Usefulness of dual-layer detector spectral CT multi-parameter imaging in preoperative prediction of lymphovascular invasion and perineural invasion in patients with colorectal cancer

Abstract Background We aimed to investigate the usefulness of combining quantitative parameters obtained with dual-layer detector spectral CT (DLSCT) and clinical risk factors for preoperative prediction of lymphovascular invasion (LVI) and perineural invasion (PNI) in patients with colorectal cancer (CRC). Materials and methods From June 2022 to August 2023, 122 patients with clinically suspected CRC were prospectively enrolled in the study for spectral CT scanning, 49 patients were confirmed with CRC by histopathology. Using the pathological results, the patients were divided into LVI-positive and LVI-negative groups and PNI-positive and PNI-negative groups, and their clinical and imaging characteristics were determined. The spectral parameters of arterial-phase (AP) and venous-phase (VP) images in the CRC patients analyzed in this study. Intergroup comparisons of parameters were performed using the independent-sample t-test, Mann–Whitney U test, and chi-square test. Independent predictors of LVI group and PNI group were determined by univariate and multifactorial logistic regression analysis, and single parameter and combined parameter models were constructed accordingly. Receiver operating characteristic curve (ROC) analysis was performed to evaluate the prediction effect of different models. Results Tumor maximum diameter (Tdia) and normalized iodine density (NID) in the AP (NIDAP) were independent predictors of LVI (P < 0.05), while Tdia, carcinoembryonic antigen 19 − 9 (CA19-9) level, and NID at the VP (NIDVP) were independent predictors of PNI (P < 0.05) in CRC patients. The area under the curve (AUC) values of Tdia and NIDAP for predicting LVI status in CRC patients were 0.795 and 0.776, respectively. Similarly, the AUC values of Tdia, CA19-9 level, and NIDVP for predicting PNI status among CRC patients were 0.804, 0.701, and 0.735, respectively. Models combining these independent predictors yielded AUC values of 0.899 (95% confidence interval [CI]: 0.781–0.966) and 0.871 (95% CI: 0.771–0.971) for predicting LVI and PNI status, respectively. Thus, the combined model was significantly better than any single independent predictor alone. Conclusion The combined models, which integrated quantitative DLSCT and clinical parameters, demonstrated good predictive capability for determining the LVI and PNI status among CRC patients, in order to provide imaging references for clinical treatment decision-making.