Click here to


Are you sure ?

Yes, do it No, cancel

A Method To Identify Genetic Signature For Radiosensitivity Of Esophagus And To Model Esophagitis

H Yao1*, W Wang1 , N Bi2 , S Jolly3 , P Fu1 , J Jin1,4 , F Kong1 , (1) Case Western Reserve University, Cleveland, OH, (2) National Cancer Center/Cancer Hospital, Chinese Academy of Medical Sciences, Beijing, (3) University of Michigan, Ann Arbor, MI, (4) University Hospitals, Case Medical Center, Cleveland, OH


(Sunday, 7/14/2019)  

Room: ePoster Forums

Purpose: To identify single nucleotide polymorphisms (SNPs) signatures for radiosensitivity of esophagus and to model radiation induced esophagitis (ES).

Methods: A total of 178 patients with 71 SNPs were included in this study. The end point was set to be ES grade ≥2. The dose-volume-histogram of esophagus was converted to gEUD for each patient. A logistic NTCP model was used to determine the optimal gEUD and the mean radiosensitivity represented by TD50, the dose having 50% risk of ES. The receiver operating characteristic (ROC) curve with the area under the curve (AUC) was used to assess model accuracy. Combined with a SNPs look-up table from the distribution of number of patients with/without esophagitis and applied with Bayes’ theorem, the patient-specific TD50 (TD50-PS) was generated for individual patient. A model accuracy criterion in combination of the �2 test was used to identify candidate SNPs to form the SNP signatures.

Results: 34% of patients developed the esophagitis with grade ≥2. gEUD with a=20 for α/β=10 (gEUD[a=20, α/β=10]) appeared to be the optimal parameter, with TD50 = 63 Gy. Using different model accuracy criteria, we have identified a 4-SNP signature and a 13-SNP signature. The accuracy and AUC to predict the risk of ES were 0.69 and 0.74, 0.74 and 0.79, 0.77 and 0.85 respectively, for using gEUD[a=20, α/β=10] only, 4-SNP signature and 13-SNP signature.

Conclusion: We have developed a method to identify SNPs and SNP signatures that contribute to the radiosensitivity of esophagus. The individual radiosensitivity represented by TD50-ps can be determined. These SNP signatures may not best represent the true genetic signature for radiosensitivity due to limited numbers of patients and SNPs used in this study. The SNP signatures and TD50-ps can be continuously updated when more data are added.


Bayesian Statistics, Radiosensitivity, NTCP


TH- Dataset analysis/biomathematics: Informatics

Contact Email