KEY POINTS
- This single-institution retrospective study included 45 patients with recurrent oral squamous cell carcinoma treated with accelerator-based BNCT between June 2020 and December 2022. All had prior irradiation, ECOG 0–1, measurable disease and no distant metastases, and salvage surgery had been considered inappropriate.
- Previous RT dose was a median 67 Gy, with a median interval of 16 months before BNCT. Most patients had advanced recurrent disease: 80% were stage IVA–IVB, and one-third had N3b disease.
- BNCT used intravenous borofalan with accelerator-generated neutron irradiation. Planning aimed for GTV minimum dose >25 Gy-Eq, while maximum pharyngeal mucosal dose was limited to approximately 20 Gy-Eq and oral-mucosal V20 Gy-Eq to <5 cm³.
- Objective response rate was 80%, including a complete-response rate of 35.6%. At a median follow-up of 25 months, 2-year OS was 51.1%, local control 31.8% and PFS 22.2%.
- Tumor volume was associated with survival on univariable analysis (p=0.036) but not independently significant after multivariable adjustment. Exploratory cutoffs suggested worse survival at volumes ≥20 cm³ and worse survival/local control at ≥25 cm³, but these thresholds were data-derived and require external validation.
- Early complete response carried a strong prognostic signal: patients without complete response within 90 days had substantially worse survival (HR 3.73, 95% CI 1.50–9.30; p=0.005). A six-month landmark analysis produced a similar association, reducing—but not eliminating—the concern for immortal-time bias.
- Toxicity was clinically substantial: grade ≥3 oral mucositis occurred in 49%, grade ≥3 osteoradionecrosis in 10.2%, and 33% required prolonged hospitalization for impaired oral intake. Other severe late events included trismus, brain abscess, oronasal fistula and visual impairment, with one grade 5 carotid rupture.
CLINICAL TAKEAWAY
Accelerator-based BNCT can produce meaningful responses in heavily pretreated recurrent oral cancer when surgery is no longer feasible, but durable local control remains modest and toxicity is far from trivial. These data support careful selection—particularly by tumor volume and proximity to vulnerable structures—rather than viewing BNCT as a low-toxicity re-irradiation option.