AUTHOR=Li Kaixin , Ni XiaoLei , Lin Duanyu , Li Jiancheng 

TITLE=Incorporation of PET Metabolic Parameters With Clinical Features Into a Predictive Model for Radiotherapy-Related Esophageal Fistula in Esophageal Squamous Cell Carcinoma

JOURNAL=Frontiers in Oncology

VOLUME=Volume 12 - 2022

YEAR=2022

URL=https://www.frontiersin.org/journals/oncology/articles/10.3389/fonc.2022.812707

DOI=10.3389/fonc.2022.812707

ISSN=2234-943X

ABSTRACT=Purpose: To determine whether the addition of metabolic parameters from fluorin-18-fluorodeoxyglucose positron emission tomography/computed tomography (18F-FDG PET/CT) scans to clinical factors could improve risk prediction model for radiotherapy-related esophageal fistula (EF) in esophageal squamous cell carcinoma (ESCC).
Methods and Materials: Anonymized data from 185 ESCC patients (20 radiotherapy-related EF positive cases) were collected including pre-therapy PET/CT scans and EF status. 29 clinical features and 15 metabolic parameters from PET/CT were included in analysis, least absolute shrinkage and selection operator (LASSO) logistic regression model was used to construct a risk score (RS) system. The predictive capabilities of the models were compared using receiver operating characteristic (ROC) curves.  
Results: In univariate analysis, the MTV_40% is a risk factor for RT-related EF, with OR 1.036 (95%CI: 1.009-1.063, p=0.007). However, it was screened out from the predictive model by multivariate logistic regression. Predictive models were built based on clinical features in training cohort. The model includes diabetes, tumor length and thickness, adjuvant chemotherapy, eosinophil, and monocyte-to-lymphocyte ratio, RS= 0.2832-(7.1369×Diabetes) + (1.4304 × Tumor length) + (2.1409 ×Tumor thickness) - (8.3967 × ACT)-(28.7671×Eosinophil) + (8.2213×MLR). The cut-off of RS is -1.415, with an area under curve (AUC) of 0.977 (95%CI: 0.9536-1), a specificity of 0.929 and a sensitivity of 1. Analysis in testing cohort shown lower AUC of 0.795 (95%CI: 0.577-1), a specificity of 0.925 and a sensitivity of 0.714. Delong’s test for two correlated ROC curves showed no significant difference between training set and testing set (p=0.109). 
Conclusions: MTV_40% is a risk factor for RT-related EF in univariate analysis, was screen out by multivariate logistic regression. A model with clinical features could perform the prediction of RT-related EF.