Hypofractionated Versus Normofractionated Accelerated Radiation Therapy With or Without Cisplatin for Locally Advanced Head and Neck Squamous Cell Carcinoma (HYPNO): A Randomized, Open-Label, Phase 3, Noninferiority Trial

Purpose Based on bioeffect modeling of published outcomes after radiation therapy for head and neck squamous cell carcinoma with various time-dose-fractionation, we hypothesized that a 20-fraction hypofractionated (HFX) schedule delivering 55 Gy in 20 fractions, 5 fractions per week, over 4 weeks would be noninferior to a 33-fraction accelerated, normofractionated (NFX) 2 Gy per fraction schedule, delivering 66 Gy in 2-Gy fractions, 6 fractions per week over 5.5 weeks with respect to both local tumor control and late adverse events. Methods and Materials The HYPNO (HYPo-fractionated vs NOrmo-fractionated radiation therapy for head and neck squamous cell carcinoma) trial was designed as a multicenter, pragmatic, embedded, 2-arm, unblinded, randomized controlled noninferiority trial with dual primary endpoints, loco-regional tumor control, and grade 3 or higher late adverse events with a 10% noninferiority margin for both endpoints. The trial was open for enrollment in 12 centers, each adhering to their standard of care to the extent that it was consistent with the requirements of the trial protocol. Concurrent chemoradiation therapy with 35 mg/m 2 cisplatin weekly was permitted. Results Between March 2014 and February 2020, 792 patients were centrally randomized: 395 to HFX and 397 to NFX. Accrual closed, with all outcome data still blinded, with 792 of a planned 836 patients (94.7%) enrolled, in part due to the emerging COVID-19 pandemic. The HYPNO test arm passed the separate noninferiority tests for both loco-regional tumor control ( P = .04) and grade 3+ late adverse events ( P = .004). At 3 years, the absolute difference in outcome between the 2 arms was ≤1.4 percentage points for overall survival, progression-free survival, loco-regional control, and grade 3+ late adverse events. The planned subgroup analyses showed no statistically significant heterogeneity of effect estimates for loco-regional control between the 2 trial arms. Conclusions The HYPNO test arm schedule was shown to be noninferior with respect to both loco-regional tumor control and grade 3+ late adverse events.