Preoperative prediction of early recurrence in hepatocellular carcinoma using simultaneous multislice diffusion kurtosis imaging
摘要
This study aimed to evaluate the value of simultaneous multislice diffusion kurtosis imaging (SMS-DKI) for predicting early recurrence (within 2 years) in hepatocellular carcinoma (HCC) and to develop a predictive model.
Materials and methodsThis prospective study included 67 HCC patients who underwent SMS-DKI on a 3-T MRI between June 2021 and January 2023. Diffusion parameters, including the apparent diffusion coefficient (ADC), SMS-mean kurtosis (SMS-MK), and SMS-mean diffusivity (SMS-MD), along with radiological features, were analyzed. Logistic regression models were used to predict early recurrence, internally validated using 10-fold cross-validation, and assessed using AUC, calibration curves, and decision curve analysis (DCA).
ResultsAmong 67 patients (58 males; mean age, 53.5 ± 9.9 years), 30 (44.8%) experienced early recurrence. The early recurrence had significantly lower ADC (1.12 vs 1.22 × 10−3 mm2/s) and SMS-MD (1.45 vs 1.70 × 10−3 mm2/s), and higher SMS-MK (0.91 vs 0.75). SMS-MK showed the highest AUC (0.90, 95% CI: 0.80–0.96). Multivariate analysis identified SMS-MK (OR = 3.43 [1.31–8.89]), tumor size (OR = 4.22 [1.58–7.76]), non-smooth tumor margin (OR = 2.68 [1.58–7.96]), and complete capsule (OR = 0.22 [0.02–0.79]) as independent predictors of early recurrence. Based on these four parameters, the final model achieved an AUC of 0.94 (95% CI: 0.88–1.00). Calibration curves and DCA confirmed clinical utility.
ConclusionSMS-DKI enhances early recurrence prediction in HCC. The predictive model, incorporating SMS-MK, tumor size, and key radiological features, demonstrated good prognostic value.
Key Points