Textural Analysis Of Baseline ¹⁸F-FDG PET For Predicting Treatment Response And Prognosis In Patients With Locally Advanced Esophageal Squamous Cell Cancer

ObjectiveTextural features on baseline ¹⁸F-FDG PET have shown the potential role in predicting treatment response in mixed stage esophageal cancer. This study is aim to investigate the value of this new technique for locally advanced esophageal squamous cell cancer（ESCC） receiving chemoradiotherapy.Materials and MethodsForty-eight patients with newly diagnosed locally advanced ESCC who treated with concurrent chemoradiotherapy were retrospectively reviewed. Thirty-nine patients with early stage ESCC were included as control. All patients underwent pretreatment whole-body ¹⁸F-FDG PET. Fifty-four texture indices describing global, local, and regional features were measured in addition to 5 conventional indices, including maximum, peak, and mean SUV, metabolic tumor volume（MTV）, and total lesion glycolysis（TLG）.Firstly, to compare the ability of the imaging parameters（including 5 conventional indices and fifty-four texture indices） to differentiate clinical stage（early stage versus locally advanced disease）.Secondly, patients were classified as responders（R, complete or partial response） and non-responders（NR, stable or progressive disease） according to the CT Response Evaluation Criteria in Solid Tumors（RECIST 1.1） at 4wk. An optimum cut point for the level of the PET indices and was calculated, maximizing the sum of sensitivity and specificity from receiver-operating-characteristic curves（ROC） curves.Thirdly, Progression-free survival（PFS） and overall survival（OS） were recorded. OS and PFS were estimated using the Kaplan-Meier analysis. All PET indices and clinical factors were calculated. To evaluate the clinical factors like sex, age, treatment, et al, associated with OS, univariate and multivariate Cox proportional hazards models were adopted.Results Both intratumor heterogeneity and mean/peak intensity of FDG uptake were significantly higher in locally advanced ESCC than those in early stage.None of the SUVs could predict treatment response. Thirty-four texture indices, MV, and TLG showed the ability to differentiate R from NR. Nine texture indices showed higher sensitivity（76.7%⁸6.7%）and specificity（77.8%⁹4.4%） than MV（76.7% and 83.3%） and TLG（73.3% and 83.3%）.None of the SUVs could predict survival. Nine texture indices and MV were hazard factors of PFS and OS. Large-zone emphasis, one of the regional texture indices,was the only independent predictor of survival, with hazard ratio of 4.22 for PFS and 3.90 for OS.Conclusion FDG PET texture indices provide predictive information for locally advanced ESCC. Therefore, its clinical application could be an important step in personalized treatment of esophageal cancer.