Objective: The rare hepatic tumor can have a wide spectrum of radiologic features, representing a diagnostic challenge. Our purpose is to report the experience of a National Cancer Center, emphasizing the radiological features encountered and assessing the LR-M categories in the diagnostic performances for these lesions.
Patients and methods: We assessed 113 patients who underwent surgical resection or biopsy for rare liver lesions from May 2010 to December 2020. For these patients a computerized search of radiological records was performed to identify which had been studied with MRI and CT. For each lesion, the radiologists recorded the attenuation on CT studies and signal intensity (SI) in T1 weighted (W), in T2-W, DWI and in the related map of the apparent diffusion coefficient (ADC). We assessed the presence and the type of contrast enhancement (CE) during contrast study on CT and MRI and the enhancement was categorized according to LI-RADS 2018. We also assessed the presence of other features in LR-M categories (ancillary LR-M features) in order to classify different subgroups. The lesions were classified according to LR categories, and the gold standard was histological analysis.
Results: The final study population included 95 patients (46 females and 49 males), with a mean age of 51 years (range 38-83 years). 83 patients had solid lesions, 12 patients had cystic lesions (simple or complex). According to histological analysis, we categorized 79 patients with malignant lesions and 16 patients with benign lesions. According to radiological features we assessed as malignant 82 patients (79 true malignant and 3 false malignant), as benign 13 patients (all true benign). Therefore, sensitivity, specificity, positive predictive value, negative predictive value and accuracy of radiological features to identify benign and malignant lesions were 100.0%, 81.3%, 96.3%, 100.0% and 96.8%, respectively. We found no significant difference in signal and contrast enhancement appearance among all LR-M categories (p-value =0.34 at Chi square test). However, among LR-M categories the presence of satellite nodules was a feature typical of cHCC-CC (p-value < 0.05 at Chi square test). The presence of intra lesion necrosis and haemorrhage was suggestive of sarcoma (p-value < 0.05 at Chi square test).
Conclusions: High diagnostic accuracy was obtained by LI-RADS classification between malignant and benign lesion. The presence of ancillary features could help the radiologist towards a correct diagnosis.