and treatment-free survival over a prespecified interval. Controls: Restoration at native abundance, non-target genotype controls, equal initial tumor burden, host/apoptosis strata, inhibitor-resistant alleles, barcode collision and rare-clone detection controls, matched cessation eligibility. Supporting: Independent native models and baseline patient strata show driver-specific regression, while identified modifiers explain most initial failures and residual survival; selective residual-dependency correction adds sustained treatment-free control. Falsifying: Fully verified selective withdrawal leaves substantial bulk tumors unchanged across baseline-defined pairs, or restored driver cannot reverse a purported withdrawal effect; residual populations recur despite candidate rescue and elimination controls.

## RESEARCH TIMELINE

| Year | Milestone | Primary source | Interpretation |
|---|---|---|---|
| 1996 | Selective ABL colony suppression | [PMID8616716](https://pubmed.ncbi.nlm.nih.gov/8616716/) | Ex vivo selectivity, not stem-cell eradication. |
| 1999 | Reversible MYC hematopoietic tumors | [PMID10488335](https://pubmed.ncbi.nlm.nih.gov/10488335/) | Conditional maintenance test. |
| 1999 | RAS melanoma withdrawal | [PMID10440378](https://pubmed.ncbi.nlm.nih.gov/10440378/) | Defined genotype and tissue. |
| 2000 | BCR-ABL acute B-cell leukemia switch | [PMID10615128](https://pubmed.ncbi.nlm.nih.gov/10615128/) | Distinct from human CML. |
| 2001 | Clinical CML response | [PMID11287972](https://pubmed.ncbi.nlm.nih.gov/11287972/) | Dose-cohort hematologic effect. |
| 2001 | ABL resistance reconstruction | [PMID11423618](https://pubmed.ncbi.nlm.nih.gov/11423618/) | Restored target signaling. |
| 2001 | HER2-selected therapy | [PMID11248153](https://pubmed.ncbi.nlm.nih.gov/11248153/) | Antibody plus chemotherapy. |
| 2002 | Neu regression and recurrence | [PMID12498714](https://pubmed.ncbi.nlm.nih.gov/12498714/) | Bulk regression without residual cure. |
| 2002 | MYC osteosarcoma differentiation | [PMID12098700](https://pubmed.ncbi.nlm.nih.gov/12098700/) | Lineage-specific sustained fate. |
| 2003 | Randomized CML consequence | [PMID12637609](https://pubmed.ncbi.nlm.nih.gov/12637609/) | Progression and cytogenetic benefit. |
| 2004 | EGFR response mutation discovery | [PMID15118073](https://pubmed.ncbi.nlm.nih.gov/15118073/) | Outcome-selected discovery. |
| 2004 | MYC liver dormancy | [PMID15475948](https://pubmed.ncbi.nlm.nih.gov/15475948/) | Residual state after regression. |
| 2005 | EGFR gatekeeper resistance | [PMID15737014](https://pubmed.ncbi.nlm.nih.gov/15737014/) | Construct specificity and detection limits. |
| 2006 | EGFR genetic/drug comparison | [PMID16705038](https://pubmed.ncbi.nlm.nih.gov/16705038/) | Continued induction and KRAS control. |
| 2007 | MET feedback bypass | [PMID17463250](https://pubmed.ncbi.nlm.nih.gov/17463250/) | Combined functional inhibition. |
| 2008 | Native MYC targeting | [PMID18716624](https://pubmed.ncbi.nlm.nih.gov/18716624/) | Autochthonous KRAS and endogenous conduit. |
| 2010 | Reversible persisters | [PMID20371346](https://pubmed.ncbi.nlm.nih.gov/20371346/) | Single-cell origin and chromatin controls. |
| 2010 | Baseline EGFR-mutant trial | [PMID20573926](https://pubmed.ncbi.nlm.nih.gov/20573926/) | Includes all enrolled responders and nonresponders. |
| 2010 | CML treatment cessation | [PMID20965785](https://pubmed.ncbi.nlm.nih.gov/20965785/) | Prior deep-response selection. |
| 2011 | Target-engaged CML survival | [PMID21157039](https://pubmed.ncbi.nlm.nih.gov/21157039/) | Cytokine-dependent ex vivo exception. |
| 2011 | BRAF-melanoma trial | [PMID21639808](https://pubmed.ncbi.nlm.nih.gov/21639808/) | Strong effect with substantial nonresponse. |
| 2012 | BRAF colon tissue boundary | [PMID22281684](https://pubmed.ncbi.nlm.nih.gov/22281684/) | EGFR feedback differs by tissue. |
| 2012 | Endogenous KRAS activity switch | [PMID22606359](https://pubmed.ncbi.nlm.nih.gov/22606359/) | Native-locus maintenance test. |
| 2013 | Repeated Omomyc and recurrence | [PMID23475959](https://pubmed.ncbi.nlm.nih.gov/23475959/) | Small four-week off-treatment endpoint. |
| 2013 | Apoptosis-effector dependence | [PMID23532334](https://pubmed.ncbi.nlm.nih.gov/23532334/) | BIM/PUMA modifies withdrawal response. |
| 2014 | YAP/TEAD bypass rescue | [PMID24954535](https://pubmed.ncbi.nlm.nih.gov/24954535/) | True bypass distinguished from rtTA reactivation. |
| 2014 | Baseline ALK first-line trial | [PMID25470694](https://pubmed.ncbi.nlm.nih.gov/25470694/) | Independent molecular disease setting. |
| 2016 | Two T790M evolution routes | [PMID26828195](https://pubmed.ncbi.nlm.nih.gov/26828195/) | Barcode and kinetic discrimination. |
| 2017 | GPX4 residual dependency | [PMID29088702](https://pubmed.ncbi.nlm.nih.gov/29088702/) | Genetic and ferroptosis rescue. |
| 2018 | Baseline EGFR active-agent trial | [PMID29151359](https://pubmed.ncbi.nlm.nih.gov/29151359/) | Same cohort as later survival report. |
| 2020 | CROWN ALK trial | [PMID33207094](https://pubmed.ncbi.nlm.nih.gov/33207094/) | Later follow-ups are not new trials. |
| 2021 | Cycling persister lineages | [PMID34381210](https://pubmed.ncbi.nlm.nih.gov/34381210/) | Functional state diversity. |
| 2023 | APOBEC resistance route | [PMID37407818](https://pubmed.ncbi.nlm.nih.gov/37407818/) | Mutation-dependent delay. |
| 2024 | FAK/YAP residual intervention | [PMID38702301](https://pubmed.ncbi.nlm.nih.gov/38702301/) | Experimental combination benefit. |
| 2025 | RAS-on and host dependence | [PMID40057911](https://pubmed.ncbi.nlm.nih.gov/40057911/) | Preclinical combination responses. |
| 2026 | TROP2 persister targeting | [PMID42314664](https://pubmed.ncbi.nlm.nih.gov/42314664/) | Experimental and preliminary clinical evidence. |
| 2026 | NRAS RAS-on sensitivity/resistance | [PMID42171647](https://pubmed.ncbi.nlm.nih.gov/42171647/) | Journal replaces preprint; two clinical cases. |
| 2026 | RAS-on immune combination | [PMID41670434](https://pubmed.ncbi.nlm.nih.gov/41670434/) | Preclinical, not patient response data. |
| 2026 | ALK bypass under KRAS treatment | [PMID42329099](https://pubmed.ncbi.nlm.nih.gov/42329099/) | Sequence-specific preclinical dependence. |
| 2026 | CROWN seven-year follow-up | [PMID42217582](https://pubmed.ncbi.nlm.nih.gov/42217582/) | On-treatment durability, same cohort. |

## KNOWLEDGE GAPS

Measure native driver activity and engagement in each surviving compartment, separating kinase function from entire-protein dependence.

Resolve whether reversible survivors arise before treatment, during exposure or by rare-clone selection using calibrated barcode and mutation-detection limits.

Prospectively map primary nonresponse within each baseline genotype pair rather than retrospectively defining addiction by response.

Separate true bypass from transgene reactivation, inadequate pharmacology and persistent wild-type pathway flux.

Establish the frequency of single-driver dependence across a representative pan-cancer maintenance denominator.

Test reservoir-specific loss and matched restoration with endogenous lineage-resolved relapse rather than culture viability alone.

Validate residual-state interventions in independent autochthonous models with dose, host and toxicity controls.

Distinguish preclinical immune-dependent complete responses from controlled clinical treatment-free cure.

Define cessation eligibility and sensitive residual assays prospectively; long-term molecular negativity does not prove absolute zero cells.

Audit inaccessible bodies, linked errata and supplements before claiming complete-paper review or transferring detailed control claims to the website.

## METHODS AND CONSENSUS CONTRIBUTION

Consensus was used through the signed-in browser for discovery, counterevidence, research-gap and timeline suggestions, and attempted checks of specific indexed full-text passages. Provider statements about Methods/Results and any returned passages were compared with original-paper passages where available; failed retrievals, source substitutions and unverified aspects are recorded in the case reconciliation. Successful original-passage comparisons are distinguished from provider summaries and assertions of access. Provider screening counts, citation counts and meters were not used as primary counts or HSEF scores. This source log records access depth; it does not claim complete-paper or supplement review. The monthly API allowance was conserved: zero Consensus API calls in this batch.

## REVIEWER NOTES

66 primary reports; nine selected partial-body examinations and 57 complete abstract/status examinations. Consensus Pro hit its per-request tool limit and returned original-text material for only Politi among eight requested originals; the other provider access limits are preserved. No complete-paper audit, blinded calibration or independent peer review is claimed. Correction details attached to selected papers remain incompletely examined.

Search leads, reviews, protocols and correction notices do not pad the primary count. Downloaded XML does not earn examination credit. Linked notices remain attached; unexamined correction details limit inference. This is a coordinator source/claim consistency assessment, not independent blinded calibration.

## ADVERSARIAL REVIEW AND APPROVAL

Authoring model: **Sol 6.1**. Adversarial reviewer: **Claude Opus 5.5 (Anthropic)**. Date: **2026-10-14** (Round 2, group 12).
Scope of checks: 66/66 reference PMIDs verified against PubMed metadata (0 failed, 0 retractions or expressions of concern); 12 load-bearing claims spot-checked against abstracts or the review's declared partial-body receipts; scoring logic re-derived against `hsef-framework.md` and the v5.3 SKILL (hard caps, pre-registered aggregation, independence basis, upward ladder, M/S/D artifacts); the known upstream defects and the 9 October revision regressions screened.

| ID | Severity | Location | Original to corrected | Basis |
|---|---|---|---|---|
| AR-1 | Major | Consensus provenance reconciliation block (review) and `consensus-addendum.md` | "The earlier adversarial approval refers to its original edition" to "No earlier adversarial approval is attached to this case" | Revision-introduced regression. The sentence was spliced in unconditionally by the 9 October integration step. The earlier edition (`hsef_v53_popularity_ranks_36_40_2026-10-08/oncogene-addiction/`) contains no ADVERSARIAL REVIEW AND APPROVAL section and no adversarial banner, so there is no earlier approval to refer to. |
| AR-2 | Major | AT A GLANCE, "Best-supported rival" | Named the nominal model (M1, Scoped dominant-oncogene maintenance dependence) as the best-supported rival; now names M2 Clonal bypass and lineage plasticity (Zeta) | Builder defect: the field prints `models_evaluated[0]`, which the score card marks `is_nominal: true`, `rank_in_tournament: 1`. A rival must be a non-nominal model, and the highest-ranked one is M2. Supporting citations re-pointed to M2's own evidence (resistance-allele reconstruction, YAP/TEAD bypass with binding-defective controls); Politi's nominal-supporting contrast removed. |
| AR-3 | Major | AT A GLANCE Momentum column; "Final score, causal direction and M/S/D"; score card `extended_scoring_MSD.momentum` | "Decisive work underway" to "Mechanistic only" | The required artifact names only completed and published 2026 RAS-on preclinical programmes and then states that no recruiting trial or imminent outcome is asserted. "Decisive work underway" requires that someone is running an experiment that could cross the next threshold; the named work addresses no boundary check (prospective nonresponse mapping within each baseline pair, residual-compartment elimination with matched restoration). The framework's table defines active causal mechanistic work with no boundary-check attempt as "Mechanistic only". Momentum modifies no letter. |
| AR-4 | Minor | Score card `final_score.confidence_rationale` (3 occurrences) | "density, heterogeneous controls and incomplete access set the confidence" to a rationale stated on evidence-density and reproducibility grounds, with access explicitly excluded | The framework states that confidence reflects evidence density and reproducibility, not the reviewer's access. The Moderate band is retained because the Moderate text is met on its own terms (multiple independent studies, partial replication, persisting methodological heterogeneity: shared model programmes among the founding perturbations, active-comparator clinical arms, heterogeneous control sets). High is not met because the review's own independence analysis counts independent disease settings rather than extensive cross-modality replication of the withdrawal experiment. |
| AR-5 | Minor | Final score block Divergence line; score card `extended_scoring_MSD.divergence.justification` | Value unchanged at Contested; the checkable artifact naming the two conflicting framings added | Divergence is computed, not felt, and Contested requires showing that the field is genuinely split. Both prose and score card previously asserted only the absence of an arithmetic gap. |
| AR-6 | Minor | Evidence cards [10], [12], [52]; six score-card ledger rows | "ErratumIn: PMIDnot supplied" to "ErratumIn: PMID not supplied" | Builder rendering defect fusing the label to the next word. The flag's content is correct: PubMed supplies no erratum PMID for these records, and the cards already state that linked notice details were not examined. |

### Rejected findings (checked, no change)

1. **Level and aggregation.** Scope-matched is applied correctly. The registered central claim is scoped to prospectively defined driver-tumor pairs, so SC1 is the subclaim whose scope matches it, and the broader SC2 (pan-cancer one-driver) and SC3 (residual eradication) are reported separately at Gamma rather than being allowed to gate a headline whose scope they exceed. The alternatives are reported as the framework requires (weakest-link Gamma, primary-subclaim Eta). Theta rejected: primary nonresponse, tissue-specific feedback, host and apoptosis dependence and residual survival are major, not minor, exceptions. Zeta rejected: endogenous conditional KRAS, native MYC inhibition, orthogonal withdrawal and drug controls, and several independently genotype-enrolled trial families exceed single-line causal confirmation.
2. **Confidence band.** Moderate retained (see AR-4); only the stated basis was wrong, not the value.
3. **`v5.claim_form` WARN ("phenomenon").** Rejected per convention: the registered restatement is a declarative causal proposition naming both the mechanism (continuous dominant oncogene signaling required for maintenance) and its consequence (regression on selective withdrawal).
4. **Citation years.** References 7 (2002/2003), 28 (2017/2018), 29 (2019/2020), 52 (2010/2011), 59 (2022/2023) and 62 (2013/2014) differ from the PubMed e-pub year by one. Rejected per convention.
5. **Collective-author defect, empty `DOI: .` fields, fused numeral tokens, retrieval artifacts.** Screened across all 66 references, the timeline and the evidence cards; none present. The STIM reference carries a trailing consortium entry in PubMed but its first author is a named individual, so the leading-comma defect does not arise.
6. **REVIEWER NOTES / Consensus contradiction.** Absent here. METHODS AND CONSENSUS CONTRIBUTION states that Consensus was used through the signed-in browser with zero API calls, and REVIEWER NOTES independently records the per-request tool limit and the single comparable original-text return (Politi, of eight requested). The two blocks agree, so no scoping edit was needed.
7. **Independence basis for Eta.** Checked against the shared-failure test. The seven recorded lines (endogenous conditional driver switch; native common-conduit inhibition; orthogonal target withdrawal plus engagement; resistance reconstruction; independent molecular clinical settings; functional bypass; dynamic persister mechanisms) are not defeated by one systematic error, and the review correctly collapses the Omomyc pair, the FLAURA pair and the three CROWN reports into single lines.
8. **Clinical and numeric claims.** Verified against PubMed abstracts: IRIS 1106 randomized; NEJ002 73.7 versus 30.7 percent response and 10.8 versus 5.4 months; BRIM-3 675 enrolled, 48 versus 5 percent; FLAURA 556; PROFILE 1007 347; PROFILE 1014 343; ALEX 303; CROWN 296 with the seven-year 55 versus 3 percent progression-free survival in the same cohort; Druker phase 1, 53 of 54 at 300 mg or above; MET amplification in 4 of 18 (22 percent). The Wei and Orlen RAS-on results are correctly held to preclinical scope, and Foth's two divergent patient cases are not inflated into a response rate.

### Residual limitations

Kapoor's 28 regressing and 20 recurring tumors, the Omomyc three-of-four four-week figure and Corbin's "more than 95 percent Philadelphia positive" are Results-level numbers confirmable in this pass only against the review's own declared partial-body receipts. Access depth is unchanged (9 partial-body, 57 complete abstract/status). Linked erratum and correction details remain unexamined, as the review states.

**Approval:** Following adversarial review and the corrections listed above, the revised document is approved. — Claude Opus 5.5 (Anthropic), 2026-10-14.

## REFERENCES

1. Felsher DW, Bishop JM et al. Reversible tumorigenesis by MYC in hematopoietic lineages. Molecular cell. 1999. [PMID 10488335](https://pubmed.ncbi.nlm.nih.gov/10488335/). DOI: 10.1016/s1097-2765(00)80367-6. Access: abstract_metadata. Study: PMID10488335.

2. Chin L, Tam A, Pomerantz J et al. Essential role for oncogenic Ras in tumour maintenance. Nature. 1999. [PMID 10440378](https://pubmed.ncbi.nlm.nih.gov/10440378/). DOI: 10.1038/22788. Access: abstract_metadata. Study: PMID10440378.

3. Huettner CS, Zhang P, Van Etten RA et al. Reversibility of acute B-cell leukaemia induced by BCR-ABL1. Nature genetics. 2000. [PMID 10615128](https://pubmed.ncbi.nlm.nih.gov/10615128/). DOI: 10.1038/71691. Access: abstract_metadata. Study: PMID10615128.

4. Moody SE, Sarkisian CJ, Hahn KT et al. Conditional activation of Neu in the mammary epithelium of transgenic mice results in reversible pulmonary metastasis. Cancer cell. 2002. [PMID 12498714](https://pubmed.ncbi.nlm.nih.gov/12498714/). DOI: 10.1016/s1535-6108(02)00212-x. Access: abstract_metadata. Study: PMID12498714.

5. Politi K, Zakowski MF, Fan PD et al. Lung adenocarcinomas induced in mice by mutant EGF receptors found in human lung cancers respond to a tyrosine kinase inhibitor or to down-regulation of the receptors. Genes & development. 2006. [PMID 16705038](https://pubmed.ncbi.nlm.nih.gov/16705038/). DOI: 10.1101/gad.1417406. Access: full_text_partial. Study: PMID16705038.

6. Jain M, Arvanitis C, Chu K et al. Sustained loss of a neoplastic phenotype by brief inactivation of MYC. Science (New York, N.Y.). 2002. [PMID 12098700](https://pubmed.ncbi.nlm.nih.gov/12098700/). DOI: 10.1126/science.1071489. Access: abstract_metadata. Study: PMID12098700.

7. Karlsson A, Giuriato S, Tang F et al. Genomically complex lymphomas undergo sustained tumor regression upon MYC inactivation unless they acquire novel chromosomal translocations. Blood. 2003. [PMID 12517816](https://pubmed.ncbi.nlm.nih.gov/12517816/). DOI: 10.1182/blood-2002-10-3091. Access: abstract_metadata. Study: PMID12517816.

8. Shachaf CM, Kopelman AM, Arvanitis C et al. MYC inactivation uncovers pluripotent differentiation and tumour dormancy in hepatocellular cancer. Nature. 2004. [PMID 15475948](https://pubmed.ncbi.nlm.nih.gov/15475948/). DOI: 10.1038/nature03043. Access: abstract_metadata. Study: PMID15475948.

9. Giuriato S, Ryeom S, Fan AC et al. Sustained regression of tumors upon MYC inactivation requires p53 or thrombospondin-1 to reverse the angiogenic switch. Proceedings of the National Academy of Sciences of the United States of America. 2006. [PMID 17056717](https://pubmed.ncbi.nlm.nih.gov/17056717/). DOI: 10.1073/pnas.0608017103. Access: abstract_metadata. Study: PMID17056717.

10. Rakhra K, Bachireddy P, Zabuawala T et al. CD4(+) T cells contribute to the remodeling of the microenvironment required for sustained tumor regression upon oncogene inactivation. Cancer cell. 2010. [PMID 21035406](https://pubmed.ncbi.nlm.nih.gov/21035406/). DOI: 10.1016/j.ccr.2010.10.002. Access: abstract_metadata. Study: PMID21035406.

11. Collins MA, Bednar F, Zhang Y et al. Oncogenic Kras is required for both the initiation and maintenance of pancreatic cancer in mice. The Journal of clinical investigation. 2012. [PMID 22232209](https://pubmed.ncbi.nlm.nih.gov/22232209/). DOI: 10.1172/JCI59227. Access: abstract_metadata. Study: PMID22232209.

12. Wang Z, Feng Y, Bardeesy N et al. Temporal dissection of K-ras(G12D) mutant in vitro and in vivo using a regulatable K-ras(G12D) mouse allele. PloS one. 2012. [PMID 22606359](https://pubmed.ncbi.nlm.nih.gov/22606359/). DOI: 10.1371/journal.pone.0037308. Access: abstract_metadata. Study: PMID22606359.

13. Podsypanina K, Politi K, Beverly LJ et al. Oncogene cooperation in tumor maintenance and tumor recurrence in mouse mammary tumors induced by Myc and mutant Kras. Proceedings of the National Academy of Sciences of the United States of America. 2008. [PMID 18356293](https://pubmed.ncbi.nlm.nih.gov/18356293/). DOI: 10.1073/pnas.0801197105. Access: abstract_metadata. Study: PMID18356293.

14. Tran PT, Fan AC, Bendapudi PK et al. Combined Inactivation of MYC and K-Ras oncogenes reverses tumorigenesis in lung adenocarcinomas and lymphomas. PloS one. 2008. [PMID 18461184](https://pubmed.ncbi.nlm.nih.gov/18461184/). DOI: 10.1371/journal.pone.0002125. Access: abstract_metadata. Study: PMID18461184.

15. Bean GR, Ganesan YT, Dong Y et al. PUMA and BIM are required for oncogene inactivation-induced apoptosis. Science signaling. 2013. [PMID 23532334](https://pubmed.ncbi.nlm.nih.gov/23532334/). DOI: 10.1126/scisignal.2003483. Access: full_text_partial. Study: PMID23532334.

16. Marinkovic D, Marinkovic T, Mahr B et al. Reversible lymphomagenesis in conditionally c-MYC expressing mice. International journal of cancer. 2004. [PMID 15095297](https://pubmed.ncbi.nlm.nih.gov/15095297/). DOI: 10.1002/ijc.20099. Access: abstract_metadata. Study: PMID15095297.

17. Soucek L, Whitfield J, Martins CP et al. Modelling Myc inhibition as a cancer therapy. Nature. 2008. [PMID 18716624](https://pubmed.ncbi.nlm.nih.gov/18716624/). DOI: 10.1038/nature07260. Access: abstract_metadata. Study: Omomyc-Kras.

18. Soucek L, Whitfield JR, Sodir NM et al. Inhibition of Myc family proteins eradicates KRas-driven lung cancer in mice. Genes & development. 2013. [PMID 23475959](https://pubmed.ncbi.nlm.nih.gov/23475959/). DOI: 10.1101/gad.205542.112. Access: full_text_partial. Study: Omomyc-Kras.

19. Soda M, Takada S, Takeuchi K et al. A mouse model for EML4-ALK-positive lung cancer. Proceedings of the National Academy of Sciences of the United States of America. 2008. [PMID 19064915](https://pubmed.ncbi.nlm.nih.gov/19064915/). DOI: 10.1073/pnas.0805381105. Access: abstract_metadata. Study: PMID19064915.

20. Soda M, Choi YL, Enomoto M et al. Identification of the transforming EML4-ALK fusion gene in non-small-cell lung cancer. Nature. 2007. [PMID 17625570](https://pubmed.ncbi.nlm.nih.gov/17625570/). DOI: 10.1038/nature05945. Access: abstract_metadata. Study: PMID17625570.

21. Druker BJ, Tamura S, Buchdunger E et al. Effects of a selective inhibitor of the Abl tyrosine kinase on the growth of Bcr-Abl positive cells. Nature medicine. 1996. [PMID 8616716](https://pubmed.ncbi.nlm.nih.gov/8616716/). DOI: 10.1038/nm0596-561. Access: abstract_metadata. Study: PMID8616716.

22. Druker BJ, Talpaz M, Resta DJ et al. Efficacy and safety of a specific inhibitor of the BCR-ABL tyrosine kinase in chronic myeloid leukemia. The New England journal of medicine. 2001. [PMID 11287972](https://pubmed.ncbi.nlm.nih.gov/11287972/). DOI: 10.1056/NEJM200104053441401. Access: abstract_metadata. Study: PMID11287972.

23. O'Brien SG, Guilhot F, Larson RA et al. Imatinib compared with interferon and low-dose cytarabine for newly diagnosed chronic-phase chronic myeloid leukemia. The New England journal of medicine. 2003. [PMID 12637609](https://pubmed.ncbi.nlm.nih.gov/12637609/). DOI: 10.1056/NEJMoa022457. Access: abstract_metadata. Study: PMID12637609.

24. Lynch TJ, Bell DW, Sordella R et al. Activating mutations in the epidermal growth factor receptor underlying responsiveness of non-small-cell lung cancer to gefitinib. The New England journal of medicine. 2004. [PMID 15118073](https://pubmed.ncbi.nlm.nih.gov/15118073/). DOI: 10.1056/NEJMoa040938. Access: abstract_metadata. Study: PMID15118073.

25. Pao W, Miller V, Zakowski M et al. EGF receptor gene mutations are common in lung cancers from "never smokers" and are associated with sensitivity of tumors to gefitinib and erlotinib. Proceedings of the National Academy of Sciences of the United States of America. 2004. [PMID 15329413](https://pubmed.ncbi.nlm.nih.gov/15329413/). DOI: 10.1073/pnas.0405220101. Access: abstract_metadata. Study: PMID15329413.

26. Mok TS, Wu YL, Thongprasert S et al. Gefitinib or carboplatin-paclitaxel in pulmonary adenocarcinoma. The New England journal of medicine. 2009. [PMID 19692680](https://pubmed.ncbi.nlm.nih.gov/19692680/). DOI: 10.1056/NEJMoa0810699. Access: abstract_metadata. Study: PMID19692680.

27. Maemondo M, Inoue A, Kobayashi K et al. Gefitinib or chemotherapy for non-small-cell lung cancer with mutated EGFR. The New England journal of medicine. 2010. [PMID 20573926](https://pubmed.ncbi.nlm.nih.gov/20573926/). DOI: 10.1056/NEJMoa0909530. Access: abstract_metadata. Study: PMID20573926.

28. Soria JC, Ohe Y, Vansteenkiste J et al. Osimertinib in Untreated EGFR-Mutated Advanced Non-Small-Cell Lung Cancer. The New England journal of medicine. 2018. [PMID 29151359](https://pubmed.ncbi.nlm.nih.gov/29151359/). DOI: 10.1056/NEJMoa1713137. Access: abstract_metadata. Study: FLAURA.

29. Ramalingam SS, Vansteenkiste J, Planchard D et al. Overall Survival with Osimertinib in Untreated, EGFR-Mutated Advanced NSCLC. The New England journal of medicine. 2020. [PMID 31751012](https://pubmed.ncbi.nlm.nih.gov/31751012/). DOI: 10.1056/NEJMoa1913662. Access: abstract_metadata. Study: FLAURA.

30. Slamon DJ, Leyland-Jones B, Shak S et al. Use of chemotherapy plus a monoclonal antibody against HER2 for metastatic breast cancer that overexpresses HER2. The New England journal of medicine. 2001. [PMID 11248153](https://pubmed.ncbi.nlm.nih.gov/11248153/). DOI: 10.1056/NEJM200103153441101. Access: abstract_metadata. Study: PMID11248153.

31. Chapman PB, Hauschild A, Robert C et al. Improved survival with vemurafenib in melanoma with BRAF V600E mutation. The New England journal of medicine. 2011. [PMID 21639808](https://pubmed.ncbi.nlm.nih.gov/21639808/). DOI: 10.1056/NEJMoa1103782. Access: abstract_metadata. Study: PMID21639808.

32. Shaw AT, Kim DW, Nakagawa K et al. Crizotinib versus chemotherapy in advanced ALK-positive lung cancer. The New England journal of medicine. 2013. [PMID 23724913](https://pubmed.ncbi.nlm.nih.gov/23724913/). DOI: 10.1056/NEJMoa1214886. Access: abstract_metadata. Study: PMID23724913.

33. Solomon BJ, Mok T, Kim DW et al. First-line crizotinib versus chemotherapy in ALK-positive lung cancer. The New England journal of medicine. 2014. [PMID 25470694](https://pubmed.ncbi.nlm.nih.gov/25470694/). DOI: 10.1056/NEJMoa1408440. Access: abstract_metadata. Study: PMID25470694.

34. Peters S, Camidge DR, Shaw AT et al. Alectinib versus Crizotinib in Untreated ALK-Positive Non-Small-Cell Lung Cancer. The New England journal of medicine. 2017. [PMID 28586279](https://pubmed.ncbi.nlm.nih.gov/28586279/). DOI: 10.1056/NEJMoa1704795. Access: abstract_metadata. Study: PMID28586279.

35. Shaw AT, Bauer TM, de Marinis F et al. First-Line Lorlatinib or Crizotinib in Advanced ALK-Positive Lung Cancer. The New England journal of medicine. 2020. [PMID 33207094](https://pubmed.ncbi.nlm.nih.gov/33207094/). DOI: 10.1056/NEJMoa2027187. Access: abstract_metadata. Study: CROWN.

36. Solomon BJ, Liu G, Felip E et al. Lorlatinib Versus Crizotinib in Patients With Advanced ALK-Positive Non-Small Cell Lung Cancer: 5-Year Outcomes From the Phase III CROWN Study. Journal of clinical oncology : official journal of the American Society of Clinical Oncology. 2024. [PMID 38819031](https://pubmed.ncbi.nlm.nih.gov/38819031/). DOI: 10.1200/JCO.24.00581. Access: abstract_metadata. Study: CROWN.

37. Shaw AT, Solomon BJ, Felip E et al. Lorlatinib versus crizotinib as first-line treatment for advanced ALK-positive non-small-cell lung cancer: 7-year update from the phase III CROWN study. Annals of oncology : official journal of the European Society for Medical Oncology. 2026. [PMID 42217582](https://pubmed.ncbi.nlm.nih.gov/42217582/). DOI: 10.1016/j.annonc.2026.05.692. Access: abstract_metadata. Study: CROWN.

38. Gorre ME, Mohammed M, Ellwood K et al. Clinical resistance to STI-571 cancer therapy caused by BCR-ABL gene mutation or amplification. Science (New York, N.Y.). 2001. [PMID 11423618](https://pubmed.ncbi.nlm.nih.gov/11423618/). DOI: 10.1126/science.1062538. Access: abstract_metadata. Study: PMID11423618.

39. Pao W, Miller VA, Politi KA et al. Acquired resistance of lung adenocarcinomas to gefitinib or erlotinib is associated with a second mutation in the EGFR kinase domain. PLoS medicine. 2005. [PMID 15737014](https://pubmed.ncbi.nlm.nih.gov/15737014/). DOI: 10.1371/journal.pmed.0020073. Access: full_text_partial. Study: PMID15737014.

40. Engelman JA, Zejnullahu K, Mitsudomi T et al. MET amplification leads to gefitinib resistance in lung cancer by activating ERBB3 signaling. Science (New York, N.Y.). 2007. [PMID 17463250](https://pubmed.ncbi.nlm.nih.gov/17463250/). DOI: 10.1126/science.1141478. Access: abstract_metadata. Study: PMID17463250.

41. Regales L, Gong Y, Shen R et al. Dual targeting of EGFR can overcome a major drug resistance mutation in mouse models of EGFR mutant lung cancer. The Journal of clinical investigation. 2009. [PMID 19759520](https://pubmed.ncbi.nlm.nih.gov/19759520/). DOI: 10.1172/JCI38746. Access: abstract_metadata. Study: PMID19759520.

42. Sequist LV, Waltman BA, Dias-Santagata D et al. Genotypic and histological evolution of lung cancers acquiring resistance to EGFR inhibitors. Science translational medicine. 2011. [PMID 21430269](https://pubmed.ncbi.nlm.nih.gov/21430269/). DOI: 10.1126/scitranslmed.3002003. Access: abstract_metadata. Study: PMID21430269.

43. Prahallad A, Sun C, Huang S et al. Unresponsiveness of colon cancer to BRAF(V600E) inhibition through feedback activation of EGFR. Nature. 2012. [PMID 22281684](https://pubmed.ncbi.nlm.nih.gov/22281684/). DOI: 10.1038/nature10868. Access: abstract_metadata. Study: PMID22281684.

44. Gainor JF, Dardaei L, Yoda S et al. Molecular Mechanisms of Resistance to First- and Second-Generation ALK Inhibitors in ALK-Rearranged Lung Cancer. Cancer discovery. 2016. [PMID 27432227](https://pubmed.ncbi.nlm.nih.gov/27432227/). DOI: 10.1158/2159-8290.CD-16-0596. Access: abstract_metadata. Study: PMID27432227.

45. Yoda S, Lin JJ, Lawrence MS et al. Sequential ALK Inhibitors Can Select for Lorlatinib-Resistant Compound ALK Mutations in ALK-Positive Lung Cancer. Cancer discovery. 2018. [PMID 29650534](https://pubmed.ncbi.nlm.nih.gov/29650534/). DOI: 10.1158/2159-8290.CD-17-1256. Access: abstract_metadata. Study: PMID29650534.

46. Shaw AT, Solomon BJ, Besse B et al. ALK Resistance Mutations and Efficacy of Lorlatinib in Advanced Anaplastic Lymphoma Kinase-Positive Non-Small-Cell Lung Cancer. Journal of clinical oncology : official journal of the American Society of Clinical Oncology. 2019. [PMID 30892989](https://pubmed.ncbi.nlm.nih.gov/30892989/). DOI: 10.1200/JCO.18.02236. Access: abstract_metadata. Study: PMID30892989.

47. Moody SE, Perez D, Pan TC et al. The transcriptional repressor Snail promotes mammary tumor recurrence. Cancer cell. 2005. [PMID 16169465](https://pubmed.ncbi.nlm.nih.gov/16169465/). DOI: 10.1016/j.ccr.2005.07.009. Access: abstract_metadata. Study: PMID16169465.

48. Sotillo R, Schvartzman JM, Socci ND et al. Mad2-induced chromosome instability leads to lung tumour relapse after oncogene withdrawal. Nature. 2010. [PMID 20173739](https://pubmed.ncbi.nlm.nih.gov/20173739/). DOI: 10.1038/nature08803. Access: abstract_metadata. Study: PMID20173739.

49. Kapoor A, Yao W, Ying H et al. Yap1 activation enables bypass of oncogenic Kras addiction in pancreatic cancer. Cell. 2014. [PMID 24954535](https://pubmed.ncbi.nlm.nih.gov/24954535/). DOI: 10.1016/j.cell.2014.06.003. Access: full_text_partial. Study: PMID24954535.

50. Singh A, Greninger P, Rhodes D et al. A gene expression signature associated with "K-Ras addiction" reveals regulators of EMT and tumor cell survival. Cancer cell. 2009. [PMID 19477428](https://pubmed.ncbi.nlm.nih.gov/19477428/). DOI: 10.1016/j.ccr.2009.03.022. Access: abstract_metadata. Study: PMID19477428.

51. Sharma SV, Lee DY, Li B et al. A chromatin-mediated reversible drug-tolerant state in cancer cell subpopulations. Cell. 2010. [PMID 20371346](https://pubmed.ncbi.nlm.nih.gov/20371346/). DOI: 10.1016/j.cell.2010.02.027. Access: full_text_partial. Study: PMID20371346.

52. Corbin AS, Agarwal A, Loriaux M et al. Human chronic myeloid leukemia stem cells are insensitive to imatinib despite inhibition of BCR-ABL activity. The Journal of clinical investigation. 2011. [PMID 21157039](https://pubmed.ncbi.nlm.nih.gov/21157039/). DOI: 10.1172/JCI35721. Access: full_text_partial. Study: PMID21157039.

53. Graham SM, Jørgensen HG, Allan E et al. Primitive, quiescent, Philadelphia-positive stem cells from patients with chronic myeloid leukemia are insensitive to STI571 in vitro. Blood. 2002. [PMID 11756187](https://pubmed.ncbi.nlm.nih.gov/11756187/). DOI: 10.1182/blood.v99.1.319. Access: abstract_metadata. Study: PMID11756187.

54. Hata AN, Niederst MJ, Archibald HL et al. Tumor cells can follow distinct evolutionary paths to become resistant to epidermal growth factor receptor inhibition. Nature medicine. 2016. [PMID 26828195](https://pubmed.ncbi.nlm.nih.gov/26828195/). DOI: 10.1038/nm.4040. Access: full_text_partial. Study: PMID26828195.

55. Hangauer MJ, Viswanathan VS, Ryan MJ et al. Drug-tolerant persister cancer cells are vulnerable to GPX4 inhibition. Nature. 2017. [PMID 29088702](https://pubmed.ncbi.nlm.nih.gov/29088702/). DOI: 10.1038/nature24297. Access: full_text_partial. Study: PMID29088702.

56. Oren Y, Tsabar M, Cuoco MS et al. Cycling cancer persister cells arise from lineages with distinct programs. Nature. 2021. [PMID 34381210](https://pubmed.ncbi.nlm.nih.gov/34381210/). DOI: 10.1038/s41586-021-03796-6. Access: abstract_metadata. Study: PMID34381210.

57. Isozaki H, Sakhtemani R, Abbasi A et al. Therapy-induced APOBEC3A drives evolution of persistent cancer cells. Nature. 2023. [PMID 37407818](https://pubmed.ncbi.nlm.nih.gov/37407818/). DOI: 10.1038/s41586-023-06303-1. Access: abstract_metadata. Study: PMID37407818.

58. Haderk F, Chou YT, Cech L et al. Focal adhesion kinase-YAP signaling axis drives drug-tolerant persister cells and residual disease in lung cancer. Nature communications. 2024. [PMID 38702301](https://pubmed.ncbi.nlm.nih.gov/38702301/). DOI: 10.1038/s41467-024-47423-0. Access: abstract_metadata. Study: PMID38702301.

59. Moghal N, Li Q, Stewart EL et al. Single-Cell Analysis Reveals Transcriptomic Features of Drug-Tolerant Persisters and Stromal Adaptation in a Patient-Derived EGFR-Mutated Lung Adenocarcinoma Xenograft Model. Journal of thoracic oncology : official publication of the International Association for the Study of Lung Cancer. 2023. [PMID 36535627](https://pubmed.ncbi.nlm.nih.gov/36535627/). DOI: 10.1016/j.jtho.2022.12.003. Access: abstract_metadata. Study: PMID36535627.

60. Liao J, Yao W, Yu Y et al. Targeting TROP2 in drug-tolerant persister cells delays EGFR tyrosine kinase inhibitor resistance in non-small-cell lung cancer. Cancer cell. 2026. [PMID 42314664](https://pubmed.ncbi.nlm.nih.gov/42314664/). DOI: 10.1016/j.ccell.2026.05.020. Access: abstract_metadata. Study: PMID42314664.

61. Mahon FX, Réa D, Guilhot J et al. Discontinuation of imatinib in patients with chronic myeloid leukaemia who have maintained complete molecular remission for at least 2 years: the prospective, multicentre Stop Imatinib (STIM) trial. The Lancet. Oncology. 2010. [PMID 20965785](https://pubmed.ncbi.nlm.nih.gov/20965785/). DOI: 10.1016/S1470-2045(10)70233-3. Access: abstract_metadata. Study: PMID20965785.

62. Shalem O, Sanjana NE, Hartenian E et al. Genome-scale CRISPR-Cas9 knockout screening in human cells. Science (New York, N.Y.). 2014. [PMID 24336571](https://pubmed.ncbi.nlm.nih.gov/24336571/). DOI: 10.1126/science.1247005. Access: abstract_metadata. Study: PMID24336571.

63. Foth M, Kim W, O'Toole K et al. Genetic Drivers of Sensitivity or Resistance to RAS(ON) Multiselective Inhibitors in NRAS-Mutated Melanoma. Cancer research. 2026. [PMID 42171647](https://pubmed.ncbi.nlm.nih.gov/42171647/). DOI: 10.1158/0008-5472.CAN-26-1313. Access: abstract_metadata. Study: PMID42171647.

64. Wei X, Blaj C, Ali Al-Radhawi M et al. Abrogation of Oncogenic RAS Signaling by a RAS(ON) Inhibitor Doublet Primes Immune-Refractory KRASG12C-Mutant NSCLC for Immune Checkpoint Blockade. Cancer discovery. 2026. [PMID 41670434](https://pubmed.ncbi.nlm.nih.gov/41670434/). DOI: 10.1158/2159-8290.CD-25-1616. Access: abstract_metadata. Study: PMID41670434.

65. Scaparone P, Mira A, Cheong TC et al. EML4-ALK mediates resistance to KRAS G12C inhibition and induces an oncogenic dependency by rewiring signaling through the wild-type RAS pathway. Cancer discovery. 2026. [PMID 42329099](https://pubmed.ncbi.nlm.nih.gov/42329099/). DOI: 10.1158/2159-8290.CD-25-0477. Access: abstract_metadata. Study: PMID42329099.

66. Orlen M, Vostrejs WP, Sor R et al. T-cell Dependency of Tumor Regressions and Complete Responses with RAS(ON) Multi-selective Inhibition in Preclinical Models of Pancreatic Ductal Adenocarcinoma. Cancer discovery. 2025. [PMID 40057911](https://pubmed.ncbi.nlm.nih.gov/40057911/). DOI: 10.1158/2159-8290.CD-24-1475. Access: abstract_metadata. Study: PMID40057911.

## Source-specific evidence cards

### [1] Reversible tumorigenesis by MYC in hematopoietic lineages.

**Shows:** Conditional MYC in hematopoietic lineages yields tumors that regress after transgene inactivation.

**Does not show:** Forced expression is not native human dosage; the abstract does not establish elimination of every residual cell or all sporadic tumors.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC3.

### [2] Essential role for oncogenic Ras in tumour maintenance.

**Shows:** Conditional H-RAS withdrawal regresses INK4a-null melanoma; enforced VEGF does not sustain the tumor after withdrawal.

**Does not show:** Engineered induction and genotype constrain scope; vascular effects do not establish universal tumor dependence.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1.

### [3] Reversibility of acute B-cell leukaemia induced by BCR-ABL1.

**Shows:** Suppressing conditional BCR-ABL1 reverses acute B-cell leukemia, with repeated induction and remission cycles.

**Does not show:** This mouse acute B-cell leukemia is not human chronic myeloid leukemia; repeat reversibility is not proven sterilization.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC3.

### [4] Conditional activation of Neu in the mammary epithelium of transgenic mice results in reversible pulmonary metastasis.

**Shows:** Neu withdrawal regresses mammary tumors and pulmonary metastases to undetectability, but most animals later recur with Neu-independent tumors.

**Does not show:** Undetectable burden is not absence of residual cells; body controls were unavailable for examination.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC3.

### [5] Lung adenocarcinomas induced in mice by mutant EGF receptors found in human lung cancers respond to a tyrosine kinase inhibitor or to down-regulation of the receptors.

**Shows:** Both EGFR transgene withdrawal and erlotinib during continued doxycycline regress mutant-EGFR tumors; placebo and KRAS tumors fail the drug response.

**Does not show:** Transgene RNA is strongly induced and bulk protein is several-fold elevated; custom mouse MRI response and mostly normal histology do not establish eradication or physiological dosage.

**Access/status:** Complete primary abstract and linked-status metadata examined. Selected body sections: 0:Publisher HTML paragraphs:paragraph22, 0:Publisher HTML paragraphs:paragraph23, 0:Publisher HTML paragraphs:paragraph24, 0:Publisher HTML paragraphs:paragraph34, 0:Publisher HTML paragraphs:paragraph36, 0:Publisher HTML paragraphs:paragraph45, 0:Publisher HTML paragraphs:paragraph47, 0:Publisher HTML paragraphs:paragraph50, 0:Publisher HTML paragraphs:paragraph64, 0:Publisher HTML paragraphs:paragraph73. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC3.

### [6] Sustained loss of a neoplastic phenotype by brief inactivation of MYC.

**Shows:** Brief MYC withdrawal in osteosarcoma promotes differentiation and sustained loss of neoplastic behavior despite later reactivation.

**Does not show:** This differentiation response is lineage specific and cannot establish treatment-free cure in every baseline driver pair.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC3.

### [7] Genomically complex lymphomas undergo sustained tumor regression upon MYC inactivation unless they acquire novel chromosomal translocations.

**Shows:** Genomically complex MYC lymphomas often regress after MYC inactivation; some escape with further chromosomal changes.

**Does not show:** The engineered lymphoma lineage and heterogeneous escape preclude universal dependence or cure.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC3.

### [8] MYC inactivation uncovers pluripotent differentiation and tumour dormancy in hepatocellular cancer.

**Shows:** MYC withdrawal in liver tumors induces differentiation and dormancy; reactivation restores neoplastic properties in genetically related cells.

**Does not show:** Persistence is compatible with bulk addiction, and the abstract alone does not resolve every residual-cell dependency or reservoir-specific intervention.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC3.

### [9] Sustained regression of tumors upon MYC inactivation requires p53 or thrombospondin-1 to reverse the angiogenic switch.

**Shows:** p53 deficiency impairs sustained MYC-withdrawal regression; p53 or thrombospondin restoration restores the response in the studied setting.

**Does not show:** Thrombospondin alone is insufficient; genotype and vascular mechanisms modify dependence rather than prove all-cancer withdrawal sufficiency.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC3.

### [10] CD4(+) T cells contribute to the remodeling of the microenvironment required for sustained tumor regression upon oncogene inactivation.

**Shows:** CD4 immunity and thrombospondin contribute to sustained regression after MYC or BCR-ABL inactivation.

**Does not show:** Host-dependent sustained control differs from immediate cell-autonomous collapse; linked erratum details were not examined.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. ErratumIn: PMID not supplied. Subclaims: SC1, SC3.

### [11] Oncogenic Kras is required for both the initiation and maintenance of pancreatic cancer in mice.

**Shows:** Conditional oncogenic Kras withdrawal regresses pancreatic lesions and changes the associated stroma.

**Does not show:** The engineered pancreas and p53 settings cannot establish all sporadic pancreatic tumors or residual-cell elimination.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1.

### [12] Temporal dissection of K-ras(G12D) mutant in vitro and in vivo using a regulatable K-ras(G12D) mouse allele.

**Shows:** Switching off an endogenous conditional KRAS fusion allele regresses pancreatic tumors.

**Does not show:** The switch remains an engineered fusion and genotype-specific test; a linked author-name correction does not add an independent causal experiment.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. ErratumIn: PMID not supplied. Subclaims: SC1.

### [13] Oncogene cooperation in tumor maintenance and tumor recurrence in mouse mammary tumors induced by Myc and mutant Kras.

**Shows:** Dual MYC and KRAS mammary tumors regress more strongly when both are removed; recurrent tumors can reactivate inducible transgenes through rtTA mutations.

**Does not show:** Reactivation is different from verified driver-independent bypass; the result contests one-node generalization rather than proving every recurrent mechanism.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC2, SC3.

### [14] Combined Inactivation of MYC and K-Ras oncogenes reverses tumorigenesis in lung adenocarcinomas and lymphomas.

**Shows:** MYC and KRAS withdrawal effects differ between lung tumors and lymphomas; dual withdrawal exceeds MYC-only lung responses.

**Does not show:** Cooperating pathways and tissue context preclude a uniform one-oncogene rule.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC2.

### [15] PUMA and BIM are required for oncogene inactivation-induced apoptosis.

**Shows:** BIM or PUMA loss reduces apoptosis and slows HER2-withdrawal regression; PUMA loss also impairs EGFR-withdrawal lung regression. Constitutive AKT/MEK separates downstream apoptotic routes.

**Does not show:** Apoptotic competence modifies response, and these inducible models do not show that every residual cell must depend on the driver.

**Access/status:** Complete primary abstract and linked-status metadata examined. Selected body sections: 1:RESULTS:paragraph0, 1:RESULTS:paragraph1, 1:RESULTS:paragraph2, 1:RESULTS:paragraph6, 1:RESULTS:paragraph7, 1:RESULTS:paragraph8. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC3.

### [16] Reversible lymphomagenesis in conditionally c-MYC expressing mice.

**Shows:** Fate-labeled MYC lymphoma cells undergo arrest, differentiation and apoptosis after MYC withdrawal.

**Does not show:** A selected engineered lymphoma is not a representative pan-cancer dependency survey.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1.

### [17] Modelling Myc inhibition as a cancer therapy.

**Shows:** Omomyc inhibition of endogenous MYC regresses KRAS-driven lung tumors with reversible effects on normal regenerative tissues.

**Does not show:** MYC is a downstream common conduit rather than the mutated initiating KRAS itself; later work shares the model family.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC2.

### [18] Inhibition of Myc family proteins eradicates KRas-driven lung cancer in mice.

**Shows:** Repeated four-week Omomyc on/off cycles lower KRAS-lung tumor recurrence; after eight cycles, three of four mice have no scored tumors after four weeks off. Refractory lesions suppress Omomyc expression.

**Does not show:** All animals recur after a first cycle; the final off-treatment cohort is small, histology excludes atypical hyperplasia, and four weeks is not lifelong sterile eradication. Linked erratum details remain unexamined.

**Access/status:** Complete primary abstract and linked-status metadata examined. Selected body sections: 0:Publisher HTML paragraphs:paragraph14, 0:Publisher HTML paragraphs:paragraph15, 0:Publisher HTML paragraphs:paragraph18, 0:Publisher HTML paragraphs:paragraph19, 0:Publisher HTML paragraphs:paragraph20, 0:Publisher HTML paragraphs:paragraph25, 0:Publisher HTML paragraphs:paragraph28, 0:Publisher HTML paragraphs:paragraph32, 0:Publisher HTML paragraphs:paragraph34, 0:Publisher HTML paragraphs:paragraph35. No complete-paper or supplement audit claimed. ErratumIn: PMID42386384. Subclaims: SC1, SC3.

### [19] A mouse model for EML4-ALK-positive lung cancer.

**Shows:** Lung-specific transgenic EML4-ALK tumors respond to an ALK inhibitor in the experimental mouse setting.

**Does not show:** Forced expression and transformation assays do not by themselves establish a native universal maintenance dependence.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1.

### [20] Identification of the transforming EML4-ALK fusion gene in non-small-cell lung cancer.

**Shows:** EML4-ALK fusions are identified in human lung tumors and transform experimental cells.

**Does not show:** Discovery and transformation capacity do not test continued native tumor maintenance.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1.

### [21] Effects of a selective inhibitor of the Abl tyrosine kinase on the growth of Bcr-Abl positive cells.

**Shows:** The selective ABL inhibitor suppresses CML colony formation while largely preserving normal hematopoietic growth.

**Does not show:** Colony suppression is not elimination of quiescent leukemic stem cells or an in vivo maintenance intervention.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC3.

### [22] Efficacy and safety of a specific inhibitor of the BCR-ABL tyrosine kinase in chronic myeloid leukemia.

**Shows:** In phase 1 CML testing, 53 of 54 patients receiving at least 300 mg daily obtain complete hematologic response.

**Does not show:** Dose-cohort selection, nonrandomization and hematologic response do not establish cure or an all-dose response.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1.

### [23] Imatinib compared with interferon and low-dose cytarabine for newly diagnosed chronic-phase chronic myeloid leukemia.

**Shows:** IRIS randomizes 1106 CML patients and shows better cytogenetic response and progression control with imatinib than interferon plus cytarabine.

**Does not show:** Crossover complicates survival inference; drug treatment does not equal genetic removal or residual-cell eradication.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1.

### [24] Activating mutations in the epidermal growth factor receptor underlying responsiveness of non-small-cell lung cancer to gefitinib.

**Shows:** EGFR kinase mutations associate with gefitinib response and show functional inhibitor sensitivity.

**Does not show:** The discovery comparison selects responders and nonresponders after outcome; it is not a prospective genotype-enrolled response denominator.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1.

### [25] EGF receptor gene mutations are common in lung cancers from "never smokers" and are associated with sensitivity of tumors to gefitinib and erlotinib.

**Shows:** EGFR mutations associate with inhibitor-sensitive lung cancer and alter drug-sensitive phosphorylation in experimental systems.

**Does not show:** Outcome-based discovery and smoking-history enrichment do not establish response in every prospectively defined mutation-positive tumor.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1.

### [26] Gefitinib or carboplatin-paclitaxel in pulmonary adenocarcinoma.

**Shows:** In IPASS, gefitinib favors progression control in the mutation-positive subgroup and disadvantages mutation-negative patients relative to chemotherapy.

**Does not show:** Enrollment uses clinical features rather than EGFR genotype, and the molecular analysis is a tested subsample; subgroup estimates are not the entire enrolled cohort.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1.

### [27] Gefitinib or chemotherapy for non-small-cell lung cancer with mutated EGFR.

**Shows:** Baseline EGFR-mutant NEJ002 enrollment yields 73.7 percent response and 10.8-month median progression-free survival with gefitinib, versus 30.7 percent and 5.4 months with chemotherapy.

**Does not show:** Responses are not universal; reported overall-survival comparison is not statistically decisive, and recurrence remains compatible with bulk addiction.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1.

### [28] Osimertinib in Untreated EGFR-Mutated Advanced Non-Small-Cell Lung Cancer.

**Shows:** FLAURA randomizes 556 baseline EGFR-mutant patients; osimertinib improves progression-free survival relative to active first-generation EGFR inhibitors.

**Does not show:** Both arms inhibit the same driver, so this comparison does not isolate inhibition versus none; initial response is not eradication.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC3.

### [29] Overall Survival with Osimertinib in Untreated, EGFR-Mutated Advanced NSCLC.

**Shows:** Later FLAURA analysis reports median overall survival of 38.6 versus 31.8 months.

**Does not show:** This is follow-up of the same 556-patient trial, not independent confirmation; survival benefit is not proof of sterile cure.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC3.

### [30] Use of chemotherapy plus a monoclonal antibody against HER2 for metastatic breast cancer that overexpresses HER2.

**Shows:** HER2-selected metastatic breast cancer shows greater response and longer survival with trastuzumab plus chemotherapy than chemotherapy alone.

**Does not show:** Antibody-mediated immunity and chemotherapy contribute; this is not pure selective kinase withdrawal, and many tumors do not respond.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1.

### [31] Improved survival with vemurafenib in melanoma with BRAF V600E mutation.

**Shows:** BRIM-3 enrolls 675 baseline BRAF-mutant melanoma patients; vemurafenib yields 48 percent response versus 5 percent with dacarbazine and improves early survival.

**Does not show:** More than half lack an objective response in this report; the baseline genotype pair cannot be narrowed to responders after outcome.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1.

### [32] Crizotinib versus chemotherapy in advanced ALK-positive lung cancer.

**Shows:** PROFILE 1007 enrolls 347 ALK-positive previously treated patients and improves response and progression-free survival with crizotinib versus chemotherapy.

**Does not show:** Interim overall survival is not significantly different; a linked erratum was not audited, and the outcome is not cure.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. ErratumIn: PMID26466010. Subclaims: SC1.

### [33] First-line crizotinib versus chemotherapy in ALK-positive lung cancer.

**Shows:** PROFILE 1014 enrolls 343 untreated ALK-positive patients and improves response and progression-free survival with crizotinib versus chemotherapy.

**Does not show:** Crossover and immature survival limit causal survival claims; linked correction details remain unexamined.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. ErratumIn: PMID26466011. Subclaims: SC1.

### [34] Alectinib versus Crizotinib in Untreated ALK-Positive Non-Small-Cell Lung Cancer.

**Shows:** ALEX enrolls 303 ALK-positive patients and shows better progression and CNS control with alectinib than crizotinib.

**Does not show:** Two active ALK agents are compared; this is not a driver-off versus driver-on test or a treatment-free endpoint.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC3.

### [35] First-Line Lorlatinib or Crizotinib in Advanced ALK-Positive Lung Cancer.

**Shows:** CROWN enrolls 296 ALK-positive patients and improves response and progression control with lorlatinib versus crizotinib.

**Does not show:** Both arms inhibit ALK; toxicity and ongoing treatment remain, and later reports reuse this cohort.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC3.

### [36] Lorlatinib Versus Crizotinib in Patients With Advanced ALK-Positive Non-Small Cell Lung Cancer: 5-Year Outcomes From the Phase III CROWN Study.

**Shows:** Five-year CROWN follow-up reports 60 percent versus 8 percent progression-free survival.

**Does not show:** Post hoc long follow-up of the same 296 patients is not a new trial or treatment-free eradication.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC3.

### [37] Lorlatinib versus crizotinib as first-line treatment for advanced ALK-positive non-small-cell lung cancer: 7-year update from the phase III CROWN study.

**Shows:** Seven-year CROWN follow-up reports 55 percent versus 3 percent progression-free survival; overall survival remains immature.

**Does not show:** Durable control occurs on treatment and does not prove cure; neither is relapse inevitable by seven years for every patient. The same CROWN cohort is reused.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC3.

### [38] Clinical resistance to STI-571 cancer therapy caused by BCR-ABL gene mutation or amplification.

**Shows:** Resistant CML samples contain ABL gatekeeper mutations or amplification; reconstructed mutant kinase resists imatinib.

**Does not show:** Drug resistance with restored signaling supports continuing target dependence, not independence after successful blockade; nine selected resistant cases do not define population frequency.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1.

### [39] Acquired resistance of lung adenocarcinomas to gefitinib or erlotinib is associated with a second mutation in the EGFR kinase domain.

**Shows:** T790M appears in three of six progressing EGFR-inhibitor cases; independent RFLP and matched mutant constructs support reduced drug-sensitive phosphorylation.

**Does not show:** One pretreatment sample is scant; sequencing sensitivity cannot exclude rare pre-existing cells. Cell-line viability contrasts have other background differences.

**Access/status:** Complete primary abstract and linked-status metadata examined. Selected body sections: 2:Results:paragraph15, 2:Results:paragraph16, 2:Results:paragraph17, 2:Results:paragraph18, 2:Results:paragraph19, 2:Results:paragraph26, 2:Results:paragraph29, 2:Results:paragraph31, 2:Results:paragraph32, 2:Results:paragraph34. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC3.

### [40] MET amplification leads to gefitinib resistance in lung cancer by activating ERBB3 signaling.

**Shows:** MET amplification activates ERBB3-mediated bypass; combined MET and EGFR inhibition restores response in resistant cells, with amplification in four of eighteen resistant patients.

**Does not show:** Culture rescue and selected progressing biopsies do not establish universal clinical bypass frequency or durable cure.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC3.

### [41] Dual targeting of EGFR can overcome a major drug resistance mutation in mouse models of EGFR mutant lung cancer.

**Shows:** Combined irreversible EGFR inhibition and cetuximab regresses resistant EGFR-mutant mouse tumors and lowers receptor signaling and abundance more effectively than either alone.

**Does not show:** Combination activity includes antibody and abundance effects; the model does not establish efficacy in every human resistance state.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC3.

### [42] Genotypic and histological evolution of lung cancers acquiring resistance to EGFR inhibitors.

**Shows:** Repeat biopsies from 37 progressing EGFR-mutant cancers retain EGFR mutations while showing heterogeneous mechanisms, including five small-cell transformations.

**Does not show:** This selected progression cohort is not all treated patients; retained mutation does not itself prove the new lineage remains EGFR-dependent.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC3.

### [43] Unresponsiveness of colon cancer to BRAF(V600E) inhibition through feedback activation of EGFR.

**Shows:** BRAF-mutant colon models resist BRAF inhibition through EGFR feedback; combined inhibition restores response, and ectopic EGFR can confer resistance in melanoma cells.

**Does not show:** The same driver behaves differently by tissue; experimental combination rescue does not prove all baseline BRAF tumors share one maintenance rule.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC2.

### [44] Molecular Mechanisms of Resistance to First- and Second-Generation ALK Inhibitors in ALK-Rearranged Lung Cancer.

**Shows:** ALK resistance biopsies and functional tests identify inhibitor-dependent spectra of secondary mutations and bypass sensitivity.

**Does not show:** Progressor selection and the shared clinical programme limit population estimates and independence; response after resistance is context specific.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC3.

### [45] Sequential ALK Inhibitors Can Select for Lorlatinib-Resistant Compound ALK Mutations in ALK-Positive Lung Cancer.

**Shows:** Experimental ALK selection and serial patient specimens show sequential compound mutations during inhibitor exposure.

**Does not show:** Ba/F3 selection and selected progressing samples cannot establish that all resistance follows this genetic route.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC3.

### [46] ALK Resistance Mutations and Efficacy of Lorlatinib in Advanced Anaplastic Lymphoma Kinase-Positive Non-Small-Cell Lung Cancer.

**Shows:** Exploratory lorlatinib analysis links baseline secondary ALK mutations after prior inhibitors to higher response rates in tissue and plasma subgroups.

**Does not show:** This is a biomarker association within a treated phase-2 programme, not a randomized genotype intervention or an independent trial for each assay.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC3.

### [47] The transcriptional repressor Snail promotes mammary tumor recurrence.

**Shows:** Snail expression promotes recurrent mammary tumor behavior after Neu withdrawal and relates to a mesenchymal recurrence phenotype.

**Does not show:** Forced Snail gain does not establish native necessity of Snail in every residual tumor.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC3.

### [48] Mad2-induced chromosome instability leads to lung tumour relapse after oncogene withdrawal.

**Shows:** Kras withdrawal initially regresses Kras/Mad2 lung tumors, followed by frequent recurrence with mostly absent oncogenic Kras transgene re-expression.

**Does not show:** The present case examines the abstract; original-body credit belongs to the preceding genome-instability review. Recurrence and aneuploidy do not directly prove a persister lineage or zero remaining instability.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC3.

### [49] Yap1 activation enables bypass of oncogenic Kras addiction in pancreatic cancer.

**Shows:** After Kras withdrawal, pancreatic tumors regress and twenty of twenty-eight recur; a subset lacks transgene signaling. YAP gain substitutes for Kras, while YAP/TEAD loss or binding-defective controls suppress bypass growth.

**Does not show:** About half reactivate the transgene through rtTA amplification; basal wild-type RAS remains. Relapse YAP-amplification and the engineered line constrain universal bypass claims; linked erratum details are unaudited.

**Access/status:** Complete primary abstract and linked-status metadata examined. Selected body sections: 1:RESULTS:paragraph0, 1:RESULTS:paragraph1, 1:RESULTS:paragraph2, 1:RESULTS:paragraph3, 1:RESULTS:paragraph4, 1:RESULTS:paragraph5, 1:RESULTS:paragraph6, 1:RESULTS:paragraph8, 1:RESULTS:paragraph9, 1:RESULTS:paragraph16. No complete-paper or supplement audit claimed. ErratumIn: PMID31730860. Subclaims: SC1, SC3.

### [50] A gene expression signature associated with "K-Ras addiction" reveals regulators of EMT and tumor cell survival.

**Shows:** KRAS-mutant cancer cell lines differ in KRAS dependency and epithelial state after RNA interference.

**Does not show:** Cell survival screens do not establish representative native tumor maintenance; a linked correction was not audited.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. ErratumIn: PMID33689705. Subclaims: SC1, SC2.

### [51] A chromatin-mediated reversible drug-tolerant state in cancer cell subpopulations.

**Shows:** Rare PC9 persisters survive concentrations far above drug IC50 with EGFR suppression; single-cell clones regenerate tolerance, withdrawal restores sensitivity, and KDM5A/HDAC/IGF1R controls separate reversible states from later T790M colonies.

**Does not show:** Cell-line state switching does not establish every patient reservoir; expanded persisters reverse more slowly, and pharmacologic prevention is not clinical treatment-free cure.

**Access/status:** Complete primary abstract and linked-status metadata examined. Selected body sections: 1:RESULTS. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC3.

### [52] Human chronic myeloid leukemia stem cells are insensitive to imatinib despite inhibition of BCR-ABL activity.

**Shows:** CML stem/progenitor cultures survive strong BCR-ABL inhibition with cytokine support; phospho-CRKL/STAT5 controls, persistent inhibition and mostly Philadelphia-positive survivors address failed engagement or normal-cell carryover.

**Does not show:** Residual phospho-signal is not literally zero activity; culture cytokines and short exposure do not prove all in vivo stem cells are wholly driver independent or dispensability of the entire BCR-ABL protein.

**Access/status:** Complete primary abstract and linked-status metadata examined. Selected body sections: 0:Publisher HTML paragraphs:paragraph13, 0:Publisher HTML paragraphs:paragraph15, 0:Publisher HTML paragraphs:paragraph16, 0:Publisher HTML paragraphs:paragraph21, 0:Publisher HTML paragraphs:paragraph23, 0:Publisher HTML paragraphs:paragraph24, 0:Publisher HTML paragraphs:paragraph25, 0:Publisher HTML paragraphs:paragraph30, 0:Publisher HTML paragraphs:paragraph31, 0:Publisher HTML paragraphs:paragraph47, 0:Publisher HTML paragraphs:paragraph58. No complete-paper or supplement audit claimed. ErratumIn: PMID not supplied. Subclaims: SC1, SC3.

### [53] Primitive, quiescent, Philadelphia-positive stem cells from patients with chronic myeloid leukemia are insensitive to STI571 in vitro.

**Shows:** Quiescent Philadelphia-positive CML progenitors survive short imatinib culture exposure.

**Does not show:** Without matched complete target-engagement and niche controls, survival alone cannot distinguish insufficient inhibition from signaling-independent persistence.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC3.

### [54] Tumor cells can follow distinct evolutionary paths to become resistant to epidermal growth factor receptor inhibition.

**Shows:** Barcoding, early resistant clones, persister pools and spiked-clone kinetic controls distinguish pre-existing and later T790M routes. Late-route cells resist apoptosis despite similar EGFR/ERK inhibition; navitoclax restores death and tumor response.

**Does not show:** Human line-response correlations do not trace original patient clones; low-frequency pretreatment mutations have detection limits, and the later route is not the only clinical route.

**Access/status:** Complete primary abstract and linked-status metadata examined. Selected body sections: 1:Results. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC3.

### [55] Drug-tolerant persister cancer cells are vulnerable to GPX4 inhibition.

**Shows:** Reversible persisters selectively depend on GPX4; knockout, ferrostatin rescue and iron/lipid controls identify ferroptotic death. Unmasking GPX4 deletion after tumor shrinkage prevents relapse in the tested melanoma xenograft.

**Does not show:** The in vivo intervention is genetic because compounds lack systemic bioavailability; continued targeted treatment, one xenograft and absent clinical safety testing do not establish treatment-free human cure.

**Access/status:** Complete primary abstract and linked-status metadata examined. Selected body sections: 3:Main text:paragraph0, 3:Main text:paragraph2, 3:Main text:paragraph3, 3:Main text:paragraph4, 3:Main text:paragraph6, 3:Main text:paragraph7, 3:Main text:paragraph8. No complete-paper or supplement audit claimed. none_found. Subclaims: SC3.

### [56] Cycling cancer persister cells arise from lineages with distinct programs.

**Shows:** Barcoded cycling persister lineages show distinct antioxidant and fatty-acid-oxidation requirements under EGFR inhibition, with related states in human residual samples.

**Does not show:** Culture lineage dynamics and patient transcriptional resemblance do not quantify every patient's persister fraction or prove all clinical recurrence is state driven.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC3.

### [57] Therapy-induced APOBEC3A drives evolution of persistent cancer cells.

**Shows:** APOBEC3A loss reduces treatment-associated mutagenesis and delays resistance in the experimental systems.

**Does not show:** Delay does not establish a sole universal resistance mechanism or spontaneous cancer initiation by APOBEC.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC3.

### [58] Focal adhesion kinase-YAP signaling axis drives drug-tolerant persister cells and residual disease in lung cancer.

**Shows:** FAK-YAP inhibition with EGFR inhibitors reduces residual disease and improves tumor response in tested experimental systems.

**Does not show:** Combined intervention and selected models do not establish a safe or universally curative clinical regimen.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC3.

### [59] Single-Cell Analysis Reveals Transcriptomic Features of Drug-Tolerant Persisters and Stromal Adaptation in a Patient-Derived EGFR-Mutated Lung Adenocarcinoma Xenograft Model.

**Shows:** Residual and baseline tumors in an EGFR-mutant lung adenocarcinoma PDX retain genomic clonal architecture; rare baseline cells have DTP-like transcripts, and CAF populations shift toward IL6 production with in vitro CAF-state conversion.

**Does not show:** A single PDX, association and state similarity do not prove all patient reservoirs or native relapse dependency; linked correction details are unaudited.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. ErratumIn: PMID37999691. Subclaims: SC3.

### [60] Targeting TROP2 in drug-tolerant persister cells delays EGFR tyrosine kinase inhibitor resistance in non-small-cell lung cancer.

**Shows:** TROP2 derepression accompanies EGFR-inhibitor persisters; experimental TROP2-directed combination delays relapse, with preliminary phase-2 observations.

**Does not show:** Preliminary nonrandomized observations are not controlled treatment-free cure or universal residual-cell dependence.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC3.

### [61] Discontinuation of imatinib in patients with chronic myeloid leukaemia who have maintained complete molecular remission for at least 2 years: the prospective, multicentre Stop Imatinib (STIM) trial.

**Shows:** STIM tests stopping imatinib after at least two years of deep molecular remission; many relapse, while a selected fraction maintains molecular remission off treatment and relapsers respond to rechallenge.

**Does not show:** Enrollment selects prior deep responders, not all newly diagnosed CML; persistent molecular remission is not proof of absence of all leukemic cells. It contradicts continuous-drug necessity for every patient.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC3.

### [62] Genome-scale CRISPR-Cas9 knockout screening in human cells.

**Shows:** Genome-scale CRISPR selection identifies diverse genes whose loss confers vemurafenib resistance.

**Does not show:** One cellular resistance screen is not a representative pan-cancer tumor-maintenance map or proof of one dominant oncogene across most cancers.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC2, SC3.

### [63] Genetic Drivers of Sensitivity or Resistance to RAS(ON) Multiselective Inhibitors in NRAS-Mutated Melanoma.

**Shows:** NRAS melanoma RAS-on response and resistance depend on alterations including cyclophilin-A and MAP2K1; two clinical cases show divergent response.

**Does not show:** Two cases do not define response frequency or clinical cure. The peer-reviewed report replaces its earlier preprint without double counting.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. UpdateOf: PMID41446219. Subclaims: SC1, SC3.

### [64] Abrogation of Oncogenic RAS Signaling by a RAS(ON) Inhibitor Doublet Primes Immune-Refractory KRASG12C-Mutant NSCLC for Immune Checkpoint Blockade.

**Shows:** RAS-on inhibitor combinations regress tested tumors and require host T cells for deeper responses.

**Does not show:** These are preclinical results, not demonstrated regression or complete responses in NSCLC patients; host dependence limits purely cell-autonomous interpretation.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC3.

### [65] EML4-ALK mediates resistance to KRAS G12C inhibition and induces an oncogenic dependency by rewiring signaling through the wild-type RAS pathway.

**Shows:** Acquired EML4-ALK restores wild-type RAS signaling under KRAS-G12C inhibition; continued KRAS blockade creates a context-specific ALK sensitivity and schedule-dependent combination response.

**Does not show:** Selected models and treatment sequence do not establish population prevalence or clinical eradication.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC3.

### [66] T-cell Dependency of Tumor Regressions and Complete Responses with RAS(ON) Multi-selective Inhibition in Preclinical Models of Pancreatic Ductal Adenocarcinoma.

**Shows:** RAS-on inhibition regresses pancreatic tumors in tested models; T cells and dendritic cells are required for the deeper response, with greater complete responses after immune combination.

**Does not show:** Complete responses are preclinical and combination dependent; shared drug-class programmes are not independent clinical validation or treatment-free cure.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC3.