Most units moved to weekly cisplatin years before the randomised evidence caught up. JCOG1008 gave us the interim answer in 2022. This month brings the five-year data — and a number that is easy to misread.
In this episode:
- Why the trial was designed as a noninferiority study, and what the 1.32 margin was chosen to preserve
- The five-year efficacy results across overall survival, relapse-free survival and local control
- Why "not worse" and "better" are different claims — and why the confidence intervals matter more than the gap between the curves
- The baseline imbalances that randomisation did not stratify for
- Dose intensity as the likely explanation for why this trial and the Indian phase III trial disagree
- Late toxicity, nutrition support–free survival, and the two treatment-related deaths
Key numbers:
261 patients randomised, 132 to 3-weekly cisplatin 100 mg/m² and 129 to weekly 40 mg/m². At a median follow-up of 5.6 years, five-year overall survival was 58.7% versus 71.2%, stratified HR 0.76 (95% CI 0.52–1.12) — noninferiority confirmed. Five-year relapse-free survival 53.0% versus 64.3% (HR 0.81), local relapse-free survival 57.2% versus 68.8% (HR 0.79). Adjusting for T stage, N stage and primary site moved the hazard ratio to 0.88 (95% CI 0.60–1.29). Estimated dose intensity was 33.6 versus 29.3 mg/m² per week. No late adverse event differed by 10% or more between the arms.
Paper discussed:
Tahara M, Kiyota N, Kodaira T, et al. Long-Term Follow-Up of JCOG1008, a Randomized Phase II/III Trial of Chemoradiotherapy Comparing 3-Weekly Cisplatin With Weekly Cisplatin in Postoperative Head and Neck Cancer. J Clin Oncol. Published online June 26, 2026. doi:10.1200/JCO-25-01708
Trial registration: jRCTs031180135
Links:
Summary: hnoncology-journalclub.netlify.app
Video channel: youtube.com/@headandneckcancers