PV QA 4 - Poster Viewing Q&A 4
TU_34_3657 - A Novel Nomogram of DVH Parameters and Clinical Factors for Predicting Severe Acute Radiation Pneumonitis in NSCLC Patients Receiving Post-Operation Radiation Therapy
Tuesday, October 23
2:45 PM - 4:15 PM
Location: Innovation Hub, Exhibit Hall 3
A Novel Nomogram of DVH Parameters and Clinical Factors for Predicting Severe Acute Radiation Pneumonitis in NSCLC Patients Receiving Post-Operation Radiation Therapy
X. Tang1, X. Tian1, M. Yu1, J. Wang1, Y. Xu1, L. Zhou1, Y. Lu2, and Y. Gong1; 1Department of Thoracic Oncology, Cancer Center and State Key Laboratory of Biotherapy, West China Hospital, Sichuan University, 610041, Chengdu, China, 2Department of Thoracic Oncology, West China Hospital, Sichuan University, Chengdu, China
Purpose/Objective(s): Postoperative radiotherapy is potentially with radiation pneumonitis (RP). Grade ≥ 3 RP is generally severe and life-threatening. The aim of this study is to develop a novel nomogram to predict the occurrence of severe (grade ≥ 3) acute radiation pneumonitis (SARP) in non-small cell lung cancer (NSCLC) patients receiving post-operation radiotherapy (PORT).
Materials/Methods: Data from 109 NSCLC patients receiving PORT were retrospectively analyzed. Clinical factors including age, sex, tumor location and concurrent chemotherapy as well as dose-volume histogram (DVH) parameters including the total, ipsilateral and contralateral of the percentage of lung volume receiving ≥ 5,10,20 and 30 Gy (V5, V10, V20 and V30), and the mean lung dose (MLD) were collected. The endpoint is the occurrence of RP defined as the Common Terminology Criteria for Adverse Events, version 4.0 within 3 months after PORT. Logistic regression was used to evaluate the predictive ability of each factor in predicting the occurrence of SARP. Factors with p<0.05 in univariate analysis were further tested in multivariate analysis using backward model-selection procedure to obtain the most simplified and accurate model. Nomogram was generated based on the multivariate regression coefficients. The accuracy and discrimination ability of the model were evaluated by the area under the ROC curves (AUC). Calibration of the model was tested by calibration curves. Decision curve analyses (DCA) were conducted to show the clinical usefulness of the model.
Results: At the end of follow-up, SARP happened in 26 (23.9%) patients. Univariate analysis indicated that ipsilateral V5, total V20, ipsilateral MLD, total MLD and concurrent chemotherapy were related to the occurrence of SARP (p=0.000, p=0.017, p=0.010, p=0.000 and p=0.006, respectively). In multivariate analysis, only ipsilateral V5, total MLD and concurrent chemotherapy were statistically significant (p=0.009, p=0.013 and p=0.014, respectively). These three factors were included in the nomogram. ROC curves showed that the AUC of the nomogram was 0.842, which was much higher than any single factor alone (ipsilateral V5: 0.744; total MLD: 0.769; concurrent chemotherapy: 0.661). Also, calibration curve showed that the predicted SARP from the nomograms exhibited favorable correlation with the actual observation. DCA showed satisfactory positive net benefits in model predicting SARP among most of the threshold probabilities, indicating favorable potential clinical effect of the model.
Conclusion: We developed a novel nomogram to predict the occurrence of SARP in NSCLC patients receiving PORT. We believe this tool may improve the predictability of SARP and assist physicians in clinical works. Table 1. Multivariate analysis (backward model-selection procedure)
|Variable ||HR (95% CI) ||P value ||AUC |
|ipsilateral V5 ||1.084 (1.020-1.151) ||0.009 ||0.744 |
|total MLD ||1.003 (1.001-1.006) ||0.013 ||0.769 |
|concurrent chemotherapy ||4.091 (1.331-12.572) ||0.014 ||0.661 |
Author Disclosure: X. Tang: None. M. Yu: None.