P1.17-09 V30 May Better Predict Radiation Pneumonitis After Intensity-Modulated Radiation Therapy for Lung Cancer

V20 and MLD are the most commonly used dose constraints for radiation pneumonitis (RP) prediction. However, intensity-modulated radiation therapy (IMRT) has unrestricted beam arrangements, an infinite number of very different dose distributions could be generated in the lung volume outside the planning target volume (PTV). Conventional dose constraints from traditional 3D conformal RT may not be valid for IMRT treatment. We hypothesize that lung dosimetric parameters may have different RP predictive values from that of traditional constraints (largely generated from 3D treatment) in IMRT treated lung cancer patients. We retrospectively enrolled184 IMRT treated lung cancer patients from January 2014 to October 2017. The primary endpoint was acute grade 2 or higher symptomatic radiation pneumonitis (RP2), based on the National Cancer Institute’s Common Terminology Criteria for Adverse Events (version 4.03). Vdose (from V5 to V50) and MLD were generated from the lung volume outside PTV. Univariate and multivariate logistic regression analysis was used to evaluate the association between the dose parameters outside PTV to RP2. We employed area under the curve (AUC) for the receiver operating characteristic curve (ROC) to assess prediction accuracy for the single or multi-variate model. 26 out of 184 lung cancer patients (14.1%) developed RP2 within 3 months after the end of IMRT treatment. In univariate logistic regression, although none of the clinical parameters was significantly associated with RP2, female gender (P=0.051) and chemotherapy (P=0.151) had a trend of correlation. V5 (P=0.007), V10(P=0.012) V20 (P=0.004), V30 (P=0.003) and MLD (P=0.004) were significantly associated with RP2 incidence. From ROC curve, the largest AUC of 0.67 was generated from V30, which showed a better predictive value compared with other dosimetric factors. Multivariate logistic regression analysis showed the only significant dosimetric factor is V30 (P=0.021). Combining gender and chemotherapy factors, V30 has an AUC of 0.71 which is the largest among all the other dosimetric factors. For IMRT treated lung cancer patients, V30 generated from lung volume outside PTV may predict RP more accurately than traditional dosimetric parameters.