Esophageal wall thickness on CT scans: can it predict the T stage of primary thoracic esophageal squamous cell carcinoma?

ABSTRACT

BACKGROUND: CT is the most commonly used method to stage esophageal cancer (EC). However, the reported CT T-staging criteria for EC are controversial. PURPOSE: To determine and validate the optimal esophageal wall thickness (EWT) threshold on CT to distinguish lesions with different T stages in esophageal squamous cell carcinoma (ESCC) patients. METHODS: One thousand, one hundred-two consecutive patients with histopathologically confirmed ESCC between July 2014 and April 2020 were retrospectively reviewed. All patients underwent a preoperative CT examination and surgical treatment. The maximal EWT of the lesions on CT was measured. Patients were divided into pT1, pT2, pT3 and pT4 subgroups according to the pathologic stage. We employed the support vector machine, where linear kernels were leveraged to determine the optimal threshold to classify samples with different T stages. 90% of samples from each subgroup were randomly selected as the training set, while the remainder comprised the testing set. RESULTS: The mean EWTs of the pT1, pT2, pT3 and pT4 subgroups were 4.9 ± 2.6 mm, 8.1 ± 2.3 mm, 12.4 ± 3.6 mm, and 18.6 ± 4.4 mm, respectively. Differences in the EWT between the four subgroups or between adjacent subgroups were significant (p < 0.001), and esophageal wall became thicker with increasing pT stage. We utilized MATLAB 2020a to implement the SVM model and ran the code 10 times. The accuracy of the model was 60.29 ± 2.33%. The thresholds between samples from pT1/pT2, pT2/pT3 and pT3/pT4 lesions were 5.5 ± 0.3 mm, 10.8 ± 0.8 mm and 15.9 ± 0.5 mm, respectively. CONCLUSIONS: Possibility of predicting T stage of ESCC by EWT on CT scans was limited to 60% by model examination with large sample size.