2445 - Impact of Fractionation Regimens on Radiation-Induced Brain Injury in Nasopharyngeal Carcinoma: Analysis Based on a Novel Equivalent Dose Conversion Model
Presenter(s)
X. F. Qin1, X. Ou1, J. X. Su1, W. Liu1, J. Li1, Q. Zhong1, D. Luo1, Y. Ou2, and Q. Du1; 1The Second Affiliated Hospital of Guangxi Medical University, Nanning, Guangxi, China, 2Department of Radiation Oncology; State Key Laboratory of Oncology in South China; Guangdong Provincial Clinical Research Center for Cancer; Sun Yat-Sen University Cancer Center, Guangzhou, China
Purpose/Objective(s): To evaluate the impact of fractionation regimens on radiation-induced brain injury (BI) in nasopharyngeal carcinoma (NPC) and propose a novel radiobiological model to address potential limitations of the linear-quadratic (LQ) framework.
Materials/Methods: A simplified-exponent (SE) model was proposed as E=n*?*dm, where n is fraction number, d is dose per fraction and m is equivalent exponent. The equivalent dose in 2 Gy fractions (EQD2) for SE model is derived as SE-EQD2=D(d/2)m-1, where D is total dose. A retrospective analysis was performed on 224 NPC patients (98 BIs in 448 half-brains) treated with 30-33 fractions radiotherapy. The minimum dose delivered to the most irradiated 1.2 cm3 volume (D1.2cc) and its corresponding dose per fraction (d1.2cc) for each half-brain were extracted. Predictive performance of D1.2cc and their respective LQ- and SE-EQD2 conversions was assessed using the area under the receiver operating characteristic curve (AUC). A logistic dose-response mode, P(Y)=1/(1+exp(-B0-B1Y)), was fitted. Analyses were conducted for the entire cohort and four subgroups stratified by fraction number.
Results: Increasing the fraction number from 30 to 33 significantly reduced BI incidence rates (for D1.2cc at 66/68/70 Gy: from 23.0/33.6/46.2% to 2.3/4.4/8.3%). In pooled analysis, the LQ-EQD2 AUC increased as the a/ß ratio decreased, ranging from 0.819 (a/ß=8Gy) to 0.836 (a/ß=0Gy). The SE-EQD2 AUC improved with increasing m, reaching a maximum of 0.842 at m=20, a value identical to using d1.2cc alone. When fraction number increased from 30 to 33, the equivalent-physical dose for a given probability increased by about 2 Gy in LQ-EQD2 model (a/ß=3Gy), compared to a around 6-7 Gy increase observed in d1.2cc dose-response curve, which aligns more closely with clinical observations.
Conclusion: The LQ model may underestimate the influence of fraction dose on late effect of the brain. Decreasing the fraction dose appears to provide greater protection against BI than traditionally predicted. For late BI, an a/ß value of 0 Gy may be optimal within the LQ model framework. The SE-model offers an extended conversion range that may better capture this fractionation effect.