Predicting infiltrative hepatocellular carcinoma and efficacy of combination treatment for conversion therapy based on pretreatment CT/MR features.

e16290 Background: Infiltrative HCC (iHCC) has been demonstrated to exhibit the highest malignant potential and poorest prognosis. Unified imaging diagnostic criteria are lacking, and a practical predictive model may help to identify iHCC and guide clinical decision-making. This study aimed to compare clinicopathological features and outcomes between iHCC and non-iHCC after resection, develop and validate a imaging-based diagnostic model for iHCC, and assess its utility in predicting response to conversion therapy. Methods: This retrospective multicenter study analyzed 1158 patients with newly diagnosed large HCC (> 5cm) from three tertiary hospitals. Patients were allocated into training/test, external validation and application cohorts. Clinical, imaging (CT/MRI assessed per LI-RADS), and pathological data (classified per Japanese Society of Hepatology criteria), overall survival (OS), recurrence-free survival (RFS) and progression-free survival (PFS) were analyzed and compared. A pretreatment CT/MRI-based predictive model for iHCC was developed and validated. Diagnostic performance was calculated and the efficacy of hepatic arterial infusion chemotherapy (HAIC) combined with targeted and immune checkpoint inhibitors in predicted iHCC and non-iHCC patients was compared in the application cohort. Results: The iHCC exhibited higher malignant features and shorter RFS after hepatectomy compared to non-iHCC. The imaging predictive model (SWIRP score) incorporated ill-defined margin, wedged-shaped peritumoral arterial phase enhancement, irregular shape, pseudo-capsule enhancement at delayed phase and rim arterial enhancement. The AUC values in training, test and validation sets were 0.890 (95% CI: 0.857–0.923), 0.851 (95% CI: 0.781–0.920) and 0.835 (95% CI: 0.767–0.902) . According to the SWIRP score, the overall objective response rates (ORR, per mRECIST) of predicted iHCC and non-iHCC in the application cohort were 82.7% and 66.7% (p = 0.044). While the median PFS in the predicted iHCC group (16.6 months, 95%CI: 12.4-20.7) was shorter (10.4 months vs. 16.6 months, p = 0.003). Conclusions: Infiltrative HCC exhibites poorer malignancy and prognosis compared to non-infiltrative types. The SWIRP score demonstrated good performance for identifying iHCCs. While combined HAIC, targeted, and immunotherapy yielded higher ORR in iHCC, it correlated with shorter PFS, suggesting the need for optimized treatment strategies in this aggressive subtype. Baseline clinicopathological characteristics in the initial cohort. Characteristics Non-iHCC (n=722) iHCC (n=158) Bilobar tumor distribution 40 (5.5%) 19 (12.0%) AFP>200ng/ml 273 (37.8%) 90 (57.0%) MVI (+) 316 (43.8%) 86 (54.4%) Satellite Nodule (+) 100 (13.9%) 33 (20.9%) Hepatic capsule invasion (+) 484 (67.0%) 134 (84.8%) Poor differentiation tumor 64 (8.9%) 27 (17.1%)