Computed tomographic features of hepatocellular carcinoma predict long-term survival after hepatic resection

Aim: To evaluate the potential of computed tomographic features of hepatocellular carcinoma as prognostic factors. Methods: Medical records for 112 patients who had undergone a partial hepatectomy for hepatocellular carcinoma were retrospectively analyzed. The largest hepatic tumour in each patient was classified by pre-operative computed tomographic features as lobular configurations with indentations showing an acute angle, or non-lobular configuration without such indentations. Results: Twenty-six tumours were lobular and 86 were non-lobular. The outcome after hepatectomy was significantly worse in patients with lobular tumours (cumulative 5-year survival rate of 19.9%) than with non-lobular ones (that of 75.2%) (P<0.001). Coxʹs proportional hazards model showed computed tomographic features (P=0.0025), cirrhosis (P=0.0033), and tumour size (P=0.0412) to be independent prognostic factors. A lobular configuration was associated with satellite nodules (P<0.001), portal vein invasion (P=0.021), and extrahepatic tumour relapse (P=0.006). Conclusions: Computed tomographic features represent a strong prognostic factor in patients undergoing partial hepatectomy for hepatocellular carcinoma, and are likely to accurately reflect the tumour biology. Configuration, size of the hepatic tumour and presence of cirrhosis are the most important prognostic imaging findings in these patients.