What changed: Condition seeded: INDEX.md (portrait + 17-page proposed canonical list, sibling cross-reference to lung-adenocarcinoma), wiki/overview.md (draft, 19 PMID-verified references), OPEN-QUESTIONS.md (7 seed questions), this log.
What was searched: Live PubMed queries on: TCGA 2012 squamous characterization, FGFR1 amplification (Weiss 2010), Lung-MAP master protocol, CheckMate 017, KEYNOTE-407 (primary + 5-yr update), KEYNOTE-024, CheckMate 816, pemetrexed histology interaction, squamous therapeutic-landscape reviews (Langer 2016; Yuan 2021), plus the shared screening set (NLST, NELSON, USPSTF 2021), classification (Travis 2011), testing guideline (Lindeman 2018), never-smoker genomics (Díaz-Gay 2025), and population mortality (Howlader 2020) — verified once this session and shared with the adenocarcinoma sibling. All PMIDs verified via search + metadata retrieval this session.
Check next time: (1) DDR2 mutation and PI3K-pathway trial outcomes for failed-and-frontier-targets.md; (2) necitumumab SQUIRE and ramucirumab REVEL primary papers (currently only covered via the Langer 2016 review); (3) CheckMate 227/9LA (IO-doublet regimens including squamous patients); (4) squamous-subset results of perioperative IO trials; (5) TROP2/other ADC trials in squamous disease; (6) histology-stratified screening performance data.
What changed: Completed all 17 canonical wiki pages and replaced the seeded overview. Built literature/BIBLIOGRAPHY.md, six landmark-study notes, guidelines/REGISTRY.md, statistics/STATISTICS.md, and the four-file patient-voice layer. Rewrote OPEN-QUESTIONS.md as a tiered, study-shaped agenda with a 12-row “Dots not yet connected” table; updated the master index and roadmap status.
PubMed work: Ran live PubMed E-utilities searches across burden and exposures; airway premalignancy; pathology and mimics; screening; TNM/staging; TCGA and later multi-omic/spatial biology; FGFR, DDR2, PI3K, cell-cycle, redox and ADC programs; chemotherapy; immunotherapy; perioperative and stage-III therapy; biomarkers; safety; palliative care; stigma; caregivers; financial toxicity; patient-reported outcomes; and trial participation. Performed a final ESummary re-query of every unique PMID used in the wiki and generated the bibliography from returned metadata. The wiki contains 321 unique PubMed records; no PMID was accepted from memory.
Important corrections: Replaced incorrect ninth-edition TNM identifiers with PMIDs 38447919 and 38885896; replaced an incorrect Lung-MAP design identifier with PMID 26408303; removed the seeded unsupported claim of roughly 400,000 annual LUSC deaths and used modelled 2022 incidence (616,769 cases; PMID 39914442) with an explicit mortality-data limitation. Final title-to-claim QA also caught four mis-mapped PMIDs in the statistics/open-question extraction and replaced them with the live-query-confirmed records: dovitinib 27315356, AZD4547 31195180, TAK-228 36240971, and postoperative complications 38975138.
Evidence boundaries: Labelled NSCLC-wide results rather than implying LUSC-specific effects; treated mechanistic targets as hypotheses unless clinically validated; preserved exploratory/subgroup qualifiers; separated relative from absolute effects; and described registry records as mutable sponsor-entered data.
Patient voice: Used only public organizational pages and PubMed-indexed research; no private communities, composite quotations, or cross-source patient dossiers. The main limitation is the scarcity of qualitative research stratified specifically by squamous histology and the geographic concentration of published evidence.
Next update triggers: New ASCO living-guideline version or jurisdictional authorization; ninth-edition TNM implementation data; overall-survival maturity for perioperative trials; randomized evidence isolating the postoperative checkpoint component; a practice-changing LUSC ADC/redox/FGFR result; definitive post-chemo-immunotherapy sequence data; or a new global histology-specific burden release.
Validation: Confirmed 17 canonical wiki files, each 150–400 lines and 25–60 unique PubMed plus ClinicalTrials.gov identifiers; 322 unique condition-wide PMIDs fully covered by the bibliography; 25 unique NCT IDs live-queried; all local Markdown links resolved; no stale scaffold markers, temporary build files, or non-Markdown artifacts remained in the condition directory. Validation matched PMID labels to destination URLs and included title-to-claim review, so an existent but unrelated PubMed record could not pass merely by resolving.
Scope checked: Audited every one of the 17 canonical wiki pages and every literature artifact: the 326-entry master bibliography, six landmark notes, guideline registry, statistics ledger, four-file patient-voice layer, open-question agenda, index, and prior curation log. The final wiki pages remain 153–247 lines each and contain 25–36 unique PMID plus NCT identifiers each.
Live PubMed verification: Re-extracted all identifiers from the finished condition corpus and submitted all 326 unique PMIDs to PubMed E-utilities EFetch during this audit session. All 326 resolved. Checked 765 full reference lines across the wiki reference sections and master bibliography against returned first-author and title metadata; the final check produced zero author/title mismatches. Every PMID used by a literature artifact is present in the bibliography, and the bibliography's canonical-wiki cited by mappings have zero missing or stale mappings.
Live registry verification: Submitted every one of the 29 unique NCT identifiers in the finished corpus to the ClinicalTrials.gov API v2 during this audit session; all 29 resolved. This includes the 25 treatment-trial identifiers already catalogued on the trials page and four anti-stigma studies added while correcting a stale evidence-gap claim (NCT05456841, NCT06191939, NCT03750864, NCT06286592). Registry state is dated 2026-08-30 and remains separated from published efficacy.
Citation errors found and fixed: Canonicalized 438 pre-existing abbreviated wiki references to the live PubMed metadata, including 62 detected wrong first-author attributions. Replaced PMID 17167000 as support for post-obstructive infection because that paper concerns outcomes after central-airway-obstruction treatment, not pneumonia; the infection claim now uses the live-verified cancer post-obstructive-pneumonia review PMID 29392577. Added primary TCGA support (PMID 22960745) to the 72% CDKN2A/RB1-pathway prevalence claim instead of leaving the negative palbociclib trial as its only citation. No wrong citation or unresolved claim remains flagged.
Stale-absence searches: Re-ran PubMed and ClinicalTrials.gov searches for redox-directed therapy, durable LUSC immunotherapy prediction, bronchial-premalignancy risk models/interception, postoperative checkpoint component isolation, central-airway patient-reported scales, histology-specific global mortality, digital-community outcomes, and stigma interventions. Converted genuine gaps into positively stated, dated search findings. Corrected the outdated implication that stigma interventions do not exist: a 2024 systematic review found 11 mostly pilot studies (PMID 39197098), a communication-training pilot reported provider-level changes (PMID 38836528), and four registry studies confirm an intervention pipeline; the remaining gap is adequately powered evidence for diagnostic interval, treatment uptake, or survival. Added 2026 retrospective pCR evidence (PMID 42055888; DFS HR 1.22, 95% CI 0.51–2.95) while preserving the narrower conclusion that no randomized study isolates the postoperative component.
Mutable sources: Rechecked all 25 treatment-trial records, the four added stigma-trial records, the guideline sources available without authentication, and the five public patient-organization pages. Updated registry, guideline, patient-source, and evidence-cutoff dates to 2026-08-30. Retained the NCCN authenticated-access caveat rather than inferring a version from a secondary source.
Final QA and disposition: All 17 wiki pages passed: 326/326 PMIDs resolved, 29/29 NCT IDs resolved, 0 reference-metadata mismatches, 0 bibliography-mapping failures, 0 broken local links, and 0 unverified flags. Every wiki page was therefore promoted to status: curated; literature assets are marked audited in the index, and the condition roadmap is now audited. Remaining substantive flags: none. Evidence gaps remain explicitly labelled as dated gaps rather than unresolved verification failures.
2026-08-31 — Completion audit of the remaining eight pages¶
Scope and disposition: Independently re-audited the eight pages identifiable as the unfinished 9/17 split: biomarkers.md, clinical-trials-landscape.md, early-stage-and-perioperative-therapy.md, guidelines.md, overview.md, patient-experience-and-advocacy.md, red-flags-and-safety-concerns.md, and systemic-therapy.md. The on-disk frontmatter had already been changed to curated before this pass, so those declarations were treated as untrusted until the checks below passed. All eight now pass and remain status: curated; combined with the nine pages that passed the earlier audit, all 17 canonical pages pass.
Live PubMed verification: Extracted the final eight-page identifier set and fetched all 153 unique PMIDs through PubMed E-utilities in this session. All 153 resolved. Reviewed 226 inline PMID links across 120 cited claim blocks against the fetched titles and abstracts, including every quoted effect size, sample size, confidence interval, and event rate. Final automated reference QA found 0 first-author mismatches, 0 title mismatches, 0 PMID label/URL mismatches, and 0 unresolved records.
Live ClinicalTrials.gov verification: Fetched all 23 unique NCT records used by the final pages individually through the API v2 in this session; all 23 resolved. Rechecked overall status, enrollment type/count, phase, histology/condition, interventions, primary endpoints, country count where stated, and last-update date. Also reran active-registry searches for squamous NSCLC, Lung-MAP, NFE2L2/KEAP1, FGFR, TROP2, ctDNA-guided resected disease, ECOG 2 disease, stigma interventions, and central-airway patient-reported outcomes.
Registry errors fixed: Corrected the description of NCT04310007. Despite its generic NSCLC title and Lung-MAP relationship, the current record is restricted to non-squamous NSCLC and is not an active LUSC option. Added the directly relevant, not-yet-recruiting PANKU-Lung05 phase II/III study (NCT07739199; 200 estimated participants), while preserving that both arms contain the investigational PD-1/VEGF bispecific and that the phase III primary endpoint is PFS. Updated all registry/access dates to 2026-08-31 and removed the page's inconsistent 29/30 August wording.
Citation and claim repairs: Added ten full PubMed references that were cited inline but missing from the clinical-trials page reference list and two missing from the patient-experience page. Connected four previously reference-only systemic-therapy papers to the claims they support: KEYNOTE-024 for PD-L1-high monotherapy (PMID 27718847), nab-paclitaxel response by histology (PMID 22547591), cisplatin/gemcitabine versus cisplatin/pemetrexed in squamous disease (PMID 18506025), and the selected bevacizumab phase III bleeding/exclusion evidence (PMID 17167137). Final inline/reference sets are identical on every audited page, with no unused reference entries.
Overstatement and mutable-source repairs: Replaced the unsupported assertion that the ASCO web guideline was version 2026.3.3 with the live PubMed-verified 2026.3.1 publication (PMID 42190141); the ASCO landing page returned HTTP 403 during this audit, so the page now states that limitation and requires version rechecking at use. Softened “no cross-trial comparison” of postoperative checkpoint therapy to the accurate claim that no randomized comparison isolates the postoperative component and that indirect comparisons remain confounded. Removed the Lung Cancer Research Foundation resource row because its cited page and site root returned HTTP 403 and its resource description could not be re-verified; the other listed patient-organization pages returned HTTP 200.
Stale-absence searches: Reran live PubMed searches for randomized ctDNA-guided escalation/de-escalation after NSCLC resection, randomized isolation of adjuvant checkpoint therapy after neoadjuvant chemo-immunotherapy, LUSC-specific qualitative patient experience, histology-specific global mortality, serum-marker screening outcomes, post-chemo-immunotherapy LUSC sequencing, stigma interventions with clinical endpoints, and central-airway LUSC patient-reported measures. Reran corresponding ClinicalTrials.gov searches. No randomized PubMed record was found for ctDNA-guided resected-NSCLC treatment or post-chemo-immunotherapy LUSC sequencing; perioperative searches found combined-pathway trials and indirect comparisons, not randomized isolation of the postoperative component. The qualitative search produced seven raw hits but no inspected record directly supplied a LUSC-specific qualitative experience study. Stigma is not an empty field: added the 11-study systematic review (PMID 39197098) and 30-clinician Nigerian communication-training pilot (PMID 38836528), while retaining the narrower gap in patient outcomes, diagnostic interval, treatment uptake, and adequately powered randomized evidence.
Non-PubMed source checks: NICE NG122 and its cited diagnosis, management, suspected-cancer, and palliative-care pages; the USPSTF recommendation; IASLC molecular-testing resource; GO2 stigma/help-line pages; LUNGevity patient hub; and Roy Castle site all resolved live. The ACS web page returned HTTP 406, so its eligibility statement is supported only by the live PubMed guideline record (PMID 37909877), not asserted as a separately retrieved web citation.
Final QA: All eight pages are 153–247 lines. Final checks found 17 canonical pages, 0 draft statuses, 0 unresolved verification markers, 0 broken local links, 0 bibliography mapping failures, and 0 citation/reference set differences. The roadmap remains audited, with the completion date advanced to 2026-08-31. Remaining substantive flags: none. The ASCO web-version access block is an explicitly documented update-at-use limitation, not an unsupported content claim.