Purpose: This prospective study seeks to extract semiquantitative positron emission tomography (PET) features from 18F-fluorodeoxyglucose PET scans performed before and during neoadjuvant chemoradiotherapy for esophageal cancer and to compare their accuracy in predicting histopathologic response.
Methods and materials: From 2012 to 2016, 26 patients with esophageal cancer underwent pretreatment and intratreatment PET scans during chemoradiotherapy followed by surgery. Median patient age was 63 years (interquartile range, 58-68 years); 26 patients had esophageal adenocarcinoma, and 3 had esophageal squamous cell carcinoma. The intratreatment PET scan was performed at a median of 32.4 Gy (interquartile range, 30.6-32.4 Gy). PET features of the primary site including maximum standardized uptake value (SUV), SUV mean, metabolic tumor volume, and total lesion glycolysis were extracted from the pretreatment and intratreatment PET scans. Patients were histopathologic responders if there was complete or near-complete tumor response by modified Ryan scheme. Mean values of PET features were compared between histopathologic responders and nonresponders. The area under the receiver operating characteristic curve (AUC) was used to compare the accuracy of PET features in predicting histopathologic response.
Results: Eleven patients (42%) were histopathologic responders. PET features most discriminatory of histopathologic response on AUC analysis were volumetric PET features from the intratreatment PET including metabolic tumor volume based on manual contour (AUC, 0.73; 95% confidence interval, 0.52-0.93) and total lesion glycolysis based on semiautomatic 40% SUV threshold (AUC, 0.73; 95% confidence interval, 0.53-0.94).
Conclusions: Volumetric PET features from the intratreatment PET were the most accurate predictors of histopathologic response.
Copyright © 2018. Published by Elsevier Inc.