A data-driven model for personalised risk prediction of radiation-induced oral mucositis in oral cavity cancer patients

Abstract

Aim: Radiation-induced oral mucositis (OM) is the most common complication presenting with chemo radiation therapy of oral cavity cancer. Tobacco use, oral hygiene and nutritional status are important entities that impact the incidence of OM. These entities must also be studied along with treatment planning strategies, to alleviate its incidence. Our study aims to present a novel method to model the OM incidence using a mucosal surface contour (MSC). Methods: Computed tomography (CT) images of 60 oral cavity patients who have started their intensity-modulated radiation therapy (IMRT)/volumetric-modulated arc therapy (VMAT) with concomitant chemotherapy (Cisplatin) were delineated with MSC as one of the organs at risk by three expert radiation oncologists. V30, V50 and Dmean doses of MSC and the PTV 60 (planning target volume for 60 Gy), along with Dmax of PTV60, were extracted from the dose volume histograms. OM toxicity was assessed once weekly, and the outcome was scored using CTCAE v5·0 grading. Tobacco use (Tb), oral hygiene (OHy) and nutritional status (Ns) were also numerically scored. A multiple linear regression analysis was done using the patient parameters and the outcome scores as predictor variables and response variables, respectively. Optimal dose volume constraints (Dmean, V30, V50) for a 20% reduction of OM were derived from the mathematical equations. Another 20 patients were planned prospectively using IMRT/VMAT with the above resulted in dose constraints. Clinical outcome was scored for these patients using CTCAE v5·0. Outcome results of the two phases (60 patients and 20 patients) were statistically compared with two-sample t-test. Results: For MSC, three mathematical equations were formulated using multiple linear regression analysis. Derived values of V30, V50 and Dmean constraints were used for dose optimisation in the second phase of treatment planning. It has showed a statistically significant deviation from the first phase of the study, with a confidence interval of 95% (p value: 0·0348) by introducing calculated dose constraints for MSC in dose optimisation. Conclusion: In this study, the feasibility of using multiple linear regression analysis to model OM incidence in radiation therapy clinics was explored. Derived dose-volume constraints for MSC could be used in IMRT/VMAT optimization to reduce its incidence. Patient treatment could be individualised by incorporating dose-volume parameters, nutritional status, tobacco use and oral hygiene status in the treatment planning procedure.