Preoperative Diagnostic Value of Spectral CT for Predicting Perineural Invasion in Esophageal Cancer

TL;DR

This study shows that preoperative spectral CT can accurately predict perineural invasion in esophageal cancer, helping guide treatment decisions.

Contribution

The study introduces a diagnostic nomogram using spectral CT parameters to predict perineural invasion in esophageal cancer.

Findings

Spectral CT parameters like 40 keV, Zeff, and IC significantly differ between PNI-positive and PNI-negative groups.

A nomogram combining clinical and CT parameters achieved an AUC of 0.971 for predicting PNI.

40 keV had the highest predictive accuracy with an AUC of 0.943.

Abstract

To assess the diagnostic value of preoperative spectral CT quantitative parameters in predicting perineural invasion (PNI) in esophageal squamous cell carcinoma (ESCC), which is a critical prognostic factor associated with increased recurrence and poor survival. Preoperative identification of PNI can guide individualized treatment strategies.

A retrospective analysis was conducted on 78 patients with EC who underwent preoperative spectral CT. Patients were classified into PNI‐positive and ‐negative groups on the basis of histopathological findings. Spectral CT parameters, including conventional single‐energy CT value (Sect), virtual monochromatic images, effective atomic number (Zeff), and iodine concentration (IC), were compared between groups. Statistical analyses were performed through t, rank sum, and chi‐squared tests. A diagnostic nomogram was constructed by employing independent predictors and validated via receiver operating characteristic curve analysis with DeLong's test for the pairwise comparison of the area under the curve (AUC), ensuring the robust evaluation of discriminative performance.

Significant differences in spectral CT parameters were observed between the PNI‐positive and PNI‐negative groups. Specifically, the PNI‐positive group exhibited higher values of 40–70 keV, Zeff, and IC (all p < 0.05) than the PNI‐negative group. Among parameters, 40 keV demonstrated the highest predictive accuracy for PNI, with an AUC of 0.943. Binary logistic regression identified CYF, Sect, and 40 keV as independent predictors of PNI status. A nomogram incorporating these variables achieved the optimal diagnostic performance with an AUC of 0.971.

Preoperative spectral CT quantitative parameters, particularly 40 keV, Zeff, and IC, provide valuable insights for assessing PNI in ESCC. The integration of spectral CT parameters with clinical features can significantly improve the accuracy of PNI diagnosis.

Figures