Objective To investigate the clinical value of preoperative gadolinium ethoxy-benzyldiethylenetriamine pentaacetic acid (Gd-EOB-DTPA)-enhanced magnetic resonance imaging (MRI) in predicting microvascular invasion (MVI) and intratumoral tertiary lymphoid structures (TLSs) in hepatocellular carcinoma (HCC).

Methods The retrospective cohort study was conducted. The clinicopathological data of 304 HCC patients who were admitted to The First Affiliated Hospital of Army Medical University and 10 HCC patients who were admitted to The Second Affiliated Hospital of Chongqing Medical University from June 2021 to June 2023 were collected. There were 272 males and 42 females, aged (56±11)years. Using a random number table method, patients were divided into a training set including 220 cases and a validation set including 94 cases in a 7:3 ratio. Among the 314 patients, 106 cases had MVI and TLSs-positive HCC (MT-HCC), and 208 cases had non-MT-HCC. All patients underwent preoperative Gd-EOB-DTPA-enhanced MRI and radical resection. Observation indicators: (1) clinicopathological characteristics of MT-HCC and non-MT-HCC patients; (2) imaging characteristics of MT-HCC and non-MT-HCC patients; (3) imaging features associated with MT-HCC diagnosis; (4) nomogram predictive model construction and evaluation for MT-HCC. Comparison of measurement data with normal distribution between groups was analyzed using the t test. Comparison of measurement data with skewed distribution between groups was analyzed using the nonpara-meter rank sum test. Univariate analysis was conducted using the corresponding statistical methods based on data type. Multivariate analysis was conducted using the logistic regression model. A nomo-gram predictive model was constructed based on results of multivariate analysis, and receiver operating characteristic (ROC) curves were plotted to evaluate the model's performance with the area under curve (AUC). Calibration curve and decision curve analyses were used to assess the calibration and clinical validity of nomogram predictive model.

Results (1) Clinicopathological characteristics of MT-HCC and non-MT-HCC patients. In the training set, there were significant differences between MT-HCC and non-MT-HCC patients in terms of age, white blood cell count, and alpha fetoprotein level (t=2.488, Z=-2.515, χ²=4.014, P<0.05). (2) Imaging characteristics of MT-HCC and non-MT-HCC patients. In the training set, there were significant differences in tumor morphology, intratumoral hemorrhage, peritumoral abnormal enhancement in arterial phase, capsule presence, intratumoral necrosis or ischemia >20%, intratumoral necrosis or ischemia >50%, peritumoral hypointensity in the hepatobiliary phase, intravascular tumor thrombus, arterial phase rim-like hyperenhancement, and mosaic architecture between MT-HCC and non-MT-HCC patients (χ²=8.811, 5.586, 13.962, 31.616, 10.154, 4.835, 5.111, 14.425, 7.112, 5.526, P<0.05). (3) Imaging features associated with MT-HCC diagnosis. Results of multivariate analysis identified the absence of intratumoral hemorrhage, incom-plete capsule, and mosaic architecture as independent risk factors for diagnosing MT-HCC (hazard ratio=3.846, 7.827, 2.345, P<0.05). (4) Nomogram predictive model construction and evaluation for MT-HCC. A nomogram predictive model for MT-HCC was constructed based on the independent risk factors (absence of intratumoral hemorrhage, incomplete capsule, and mosaic architecture) iden-tified in the multivariate analysis. The ROC curve analysis showed that AUC of nomogram predictive model was 0.778 (95% confidence interval as 0.714-0.843), with sensitivity and specificity of 0.857 and 0.573 in the training set. In the validation set, the area under the curve, sensitivity, and specifi-city were 0.825 (95% confidence interval as 0.745-0.926), 0.655, and 0.877, respectively. The calibra-tion curves for both the training set and the validation set closely aligned with the standard curve, indicating high calibration accuracy. The decision curve analysis demonstrated net clinical benefits at thresholds of 0.130-0.690 in the training set and 0.060-0.750 in the validation set.

Conclusions The absence of intratumoral hemorrhage, incomplete capsule, and mosaic architecture are independent risk factors for diagnosing MT-HCC. A nomogram model based on imaging features can predict MT-HCC in HCC patients.