Science's Data Search Engine

Selected high‑quality evidence behind the figures (concise extracts)

• Urological survival framing (testis/prostate)
• Screen-detected early breast cancer survival evidence
• MOSAIC/FOLFOX colon cure estimate
• D2 gastric surgery 5‑yr survival
• ELCAP: CT screening → earlier stage detection
• Global cancer burden & survival context

Interpretation — how to read these survival / 'cure' numbers (critical points)

1. "Cure" ≠ single number: Most published 'cure' claims are operationally 5‑year relative survival or trial-derived disease‑free survival; these are stage- and cohort-dependent and influenced by lead‑time / overdiagnosis biases introduced by screening (ELCAP caution on overdiagnosis).
2. Stage and selection matter: High cure rates (e.g., testicular ~97% 5‑yr) derive from cases detected at early stages amenable to orchiectomy/chemo and from registry aggregates where stage distribution and age skew favourable (Registry survival framing).
3. Interventions change curability locally: RCTs and trials (e.g., MOSAIC / FOLFOX) show adjuvant therapy materially raises 5‑yr disease‑free survival in stage III colon cancer (modelled cure ~71.5% with FOLFOX in a Japanese cost model), but these are trial cohorts and require careful external validation (MOSAIC/FOLFOX estimate).
4. Regional practice differences: Surgical series from high-volume East Asian centers report 5‑yr OS after curative gastric resection ~69–70%; generalizability to other settings depends on selection and perioperative care (Gastric D2 survival).
5. Population averages hide heterogeneity: For lung cancer, population 5‑yr survival remains low overall (single-digit percent for advanced disease) despite CT screening shifting stage distribution; mortality benefit requires RCT mortality endpoints (ELCAP highlighted stage shift but not definitive mortality reductions at publication) (ELCAP stage-shift vs mortality).

Practical summary (evidence‑weighted)

• Strong, reproducible evidence that some cancers have high 5‑year survival when detected early and treated with curative intent (testicular cancer; many localized skin cancers; early breast, early colorectal; some lymphomas) — but these statements are stage-dependent and not universal (Global survival heterogeneity).
• Randomized trial evidence (e.g., MOSAIC → FOLFOX) provides the strongest basis to claim improved curability from specific interventions in defined stage groups (stage III colon); such claims should be tied to trial definitions and follow‑up (MOSAIC trial outputs).
• Screening increases detection of earlier-stage disease (ELCAP for lung CT; mammography/MRI for breast) which correlates with higher survival but introduces biases (lead time/overdiagnosis); therefore, population-level mortality RCTs or high-quality registry mortality trends are needed to validate screening → net lives saved (ELCAP caution).

Primary blindspots, biases & how to falsify these 'curability' claims

• Overdiagnosis bias: If a screening program primarily detects indolent lesions that would never progress, 5‑yr survival rises without mortality benefit; falsification: demonstrate stable or reduced disease-specific mortality despite rising incidence and improved 5‑yr survival (ELCAP warns of this) (Overdiagnosis caution).
• Stage-migration (Will Rogers) and selection bias: Improved imaging or referral patterns can increase apparent survival without altering natural history; falsify by adjusting for stage and lead-time in long-term randomized cohorts.
• Population vs trial differences: RCT-derived cure rates (narrow eligibility) may not generalize; falsify by reproducing trial outcomes in broad registry-based real-world cohorts with similar stage mix and comorbidity distributions.
• Data quality heterogeneity: Global registry numbers vary in completeness; robust inference requires high-quality, population-based registries or RCT mortality endpoints (Registry data caveats).

Actionable next steps (if you want an authoritative, evidence‑graded list)

1. Construct a structured table of cancers by stage with: (a) population 5‑yr relative survival (by registry), (b) RCT-derived cure rates when available, (c) quality rating (RCT/registry/model), and (d) key limitations. This requires extracting stage‑specific survival from high-quality registries (SEER, EUROCARE, National registries) and linking to trial outputs (MOSAIC, relevant surgical series, large screening RCTs).
2. Run sensitivity analyses to show how lead‑time and overdiagnosis affect apparent 'cure' percentages for screen‑detected cancers (lung CT, prostate PSA, breast mammography/MRI).
3. Produce a reproducible visual: stage-by-stage stacked survival curves and evidence-strength heatmap to communicate which cancer-stage pairs have robust proof of curability versus those that are conditional/uncertain.

Key citations used (brief explainers)

• Urological survival framing
• Breast screening detection / survival
• MOSAIC/FOLFOX cure modelling
• Gastric D2 outcomes
• ELCAP stage-shift evidence
• Global cancer burden (GLOBOCAN 2002)