MENU

Click here to

×

Are you sure ?

Yes, do it No, cancel
  1. Home
  2. Meeting Program
  3. All
  4. A Generic Normal Tissue Complication Probability Model for Predicting Radiation-Induced Esophagitis Observed in Non-Small Cell Lung Cancer Patients

A Generic Normal Tissue Complication Probability Model for Predicting Radiation-Induced Esophagitis Observed in Non-Small Cell Lung Cancer Patients

M Chen1,2*, Z Wang2, J Sun2, S Jiang2, B Gunn3, S Frank3, C Xu1, J Chen1, Q Nguyen3, J Chang3, Z Liao3, N Sahoo2, X Zhu2, X Zhang2, (1) Department of Radiation Oncology, Ruijin Hospital, Shanghai Jiao Tong University School of Medicine Shanghai, CN, (2) Department of Radiation Physics, The University Of Texas MD Anderson Cancer Center,Houston,TX,(3) Department of Radiation Physics, The University Of Texas MD Anderson Cancer Center,Houston,TX,

Presentations

(Sunday, 7/12/2020)   [Eastern Time (GMT-4)]

Room: AAPM ePoster Library

Purpose:
To develop a generic normal tissue complication probability model for predicting radiation-induced esophagitis (RE) observed in non-small cell lung cancer patients receiving either photon or proton radiotherapy.

Methods:
A total of 324 patients treated at our institution were included: 221 patients received intensity-modulated radiation therapy, 103 patients received passive-scattering proton therapy. Then endpoint of the study was grade =2 RE within 6 months from the first treatment. Multivariable logistic regression models were developed for the whole cohort (Generic-LR), photon-treated cohort (Photon-LR), and proton-treated cohort (Proton-LR), respectively. Significant predictors were selected using the LASSO penalized method. Predictive performance of the models was evaluated across treatment modality. The discriminative ability, and calibration were assessed using the area under the receiver operator curve (AUC), and Hosmer–Lemeshow test, respectively.

Results:
Grade 2 or higher RE was observed in 184 (56.79%) patients (photon: 133 (60.18%), proton: 51 (49.51%)), and no grade 4-5 was reported. The Generic-LR predicted an increased risk of grade =2 RE with Dmax, V45, and V75, showing a better performance than the cross-modality performance of Photon-LR (Predictor: Dmax, V40, V45) and Proton-LR (Predictor: Dmax, V5, V55, V75). The AUCs of Generic-LR for photon-treated and proton-treated cohorts were 0.6632 and 0.7751, whereas the AUC of Proton-LR in photon-treated cohort was 0.5741 and the AUC of Photon-LR in proton-treated cohort was 0.7351. The Hosmer-Lemeshow test of Generic-LR showed significant agreement between predicted and observed morbidity for both cohorts. The predictive performance of Generic-LR is comparable to Photon-LR in photon-treated cohort (AUC=0.6806), while is worse than Proton-LR in proton-treated cohort (AUC=0.8209).

Conclusion:
A generic NTCP model including Dmax,V45, and V75 to predict grade =2 RE in NSCLC patients showed good predictive performance both in patients receiving photon and proton radiotherapy. Proton-specific NTCP model is recommended for better prediction in proton-treated cohort.

Funding Support, Disclosures, and Conflict of Interest: The University of Texas MD Anderson Cancer Center was supported in part by the National Cancer Institute Cancer Center Support Grant P30CA016672. This study was supported in part by the National Key Research and Development Program of China (Grant 2016YFC0105409).

Keywords

Not Applicable / None Entered.

Taxonomy

Not Applicable / None Entered.

Contact Email

Additional Links