CNS Necrosis after Proton Therapy for Sinonasal Carcinoma

Abstract

<h3>Purpose/Objective(s)</h3> Sinonasal cancers (SC) are a rare, diverse disease group with a relatively poor prognosis. Local control (LC) has improved with advances in imaging, surgery, and radiotherapy including proton therapy (PT), albeit with increased risk of toxicity such as central nervous system (CNS) necrosis. Our objective was to assess CNS necrosis risk in patients who receive PT for SC and evaluate prognostic factors and DVH metrics. <h3>Materials/Methods</h3> We reviewed 173 patients (age 18+ yrs) treated with PT (+/- chemo) for SC (excluding melanoma, lymphoma, sarcoma, distant metastasis, ≤ 6 mos follow-up, and reirradiation) at a single institution. We assessed CNS necrosis using CTCAEv5 and censored events at last primary site imaging, local recurrence, distant metastasis, or death. We estimated survival, disease control, and CNS necrosis rates using the cumulative incidence method. We used a Fine-Gray test to correlate prognostic univariate factors, competing risks proportional hazards regression for multivariate analysis, and correlated select whole-brain DVH parameters (V40-80, 5 GyE intervals, Dmax, and Dmean) with CNS necrosis using univariate logistic regression on 93 patients with >1-yr imaging follow-up and no competing events. <h3>Results</h3> Median age was 60 years (range, 18-86) with a median follow-up of 5 yrs among survivors (range, 0.5 – 13.5), most had T3-4 tumors (91%) and naso-ethmoid tumors (77%); 42% had intracranial and 34% had cranial nerve invasion. Most (80%) had surgery before PT; 69% received chemo. Most received passive-scatter PT (84%) at 1.2 GyE/fraction twice-daily (85%) to ≥70 GyE (79%; median, 73.2; range, 54 – 74.4). 5-yr LC (95% CI), disease-free, and overall survival rates were 84% (78-89%), 63% (56-71%), and 59% (51-77%). After gross-total resection, the 5-yr LC rate was 92% (95% CI, 86-96%). There were 72 (40%) CNS necrosis events at a median 2.6 yrs from end of PT (range, 1.1 – 9.1) with 14 (8%) grade 3+ events and 3 deaths. Most events (53%) were asymptomatic (grade 1). Freedom from CNS necrosis (grade 1+, 2+, and 3+ [95% CI]) at 5 years was 78% (72-85), 90% (85-94), and 92% (87-95). T3/4 disease, intracranial invasion, biopsy alone, total dose, and chemotherapy were negatively correlated prognostic factors. Nearly all DVH metrics showed strong correlation with CNS necrosis risk except Dmean. The receiver-operator curve characteristics for V65 showed the strongest correlation (AUC 0.83 grade 2+, AUC 0.78 grade 3+) and the model provided candidate DVH goals (V65 < 1mL, 5% risk of grade 3) with an acceptable deviation (V65 < 17mL, 15% risk of grade 3+) for those with gross skull base/intracranial tumor. <h3>Conclusion</h3> PT +/- chemo in the primary and adjuvant settings yields high rates of LC for locally advanced SC. CNS toxicity is a significant risk, but most events are asymptomatic and self-limited. We intend to decrease treatment intensity after gross total resection given the favorable LC in this subset. These data provide updated DVH goals to decrease the risk of CNS toxicity.