Ki-67 expression correlates with hepatocellular carcinoma recurrence and is predictable using radiomics features.

Area of Science:

• Hepatocellular Carcinoma (HCC) Research
• Radiomics and Medical Imaging
• Tumor Biology and Biomarkers

Background:

• Hepatocellular carcinoma (HCC) recurrence post-resection remains a significant clinical challenge.
• Accurate preoperative prediction of tumor aggressiveness, particularly Ki-67 expression, is crucial for treatment planning.
• Current methods for Ki-67 assessment are invasive and may not reflect the overall tumor heterogeneity.

Purpose of the Study:

• To determine the optimal Ki-67 expression cutoff for predicting HCC recurrence.
• To develop and validate a radiomics model using CT arterial phase imaging for non-invasive preoperative Ki-67 assessment.
• To compare the predictive performance of the radiomics model against clinical factors and combined models.

Main Methods:

• Analysis of 180 HCC patients undergoing radical resection.
• Multivariate Cox regression and ROC curve analysis to identify Ki-67 prognostic value and optimal cutoff.
• Extraction of radiomics features from intratumoral and peritumoral regions on CT arterial phase images.
• Development of a radiomics prediction model using the random forest algorithm and SHAP analysis for feature importance.

Main Results:

• High Ki-67 expression (≥25%) was an independent predictor of postoperative HCC recurrence (HR=9.923, P<0.001).
• The radiomics model achieved an AUC of 0.869 for predicting Ki-67 expression, outperforming the clinical model (AUC=0.540, P=0.005).
• Peritumoral texture heterogeneity (peri_wavelet_LLH_firstorder_Kurtosis) was the most significant feature in the radiomics model (SHAP value=+0.17).

Conclusions:

• High Ki-67 expression (≥25%) is a critical prognostic factor for HCC recurrence.
• CT arterial phase radiomics offers a promising non-invasive approach for preoperative Ki-67 assessment.
• Peritumoral heterogeneity plays a key role in predicting HCC recurrence risk via radiomics.