Construction of a novel radiomics nomogram for the prediction of aggressive intrasegmental recurrence of HCC after radiofrequency ablation.

Objectives: To construct a precise prediction model of preoperative magnetic resonance imaging (MRI)-based nomogram for aggressive intrasegmental recurrence (AIR) of hepatocellular carcinoma (HCC) patients treated with radiofrequency ablation (RFA).

Methods: Among 891 patients with HCC treated by RFA, 22 patients with AIR and 36 patients without AIR (non-AIR) were finally enrolled in our study, and each patient was followed up for more than 6 months to determine the occurrence of AIR. The laboratory indicators and MRI features were compared and assessed. Preoperative contrast-enhanced T1-weighted images (CE-T1WI) were used for radiomics analysis. The selected clinical indicators and texture features were finally screened out to generate the novel prediction nomogram.

Results: Tumor shape, ADC Value, DWI signal intensity and ΔSI were selected as the independent factors of AIR by univariate and multivariate logistic regression analysis. Meanwhile, two radiomics features were selected from 396 candidate features by LASSO (P < 0.05), which were further used to calculate the Rad-score. The selected clinical factors were further integrated with the Rad-score to construct the predictive model, and the AUCs were 0.941 (95% CI: 0.876-1.000) and 0.818 (95% CI: 0.576-1.000) in the training (15 AIR and 25 non-AIR) and validation cohorts (7 AIR and 11 non-AIR), respectively. The AIR predictive model was further converted into a novel radiomics nomogram, and decision curve analysis showed good agreement.

Conclusions: The predictive nomogram integrated with clinical factors and CE-T1WI -based radiomics signature could accurately predict the occurrence of AIR after RFA, which could greatly help individualized evaluation before treatment.