Open questions — melanoma¶
Last curated: 2026-09-01. Build set. This file replaces the seed set written during scoping. Every question below was re-grounded against the literature retrieved live in the build session, and every asserted absence was re-run and dated on 2026-09-01. Stable IDs (OQ-n) are preserved from the seed where the question survived; seed questions that the literature has answered are closed explicitly at the end rather than deleted.
How to read priority¶
Tier 1 = answering it would change practice or reorder counselling, and a study is designable today. Tier 2 = important, but blocked on tools, numbers, or a Tier-1 answer. Tier 3 = mechanistic or infrastructural; would reshape how the field asks questions. Tiers are curator judgment.
Dots not yet connected¶
Cross-domain junctions where two bodies of evidence in this knowledge base both exist and no study has joined them.
| # | Dot A | Dot B | The missing junction | Powers |
|---|---|---|---|---|
| D1 | Incidence is projected to rise ~50% to 510,000 cases by 2040 on 2020 rates, and ~88% of cases are attributed to UV exposure (PMID 35353115; PMID 42502459) | Overdiagnosis is estimated at 59–60% of US White melanoma diagnoses in 2014, and 29–60% across studies (PMID 35293957; PMID 37793786) | Burden projections and UVR-attributable fractions use the same incidence series the overdiagnosis literature argues is inflated by detection. Neither has reconciled its denominator with the other's | OQ-1 |
| D2 | MSLT-II and DeCOG-SLT found no melanoma-specific survival benefit from completion dissection (PMID 28591523; PMID 31557067) | Sentinel node status remains the AJCC 8th-edition stage III determinant and the eligibility gate for every modern adjuvant and neoadjuvant trial (PMID 29028110; PMID 33857412; PMID 38828984) | A staging variable whose therapeutic action was withdrawn still directs systemic-treatment eligibility; the inheritance has never been re-derived. An EORTC-DeCOG nomogram and a validated 31-gene profile both exist as alternatives (PMID 32454396; PMID 29402264) | OQ-2 |
| D3 | NADINA gave 12-month event-free survival 83.7% vs 57.2%, HR 0.32 (PMID 38828984); S1801 gave 2-year EFS 72% vs 49% (PMID 36856617) | Adjuvant trials established the current standard on a different population, with recurrence endpoints whose trial-level surrogacy for overall survival is R² = 0.59 (0.08–1.00) (PMID 32777716) | No trial compares modern neoadjuvant against modern adjuvant with overall survival as the primary endpoint, and none appears in the active phase 3 landscape re-checked 2026-09-01 | OQ-3 |
| D4 | Checkpoint blockade produced 10-year melanoma-specific survival of 96% among cutaneous patients progression-free at 3 years (PMID 39282897) | Uveal melanoma's 5-year relative survival was 82.8% and unchanged from 1975 to 2016, and its first-line agent works by an entirely different mechanism (PMID 40225965; PMID 34551229) | The mechanistic explanation — mutational burden, driver divergence, ocular immune privilege — is described but never tested as a predictive framework that would say in advance which other tumours behave which way | OQ-4 |
| D5 | Heritable retinoblastoma survivors have 50-year cumulative melanoma incidence 4.5% versus 0.7% in non-heritable survivors, with onset ~20 years earlier (PMID 34153328) | Melanoma germline testing pathways are organised around CDKN2A, MC1R, BAP1, POT1, MITF and the shelterin genes; RB1 is not among the eight genes in the genome-first prevalence analysis (PMID 42201696) | Whether RB1-carrier survivors are captured by, or fall outside, melanoma surveillance pathways is unmapped in both directions | OQ-5 |
| D6 | Acral and mucosal melanoma have distinct genomics — structural-variant-driven, KIT-enriched, pigmentation-unrelated (PMID 31320640; PMID 29746316; PMID 32341527) | Registrational trials were conducted in overwhelmingly cutaneous, overwhelmingly White populations; first-line immunotherapy response was 15.0% in acral versus 39.1% in cutaneous melanoma in one registry (PMID 40841506; PMID 35293617) | Standard-of-care effect sizes are applied to subtypes barely represented in the trials that established them, and no pooled registry analysis with subtype recorded has been designed to test transportability | OQ-6 |
| D7 | BRAF V600E occurs in unequivocally benign lesions and TERT promoter mutation in 77% of intermediate lesions and melanoma in situ (PMID 26559571) | Diagnostic accuracy against expert consensus is 25–43% in the moderate-atypia to pT1a range, and second opinions reduce misclassification only from 52.8% to 36.7% (PMID 28659278; PMID 31603483) | The molecular and morphological literatures independently show the benign–malignant boundary is not sharp, and no assay has been validated to place a lesion on the correct side of it | OQ-7 |
| D8 | KEYNOTE-716's PRFS2 hazard ratio is 0.75 (95% CI 0.56–1.01) — progression-free survival on next-line therapy, which crosses 1.0 (PMID 40198940) | Chronic immune-related events persist beyond 12 weeks after adjuvant anti-PD-1 in 43.2% of patients, 96.4% grade 1–2 and mostly unresolved (PMID 33764387) | Nobody has expressed the stage IIB/IIC adjuvant decision as one quantity per 100 patients treated: recurrences prevented against permanent toxicities caused, with PRFS2 rather than RFS as the benefit term | OQ-8 |
| D9 | Nodal ultrasound has pooled sensitivity 0.879 and specificity 0.969, and MSLT-II's observation arm kept 80.2% of basins recurrence-free at 10 years (PMID 38635019; PMID 35921122) | Adherence to planned ultrasound surveillance is 77.0% overall and significantly lower in the US (PMID 38635019) | An accurate test is deployed in a strategy whose realised performance is capped by adherence, and no trial has measured the strategy rather than the test | OQ-9 |
| D10 | High IFN-γ signature plus high tumour mutational burden gives a 100% neoadjuvant pathological response rate; low/low gives 39% (PMID 33558721) | No predictive biomarker selects checkpoint therapy anywhere in melanoma practice, and high TMB fails to predict response across cancers generally (PMID 33736924) | A two-marker combination separating pathological response rates by 61 percentage points has never been tested prospectively as a selection tool | OQ-10 |
| D11 | Rapid tumour growth (≥0.5 mm/month) occurs more often in men, people ≥70, and those with fewer nevi and freckles (PMID 17178980) | Risk-prediction tools and surveillance targeting are built on nevus count, freckling and phenotype, with ≥25 common nevi carrying a population attributable fraction of 0.42 (PMID 20086181; PMID 25713022) | The people whose melanomas grow fastest are systematically the people risk models flag least, and no surveillance strategy has been designed around growth rate | OQ-11 |
| D12 | Tebentafusp prolongs survival in metastatic uveal melanoma even in patients whose best RECIST response was progression with >20% target growth (PMID 42162665) | Response criteria gate trial design, drug approval and clinical decision-making throughout melanoma | An agent whose benefit is invisible to the standard response endpoint has been licensed, and no methodological work has followed on what endpoint should replace RECIST for this drug class | OQ-12 |
Tier 1 — highest value, designable today¶
OQ-1. How much of the projected rise in melanoma incidence is disease, and how much is detection?¶
Arnold projects 510,000 cases and 96,000 deaths by 2040 on unchanged 2020 rates (PMID 35353115) and UVR-attributable modelling assigns ~88% of the 2022 case count to ultraviolet exposure (PMID 42502459). Against this, US county-level melanoma incidence correlates with median household income at r = 0.43 and with UV daily dose at r = 0.03, and with melanoma mortality at r = 0.09 against r = 0.96 for lung cancer in the same counties (PMID 36190719); overdiagnosis is estimated at 59–60% for 2014 US White diagnoses (PMID 35293957) and 29–60% across a 35-study scoping review (PMID 37793786). The counter-case is that increases occur in thick, ulcerated and advanced disease including among low-SES men (PMID 28736233) and that registry-level granularity implicates factors beyond overdiagnosis (PMID 38957842). A designable study: decompose the projected 2040 burden by thickness stratum and reconcile it against the observed mortality series, rather than projecting a pooled incidence rate forward. → wiki/screening-and-overdiagnosis.md, wiki/epidemiology-and-global-burden.md
OQ-2. If completion dissection does not improve survival, what is sentinel node status now for?¶
MSLT-II found 3-year melanoma-specific survival of 86% in both arms (P = .42) with 24.1% lymphoedema in the dissection arm (PMID 28591523), and DeCOG-SLT agreed (PMID 31557067). Sentinel node status nonetheless determines AJCC 8th-edition stage III subgrouping (PMID 29028110) and gates eligibility for KEYNOTE-054, COMBI-AD, CheckMate 238, KEYNOTE-716, CheckMate 76K, S1801 and NADINA. MSLT-II's observation-arm analysis suggests the biopsy may itself be therapeutic in some patients, but that is a single-arm analysis (PMID 35921122). A designable study: randomise adjuvant-therapy eligibility defined by the validated EORTC-DeCOG nomogram or the 31-gene profile against eligibility defined by nodal status (PMID 32454396; PMID 29402264). → wiki/sentinel-node-and-nodal-management.md, wiki/staging.md, wiki/adjuvant-therapy.md
OQ-3. Does neoadjuvant sequencing improve overall survival, or the endpoints trials have chosen to measure?¶
NADINA is phase 3 with HR 0.32 (99.9% CI 0.15–0.66) for event-free survival at a median 9.9 months (PMID 38828984); S1801 is phase 2 with 2-year EFS 72% vs 49% (PMID 36856617) and a published critique identifying an arbitrary event-assignment rule, differential early censoring raising the possibility of informative censoring, and the general risk of chance findings in phase 2 (PMID 38621314). NADINA's phase 3 design answers the third objection and not the first two, since its primary endpoint is also event-free survival. The 2024 European guideline already recommends both regimens (PMID 39709737). The disagreement is unresolved and is presented as such. → wiki/neoadjuvant-therapy.md, wiki/adjuvant-therapy.md
OQ-8. What is the absolute benefit of adjuvant therapy in stage IIB/IIC against its permanent toxicity, per 100 patients treated?¶
KEYNOTE-716 at a median 52.8 months gives 48-month recurrence-free survival 71.3% vs 58.3% (HR 0.62, 0.50–0.78) and DMFS 81.0% vs 70.1% — but PRFS2, which incorporates what happens after recurrence, is HR 0.75 (0.56–1.01) with 48-month rates 82.5% vs 76.7% (PMID 40198940). CheckMate 76K gives 12-month RFS 89.0% vs 79.4% (PMID 37845511) and an indirect comparison finds no difference between the two agents (PMID 42101794). On the harm side, 43.2% of adjuvant anti-PD-1 recipients had an event persisting beyond 12 weeks after cessation, 96.4% grade 1–2 and most unresolved at last follow-up (PMID 33764387), and endocrine events usually require lifelong replacement with no clear benefit from high-dose glucocorticoids (PMID 36149449). A published critique argues the whole strategy is "non-personalized medicine with no overall survival benefit" (PMID 39025250). A designable study: a decision analysis expressing benefit as PRFS2 rather than RFS, against permanent-toxicity incidence, per 100 patients. → wiki/adjuvant-therapy.md, wiki/immune-related-adverse-events.md, wiki/staging.md
OQ-13. Is adjuvant therapy justified in stage IIIA, whose survival exceeds stage IIB and IIC?¶
Five-year melanoma-specific survival was 89% for AJCC-8 stage IIIA against 80% for IIB and 67% for IIC in one validation cohort, whose authors call the stage IIIA indication questionable (PMID 31401470); the Japanese cohort gives 87.5% for IIIA against 84.4% and 72.2% (PMID 31023613). Retrospective multicentre data in 628 stage IIIA patients gave 2-year RFS 79.3% with anti-PD-1, 98.6% with dabrafenib–trametinib and 84.3% with observation — with poor-prognostic variables significantly more common in the anti-PD-1 arm, so the comparison is confounded by indication (PMID 40204154). A designable study: a randomised de-escalation trial of observation versus adjuvant therapy in stage IIIA. → wiki/adjuvant-therapy.md, wiki/staging.md
OQ-14. Does an AI-assisted diagnostic workflow improve any patient outcome?¶
Deep learning matches or exceeds dermatologists on curated image sets (pooled sensitivity 82%, specificity 87%, AUC 0.92; DL-assisted dermatologists AUC 0.87 vs 0.76 unassisted) (PMID 39088883; PMID 29846502; PMID 31401469), but in the one prospective in-clinic accuracy study the local dermatologist had the highest sensitivity (96.6%) and by far the lowest specificity (32.2%) — a trade-off no reader study reproduces (PMID 32289389). The one randomised imaging intervention in melanoma increased excisions (5.73 vs 3.99 per person, P = .02) while detecting fewer melanomas, at US$945 incremental cost per person with no QALY difference (PMID 40136310; PMID 40136266). No AI tool has been shown to improve a patient outcome in a prospective workflow. → wiki/clinical-diagnosis-and-dermoscopy.md
OQ-15. Can ctDNA-guided management improve melanoma outcomes?¶
ctDNA is prognostic: droplet digital PCR assays for BRAF V600 ctDNA were clinically validated in the COMBI-AD intention-to-treat population (PMID 40250457), ctDNA was detected in 43.8% of 48 stage III patients and associated with distant and multi-site recurrence (PMID 40387867), and undetectable ctDNA associated with longer survival on tebentafusp (PMID 42162665). No melanoma trial has randomised management on ctDNA. The design exists in colon cancer, where ctDNA-guided adjuvant therapy has five-year randomised outcomes (PMID 40055522). → wiki/surveillance-and-survivorship.md, wiki/adjuvant-therapy.md
OQ-16. Is a 1 cm excision margin adequate for AJCC stage II melanoma?¶
Seven RCTs in 4,579 patients found no significant difference between narrow and wide margins on any endpoint (PMID 33722422), and 2 cm was as safe as 4 cm at a median 19.6 years (PMID 31280965) — but the only trial testing 1 cm in melanoma >2 mm found it inferior to 3 cm for melanoma-specific survival (HR 1.24, 1.01–1.53, P = .041) (PMID 26790922). MelMarT showed 2 cm margins double the reconstruction requirement (34.9% vs 13.6%, P < .0001) (PMID 29850955). MelMarT-II (NCT03860883, 2,998 planned, verified recruiting 2026-09-01) is powered for the question and reports in the 2030s. → wiki/surgical-management-and-margins.md
Tier 2 — important but blocked¶
OQ-4. Why does checkpoint blockade transform cutaneous but not uveal melanoma, and does that explanation predict anything?¶
Blocked on the absence of a testable predictive framework. Cutaneous melanoma shows 10-year melanoma-specific survival of 96% among 3-year progression-free patients (PMID 39282897); uveal melanoma's 5-year relative survival was flat at 82.8% across 1975–2016 (PMID 40225965) and its first-line agent is an HLA-restricted bispecific (PMID 34551229) delivering 1-year OS 73% vs 50% against combination checkpoint blockade in a propensity-weighted comparison (PMID 38048850). Mutational burden and the GNAQ/GNA11 versus BRAF divide are the standard explanations. What would unblock it: a prospective statement of which tumours the framework predicts will and will not respond, tested against outcomes. → wiki/uveal-melanoma.md, wiki/immunotherapy-advanced-disease.md
OQ-5. Are RB1-carrier retinoblastoma survivors captured by melanoma surveillance pathways?¶
Now partly answered on the risk side. Kleinerman supplies the melanoma-specific figures the Schonfeld abstract does not: 50-year cumulative melanoma incidence 4.5% in heritable versus 0.7% in non-heritable survivors, melanomas distributed in patterns resembling melanoma-prone families, and sun sensitivity not varying by skin-cancer status — pointing to genetic rather than exposure mechanism (PMID 34153328). Still blocked on the surveillance side: RB1 is not among the eight genes assessed in the genome-first familial-melanoma prevalence analysis (PMID 42201696), and no melanoma surveillance guideline located in this build names RB1 carriers as a target group. → wiki/germline-predisposition.md, wiki/special-populations.md
OQ-6. Do standard-of-care effect sizes hold in acral and mucosal melanoma?¶
Blocked on trial representation. First-line immunotherapy response was 15.0% in 69 acral versus 39.1% in 724 cutaneous melanomas in a predominantly non-Hispanic White Spanish registry, with median OS 17.3 vs 43.0 months (P = .007) (PMID 40841506); anti-PD-1 response was 54% in White versus 20% in East Asian, Hispanic and African patients across 1,135 patients at five institutions (PMID 35293617). KIT inhibition gives a pooled 15% response rate (PMID 34562816). One Chinese-led phase 3 in acral melanoma is recruiting (NCT05789043, verified 2026-09-01). What would unblock it: subtype-stratified pooled registry analysis, or dedicated subtype trials. → wiki/acral-and-mucosal-melanoma.md, wiki/clinical-trials-landscape.md
OQ-7. Can any assay separate indolent from progressive early melanocytic lesions?¶
Blocked on biology. The driver events are shared across the benign–intermediate–malignant boundary: BRAF V600E in unequivocally benign lesions, TERT promoter mutation in 77% of intermediate lesions and melanoma in situ, with biallelic CDKN2A inactivation appearing only at invasion (PMID 26559571) and no mutation specific to metastatic progression (PMID 29990500). Methylation profiling plus targeted sequencing distinguished Spitz nevi from both nevi and melanoma and changed the histological diagnosis in 47% of equivocal spitzoid cases (PMID 35737508; PMID 35104769), which is the most promising direction. The WHO's response was nomenclature — melanocytoma for intermediate lesions (PMID 32057276). → wiki/molecular-subtypes-and-genomics.md, wiki/histopathology-and-prognostic-factors.md, wiki/screening-and-overdiagnosis.md
OQ-10. Can the IFN-γ/TMB combination select patients for checkpoint therapy?¶
Blocked on prospective validation. In OpACIN/OpACIN-neo, pathological response rate was 100% with high IFN-γ score and high TMB, 91% and 88% with one high marker, and 39% with both low (PMID 33558721); IFN-γ signalling is the dominant response axis in advanced disease (PMID 32916126) and the T-cell-inflamed gene expression profile is necessary but not always sufficient for benefit (PMID 28650338). High TMB alone fails to predict response across cancers (PMID 33736924). What would unblock it: a biomarker-stratified randomised trial. → wiki/immunotherapy-advanced-disease.md, wiki/neoadjuvant-therapy.md
OQ-17. What intervention reduces fear of recurrence in melanoma survivors?¶
Blocked on a negative trial. MELACARE randomised 153 patients with stage IA–IIA melanoma to a nurse-led intervention combining self-examination training with metacognitive strategies; at 6 months fear of recurrence was lower but not significantly (−0.86, 95% CI −3.34 to 1.62), while quality of life, patient activation and self-examination confidence improved significantly (PMID 40504479). MELFO showed that reducing follow-up frequency does not worsen patient-reported outcomes, so more contact is not the answer (PMID 35866644). → wiki/patient-experience-and-advocacy.md, wiki/surveillance-and-survivorship.md
OQ-18. Can checkpoint inhibitors be given safely to solid organ transplant recipients?¶
Blocked on the absence of a prospective strategy. Individual-participant-data meta-analysis of 343 transplant recipients across 128 studies gives 31.6% objective response at 1 year against 36.2% acute rejection and 18.4% graft loss, with 52.8% dying of cancer within 3 years (PMID 40545616). Transplant recipients have melanoma SIR 2.29, present at more advanced stage (adjusted OR 4.29) and have higher melanoma-specific mortality (adjusted HR 1.93) (PMID 32111554). What would unblock it: a prospective protocol modifying immunosuppression around checkpoint therapy, of the kind reported for liver transplantation (PMID 41082711). → wiki/special-populations.md, wiki/immune-related-adverse-events.md
Tier 3 — mechanistic and infrastructural¶
OQ-9. Does a surveillance strategy, rather than a surveillance test, improve outcomes?¶
Nodal ultrasound has pooled sensitivity 0.879 and specificity 0.969 with a diagnostic odds ratio of 224.5 across 36 studies and 18,273 patients, but 77.0% adherence overall and significantly lower in the US (PMID 38635019). MELFO showed reduced-intensity follow-up is safe and that self-examination detects ~76% of recurrences (PMID 35866644). Number needed to excise varies from 3.0 to 17.2 depending on setting and operator (PMID 35648432; PMID 32114083; PMID 39682200). Every one of these is a property of the strategy, not the test. → wiki/surveillance-and-survivorship.md, wiki/clinical-diagnosis-and-dermoscopy.md
OQ-11. Should surveillance be targeted by growth rate rather than by nevus phenotype?¶
One third of melanomas grow ≥0.5 mm/month, and rapid growth is associated with male sex (GMR 1.7), age ≥70 (2.8), and fewer nevi (<50, GMR 2.0) and fewer freckles (2.5) (PMID 17178980). Nodular melanoma grows at a median 0.49 mm/month against 0.12 for superficial spreading (PMID 17178980) and often presents unremarkably (PMID 20231503). Yet risk prediction and surveillance targeting are built on nevus count and phenotype (PMID 25713022; PMID 21605094; PMID 33420020). Population data also show no significant association between thickness and reported time to diagnosis except for post-presentation delay in physician-detected nodular melanoma (PMID 17116832) — so thickness at presentation is largely a growth-rate phenomenon. → wiki/clinical-diagnosis-and-dermoscopy.md, wiki/risk-factors-and-prevention.md, wiki/red-flags-and-safety-concerns.md
OQ-12. What endpoint should replace RECIST for agents whose benefit is invisible to it?¶
Tebentafusp improves overall survival in metastatic uveal melanoma even among patients whose best RECIST response was progressive disease with target growth exceeding 20%, and treatment beyond radiographic progression was associated with longer survival in post-hoc analysis (PMID 42162665). Treatment-free survival was developed for checkpoint blockade because conventional endpoints miss the distinctive feature of that class (PMID 31498030; PMID 34799400), and is itself sensitive to subgroup definition (PMID 34855329). Neoadjuvant trials introduced pathological response as an endpoint whose meaning differs by modality — NeoCombi's high pathological response rates were followed by 60% recurrence at five years (PMID 38754780). → wiki/uveal-melanoma.md, wiki/immunotherapy-advanced-disease.md, wiki/neoadjuvant-therapy.md
OQ-19. Why did SCREEN's mortality signal not reproduce at national scale in the same health system?¶
The Schleswig-Holstein pilot reached 360,288 of 1.88 million eligible citizens (19% participation), raised invasive melanoma incidence 34% during its 12 months, and was followed by melanoma mortality of 0.79 versus 2.00 per 100,000 expected in men and 0.66 versus 1.30 in women five years later (PMID 22074699). A national program followed in 2008; a ten-year review found no mortality reduction and noted lower screening intensity than in the pilot (PMID 30411137); and an ecological comparison of 15 German federal states against nine neighbouring countries 2009–2022 found pooled annual mortality change −1.8% versus −2.2%, P = .42 (PMID 42268621). Candidate explanations — screening intensity, regression to the mean, ecological-design artefact — have not been distinguished. → wiki/screening-and-overdiagnosis.md, wiki/guidelines.md
OQ-20. What is the patient experience of acral, mucosal and skin-of-colour melanoma?¶
Recorded as an infrastructural question because the absence is systematic. The principal qualitative syntheses draw on Australian, British, North American and Northern European studies (PMID 27775838; PMID 32342223), and the largest patient-authored corpus covers four countries (PMID 29173015). The populations with the worst outcomes — later stage, worse survival, different subtype mix (PMID 35533771; PMID 34363907) — are the least documented. No acral- or mucosal-specific patient organisation was located in this build. → wiki/patient-experience-and-advocacy.md, wiki/acral-and-mucosal-melanoma.md
OQ-21. Does mucosal melanoma need a single staging system, and which one?¶
Three systems compete: mucosal-melanoma TNM, Ballantyne/Prasad, and squamous-cell TNM. In 96 consecutive head-and-neck mucosal melanomas both mmTNM and sccTNM correlated with overall, progression-free and distant-metastasis-free survival (PMID 31421470); in 35 sinonasal cases only the squamous-carcinoma TNM correlated with overall (P = .012) and disease-free survival (P = .041) (PMID 23729399). Five-year relative survival ranges 3–69% across mucosal sites (PMID 24272143), so a system that does not stratify by site may not be improvable. → wiki/staging.md, wiki/acral-and-mucosal-melanoma.md
Seed questions closed by this build¶
| Seed ID | Status | Resolution |
|---|---|---|
| Seed OQ-5 (partial) | Partly closed | The melanoma-specific figures the Schonfeld abstract lacked were located in a separate cohort: 50-year cumulative melanoma incidence 4.5% (heritable) vs 0.7% (non-heritable) (PMID 34153328). The surveillance half of the question remains open and is carried forward as OQ-5. |
| Seed OQ-7 | Reframed and carried forward as OQ-8 | The stage IIB/IIC adjuvant trials were located (KEYNOTE-716, PMID 40198940; CheckMate 76K, PMID 37845511), so the question is no longer blocked on missing evidence. It is now a quantified benefit-versus-permanent-harm question, sharpened by the PRFS2 result. |
| Seed anchor gap: DeCOG-SLT | Closed | Located and incorporated: final analysis, no survival benefit for completion dissection (PMID 31557067). |
| Seed anchor gap: KIT-directed therapy | Closed | Located: pooled objective response 15% (12–18) across 19 studies and 601 patients (PMID 34562816). |
| Seed anchor gap: melanoma in pregnancy | Closed | Located: incidence and outcomes (PMID 25349945) and placental/fetal metastasis (PMID 12775744). |
| Seed anchor gap: transplant and immunosuppressed cohorts | Closed | Located: population incidence and stage (PMID 32111554) and checkpoint-inhibitor outcomes with rejection rates (PMID 40545616). Carried forward as OQ-18. |
| Seed anchor gap: ctDNA surveillance | Closed as an evidence gap, opened as OQ-15 | Located: prognostic validation exists (PMID 40250457; PMID 40387867); no management-guiding randomised trial exists. |
| Seed anchor gap: current NCCN and ESMO documents | Closed | ESMO 2025 cutaneous (PMID 39550033) and ESMO-EURACAN 2026 uveal (PMID 42020098) located; NCCN cutaneous and uveal resources verified by URL on 2026-09-01. |
| Seed anchor gap: patient-organisation material | Closed | Eighteen organisations verified live on 2026-09-01; directory in literature/patient-voice/organizations.md. |
Note to the next sweep. Every asserted absence in this file was re-run on 2026-09-01 and is dated. Stale absences are the most common real error in a file of this kind: re-run each one rather than inheriting it. In particular, re-check (i) whether any trial has been registered comparing neoadjuvant against adjuvant therapy with overall survival as primary (OQ-3), (ii) whether any melanoma ctDNA management trial has opened (OQ-15), (iii) whether MelMarT-II (NCT03860883) has reported (OQ-16), and (iv) whether any guideline has addressed the KEYNOTE-716 PRFS2 result (OQ-8).