# HSEF v5.3 review: Cancer Stem Cell Hypothesis

> Sources: 68 primary reports; claims: 74; verified: 74; unverified: 0. Verification is bounded attribution. Access: 5 partial-body and 63 complete abstract/status examinations. Literature cutoff: 2026-10-05; source examination UTC date: 2026-10-06. No complete-paper audit or blinded calibration claimed.
> **Adversarial review:** Generated by Sol 6.1; adversarially reviewed by Claude Opus 5.5 (Anthropic) on 2026-10-06. 0 corrections applied (see ADVERSARIAL REVIEW AND APPROVAL). Status: **Approved as revised.**

## AT A GLANCE

**Functional stem-like compartments can sustain specific cancers; a fixed, universally necessary CSC hierarchy is not established.**

| Level | Confidence | Causal role | Momentum | Stability | Divergence |
|---|---|---|---|---|---|
| Γ — Mechanistically plausible, partial system support | Moderate | reinforcing / stabilizing factor | Mechanistic only | Stable | Contested |

**Deciding experiment:** In prespecified AML and epithelial tumour cohorts, combine longitudinal clone/state tracking with selective depletion of the putative stable apex, an equally cytotoxic bulk-killing comparator and population-specific rescue; quantify replacement from negative states and durable maintenance/relapse. Use native human serial sampling and independent disease-relevant perturbation platforms rather than relying on one marker or host.

**Belief versus evidence:** Γ for the exact registered broad fixed-hierarchy claim. The cancer rubric explicitly permits Ε for context-specific CSC validity; that threshold is awarded to SC2 and SC3 rather than silently replacing SC5 with a weaker functional-stemness statement. Across-context causal studies are credited as a convergent set, but their results include genuine population replacement, stage-specific dispensability and alternative lineage routes. They therefore do not cross the fixed cross-human necessity gate. Scope-matched and primary-subclaim aggregation both return Γ; weakest-link also returns Γ. Narrow existence is Ζ, and defined maintenance/relapse dependencies are Ε.

**Best-supported rival:** Context-dependent hybrid hierarchy/plasticity. A prospective native lineage study coupling clone ancestry, reversible state measurement, selective ablation and restoration under identical treatment could separate a plastic functional hierarchy from pure state-equilibrium explanations. Neither rival is awarded universal necessity.

## COMPETING MECHANISTIC MODELS

**Rank 1: Context-dependent hybrid hierarchy/plasticity (Ζ).** Functional hierarchies coexist with changing membership and genetic/niche constraints. Native tracing/ablation and human CRC regrowth jointly support important stem states with replacement; PMID22854781, PMID28355176 and PMID41565826. A hybrid is broad and requires prospective transition/dependency predictions to avoid explaining every possible outcome.

**Rank 2: Reversible state plasticity (Ζ).** Tumour cells can enter or leave self-renewing/therapy-resistant states. Mammary intravital tracing, AML reversal, CRC negative-to-positive recovery and orthotopic HPCS re-entry exclude permanent one-way membership in these settings. Not every tumour cell is equipotent and BCC shows limited short-term recovery; plastic capacity is not sufficient to explain every relapse.

**Rank 3: Microenvironment-dependent stemness (Ε).** Location and stromal or metabolic signals determine functional stemness. Marker-free pancreatic tracing, CRC edge clones and lactate/COL17A1 perturbations support niche-dependent behaviour. CRC metastasis and genetic driver history also supply cell-intrinsic mechanisms; one niche model does not explain all cancer classes.

**Rank 4: Clonal evolution (Ε).** Genetic subclones selected by therapy generate propagation and resistance. Human AML paired sampling and single-cell genotypes, and pancreatic replica clones, establish selected resistant lineages. Genetically stable CRC clones show functional variability and multiple states can interconvert; mutation alone does not explain membership.

**Rank 5: Fixed CSC hierarchy (Γ).** A discrete self-renewing apex has stable membership and is irreplaceable across human cancers. Historical AML/breast/brain transplantation and native Sox2/glioma/BCC dependencies provide the strongest local case. Permissive melanoma assays, mature AML reversal, Lgr5-negative CRC maintenance, EMP1 timing and HPCS replenishment contradict universal fixed membership.

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

Discrete self-renewing cancer stem-cell populations organized in stable hierarchies are necessary for tumour maintenance and post-treatment relapse across human cancers.

The registered gate is cross-cancer generalization SC5. Stable hierarchy and necessary maintenance/relapse are separate; strong endogenous context-specific results do not establish the universal fixed-population model.

Selection: public-interest rank 16 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 early AML, breast, brain and colon studies prospectively separated cells with enriched serial transplantation activity. That evidence established functional heterogeneity and self-renewal potential, while leaving host permissiveness and sorting as major limits (PMID7509044, PMID9212098, PMID12629218, PMID15549107, PMID18765800). Optimized melanoma assays then detected abundant tumour-initiating cells and reversible marker expression; an assay-defined rare apex is therefore not a universal property (PMID19052619, PMID21075313).

Native tracing materially strengthens the causal case. Intestinal adenomas and skin tumours contain cells that contribute persistent progeny; glioma regrowth after temozolomide originates from a marked subset and is impaired by its ablation. Sox2 lineage ablation/deletion in skin and Prom1 depletion in liver provide context-specific functional dependencies (PMID22855427, PMID22854777, PMID22854781, PMID24909994, PMID34588223). This evidence is not demoted to marker association, nor does it establish permanence in every tumour.

CRC offers a direct distinction between functional importance and fixed identity. Human organoid-derived tumours regress after LGR5 ablation yet regain stem cells through differentiated-lineage reversion. Mouse CRC can be maintained by Lgr5-negative cells, and disseminating negative cells can restore a positive compartment for outgrowth. Tissue origin, transplantation and anatomical phase remain explicit (PMID28355176, PMID28358093, PMID32169167).

Relapse is not a single endpoint. Original EMP1 Results show that ablation before CRC resection prevented recurrence in most mice, whereas post-seeding ablation failed; independent pre-resection Lgr5 ablation did not prevent relapse. Mex3a, p27/COL17A1 and p57 studies identify dormant/adaptive therapy survivors with different functional boundaries (PMID36352230, PMID35773527, PMID35798028, PMID36880956). BCC combined-treatment studies demonstrate short-horizon relapse suppression, with normal-liver toxicity constraining sustained Lgr5 ablation; four-week absence of recurrence is not long-term cure (PMID30297799).

Human AML supplies clinical ancestry and relapse linkage, but not one invariant primitive phenotype. Paired diagnosis/relapse work finds rare primitive and larger committed-phenotype origins sharing stemness. Mature AML cells can recover leukemogenic properties after differentiation-program perturbation or therapy withdrawal. Pre-leukemic persistence, heterogeneous genotype routes and expression prognostics are distinct evidence classes (PMID28658204, PMID31374198, PMID24522528, PMID30827681, PMID37098346).

Recent work makes the stable-population assumption less tenable without eliminating therapeutic opportunity. Lung HPCS tracing/ablation establishes a potent progression and resistance hub; non-HPCS cells reacquire it in transplantation tests, and residual reporter-positive cells can escape incomplete recombination. TROP2 therapy remodels CRC states reversibly, while ZFP36L2 loss can suppress canonical LGR5 restoration yet select alternative metastatic fates (PMID41565826, PMID42386981, PMID42557330). These findings favour targeting functions/transitions with measured escape, rather than assuming immutable membership.

The breadth of the corpus tests the conjunction rather than merely accumulating positive marker papers. Pancreatic marker-free tracing and fibroblast-dependent growth, breast/melanoma plasticity, ALL equipotent founders and prostate/CRC lineage reprogramming show that histology, mutation, niche and treatment are mechanistically meaningful limits. Strong narrow dependencies and credible exceptions coexist (PMID34686335, PMID30177776, PMID21498687, PMID28487542, PMID29228234, PMID42330962).

## 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_16_20_2026-10-05/cancer-stem-cell-hierarchy/consensus-leads-search-log.json`
- `deliverables/hsef_v53_popularity_ranks_16_20_2026-10-05/cancer-stem-cell-hierarchy/consensus/browser-provenance.json`
- `deliverables/hsef_v53_popularity_ranks_16_20_2026-10-05/cancer-stem-cell-hierarchy/consensus/CONSENSUS_PROOF.png`
- `deliverables/hsef_v53_popularity_ranks_16_20_2026-10-05/cancer-stem-cell-hierarchy/consensus/deep-dom.txt`
- `deliverables/hsef_v53_popularity_ranks_16_20_2026-10-05/cancer-stem-cell-hierarchy/consensus/deep-response.txt`

The located artifacts do not state a reliable source-comparison count. No count is inferred.

A dated adversarial approval record exists for the separately identified cancer-stem-cell-hierarchy.review.md (2026-10-06; see edition-approval-audit.json). Its scope is that recorded edition. This supplement has no separate approval and no coverage is inferred from topic identity. No score or confidence change is made by this reconciliation.
<!-- end-consensus-consistency -->


### Subclaim matrix and aggregation

| Subclaim | Scope | Level | Decision |
|---|---|---|---|
| SC1 Hierarchical organization with functionally distinct stem-like cells exists. | prespecified tumour histologies and endogenous experimental contexts | Ζ | Endogenous skin, intestinal and glioma tracing, supported by human-origin functional assays, independently establishes functional hierarchy/stem-like contribution in defined settings. It does not establish permanent marker identity in every cancer. |
| SC2 A discrete CSC population is necessary for tumour maintenance. | defined native or patient-derived tumours | Ε | Precise Sox2, Prom1 and HPCS perturbations impair growth in defined native models. Lgr5-negative CRC maintenance and context-specific niche dependence delimit the necessity claim; CSC-like function can be required without an irreplaceable fixed population. |
| SC3 CSCs are necessary for relapse after treatment. | defined treated tumours and relapse horizons | Ε | Glioma, BCC, p57 and residual CRC tests demonstrate treatment/stage-dependent relapse dependencies. EMP1 ablation before surgery and after seeding differ; selective Lgr5 depletion can fail to prevent CRC relapse. Clinical universal relapse prevention is not demonstrated. |
| SC4 The hierarchy is stable rather than a reversible cell state. | native tumours under baseline and treatment conditions | Γ | Permanent one-way membership remains a plausible local description, but mammary intravital tracing, AML differentiation reversal, CRC regrowth and HPCS re-entry directly contradict its broad stability prediction. The level records the unestablished general mechanism, not endorsement of stability. |
| SC5 A fixed discrete CSC hierarchy is necessary across human cancers. | human haematological and solid cancers | Γ | The preregistered cross-human-cancer fixed-population gate is not crossed. Strong causal successes in defined models do not establish the conjunction of stable hierarchy, irreplaceable membership and necessary maintenance/relapse across human haematological and solid cancers. |

Preregistered aggregation: scope-matched. Γ for the exact registered broad fixed-hierarchy claim. The cancer rubric explicitly permits Ε for context-specific CSC validity; that threshold is awarded to SC2 and SC3 rather than silently replacing SC5 with a weaker functional-stemness statement. Across-context causal studies are credited as a convergent set, but their results include genuine population replacement, stage-specific dispensability and alternative lineage routes. They therefore do not cross the fixed cross-human necessity gate. Scope-matched and primary-subclaim aggregation both return Γ; weakest-link also returns Γ. Narrow existence is Ζ, and defined maintenance/relapse dependencies are Ε. Sensitivities: weakest-link Γ; primary-subclaim Γ; scope-matched Γ.

### Decisive-perturbation inventory and causal tally

[6] More permissive host conditions detected tumour formation from many unselected human melanoma cells, including single-cell inocula. High transplantation competence challenges rarity and assay exclusivity but alone does not quantify endogenous fate or refute all functional hierarchies.

[7] Single-cell melanoma transplants revealed multiple propagating phenotypes with different capacities to reconstruct heterogeneity and resist chemotherapy. Transplanted mouse tumour populations do not establish patient recurrence mechanisms; self-renewal need not require multilineage differentiation.

[8] Inducing EMT in mammary epithelial systems increased stem-associated properties and tumour-forming activity after transformation. Induced EMT and sphere readouts do not measure natural bidirectional transitions or establish universal therapy-resistant CSC identity.

[9] Purified mammary non-stem populations spontaneously generated stem-like cells, with enhanced conversion after transformation in vitro and in vivo. Selected mammary systems do not establish transition rates across all tumours; spontaneous conversion undermines a permanently one-way hierarchy.

[10] Sorted breast cancer cell-line states returned toward equilibrium proportions, consistent with stochastic transitions and de novo stem-like states. A Markov fit and cultured state recovery do not uniquely identify endogenous patient fate dynamics; two linked errata remain unresolved.

[14] Paired diagnosis/relapse genetics and xenografts traced resistant origins to either rare primitive LSCs or larger committed-phenotype subclones retaining stemness signatures. The two relapse patterns support functional stemness while contradicting a universal rare primitive phenotype; no randomized selective patient LSC ablation was tested.

[15] Lentiviral tracking within genetically stable colorectal clones revealed variable propagation and chemotherapy-selected minor or dormant lineages. Serial xenografts select engraftment-capable cells and do not distinguish every source of state variability in untreated patients.

[16] Endogenous multicolour lineage retracing showed Lgr5-positive adenoma cells self-renewing and producing the other adenoma cell types. A lineage contribution in mouse precursor adenomas is not necessity, patient carcinoma relapse, or proof that negative cells cannot regenerate stemness.

[17] Genetic clonal tracing in native skin tumours distinguished persistent stem-like and transient compartments in papillomas, with different dynamics in invasive carcinoma. Contribution and stage-dependent clonal behaviour do not prove universal rare-CSC necessity or an invariant hierarchy through progression.

[18] Pulse-chase and transgene-directed ablation identified an endogenous glioma subset sustaining regrowth after temozolomide; chronic ganciclovir arrested growth and combination treatment impeded development. The engineered glioma and nestin-TK targeting define a context-specific dependency, not universal patient CSC identity or selective clinical relapse prevention.

[19] Human CRC organoid lineage tracing and selective LGR5 ablation caused regression followed by regrowth involving KRT20-lineage conversion into LGR5-positive cells. Engineered organoid xenografts show a plastic functional hierarchy rather than necessity of a permanently distinct pre-existing CSC population in patients.

[20] Selective Lgr5 ablation restricted primary CRC growth without regression while negative cells replenished the pool; liver metastases showed stronger Lgr5 dependence. Dependency differs by anatomical stage; primary maintenance despite ablation contradicts universal fixed-population necessity while preserving a metastatic dependency.

[21] Edited patient-derived CRC organoid xenografts showed long-term LGR5-lineage self-renewal and differentiated descendants. Human-origin material in an engineered mouse host is not intact-patient fate tracing, and lineage contribution alone does not demonstrate necessity.

[22] CRC cells downregulated initiating-cell markers under differentiating adherent conditions yet retained serial xenograft tumour initiation and clonal contribution. Culture-induced marker differentiation does not identify every endogenous transition, but marker-defined phenotype was not necessary for transplantation activity.

[23] Mouse CRC and human xenograft tracking showed many disseminating Lgr5-negative cells restoring positive stem-like cells; plasticity supported metastatic outgrowth. Seeding, establishment and outgrowth are distinct phases; a plastic stem state requirement is not a fixed originating cell population requirement.

[24] In orthotopic CRC relapse models, pre-resection EMP1-high ablation prevented recurrence in most mice; post-seeding ablation failed, and pre-resection Lgr5 ablation did not prevent recurrence. Timing and metastatic stage determine dependency; patient signatures were associative, and EMP1 residual states are not an immutable LGR5 apex.

[25] Mex3a-lineage persisters expanded after chemotherapy and transiently shifted away from the WNT/stem program; deficient cells differentiated and failed to resist treatment. A therapy-dependent, niche-adaptive persister state is not an invariant CSC phenotype; clinical selective targeting and durable patient relapse prevention were not tested.

[26] Mouse and human neoplasia showed variable conventional/regenerative stem-state admixture that adapted under acute selective pressure and tracked chemoresistance. The fitness-landscape interpretation is not a unique proof of transition causation; a linked corrigendum remains part of the access limit.

[27] Marker-independent in-situ pancreatic clonal tracing found widespread clonogenic potential controlled by fibroblast proximity; stromal Hedgehog inhibition altered growth mode. Spatially instructed clonogenicity challenges a fixed rare apex, but the perturbation can change multiple stromal functions and does not cover every PDAC subtype.

[28] Serial pancreatic xenografts were maintained by successive transiently active initiating clones with little overlap between generations. Transplant bottlenecks may contribute to clonal succession; these data challenge fixed continual activity without establishing native patient dynamics.

[29] Clonal replica pancreatic xenografts exposed pre-existing genetic and functional diversity with different gemcitabine responses. Drug sensitivity and predictive signatures do not show that one stable CSC compartment is necessary for patient relapse.

[32] Pre/post-mutation STRACK tracing showed driver-specific expansion of differentiation-primed stem cells and Npm1c-associated reversal toward primitive malignant states. Early engineered murine evolution is not relapse necessity across cancers, but demonstrates that cell history and mutation alter the apparent hierarchy.

[33] Prom1 lineage tracing and depletion in two intact HCC models linked clonal expansion and dynamic dedifferentiation to tumour growth; depletion impeded growth. A lineage-state dependency in these models is not a permanently uniform marker compartment or patient-wide relapse necessity.

[34] Integrated SCC lineage/state analyses showed chemotherapy suppressing proliferation and activating plasticity; manipulating differentiation enhanced cell-cycle-inhibitor responses. Treatment-induced state movement challenges stability but does not establish that all tumour cells are equivalently plastic or that the same switch governs every cancer.

[36] TNBC profiling identified FXYD3-positive ancestral/persister-like cells that persisted during neoadjuvant therapy and were sensitive to Na/K-pump inhibitors. Clinical persistence is associative and pump inhibitors affect more than CSC-specific functions; selective CSC elimination improving patient relapse-free survival was not established.

[37] Anti-CD123 treatment reduced human AML xenograft engraftment and established disease burden, impaired secondary transplantation and improved mouse survival. Homing, cytokine blockade and innate effector activation are intertwined; a mouse efficacy experiment does not isolate a fixed population's necessity for human cure.

[38] Perturbing endogenous PU.1 or withdrawing differentiation treatment allowed mature AML cells in murine and human models to regain clonogenic/leukemogenic activity. This does not show unrestricted reversal in all AML genotypes, but directly challenges irreversible differentiation even in a prototypical hierarchical malignancy.

[39] Ovarian PDX tumours retained heterogeneity and clinical response patterns; carboplatin/paclitaxel enriched stem-associated populations without making persisters exclusively those populations. Marker enrichment after treatment does not identify uniquely causal relapse cells or distinguish selection from induced plasticity.

[40] Clonal HRAS-mutant skin models and functional pathway tests linked dynamic CSC-stromal signalling, angiogenesis and metabolic signalling to malignant transition. Interdependent niche signals support a contextual mechanism rather than a cell-intrinsic immutable hierarchy or pan-cancer necessity.

[42] Functional miR-126 manipulation regulated primary AML self-renewal, differentiation and quiescence through PI3K/AKT/MTOR-associated signalling in vivo. The pathway also affects normal HSCs differently; functional perturbation in transplanted systems is not exclusive clinical LSC necessity, and an erratum is linked.

[43] Barcoded primary ALL blasts showed many approximately equipotent founders contributing through serial transplants rather than a rare dominant apex. Transplant activity is host-conditioned; this challenges rarity/exclusivity in ALL without showing that every native blast has identical fate in patients.

[44] Prostate small-cell PDXs expressed reprogramming factors, and factor transfection altered expression and stem-like colony phenotypes. Forced transcription-factor expression and colonies establish plausibility, not natural lineage conversion or fixed CSC dependence.

[45] Ly6d genetic perturbation in KRAS-driven pancreatic models affected a persistent gastric-like lineage and delayed tumourigenesis; human expression correlated with prognosis. A genetic lineage program and OXPHOS dependency do not prove that a discrete cell population is universally necessary or that patient relapse is prevented by targeting it.

[46] TNBC organoids retained EMT-associated ciliated stem-like survivors after chemotherapy; genetic cilia ablation and inhibitors suppressed resistance. Organoid chemoresistance and NFkB-associated killing are not selective in-patient CSC eradication or a demonstration of permanently fixed hierarchy.

[47] GATA6 loss in orthotopic CRC organoid models increased metastatic plasticity and LGR5-negative states with little primary-growth change. Genetic reprogramming shows context-specific metastatic regulation; serial transplant selection and native patient translation remain limits.

[48] POLR1A-high biosynthetic CRC compartments spanned LGR5-positive and negative cells; genetic population ablation imposed irreversible growth arrest in the reported systems. Abstract-level access does not establish every model's endogenous status or exclude biosynthesis-dependent general toxicity; this functional compartment is broader than LGR5 identity.

[49] Combined Pten/Tgfbr2 perturbation increased reprogramming, rapid prostate tumour development and early mortality with BMP/stemness-associated signalling. Multi-pathway genetic changes are not selective ablation of a fixed CSC population; survival associations do not establish human mechanistic mediation.

[50] ABCB5-enriched melanoma cells showed serial xenograft self-renewal and antibody targeting inhibited established tumour growth. Host and assay conditions constrain rarity; antibody-mediated cytotoxicity and ABCB5 functions do not establish universal exclusive endogenous CSC maintenance.

[51] FAK inhibition depleted a transient neural-crest-like residual melanoma state and delayed xenograft relapse, with escape favouring genetic resistance. A drug acting on a signalling network is not perfectly state-specific; delayed rather than abolished relapse illustrates alternate resistance routes.

[52] Marker-sorted melanoma populations transiently overshot stem-marker equilibrium, consistent with regulated phenotypic switching after depletion. Marker dynamics and a regulatory model do not alone prove tumour-propagating function or endogenous relapse prevention failure.

[53] ABCB5 blockade altered an IL1/IL8/CXCR1 paracrine circuit, induced differentiation, reversed drug resistance and impaired melanoma growth in vivo. Cytokine interactions with negative cells favour a network dependency; inhibition does not establish permanently discrete CSC identity or clinical cure.

[54] Chronic extracellular acidosis increased melanoma stem-associated and tumour-initiating properties, with a reversible phenotype linked to EMT. Artificial environmental adaptation and initiation assays establish plastic potential rather than the rate or necessity of native patient state switching.

[55] Reporter-tracked human intestinal tumour organoids showed lactate suppressing differentiation and inducing a proliferative stem state through acetylation/MYC-associated regulation. Organoid transitions and candidate BRD4 targeting do not demonstrate prevention of patient relapse or a unique physiological lactate mechanism.

[56] Intravital human CRC organoid xenograft tracing linked pre-existing LGR5/p27 dormancy to survival and post-chemotherapy expansion; COL17A1 loss and YAP interference altered sensitivity/regrowth. Human-origin transplanted systems and niche signalling do not prove a fixed population's necessity in patients; delaying regrowth is not universal eradication.

[57] Anti-CD276 eliminated stem-enriched SCC cells through CD8-dependent immunity and reduced tumour growth and nodal metastases in mouse HNSCC. Immune-mediated targeting remodels several tumour states; human CD276 expression is not patient evidence that a stable CSC compartment is uniquely necessary.

[58] SHH medulloblastoma studies identified OLIG2 progenitors that changed to quiescent stem-like states; depletion/deletion impeded initiation and pathway combinations arrested growth. Therapy-resistant enrichment is not direct selective relapse prevention in children; stem-like function is developmentally and disease-context dependent.

[59] Direct patient melanoma cells frequently initiated serially transplantable tumours in permissive hosts; diverse markers failed to separate a large non-tumorigenic compartment and were reversible. A transplantation counterexample challenges rarity and marker exclusivity, but does not establish identical native fate for every melanoma cell.

[60] A syngeneic orthotopic intestinal organoid model identified quiescent p57-positive lineages surviving chemotherapy; their ablation suppressed post-treatment regrowth. This is a transplanted model and demonstrates suppression, not universal complete relapse prevention or an invariant CSC pool across tumours.

[61] In engineered BCC models, Lgr5-positive persisters drove regrowth after vismodegib withdrawal; combined treatment removed most lesions and dual Wnt/Hedgehog inhibition prevented recurrence over the tested four-week follow-up. Human biopsies supplied phenotype/withdrawal linkage; mouse Lgr5-ablation duration was limited by normal liver toxicity, and long-term eradication was not established.

[62] Autochthonous lung tumour reporters, tracing and two ablation approaches showed a high-plasticity hub driving progression, burden and resistant states. Non-HPCS cells can acquire HPCS; suppression does not establish an immutable population, pan-cancer clinical necessity or durable human cure. The linked earlier version is not a second independent report.

[63] Patient-derived CRC models showed context-dependent TROP2-positive stemness, chemotherapy-induced state redistribution and enhanced efficacy of combination TROP2-directed treatment. State-specific vulnerability and model efficacy do not demonstrate permanent cell identity or relapse-free survival benefit in patients.

[64] ZFP36L2 loss impaired canonical LGR5 re-entry/metastatic growth while promoting alternative heterogeneous fates; mouse regeneration tests also implicated stress-responsive dedifferentiation. Canonical stem-state dependency can be bypassed by alternative plasticity; poor-outcome associations are not a pan-cancer necessity/sufficiency test.

[65] Intravital lineage tracing in unperturbed mammary tumours documented disappearance of existing stem-like contributors and formation of new ones. A dynamic lineage contribution supports plastic stemness, without measuring universal patient transition rates or showing every cancer lacks a hierarchy.

[66] GBM stem-marker populations exhibited reversible microenvironment-instructed transitions; faster reconstitution of heterogeneity conferred an in-vivo growth advantage. Model-based and transplanted growth tests challenge discrete marker identity but do not exclude functionally important stem-like states in native gliomas.

[67] Time-resolved colon xenograft clones linked functional stemness to edge location and fibroblast-derived osteopontin, including after chemotherapy. Quantitative model fit is not unique proof; transplanted tissue and stromal interactions limit inference about a stable intrinsic patient hierarchy.

[68] Sox2 deletion reduced chemical skin tumour formation; lineage ablation and conditional deletion in established papillomas/SCCs caused regression. SOX2 regulates multiple survival and lineage functions; a strong skin-model dependency does not establish pan-cancer fixed-CSC necessity.

Tally: 56 discriminating contrasts. Null and negative contrasts are discriminators, not positive threshold crossings. Reports and replications are not independent causal lines. Selective killing of a state can affect multiple functions; it is not identical to a mechanism-specific rescue. Prevention over a short mouse follow-up is not durable clinical cure. Null/contrary perturbations and fate tests are discriminators but not positive universal threshold crossings.

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

The cancer rubric gives endogenous lineage tracing and precise disease-relevant perturbation substantial causal weight; SC1 reaches Ζ and defined maintenance/relapse reach Ε. Its statement that CSC models often have context-specific validity is applied explicitly to those scopes. Γ restrictions on sphere/marker/xenograft-only inference do not erase the assembled native evidence. SC4 and SC5 stop on unmet broader threshold burdens and contradictory stability predictions, without an invented literal hard cap or treating the absence of an RCT as dispositive.

Gamma hard-rule audit: Sphere or marker evidence alone does not establish endogenous dependency. Transplant competence is distinct from native lineage contribution. These hard rules apply to the indicated assay-only sources; they are not re-applied over the convergent endogenous set. The global stable-membership gate fails its positive scope-matched perturbation test and encounters counterevidence; this is not a blanket no-RCT or no-human-ablation cap.

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. Endogenous skin, intestinal and glioma tracing, supported by human-origin functional assays, independently establishes functional hierarchy/stem-like contribution in defined settings. It does not establish permanent marker identity in every cancer.

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: yes. Precise Sox2, Prom1 and HPCS perturbations impair growth in defined native models. Lgr5-negative CRC maintenance and context-specific niche dependence delimit the necessity claim; CSC-like function can be required without an irreplaceable fixed population.

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: yes. Glioma, BCC, p57 and residual CRC tests demonstrate treatment/stage-dependent relapse dependencies. EMP1 ablation before surgery and after seeding differ; selective Lgr5 depletion can fail to prevent CRC relapse. Clinical universal relapse prevention is not demonstrated.

SC4 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. Permanent one-way membership remains a plausible local description, but mammary intravital tracing, AML differentiation reversal, CRC regrowth and HPCS re-entry directly contradict its broad stability prediction. The level records the unestablished general mechanism, not endorsement of stability.

SC5 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. The preregistered cross-human-cancer fixed-population gate is not crossed. Strong causal successes in defined models do not establish the conjunction of stable hierarchy, irreplaceable membership and necessary maintenance/relapse across human haematological and solid cancers.

### 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: no. Replication of native human fate hierarchies with orthogonal functional perturbation and controls excluding lineage-marker induction artefacts.

SC2 Δ: “Requires at least one meaningful causal perturbation showing the proposed mechanism contributes to initiation, maintenance, or relapse.” Satisfied: yes. 

SC2 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. 

SC2 Ζ: “Requires:
- Independent causal confirmation across models
- Reduced viability of major alternatives
- Relevance beyond one assay class” Satisfied: no. Across predefined human tumour classes, selective elimination of the alleged fixed CSC population abolishes maintenance without generalized killing, followed by population-specific rescue and tests of replacement from negative cells.

SC3 Δ: “Requires at least one meaningful causal perturbation showing the proposed mechanism contributes to initiation, maintenance, or relapse.” Satisfied: yes. 

SC3 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. 

SC3 Ζ: “Requires:
- Independent causal confirmation across models
- Reduced viability of major alternatives
- Relevance beyond one assay class” Satisfied: no. Prospective state/clone tracing plus state-specific targeting improves durable relapse-free survival over matched bulk killing, with pre-existing versus induced states distinguished.

SC4 Δ: “Requires at least one meaningful causal perturbation showing the proposed mechanism contributes to initiation, maintenance, or relapse.” Satisfied: no. Bidirectional native fate mapping under baseline and treatment, with adequate detection limits, demonstrates irreversible membership and excludes state regeneration across the registered settings.

SC5 Δ: “Requires at least one meaningful causal perturbation showing the proposed mechanism contributes to initiation, maintenance, or relapse.” Satisfied: no. An independently replicated cross-histology causal dependency series satisfies stable membership, necessity for maintenance and durable relapse, and matched rescue in human disease or convergent field-valid equivalents.

Upper ladder: Withdrawal/rechallenge and combined pathway treatment have tumour-burden/relapse consequences in BCC models, with human biopsy/treatment-withdrawal linkage. Precise native skin/glioma/lung perturbations and patient-origin clone tracking provide upper-ladder-relevant convergence. They earn narrow causal credit, not a cap at association. No examined genetic natural experiment, cross-population dose response, human state-selective durable clinical endpoint or matched restoration establishes the full fixed hierarchy across human cancers. Actual recovery/dispensability observations constrain advancement more than an absence of randomized trials.

Independence: Independence is assessed by failure mode, not four papers equalling four causal lines. NSG xenografts, Lgr5 reporters, organoid/transplant platforms and common mutation backgrounds are explicitly shared. Distinct native tissue models, precise perturbations and patient ancestry address different errors; no complete multi-cohort provenance or blinded replication audit is claimed.

Native lineage contribution [17, 16, 18]: Excludes the assumption that stem-like contribution exists only because cells were transplanted; related fluorescent/Cre tracing retains marker and recombination limitations.

Precise native state/function perturbation [68, 33, 62]: Tests dependency beyond expression and engraftment; different skin, liver and lung backgrounds reduce one-model failure, while generalized state killing remains a shared limitation.

Longitudinal human ancestry and functional sampling [14, 31, 41]: Connects patient clonal history with phenotype and function, reducing wholly artificial lineage origin; sampling and transplantation selection still limit universal necessity.

Replacement and state reversal [65, 38, 19, 59]: Intravital tracing, endogenous transcriptional/differentiation perturbation and human-origin state assays have different principal artefacts; together challenge fixed identity rather than simply adding more marker papers.

### Falsification and pitfalls

Proposed CSC fraction is not required for maintenance/relapse. Lgr5-negative cells maintain primary CRC; pre-resection Lgr5 depletion in EMP1 relapse models fails to prevent recurrence. Additional downgrade: False.

Non-CSC populations reacquire equivalent function robustly. KRT20 CRC reversion, mature AML dedifferentiation, mammary contributor turnover and non-HPCS re-entry. Additional downgrade: False.

Marker identity confused with function. CD133/CD44/ABCB5/LGR5 sorting and transcriptomic stemness programs.

Xenograft capacity equated with endogenous hierarchy. Historical solid-tumour initiation assays and engineered human organoid hosts.

Relapse association mistaken for relapse causation. Patient survival signatures and persistent marker enrichment.

Plasticity ignored. Negative-to-positive reconstitution and therapy-induced TROP2/HPCS states.

Broad generalization across tumor types. Fixed necessary apex extrapolated from one tissue or treatment phase.

### Adversarial steelman: above, then below

One rung higher, Δ, is compelling if the claim is read as stem-like states contributing to cancer behaviour: several native ablations exceed that boundary. For the registered SC5 fixed cross-human necessity gate, Δ would require that the same fixed-population mechanism actually passes a relevant broad-scope dependency test; present successes and replacement failures do not supply it.

One rung lower, Β, would fit expression/prognosis alone. It is rejected because native tracing, precise ablation, patient-origin functional studies and state reversal supply substantially more than association. Γ is not assigned by ignoring these sources.

Reject Β for undercrediting native causal studies and reject global Δ because it silently weakens the fixed cross-human claim. Preserve Ζ existence and Ε context-specific maintenance/relapse, while the exact nominal gate remains Γ.

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

Γ for the exact registered broad fixed-hierarchy claim. The cancer rubric explicitly permits Ε for context-specific CSC validity; that threshold is awarded to SC2 and SC3 rather than silently replacing SC5 with a weaker functional-stemness statement. Across-context causal studies are credited as a convergent set, but their results include genuine population replacement, stage-specific dispensability and alternative lineage routes. They therefore do not cross the fixed cross-human necessity gate. Scope-matched and primary-subclaim aggregation both return Γ; weakest-link also returns Γ. Narrow existence is Ζ, and defined maintenance/relapse dependencies are Ε.

Step1: tumour-generated stress/niches can induce stem states, but those states feed back on growth; the complete mechanism is not solely a downstream consequence. Step2: ablation and functional state tests establish dependency in some contexts, excluding mere co-occurrence. Step3: no universal initiation-plus-removal-plus-rate-limiting case for the fixed hierarchy. Step4: universal irreplaceable membership is contradicted by recovery and alternate states. Step5 therefore applies to the registered broad mechanism: stem-like functions reinforce/maintain malignancy initiated by upstream events, with permissive or stronger dependencies in explicitly bounded settings.

Momentum: Mechanistic only. The 2026 HPCS, TROP2 and ZFP36L2 programmes test causal state plasticity and therapeutic vulnerabilities. They do not constitute a verified current cross-human fixed-population necessity programme; no ongoing/recruiting clinical trial status is asserted.

Stability: Stable. No verified pending result matching fixed membership and cross-human necessary maintenance/relapse is identified; the already published 2026 findings are included rather than treated as future triggers.

Divergence: Contested; no single implied-level arithmetic. These dimensions do not change level or confidence.

## PROPOSED DECISIVE EXPERIMENTS

In prespecified AML and epithelial tumour cohorts, combine longitudinal clone/state tracking with selective depletion of the putative stable apex, an equally cytotoxic bulk-killing comparator and population-specific rescue; quantify replacement from negative states and durable maintenance/relapse. Use native human serial sampling and independent disease-relevant perturbation platforms rather than relying on one marker or host.

Readout: Clone ancestry, bidirectional transition frequencies, target engagement, total burden and durable relapse-free survival with prespecified state-replacement detection limits. Controls: Marker-independent tracking; lineage-label induction controls; intact niche; normal stem-cell toxicity; orthogonal targeting; sham and matched bulk cytotoxicity; blinded relapse assessment; rescue without changing oncogenic drivers. Supporting: Irreplaceable fixed-apex depletion abolishes maintenance and recurrence across predefined classes, with selective rescue restoring both and no regeneration from negative cells. Falsifying: Negative cells or alternate states sustain disease or regenerate the apex; selective depletion fails despite verified engagement, or benefit tracks generalized killing rather than state dependency.

## RESEARCH TIMELINE

| Year | Milestone | Primary source | Interpretation |
|---|---|---|---|
| 1994 | Human AML initiating activity isolated by SCID transplantation | [PMID7509044](https://pubmed.ncbi.nlm.nih.gov/7509044/) | Functional assay milestone, not native patient necessity. |
| 1997 | AML self-renewing hierarchy in NOD/SCID assay | [PMID9212098](https://pubmed.ncbi.nlm.nih.gov/9212098/) | Strong historical foundation with host-selection limits. |
| 2003 | Prospective breast tumour-initiating cell isolation | [PMID12629218](https://pubmed.ncbi.nlm.nih.gov/12629218/) | Extends assay-defined hierarchy to a solid tumour. |
| 2004 | Human brain tumour initiating cells | [PMID15549107](https://pubmed.ncbi.nlm.nih.gov/15549107/) | Serial xenograft phenocopy; marker exclusivity is assay-conditioned. |
| 2008 | Permissive assays reveal common melanoma initiating cells | [PMID19052619](https://pubmed.ncbi.nlm.nih.gov/19052619/) | Rarity depends on assay conditions. |
| 2010 | Reversible non-hierarchical melanoma marker states | [PMID21075313](https://pubmed.ncbi.nlm.nih.gov/21075313/) | A direct challenge to universal stable identity. |
| 2011 | Mammary non-stem to stem-like conversion | [PMID21498687](https://pubmed.ncbi.nlm.nih.gov/21498687/) | Bidirectional state potential emerges. |
| 2012 | Native intestinal adenoma lineage tracing | [PMID22855427](https://pubmed.ncbi.nlm.nih.gov/22855427/) | Functional hierarchy without transplantation selection. |
| 2012 | Glioma regrowth subset traced and ablated | [PMID22854781](https://pubmed.ncbi.nlm.nih.gov/22854781/) | Endogenous treatment-response dependency in a defined model. |
| 2013 | Intravital mammary contributor turnover | [PMID23225641](https://pubmed.ncbi.nlm.nih.gov/23225641/) | Native hierarchy membership is dynamic. |
| 2014 | Sox2 ablation/deletion in skin cancer | [PMID24909994](https://pubmed.ncbi.nlm.nih.gov/24909994/) | Context-specific maintenance dependency. |
| 2017 | Human CRC LGR5 ablation and differentiated-lineage reversion | [PMID28355176](https://pubmed.ncbi.nlm.nih.gov/28355176/) | Functional apex can be regenerated. |
| 2017 | Two patient AML relapse-origin patterns | [PMID28658204](https://pubmed.ncbi.nlm.nih.gov/28658204/) | Clinical stemness linkage with phenotype heterogeneity. |
| 2018 | BCC persister targeting and withdrawal recurrence | [PMID30297799](https://pubmed.ncbi.nlm.nih.gov/30297799/) | Combination efficacy with toxicity and follow-up limits. |
| 2019 | Mature AML state reversal | [PMID31374198](https://pubmed.ncbi.nlm.nih.gov/31374198/) | Challenges irreversible differentiation. |
| 2021 | Marker-free pancreatic niche-instructed clonogenicity | [PMID34686335](https://pubmed.ncbi.nlm.nih.gov/34686335/) | Spatial context can determine function. |
| 2022 | EMP1 residual CRC cells and timed recurrence prevention | [PMID36352230](https://pubmed.ncbi.nlm.nih.gov/36352230/) | Pre-seeding and post-seeding dependencies differ. |
| 2026 | Autochthonous lung high-plasticity hub ablation | [PMID41565826](https://pubmed.ncbi.nlm.nih.gov/41565826/) | State dependency coexists with replenishment. |
| 2026 | TROP2-directed treatment remodels CRC states | [PMID42386981](https://pubmed.ncbi.nlm.nih.gov/42386981/) | Model therapeutic benefit does not imply fixed identity. |
| 2026 | ZFP36L2 loss selects alternative lineage outcomes | [PMID42557330](https://pubmed.ncbi.nlm.nih.gov/42557330/) | Canonical stem-state blockade can be bypassed. |

## KNOWLEDGE GAPS

Native human fate and dependency: distinguish engraftment potential from in-patient contribution, using safe serial lineage/genotype/state reconstruction rather than assuming reporter xenografts are native evidence (PMID28355176, PMID28658204, PMID37098346).

Selection versus induction: measure the origin of post-treatment stem states with simultaneous ancestry and state tracking; pseudotime alone cannot determine dedifferentiation (PMID23225641, PMID42386981, PMID40010350).

Ablation-resistant replacement and timing: determine which reservoirs regenerate an apex and when a state becomes dispensable after seeding; prespecify primary, micrometastatic and established-metastatic phases (PMID28358093, PMID36352230, PMID42557330).

Mechanism-specific clinical benefit: establish whether selectively targeting a functional state improves durable survival beyond equivalent bulk cytotoxicity and multicomponent pathway effects (PMID19570512, PMID33945793, PMID42386981).

Normal-tissue therapeutic window and durability: quantify normal stem-cell injury, incomplete targeting, escape and post-withdrawal relapse over longer horizons (PMID30297799, PMID41565826).

Prospective cross-histology predictions: define before observation which genotypes/niches retain a hierarchy, which replenish it and which bypass it; prevent a hybrid model from becoming unfalsifiable (PMID34686335, PMID21075313, PMID41545197).

## 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. Returned Methods/Results passages were reconciled with original-paper passages where available; failed retrievals, source substitutions and unverified aspects are recorded in the case reconciliation. Successful passage checks are distinguished from provider summaries. 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

68 primary reports were actually presented and manually appraised; five have selected original body-section examinations. One opened methods-only protocol was excluded. Body access and correction notices remain explicit. Consensus supplied four additional verified leads and a reconciled HPCS body check; its review-heavy summaries, counts and citation-weighted timeline do not establish field consensus. Exact v5.3 framework/validator are used; structural validation is not scientific peer review or a completed blinded calibration.

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:** 2026-10-06.

**Scope of checks.** All 68 reference PMIDs re-resolved against PubMed (68/68 correct for first author, title, journal and DOI; no retraction or expression of concern; four linked errata already recorded in the ledger's `status_flags`). Twelve load-bearing claims spot-checked against primary abstracts or the stored body passages (Chen 2012 glioma ablation, Boumahdi 2014 SOX2 deletion/lineage ablation, Zhou 2021 Prom1 depletion, Shimokawa 2017 LGR5 ablation and KRT20 reversion, Fumagalli 2020 Lgr5-negative dissemination, Cañellas-Socias 2022 EMP1 pre-resection ablation, Sánchez-Danés 2018 BCC relapse, Chan 2026 HPCS, Vaquero-Siguero 2026 TROP2, Jiang 2026 ZFP36L2, Quintana 2008/2010 melanoma permissiveness). Scoring logic re-derived independently before the existing letter was read: cap precedence, the Β/Γ/Δ boundary, the declared aggregation rule and the subclaim scopes were checked against `SKILL.md` Step 3 and `references/hsef-framework.md`.

**Findings.**

| ID | Severity | Location | Original → corrected | Evidence |
|---|---|---|---|---|
| — | — | — | No corrections required | See rejected findings below |

**Rejected findings (checked, no change).**

1. *Scope-matched aggregation (the error found in the 2026-09-30 red team of an earlier cancer-stem-cell review).* Checked specifically. That pass found an Ε headline produced by reading "scope-matched" as "the highest subclaim that reached Ε" while the broad-scope subclaims sat at Γ. This review does not repeat it: the subclaim whose scope matches the registered claim ("human haematological and solid cancers") is SC5, SC5 is Γ, and Γ is what the headline reports. The narrower Ζ existence and Ε maintenance/relapse findings are reported separately, as the rule requires, and weakest-link and primary-subclaim sensitivities are also Γ, so no rule choice moves the letter.
2. *Γ understates the evidence.* Considered and rejected: the registered claim is the conjunction of stable hierarchy, irreplaceable membership and necessary maintenance and relapse across human cancers, and the review's own inventory records population replacement (KRT20 reversion, Lgr5-negative maintenance), stage-dependent dispensability (EMP1 before versus after seeding) and alternative lineage routes, which block the broad-scope dependency test.
3. *Citation-year differences of ±1 between the printed issue year and the PubMed e-pub year* (references 15, 22, 33, 41). Publication-convention difference, not an error.
4. *Decisive-perturbation tally (56).* Recounted against the inventory entries; it reconciles.

**Residual limitations.** Verification depth is bounded by access: 5 partial-body examinations and 63 abstract/status examinations, as the header states. Claims resting on figures inside unexamined full texts were checked only to the depth the record allowed, and no image or raw-data audit was performed.

**Approval:** Following adversarial review and the corrections listed above, the revised document is approved. — Claude Opus 5.5 (Anthropic), 2026-10-06.

## REFERENCES

1. Lapidot T, Sirard C, Vormoor J et al. A cell initiating human acute myeloid leukaemia after transplantation into SCID mice. Nature. 1994. [PMID 7509044](https://pubmed.ncbi.nlm.nih.gov/7509044/). DOI: 10.1038/367645a0. Access: abstract_metadata. Study: PMID7509044.

2. Bonnet D, Dick JE et al. Human acute myeloid leukemia is organized as a hierarchy that originates from a primitive hematopoietic cell. Nature medicine. 1997. [PMID 9212098](https://pubmed.ncbi.nlm.nih.gov/9212098/). DOI: 10.1038/nm0797-730. Access: abstract_metadata. Study: PMID9212098.

3. Al-Hajj M, Wicha MS, Benito-Hernandez A et al. Prospective identification of tumorigenic breast cancer cells. Proceedings of the National Academy of Sciences of the United States of America. 2003. [PMID 12629218](https://pubmed.ncbi.nlm.nih.gov/12629218/). DOI: 10.1073/pnas.0530291100. Access: abstract_metadata. Study: PMID12629218.

4. Singh SK, Hawkins C, Clarke ID et al. Identification of human brain tumour initiating cells. Nature. 2004. [PMID 15549107](https://pubmed.ncbi.nlm.nih.gov/15549107/). DOI: 10.1038/nature03128. Access: abstract_metadata. Study: PMID15549107.

5. Vermeulen L, Todaro M, de Sousa Mello F et al. Single-cell cloning of colon cancer stem cells reveals a multi-lineage differentiation capacity. Proceedings of the National Academy of Sciences of the United States of America. 2008. [PMID 18765800](https://pubmed.ncbi.nlm.nih.gov/18765800/). DOI: 10.1073/pnas.0805706105. Access: abstract_metadata. Study: PMID18765800.

6. Quintana E, Shackleton M, Sabel MS et al. Efficient tumour formation by single human melanoma cells. Nature. 2008. [PMID 19052619](https://pubmed.ncbi.nlm.nih.gov/19052619/). DOI: 10.1038/nature07567. Access: abstract_metadata. Study: PMID19052619.

7. Held MA, Curley DP, Dankort D et al. Characterization of melanoma cells capable of propagating tumors from a single cell. Cancer research. 2010. [PMID 20048081](https://pubmed.ncbi.nlm.nih.gov/20048081/). DOI: 10.1158/0008-5472.CAN-09-2153. Access: abstract_metadata. Study: PMID20048081.

8. Mani SA, Guo W, Liao MJ et al. The epithelial-mesenchymal transition generates cells with properties of stem cells. Cell. 2008. [PMID 18485877](https://pubmed.ncbi.nlm.nih.gov/18485877/). DOI: 10.1016/j.cell.2008.03.027. Access: abstract_metadata. Study: PMID18485877.

9. Chaffer CL, Brueckmann I, Scheel C et al. Normal and neoplastic nonstem cells can spontaneously convert to a stem-like state. Proceedings of the National Academy of Sciences of the United States of America. 2011. [PMID 21498687](https://pubmed.ncbi.nlm.nih.gov/21498687/). DOI: 10.1073/pnas.1102454108. Access: abstract_metadata. Study: PMID21498687.

10. Gupta PB, Fillmore CM, Jiang G et al. Stochastic state transitions give rise to phenotypic equilibrium in populations of cancer cells. Cell. 2011. [PMID 21854987](https://pubmed.ncbi.nlm.nih.gov/21854987/). DOI: 10.1016/j.cell.2011.07.026. Access: abstract_metadata. Study: PMID21854987.

11. Eppert K, Takenaka K, Lechman ER et al. Stem cell gene expression programs influence clinical outcome in human leukemia. Nature medicine. 2011. [PMID 21873988](https://pubmed.ncbi.nlm.nih.gov/21873988/). DOI: 10.1038/nm.2415. Access: abstract_metadata. Study: PMID21873988.

12. Shlush LI, Zandi S, Mitchell A et al. Identification of pre-leukaemic haematopoietic stem cells in acute leukaemia. Nature. 2014. [PMID 24522528](https://pubmed.ncbi.nlm.nih.gov/24522528/). DOI: 10.1038/nature13038. Access: abstract_metadata. Study: PMID24522528.

13. van Galen P, Hovestadt V, Wadsworth Ii MH et al. Single-Cell RNA-Seq Reveals AML Hierarchies Relevant to Disease Progression and Immunity. Cell. 2019. [PMID 30827681](https://pubmed.ncbi.nlm.nih.gov/30827681/). DOI: 10.1016/j.cell.2019.01.031. Access: abstract_metadata. Study: PMID30827681.

14. Shlush LI, Mitchell A, Heisler L et al. Tracing the origins of relapse in acute myeloid leukaemia to stem cells. Nature. 2017. [PMID 28658204](https://pubmed.ncbi.nlm.nih.gov/28658204/). DOI: 10.1038/nature22993. Access: abstract_metadata. Study: PMID28658204.

15. Kreso A, O'Brien CA, van Galen P et al. Variable clonal repopulation dynamics influence chemotherapy response in colorectal cancer. Science (New York, N.Y.). 2013. [PMID 23239622](https://pubmed.ncbi.nlm.nih.gov/23239622/). DOI: 10.1126/science.1227670. Access: abstract_metadata. Study: PMID23239622.

16. Schepers AG, Snippert HJ, Stange DE et al. Lineage tracing reveals Lgr5+ stem cell activity in mouse intestinal adenomas. Science (New York, N.Y.). 2012. [PMID 22855427](https://pubmed.ncbi.nlm.nih.gov/22855427/). DOI: 10.1126/science.1224676. Access: abstract_metadata. Study: PMID22855427.

17. Driessens G, Beck B, Caauwe A et al. Defining the mode of tumour growth by clonal analysis. Nature. 2012. [PMID 22854777](https://pubmed.ncbi.nlm.nih.gov/22854777/). DOI: 10.1038/nature11344. Access: abstract_metadata. Study: PMID22854777.

18. Chen J, Li Y, Yu TS et al. A restricted cell population propagates glioblastoma growth after chemotherapy. Nature. 2012. [PMID 22854781](https://pubmed.ncbi.nlm.nih.gov/22854781/). DOI: 10.1038/nature11287. Access: abstract_metadata. Study: PMID22854781.

19. Shimokawa M, Ohta Y, Nishikori S et al. Visualization and targeting of LGR5+ human colon cancer stem cells. Nature. 2017. [PMID 28355176](https://pubmed.ncbi.nlm.nih.gov/28355176/). DOI: 10.1038/nature22081. Access: abstract_metadata. Study: PMID28355176.

20. de Sousa e Melo F, Kurtova AV, Harnoss JM et al. A distinct role for Lgr5+ stem cells in primary and metastatic colon cancer. Nature. 2017. [PMID 28358093](https://pubmed.ncbi.nlm.nih.gov/28358093/). DOI: 10.1038/nature21713. Access: abstract_metadata. Study: PMID28358093.

21. Cortina C, Turon G, Stork D et al. A genome editing approach to study cancer stem cells in human tumors. EMBO molecular medicine. 2017. [PMID 28468934](https://pubmed.ncbi.nlm.nih.gov/28468934/). DOI: 10.15252/emmm.201707550. Access: abstract_metadata. Study: PMID28468934.

22. Dubash TD, Hoffmann CM, Oppel F et al. Phenotypic differentiation does not affect tumorigenicity of primary human colon cancer initiating cells. Cancer letters. 2016. [PMID 26679053](https://pubmed.ncbi.nlm.nih.gov/26679053/). DOI: 10.1016/j.canlet.2015.11.037. Access: abstract_metadata. Study: PMID26679053.

23. Fumagalli A, Oost KC, Kester L et al. Plasticity of Lgr5-Negative Cancer Cells Drives Metastasis in Colorectal Cancer. Cell stem cell. 2020. [PMID 32169167](https://pubmed.ncbi.nlm.nih.gov/32169167/). DOI: 10.1016/j.stem.2020.02.008. Access: abstract_metadata. Study: PMID32169167.

24. Cañellas-Socias A, Cortina C, Hernando-Momblona X et al. Metastatic recurrence in colorectal cancer arises from residual EMP1+ cells. Nature. 2022. [PMID 36352230](https://pubmed.ncbi.nlm.nih.gov/36352230/). DOI: 10.1038/s41586-022-05402-9. Access: full_text_partial. Study: PMID36352230.

25. Álvarez-Varela A, Novellasdemunt L, Barriga FM et al. Mex3a marks drug-tolerant persister colorectal cancer cells that mediate relapse after chemotherapy. Nature cancer. 2022. [PMID 35773527](https://pubmed.ncbi.nlm.nih.gov/35773527/). DOI: 10.1038/s43018-022-00402-0. Access: abstract_metadata. Study: PMID35773527.

26. Vasquez EG, Nasreddin N, Valbuena GN et al. Dynamic and adaptive cancer stem cell population admixture in colorectal neoplasia. Cell stem cell. 2022. [PMID 35931031](https://pubmed.ncbi.nlm.nih.gov/35931031/). DOI: 10.1016/j.stem.2022.07.008. Access: abstract_metadata. Study: PMID35931031.

27. Lodestijn SC, Miedema DM, Lenos KJ et al. Marker-free lineage tracing reveals an environment-instructed clonogenic hierarchy in pancreatic cancer. Cell reports. 2021. [PMID 34686335](https://pubmed.ncbi.nlm.nih.gov/34686335/). DOI: 10.1016/j.celrep.2021.109852. Access: abstract_metadata. Study: PMID34686335.

28. Ball CR, Oppel F, Ehrenberg KR et al. Succession of transiently active tumor-initiating cell clones in human pancreatic cancer xenografts. EMBO molecular medicine. 2017. [PMID 28526679](https://pubmed.ncbi.nlm.nih.gov/28526679/). DOI: 10.15252/emmm.201607354. Access: abstract_metadata. Study: PMID28526679.

29. Seth S, Li CY, Ho IL et al. Pre-existing Functional Heterogeneity of Tumorigenic Compartment as the Origin of Chemoresistance in Pancreatic Tumors. Cell reports. 2019. [PMID 30726735](https://pubmed.ncbi.nlm.nih.gov/30726735/). DOI: 10.1016/j.celrep.2019.01.048. Access: abstract_metadata. Study: PMID30726735.

30. Wu J, Xiao Y, Sun J et al. A single-cell survey of cellular hierarchy in acute myeloid leukemia. Journal of hematology & oncology. 2020. [PMID 32977829](https://pubmed.ncbi.nlm.nih.gov/32977829/). DOI: 10.1186/s13045-020-00941-y. Access: abstract_metadata. Study: PMID32977829.

31. Beneyto-Calabuig S, Merbach AK, Kniffka JA et al. Clonally resolved single-cell multi-omics identifies routes of cellular differentiation in acute myeloid leukemia. Cell stem cell. 2023. [PMID 37098346](https://pubmed.ncbi.nlm.nih.gov/37098346/). DOI: 10.1016/j.stem.2023.04.001. Access: abstract_metadata. Study: PMID37098346.

32. Singh I, Fernandez-Perez D, Sanchez PS et al. Pre-existing stem cell heterogeneity dictates clonal responses to the acquisition of leukemic driver mutations. Cell stem cell. 2025. [PMID 40010350](https://pubmed.ncbi.nlm.nih.gov/40010350/). DOI: 10.1016/j.stem.2025.01.012. Access: abstract_metadata. Study: PMID40010350.

33. Zhou L, Yu KH, Wong TL et al. Lineage tracing and single-cell analysis reveal proliferative Prom1+ tumour-propagating cells and their dynamic cellular transition during liver cancer progression. Gut. 2022. [PMID 34588223](https://pubmed.ncbi.nlm.nih.gov/34588223/). DOI: 10.1136/gutjnl-2021-324321. Access: abstract_metadata. Study: PMID34588223.

34. Taylor MA, Kandyba E, Halliwill K et al. Stem-cell states converge in multistage cutaneous squamous cell carcinoma development. Science (New York, N.Y.). 2024. [PMID 38815020](https://pubmed.ncbi.nlm.nih.gov/38815020/). DOI: 10.1126/science.adi7453. Access: abstract_metadata. Study: PMID38815020.

35. Praktiknjo SD, Obermayer B, Zhu Q et al. Tracing tumorigenesis in a solid tumor model at single-cell resolution. Nature communications. 2020. [PMID 32080185](https://pubmed.ncbi.nlm.nih.gov/32080185/). DOI: 10.1038/s41467-020-14777-0. Access: abstract_metadata. Study: PMID32080185.

36. Li M, Nishimura T, Takeuchi Y et al. FXYD3 functionally demarcates an ancestral breast cancer stem cell subpopulation with features of drug-tolerant persisters. The Journal of clinical investigation. 2023. [PMID 37966117](https://pubmed.ncbi.nlm.nih.gov/37966117/). DOI: 10.1172/JCI166666. Access: abstract_metadata. Study: PMID37966117.

37. Jin L, Lee EM, Ramshaw HS et al. Monoclonal antibody-mediated targeting of CD123, IL-3 receptor alpha chain, eliminates human acute myeloid leukemic stem cells. Cell stem cell. 2009. [PMID 19570512](https://pubmed.ncbi.nlm.nih.gov/19570512/). DOI: 10.1016/j.stem.2009.04.018. Access: abstract_metadata. Study: PMID19570512.

38. McKenzie MD, Ghisi M, Oxley EP et al. Interconversion between Tumorigenic and Differentiated States in Acute Myeloid Leukemia. Cell stem cell. 2019. [PMID 31374198](https://pubmed.ncbi.nlm.nih.gov/31374198/). DOI: 10.1016/j.stem.2019.07.001. Access: abstract_metadata. Study: PMID31374198.

39. Dobbin ZC, Katre AA, Steg AD et al. Using heterogeneity of the patient-derived xenograft model to identify the chemoresistant population in ovarian cancer. Oncotarget. 2014. [PMID 25209969](https://pubmed.ncbi.nlm.nih.gov/25209969/). DOI: 10.18632/oncotarget.2373. Access: abstract_metadata. Study: PMID25209969.

40. Yuan S, Stewart KS, Yang Y et al. Ras drives malignancy through stem cell crosstalk with the microenvironment. Nature. 2022. [PMID 36450983](https://pubmed.ncbi.nlm.nih.gov/36450983/). DOI: 10.1038/s41586-022-05475-6. Access: abstract_metadata. Study: PMID36450983.

41. Potter N, Miraki-Moud F, Ermini L et al. Single cell analysis of clonal architecture in acute myeloid leukaemia. Leukemia. 2019. [PMID 30568172](https://pubmed.ncbi.nlm.nih.gov/30568172/). DOI: 10.1038/s41375-018-0319-2. Access: abstract_metadata. Study: PMID30568172.

42. Lechman ER, Gentner B, Ng SW et al. miR-126 Regulates Distinct Self-Renewal Outcomes in Normal and Malignant Hematopoietic Stem Cells. Cancer cell. 2016. [PMID 26832662](https://pubmed.ncbi.nlm.nih.gov/26832662/). DOI: 10.1016/j.ccell.2015.12.011. Access: abstract_metadata. Study: PMID26832662.

43. Elder A, Bomken S, Wilson I et al. Abundant and equipotent founder cells establish and maintain acute lymphoblastic leukaemia. Leukemia. 2017. [PMID 28487542](https://pubmed.ncbi.nlm.nih.gov/28487542/). DOI: 10.1038/leu.2017.140. Access: abstract_metadata. Study: PMID28487542.

44. Kanan AD, Corey E, Vêncio RZN et al. Lineage relationship between prostate adenocarcinoma and small cell carcinoma. BMC cancer. 2019. [PMID 31146720](https://pubmed.ncbi.nlm.nih.gov/31146720/). DOI: 10.1186/s12885-019-5680-7. Access: abstract_metadata. Study: PMID31146720.

45. Shi J, Wang X, Tang Y et al. LY6D identifies persistent stem-like cells driving pancreatic tumourigenesis. Gut. 2026. [PMID 41545197](https://pubmed.ncbi.nlm.nih.gov/41545197/). DOI: 10.1136/gutjnl-2025-336460. Access: abstract_metadata. Study: PMID41545197.

46. Tessier CE, Derrien J, Dupuy AMM et al. EMT-ciliary signaling in quasi-mesenchymal-stem-like cells drives therapeutic resistance and is a druggable vulnerability in triple-negative breast cancer. EMBO molecular medicine. 2025. [PMID 40859055](https://pubmed.ncbi.nlm.nih.gov/40859055/). DOI: 10.1038/s44321-025-00289-1. Access: abstract_metadata. Study: PMID40859055.

47. Goto S, Deshpande V, Yilmaz ÖH et al. Lineage plasticity driven by GATA6 loss fuels colorectal cancer metastasis. Cell stem cell. 2026. [PMID 42330962](https://pubmed.ncbi.nlm.nih.gov/42330962/). DOI: 10.1016/j.stem.2026.05.013. Access: abstract_metadata. Study: PMID42330962.

48. Morral C, Stanisavljevic J, Hernando-Momblona X et al. Zonation of Ribosomal DNA Transcription Defines a Stem Cell Hierarchy in Colorectal Cancer. Cell stem cell. 2020. [PMID 32396863](https://pubmed.ncbi.nlm.nih.gov/32396863/). DOI: 10.1016/j.stem.2020.04.012. Access: abstract_metadata. Study: PMID32396863.

49. Zhao W, Zhu Q, Tan P et al. Tgfbr2 inactivation facilitates cellular plasticity and development of Pten-null prostate cancer. Journal of molecular cell biology. 2018. [PMID 29228234](https://pubmed.ncbi.nlm.nih.gov/29228234/). DOI: 10.1093/jmcb/mjx052. Access: abstract_metadata. Study: PMID29228234.

50. Schatton T, Murphy GF, Frank NY et al. Identification of cells initiating human melanomas. Nature. 2008. [PMID 18202660](https://pubmed.ncbi.nlm.nih.gov/18202660/). DOI: 10.1038/nature06489. Access: abstract_metadata. Study: PMID18202660.

51. Marin-Bejar O, Rogiers A, Dewaele M et al. Evolutionary predictability of genetic versus nongenetic resistance to anticancer drugs in melanoma. Cancer cell. 2021. [PMID 34143978](https://pubmed.ncbi.nlm.nih.gov/34143978/). DOI: 10.1016/j.ccell.2021.05.015. Access: abstract_metadata. Study: PMID34143978.

52. Sellerio AL, Ciusani E, Ben-Moshe NB et al. Overshoot during phenotypic switching of cancer cell populations. Scientific reports. 2015. [PMID 26494317](https://pubmed.ncbi.nlm.nih.gov/26494317/). DOI: 10.1038/srep15464. Access: abstract_metadata. Study: PMID26494317.

53. Wilson BJ, Saab KR, Ma J et al. ABCB5 maintains melanoma-initiating cells through a proinflammatory cytokine signaling circuit. Cancer research. 2014. [PMID 24934811](https://pubmed.ncbi.nlm.nih.gov/24934811/). DOI: 10.1158/0008-5472.CAN-14-0582. Access: abstract_metadata. Study: PMID24934811.

54. Andreucci E, Peppicelli S, Ruzzolini J et al. The acidic tumor microenvironment drives a stem-like phenotype in melanoma cells. Journal of molecular medicine (Berlin, Germany). 2020. [PMID 32803272](https://pubmed.ncbi.nlm.nih.gov/32803272/). DOI: 10.1007/s00109-020-01959-y. Access: abstract_metadata. Study: PMID32803272.

55. Nguyen NTB, Gevers S, Kok RNU et al. Lactate controls cancer stemness and plasticity through epigenetic regulation. Cell metabolism. 2025. [PMID 39933514](https://pubmed.ncbi.nlm.nih.gov/39933514/). DOI: 10.1016/j.cmet.2025.01.002. Access: abstract_metadata. Study: PMID39933514.

56. Ohta Y, Fujii M, Takahashi S et al. Cell-matrix interface regulates dormancy in human colon cancer stem cells. Nature. 2022. [PMID 35798028](https://pubmed.ncbi.nlm.nih.gov/35798028/). DOI: 10.1038/s41586-022-05043-y. Access: abstract_metadata. Study: PMID35798028.

57. Wang C, Li Y, Jia L et al. CD276 expression enables squamous cell carcinoma stem cells to evade immune surveillance. Cell stem cell. 2021. [PMID 33945793](https://pubmed.ncbi.nlm.nih.gov/33945793/). DOI: 10.1016/j.stem.2021.04.011. Access: abstract_metadata. Study: PMID33945793.

58. Zhang L, He X, Liu X et al. Single-Cell Transcriptomics in Medulloblastoma Reveals Tumor-Initiating Progenitors and Oncogenic Cascades during Tumorigenesis and Relapse. Cancer cell. 2019. [PMID 31474569](https://pubmed.ncbi.nlm.nih.gov/31474569/). DOI: 10.1016/j.ccell.2019.07.009. Access: abstract_metadata. Study: PMID31474569.

59. Quintana E, Shackleton M, Foster HR et al. Phenotypic heterogeneity among tumorigenic melanoma cells from patients that is reversible and not hierarchically organized. Cancer cell. 2010. [PMID 21075313](https://pubmed.ncbi.nlm.nih.gov/21075313/). DOI: 10.1016/j.ccr.2010.10.012. Access: abstract_metadata. Study: PMID21075313.

60. Oka T, Higa T, Sugahara O et al. Ablation of p57+ Quiescent Cancer Stem Cells Suppresses Recurrence after Chemotherapy of Intestinal Tumors. Cancer research. 2023. [PMID 36880956](https://pubmed.ncbi.nlm.nih.gov/36880956/). DOI: 10.1158/0008-5472.CAN-22-2578. Access: abstract_metadata. Study: PMID36880956.

61. Sánchez-Danés A, Larsimont JC, Liagre M et al. A slow-cycling LGR5 tumour population mediates basal cell carcinoma relapse after therapy. Nature. 2018. [PMID 30297799](https://pubmed.ncbi.nlm.nih.gov/30297799/). DOI: 10.1038/s41586-018-0603-3. Access: full_text_partial. Study: PMID30297799.

62. Chan JE, Pan CH, Rub J et al. Critical role for a high-plasticity cell state in lung cancer. Nature. 2026. [PMID 41565826](https://pubmed.ncbi.nlm.nih.gov/41565826/). DOI: 10.1038/s41586-025-09985-x. Access: full_text_partial. Study: PMID41565826.

63. Vaquero-Siguero N, Georgakopoulos N, Puschhof MC et al. TROP2 targeting reveals therapy-driven cell state dynamics in colorectal cancer. Nature. 2026. [PMID 42386981](https://pubmed.ncbi.nlm.nih.gov/42386981/). DOI: 10.1038/s41586-026-10705-2. Access: full_text_partial. Study: PMID42386981.

64. Jiang Q, Raghavan MS, Rodriguez AM et al. ZFP36L2 orchestrates stress-adaptive plasticity in regeneration and cancer. Nature. 2026. [PMID 42557330](https://pubmed.ncbi.nlm.nih.gov/42557330/). DOI: 10.1038/s41586-026-10890-0. Access: full_text_partial. Study: PMID42557330.

65. Zomer A, Ellenbroek SI, Ritsma L et al. Intravital imaging of cancer stem cell plasticity in mammary tumors. Stem cells (Dayton, Ohio). 2013. [PMID 23225641](https://pubmed.ncbi.nlm.nih.gov/23225641/). DOI: 10.1002/stem.1296. Access: abstract_metadata. Study: PMID23225641.

66. Dirkse A, Golebiewska A, Buder T et al. Stem cell-associated heterogeneity in Glioblastoma results from intrinsic tumor plasticity shaped by the microenvironment. Nature communications. 2019. [PMID 30992437](https://pubmed.ncbi.nlm.nih.gov/30992437/). DOI: 10.1038/s41467-019-09853-z. Access: abstract_metadata. Study: PMID30992437.

67. Lenos KJ, Miedema DM, Lodestijn SC et al. Stem cell functionality is microenvironmentally defined during tumour expansion and therapy response in colon cancer. Nature cell biology. 2018. [PMID 30177776](https://pubmed.ncbi.nlm.nih.gov/30177776/). DOI: 10.1038/s41556-018-0179-z. Access: abstract_metadata. Study: PMID30177776.

68. Boumahdi S, Driessens G, Lapouge G et al. SOX2 controls tumour initiation and cancer stem-cell functions in squamous-cell carcinoma. Nature. 2014. [PMID 24909994](https://pubmed.ncbi.nlm.nih.gov/24909994/). DOI: 10.1038/nature13305. Access: abstract_metadata. Study: PMID24909994.

## Source-specific evidence cards

### [1] A cell initiating human acute myeloid leukaemia after transplantation into SCID mice.

**Shows:** Marker-fractionated human AML cells initiated disseminated leukemia in cytokine-treated SCID hosts, defining transplantation-enriched initiating activity.

**Does not show:** Engraftment selection does not establish an immutable endogenous hierarchy or necessity for maintenance and relapse in patients.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC5.

### [2] Human acute myeloid leukemia is organized as a hierarchy that originates from a primitive hematopoietic cell.

**Shows:** Primitive CD34-high CD38-negative fractions from tested AML subtypes self-renewed and generated differentiated leukemic blasts in NOD/SCID mice.

**Does not show:** Assay-defined exclusivity need not persist with different host permissiveness or show the original patient's tumour maintenance mechanism.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC5.

### [3] Prospective identification of tumorigenic breast cancer cells.

**Shows:** Prospectively isolated CD44-positive CD24-low breast tumour cells showed enriched serial tumour formation and regenerated heterogeneity in immunocompromised mice.

**Does not show:** Selected xenograft initiation in a small patient series is not universal fixed-hierarchy necessity; linked erratum details were not resolved.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. ErratumIn: PMIDnot supplied. Subclaims: SC1, SC5.

### [4] Identification of human brain tumour initiating cells.

**Shows:** CD133-positive human brain tumour fractions generated serially transplantable phenocopies at lower inocula than the tested negative fractions.

**Does not show:** CD133 enrichment in this host assay does not establish permanent marker exclusivity, endogenous maintenance necessity or relapse prevention.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC5.

### [5] Single-cell cloning of colon cancer stem cells reveals a multi-lineage differentiation capacity.

**Shows:** Single colon cancer cells self-renewed and generated differentiated tumour progeny after culture and xenotransplantation; PI3K signalling affected lineage decisions.

**Does not show:** Clonal multilineage potential is compatible with plastic hierarchies and does not prove an irreplaceable fixed population in patients.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC4.

### [6] Efficient tumour formation by single human melanoma cells.

**Shows:** More permissive host conditions detected tumour formation from many unselected human melanoma cells, including single-cell inocula.

**Does not show:** High transplantation competence challenges rarity and assay exclusivity but alone does not quantify endogenous fate or refute all functional hierarchies.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC4, SC5.

### [7] Characterization of melanoma cells capable of propagating tumors from a single cell.

**Shows:** Single-cell melanoma transplants revealed multiple propagating phenotypes with different capacities to reconstruct heterogeneity and resist chemotherapy.

**Does not show:** Transplanted mouse tumour populations do not establish patient recurrence mechanisms; self-renewal need not require multilineage differentiation.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC4, SC5.

### [8] The epithelial-mesenchymal transition generates cells with properties of stem cells.

**Shows:** Inducing EMT in mammary epithelial systems increased stem-associated properties and tumour-forming activity after transformation.

**Does not show:** Induced EMT and sphere readouts do not measure natural bidirectional transitions or establish universal therapy-resistant CSC identity.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC4.

### [9] Normal and neoplastic nonstem cells can spontaneously convert to a stem-like state.

**Shows:** Purified mammary non-stem populations spontaneously generated stem-like cells, with enhanced conversion after transformation in vitro and in vivo.

**Does not show:** Selected mammary systems do not establish transition rates across all tumours; spontaneous conversion undermines a permanently one-way hierarchy.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC4.

### [10] Stochastic state transitions give rise to phenotypic equilibrium in populations of cancer cells.

**Shows:** Sorted breast cancer cell-line states returned toward equilibrium proportions, consistent with stochastic transitions and de novo stem-like states.

**Does not show:** A Markov fit and cultured state recovery do not uniquely identify endogenous patient fate dynamics; two linked errata remain unresolved.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. ErratumIn: PMIDnot supplied; ErratumIn: PMIDnot supplied. Subclaims: SC4.

### [11] Stem cell gene expression programs influence clinical outcome in human leukemia.

**Shows:** Functionally xenograft-validated human AML stemness programs predicted patient survival, bridging transplantation activity and clinical association.

**Does not show:** Prognostic prediction is not selective LSC eradication or proof that a fixed population is necessary for relapse.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC3, SC5.

### [12] Identification of pre-leukaemic haematopoietic stem cells in acute leukaemia.

**Shows:** DNMT3A-mutant pre-leukemic HSCs lacked later NPM1 lesions, repopulated multilineage xenografts and persisted in remission samples.

**Does not show:** A pre-leukemic reservoir is not identical to the overt AML-maintaining population; presence after chemotherapy does not prove inevitable relapse.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. ErratumIn: PMIDnot supplied. Subclaims: SC3, SC4.

### [13] Single-Cell RNA-Seq Reveals AML Hierarchies Relevant to Disease Progression and Immunity.

**Shows:** Single-cell genotyping and transcriptional profiling mapped variable malignant differentiation spectra across AML patients and subclones.

**Does not show:** Expression-defined primitive states and prognosis do not by themselves establish lineage direction, necessity or a stable universal hierarchy.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC4, SC5.

### [14] Tracing the origins of relapse in acute myeloid leukaemia to stem cells.

**Shows:** Paired diagnosis/relapse genetics and xenografts traced resistant origins to either rare primitive LSCs or larger committed-phenotype subclones retaining stemness signatures.

**Does not show:** The two relapse patterns support functional stemness while contradicting a universal rare primitive phenotype; no randomized selective patient LSC ablation was tested.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC3, SC4, SC5.

### [15] Variable clonal repopulation dynamics influence chemotherapy response in colorectal cancer.

**Shows:** Lentiviral tracking within genetically stable colorectal clones revealed variable propagation and chemotherapy-selected minor or dormant lineages.

**Does not show:** Serial xenografts select engraftment-capable cells and do not distinguish every source of state variability in untreated patients.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC2, SC3, SC4.

### [16] Lineage tracing reveals Lgr5+ stem cell activity in mouse intestinal adenomas.

**Shows:** Endogenous multicolour lineage retracing showed Lgr5-positive adenoma cells self-renewing and producing the other adenoma cell types.

**Does not show:** A lineage contribution in mouse precursor adenomas is not necessity, patient carcinoma relapse, or proof that negative cells cannot regenerate stemness.

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

### [17] Defining the mode of tumour growth by clonal analysis.

**Shows:** Genetic clonal tracing in native skin tumours distinguished persistent stem-like and transient compartments in papillomas, with different dynamics in invasive carcinoma.

**Does not show:** Contribution and stage-dependent clonal behaviour do not prove universal rare-CSC necessity or an invariant hierarchy through progression.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC2, SC4.

### [18] A restricted cell population propagates glioblastoma growth after chemotherapy.

**Shows:** Pulse-chase and transgene-directed ablation identified an endogenous glioma subset sustaining regrowth after temozolomide; chronic ganciclovir arrested growth and combination treatment impeded development.

**Does not show:** The engineered glioma and nestin-TK targeting define a context-specific dependency, not universal patient CSC identity or selective clinical relapse prevention.

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

### [19] Visualization and targeting of LGR5+ human colon cancer stem cells.

**Shows:** Human CRC organoid lineage tracing and selective LGR5 ablation caused regression followed by regrowth involving KRT20-lineage conversion into LGR5-positive cells.

**Does not show:** Engineered organoid xenografts show a plastic functional hierarchy rather than necessity of a permanently distinct pre-existing CSC population in patients.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC2, SC3, SC4.

### [20] A distinct role for Lgr5+ stem cells in primary and metastatic colon cancer.

**Shows:** Selective Lgr5 ablation restricted primary CRC growth without regression while negative cells replenished the pool; liver metastases showed stronger Lgr5 dependence.

**Does not show:** Dependency differs by anatomical stage; primary maintenance despite ablation contradicts universal fixed-population necessity while preserving a metastatic dependency.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC2, SC3, SC4, SC5.

### [21] A genome editing approach to study cancer stem cells in human tumors.

**Shows:** Edited patient-derived CRC organoid xenografts showed long-term LGR5-lineage self-renewal and differentiated descendants.

**Does not show:** Human-origin material in an engineered mouse host is not intact-patient fate tracing, and lineage contribution alone does not demonstrate necessity.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC4.

### [22] Phenotypic differentiation does not affect tumorigenicity of primary human colon cancer initiating cells.

**Shows:** CRC cells downregulated initiating-cell markers under differentiating adherent conditions yet retained serial xenograft tumour initiation and clonal contribution.

**Does not show:** Culture-induced marker differentiation does not identify every endogenous transition, but marker-defined phenotype was not necessary for transplantation activity.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC4.

### [23] Plasticity of Lgr5-Negative Cancer Cells Drives Metastasis in Colorectal Cancer.

**Shows:** Mouse CRC and human xenograft tracking showed many disseminating Lgr5-negative cells restoring positive stem-like cells; plasticity supported metastatic outgrowth.

**Does not show:** Seeding, establishment and outgrowth are distinct phases; a plastic stem state requirement is not a fixed originating cell population requirement.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC3, SC4.

### [24] Metastatic recurrence in colorectal cancer arises from residual EMP1+ cells.

**Shows:** In orthotopic CRC relapse models, pre-resection EMP1-high ablation prevented recurrence in most mice; post-seeding ablation failed, and pre-resection Lgr5 ablation did not prevent recurrence.

**Does not show:** Timing and metastatic stage determine dependency; patient signatures were associative, and EMP1 residual states are not an immutable LGR5 apex.

**Access/status:** Complete primary abstract and linked-status metadata examined. Selected body sections: 5:Relapse by residual EMP1+ cells. No complete-paper or supplement audit claimed. none_found. Subclaims: SC3, SC4.

### [25] Mex3a marks drug-tolerant persister colorectal cancer cells that mediate relapse after chemotherapy.

**Shows:** Mex3a-lineage persisters expanded after chemotherapy and transiently shifted away from the WNT/stem program; deficient cells differentiated and failed to resist treatment.

**Does not show:** A therapy-dependent, niche-adaptive persister state is not an invariant CSC phenotype; clinical selective targeting and durable patient relapse prevention were not tested.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC3, SC4.

### [26] Dynamic and adaptive cancer stem cell population admixture in colorectal neoplasia.

**Shows:** Mouse and human neoplasia showed variable conventional/regenerative stem-state admixture that adapted under acute selective pressure and tracked chemoresistance.

**Does not show:** The fitness-landscape interpretation is not a unique proof of transition causation; a linked corrigendum remains part of the access limit.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. ErratumIn: PMID36332574. Subclaims: SC1, SC3, SC4.

### [27] Marker-free lineage tracing reveals an environment-instructed clonogenic hierarchy in pancreatic cancer.

**Shows:** Marker-independent in-situ pancreatic clonal tracing found widespread clonogenic potential controlled by fibroblast proximity; stromal Hedgehog inhibition altered growth mode.

**Does not show:** Spatially instructed clonogenicity challenges a fixed rare apex, but the perturbation can change multiple stromal functions and does not cover every PDAC subtype.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC2, SC4, SC5.

### [28] Succession of transiently active tumor-initiating cell clones in human pancreatic cancer xenografts.

**Shows:** Serial pancreatic xenografts were maintained by successive transiently active initiating clones with little overlap between generations.

**Does not show:** Transplant bottlenecks may contribute to clonal succession; these data challenge fixed continual activity without establishing native patient dynamics.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC2, SC3, SC4.

### [29] Pre-existing Functional Heterogeneity of Tumorigenic Compartment as the Origin of Chemoresistance in Pancreatic Tumors.

**Shows:** Clonal replica pancreatic xenografts exposed pre-existing genetic and functional diversity with different gemcitabine responses.

**Does not show:** Drug sensitivity and predictive signatures do not show that one stable CSC compartment is necessary for patient relapse.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC3, SC4.

### [30] A single-cell survey of cellular hierarchy in acute myeloid leukemia.

**Shows:** Single-cell sequencing of AML marrow resolved progenitor states and associations between ribosomal programs, subtype and remission.

**Does not show:** A transcriptomic landscape and proposed attractor do not demonstrate fate direction, selective dependency or population immutability.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC5.

### [31] Clonally resolved single-cell multi-omics identifies routes of cellular differentiation in acute myeloid leukemia.

**Shows:** CloneTracer connected genotypes with differentiation routes in AML, separating healthy/pre-leukemic dormant cells from active LSC-like and downstream malignant compartments.

**Does not show:** Inferred routes and chemotherapy associations do not constitute selective in-patient stem-cell ablation or prove a universal fixed phenotype.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC4, SC5.

### [32] Pre-existing stem cell heterogeneity dictates clonal responses to the acquisition of leukemic driver mutations.

**Shows:** Pre/post-mutation STRACK tracing showed driver-specific expansion of differentiation-primed stem cells and Npm1c-associated reversal toward primitive malignant states.

**Does not show:** Early engineered murine evolution is not relapse necessity across cancers, but demonstrates that cell history and mutation alter the apparent hierarchy.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC4.

### [33] Lineage tracing and single-cell analysis reveal proliferative Prom1+ tumour-propagating cells and their dynamic cellular transition during liver cancer progression.

**Shows:** Prom1 lineage tracing and depletion in two intact HCC models linked clonal expansion and dynamic dedifferentiation to tumour growth; depletion impeded growth.

**Does not show:** A lineage-state dependency in these models is not a permanently uniform marker compartment or patient-wide relapse necessity.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC2, SC4.

### [34] Stem-cell states converge in multistage cutaneous squamous cell carcinoma development.

**Shows:** Integrated SCC lineage/state analyses showed chemotherapy suppressing proliferation and activating plasticity; manipulating differentiation enhanced cell-cycle-inhibitor responses.

**Does not show:** Treatment-induced state movement challenges stability but does not establish that all tumour cells are equivalently plastic or that the same switch governs every cancer.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC2, SC3, SC4.

### [35] Tracing tumorigenesis in a solid tumor model at single-cell resolution.

**Shows:** Single-cell and lineage analyses in a defined salivary-gland SCC mouse model mapped a reproducible basal-to-stem-like-to-luminal trajectory.

**Does not show:** A developmental trajectory inferred with profiling is not ablation-defined necessity; a genetically defined model has restricted generalizability.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC4.

### [36] FXYD3 functionally demarcates an ancestral breast cancer stem cell subpopulation with features of drug-tolerant persisters.

**Shows:** TNBC profiling identified FXYD3-positive ancestral/persister-like cells that persisted during neoadjuvant therapy and were sensitive to Na/K-pump inhibitors.

**Does not show:** Clinical persistence is associative and pump inhibitors affect more than CSC-specific functions; selective CSC elimination improving patient relapse-free survival was not established.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC3, SC4.

### [37] Monoclonal antibody-mediated targeting of CD123, IL-3 receptor alpha chain, eliminates human acute myeloid leukemic stem cells.

**Shows:** Anti-CD123 treatment reduced human AML xenograft engraftment and established disease burden, impaired secondary transplantation and improved mouse survival.

**Does not show:** Homing, cytokine blockade and innate effector activation are intertwined; a mouse efficacy experiment does not isolate a fixed population's necessity for human cure.

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

### [38] Interconversion between Tumorigenic and Differentiated States in Acute Myeloid Leukemia.

**Shows:** Perturbing endogenous PU.1 or withdrawing differentiation treatment allowed mature AML cells in murine and human models to regain clonogenic/leukemogenic activity.

**Does not show:** This does not show unrestricted reversal in all AML genotypes, but directly challenges irreversible differentiation even in a prototypical hierarchical malignancy.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC4, SC5.

### [39] Using heterogeneity of the patient-derived xenograft model to identify the chemoresistant population in ovarian cancer.

**Shows:** Ovarian PDX tumours retained heterogeneity and clinical response patterns; carboplatin/paclitaxel enriched stem-associated populations without making persisters exclusively those populations.

**Does not show:** Marker enrichment after treatment does not identify uniquely causal relapse cells or distinguish selection from induced plasticity.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC3, SC4.

### [40] Ras drives malignancy through stem cell crosstalk with the microenvironment.

**Shows:** Clonal HRAS-mutant skin models and functional pathway tests linked dynamic CSC-stromal signalling, angiogenesis and metabolic signalling to malignant transition.

**Does not show:** Interdependent niche signals support a contextual mechanism rather than a cell-intrinsic immutable hierarchy or pan-cancer necessity.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC2, SC4.

### [41] Single cell analysis of clonal architecture in acute myeloid leukaemia.

**Shows:** Single-cell mutation analysis and xenografts showed branching AML subclones, pre-leukemic variants and preferential but variable regenerating clones.

**Does not show:** Genetic ancestry and engraftment do not equate to fixed phenotypic stem hierarchy or independently establish post-treatment patient relapse necessity.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC4.

### [42] miR-126 Regulates Distinct Self-Renewal Outcomes in Normal and Malignant Hematopoietic Stem Cells.

**Shows:** Functional miR-126 manipulation regulated primary AML self-renewal, differentiation and quiescence through PI3K/AKT/MTOR-associated signalling in vivo.

**Does not show:** The pathway also affects normal HSCs differently; functional perturbation in transplanted systems is not exclusive clinical LSC necessity, and an erratum is linked.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. ErratumIn: PMID27070706. Subclaims: SC2, SC3.

### [43] Abundant and equipotent founder cells establish and maintain acute lymphoblastic leukaemia.

**Shows:** Barcoded primary ALL blasts showed many approximately equipotent founders contributing through serial transplants rather than a rare dominant apex.

**Does not show:** Transplant activity is host-conditioned; this challenges rarity/exclusivity in ALL without showing that every native blast has identical fate in patients.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC4, SC5.

### [44] Lineage relationship between prostate adenocarcinoma and small cell carcinoma.

**Shows:** Prostate small-cell PDXs expressed reprogramming factors, and factor transfection altered expression and stem-like colony phenotypes.

**Does not show:** Forced transcription-factor expression and colonies establish plausibility, not natural lineage conversion or fixed CSC dependence.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC4.

### [45] LY6D identifies persistent stem-like cells driving pancreatic tumourigenesis.

**Shows:** Ly6d genetic perturbation in KRAS-driven pancreatic models affected a persistent gastric-like lineage and delayed tumourigenesis; human expression correlated with prognosis.

**Does not show:** A genetic lineage program and OXPHOS dependency do not prove that a discrete cell population is universally necessary or that patient relapse is prevented by targeting it.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC2, SC4.

### [46] EMT-ciliary signaling in quasi-mesenchymal-stem-like cells drives therapeutic resistance and is a druggable vulnerability in triple-negative breast cancer.

**Shows:** TNBC organoids retained EMT-associated ciliated stem-like survivors after chemotherapy; genetic cilia ablation and inhibitors suppressed resistance.

**Does not show:** Organoid chemoresistance and NFkB-associated killing are not selective in-patient CSC eradication or a demonstration of permanently fixed hierarchy.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC3, SC4.

### [47] Lineage plasticity driven by GATA6 loss fuels colorectal cancer metastasis.

**Shows:** GATA6 loss in orthotopic CRC organoid models increased metastatic plasticity and LGR5-negative states with little primary-growth change.

**Does not show:** Genetic reprogramming shows context-specific metastatic regulation; serial transplant selection and native patient translation remain limits.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC3, SC4.

### [48] Zonation of Ribosomal DNA Transcription Defines a Stem Cell Hierarchy in Colorectal Cancer.

**Shows:** POLR1A-high biosynthetic CRC compartments spanned LGR5-positive and negative cells; genetic population ablation imposed irreversible growth arrest in the reported systems.

**Does not show:** Abstract-level access does not establish every model's endogenous status or exclude biosynthesis-dependent general toxicity; this functional compartment is broader than LGR5 identity.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC2, SC4.

### [49] Tgfbr2 inactivation facilitates cellular plasticity and development of Pten-null prostate cancer.

**Shows:** Combined Pten/Tgfbr2 perturbation increased reprogramming, rapid prostate tumour development and early mortality with BMP/stemness-associated signalling.

**Does not show:** Multi-pathway genetic changes are not selective ablation of a fixed CSC population; survival associations do not establish human mechanistic mediation.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC2, SC4.

### [50] Identification of cells initiating human melanomas.

**Shows:** ABCB5-enriched melanoma cells showed serial xenograft self-renewal and antibody targeting inhibited established tumour growth.

**Does not show:** Host and assay conditions constrain rarity; antibody-mediated cytotoxicity and ABCB5 functions do not establish universal exclusive endogenous CSC maintenance.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC2, SC5.

### [51] Evolutionary predictability of genetic versus nongenetic resistance to anticancer drugs in melanoma.

**Shows:** FAK inhibition depleted a transient neural-crest-like residual melanoma state and delayed xenograft relapse, with escape favouring genetic resistance.

**Does not show:** A drug acting on a signalling network is not perfectly state-specific; delayed rather than abolished relapse illustrates alternate resistance routes.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC3, SC4.

### [52] Overshoot during phenotypic switching of cancer cell populations.

**Shows:** Marker-sorted melanoma populations transiently overshot stem-marker equilibrium, consistent with regulated phenotypic switching after depletion.

**Does not show:** Marker dynamics and a regulatory model do not alone prove tumour-propagating function or endogenous relapse prevention failure.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC4.

### [53] ABCB5 maintains melanoma-initiating cells through a proinflammatory cytokine signaling circuit.

**Shows:** ABCB5 blockade altered an IL1/IL8/CXCR1 paracrine circuit, induced differentiation, reversed drug resistance and impaired melanoma growth in vivo.

**Does not show:** Cytokine interactions with negative cells favour a network dependency; inhibition does not establish permanently discrete CSC identity or clinical cure.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC2, SC3, SC4.

### [54] The acidic tumor microenvironment drives a stem-like phenotype in melanoma cells.

**Shows:** Chronic extracellular acidosis increased melanoma stem-associated and tumour-initiating properties, with a reversible phenotype linked to EMT.

**Does not show:** Artificial environmental adaptation and initiation assays establish plastic potential rather than the rate or necessity of native patient state switching.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC4.

### [55] Lactate controls cancer stemness and plasticity through epigenetic regulation.

**Shows:** Reporter-tracked human intestinal tumour organoids showed lactate suppressing differentiation and inducing a proliferative stem state through acetylation/MYC-associated regulation.

**Does not show:** Organoid transitions and candidate BRD4 targeting do not demonstrate prevention of patient relapse or a unique physiological lactate mechanism.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC4.

### [56] Cell-matrix interface regulates dormancy in human colon cancer stem cells.

**Shows:** Intravital human CRC organoid xenograft tracing linked pre-existing LGR5/p27 dormancy to survival and post-chemotherapy expansion; COL17A1 loss and YAP interference altered sensitivity/regrowth.

**Does not show:** Human-origin transplanted systems and niche signalling do not prove a fixed population's necessity in patients; delaying regrowth is not universal eradication.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC3, SC4.

### [57] CD276 expression enables squamous cell carcinoma stem cells to evade immune surveillance.

**Shows:** Anti-CD276 eliminated stem-enriched SCC cells through CD8-dependent immunity and reduced tumour growth and nodal metastases in mouse HNSCC.

**Does not show:** Immune-mediated targeting remodels several tumour states; human CD276 expression is not patient evidence that a stable CSC compartment is uniquely necessary.

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

### [58] Single-Cell Transcriptomics in Medulloblastoma Reveals Tumor-Initiating Progenitors and Oncogenic Cascades during Tumorigenesis and Relapse.

**Shows:** SHH medulloblastoma studies identified OLIG2 progenitors that changed to quiescent stem-like states; depletion/deletion impeded initiation and pathway combinations arrested growth.

**Does not show:** Therapy-resistant enrichment is not direct selective relapse prevention in children; stem-like function is developmentally and disease-context dependent.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC2, SC3, SC4.

### [59] Phenotypic heterogeneity among tumorigenic melanoma cells from patients that is reversible and not hierarchically organized.

**Shows:** Direct patient melanoma cells frequently initiated serially transplantable tumours in permissive hosts; diverse markers failed to separate a large non-tumorigenic compartment and were reversible.

**Does not show:** A transplantation counterexample challenges rarity and marker exclusivity, but does not establish identical native fate for every melanoma cell.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC4, SC5.

### [60] Ablation of p57+ Quiescent Cancer Stem Cells Suppresses Recurrence after Chemotherapy of Intestinal Tumors.

**Shows:** A syngeneic orthotopic intestinal organoid model identified quiescent p57-positive lineages surviving chemotherapy; their ablation suppressed post-treatment regrowth.

**Does not show:** This is a transplanted model and demonstrates suppression, not universal complete relapse prevention or an invariant CSC pool across tumours.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC3, SC4.

### [61] A slow-cycling LGR5 tumour population mediates basal cell carcinoma relapse after therapy.

**Shows:** In engineered BCC models, Lgr5-positive persisters drove regrowth after vismodegib withdrawal; combined treatment removed most lesions and dual Wnt/Hedgehog inhibition prevented recurrence over the tested four-week follow-up.

**Does not show:** Human biopsies supplied phenotype/withdrawal linkage; mouse Lgr5-ablation duration was limited by normal liver toxicity, and long-term eradication was not established.

**Access/status:** Complete primary abstract and linked-status metadata examined. Selected body sections: 7:Main text (unsectioned body paragraphs):paragraph5, 7:Main text (unsectioned body paragraphs):paragraph17, 7:Main text (unsectioned body paragraphs):paragraph18, 7:Main text (unsectioned body paragraphs):paragraph19, 7:Main text (unsectioned body paragraphs):paragraph21. No complete-paper or supplement audit claimed. none_found. Subclaims: SC2, SC3.

### [62] Critical role for a high-plasticity cell state in lung cancer.

**Shows:** Autochthonous lung tumour reporters, tracing and two ablation approaches showed a high-plasticity hub driving progression, burden and resistant states.

**Does not show:** Non-HPCS cells can acquire HPCS; suppression does not establish an immutable population, pan-cancer clinical necessity or durable human cure. The linked earlier version is not a second independent report.

**Access/status:** Complete primary abstract and linked-status metadata examined. Selected body sections: 4:HPCS drives tumour growth and progression:paragraph0, 4:HPCS drives tumour growth and progression:paragraph3, 4:HPCS drives tumour growth and progression:paragraph5, 4:HPCS drives tumour growth and progression:paragraph7, 7:Discussion:paragraph0, 7:Discussion:paragraph1, 7:Discussion:paragraph2, 7:Discussion:paragraph3, 3:Non-HPCS cells can acquire the HPCS. No complete-paper or supplement audit claimed. UpdateOf: PMID41278778. Subclaims: SC1, SC2, SC3, SC4.

### [63] TROP2 targeting reveals therapy-driven cell state dynamics in colorectal cancer.

**Shows:** Patient-derived CRC models showed context-dependent TROP2-positive stemness, chemotherapy-induced state redistribution and enhanced efficacy of combination TROP2-directed treatment.

**Does not show:** State-specific vulnerability and model efficacy do not demonstrate permanent cell identity or relapse-free survival benefit in patients.

**Access/status:** Complete primary abstract and linked-status metadata examined. Selected body sections: 4:TROP2 targeting and cell state remodelling:paragraph0, 5:Chemotherapy-induced TROP2 and plasticity:paragraph0, 5:Chemotherapy-induced TROP2 and plasticity:paragraph3, 5:Chemotherapy-induced TROP2 and plasticity:paragraph4. No complete-paper or supplement audit claimed. none_found. Subclaims: SC2, SC3, SC4.

### [64] ZFP36L2 orchestrates stress-adaptive plasticity in regeneration and cancer.

**Shows:** ZFP36L2 loss impaired canonical LGR5 re-entry/metastatic growth while promoting alternative heterogeneous fates; mouse regeneration tests also implicated stress-responsive dedifferentiation.

**Does not show:** Canonical stem-state dependency can be bypassed by alternative plasticity; poor-outcome associations are not a pan-cancer necessity/sufficiency test.

**Access/status:** Complete primary abstract and linked-status metadata examined. Selected body sections: 4:ZFP36L2 loss drives non-canonical fates, 7:Discussion. No complete-paper or supplement audit claimed. none_found. Subclaims: SC3, SC4.

### [65] Intravital imaging of cancer stem cell plasticity in mammary tumors.

**Shows:** Intravital lineage tracing in unperturbed mammary tumours documented disappearance of existing stem-like contributors and formation of new ones.

**Does not show:** A dynamic lineage contribution supports plastic stemness, without measuring universal patient transition rates or showing every cancer lacks a hierarchy.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC4.

### [66] Stem cell-associated heterogeneity in Glioblastoma results from intrinsic tumor plasticity shaped by the microenvironment.

**Shows:** GBM stem-marker populations exhibited reversible microenvironment-instructed transitions; faster reconstitution of heterogeneity conferred an in-vivo growth advantage.

**Does not show:** Model-based and transplanted growth tests challenge discrete marker identity but do not exclude functionally important stem-like states in native gliomas.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC4.

### [67] Stem cell functionality is microenvironmentally defined during tumour expansion and therapy response in colon cancer.

**Shows:** Time-resolved colon xenograft clones linked functional stemness to edge location and fibroblast-derived osteopontin, including after chemotherapy.

**Does not show:** Quantitative model fit is not unique proof; transplanted tissue and stromal interactions limit inference about a stable intrinsic patient hierarchy.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC2, SC3, SC4.

### [68] SOX2 controls tumour initiation and cancer stem-cell functions in squamous-cell carcinoma.

**Shows:** Sox2 deletion reduced chemical skin tumour formation; lineage ablation and conditional deletion in established papillomas/SCCs caused regression.

**Does not show:** SOX2 regulates multiple survival and lineage functions; a strong skin-model dependency does not establish pan-cancer fixed-CSC necessity.

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