Studies in patients undergoing liver resection for hepatocellular cancer

Abstract

INTRODUCTION Hepatocellular cancer (HCC) is a leading cause of mortality and liver resection is the main treatment in suitable patients with operable HCC. While considerable research has been performed in identifying prognostic factors and molecular markers associated with poorer survival little of this research has originated in Australia. METHODS Clinicopathological and survival data was retrospectively collected from patients who underwent liver resection for HCC at Royal North Shore/North Shore Private Hospitals (North Shore) and Westmead Hospital (Westmead) from 1998 to 2012 (n=125). A survival analysis was performed and prognostic factors were identified. Subgroup prognostic factor and molecular marker research on North Shore patients was also performed using tissue proteomics (n=30) and immunohistochemistry (IHC; n=34, n=30). RESULTS In the combined group of 125 patients the 5-year overall survival rate was 56%. On multivariate analysis, only microvascular invasion (MVI) was independently associated with a poorer prognosis on both overall and disease-free survival. Further analysis revealed that age > 64, serum alpha-fetoprotein (AFP) and tumour size > 50 mm were independent preoperative risk factors for MVI and these were used to create an MVI prediction model (area under the curve (AUC) 0.721 or 0.724 depending of AFP cut-off value). Subgroup proteomic analysis identified a panel of four proteins that were differentially expressed between HCC and adjacent normal liver tissue (AUC 0.954). The decreased expression of one of these proteins -- m/z 9,961 -- correlated significantly with the presence of MVI. Subgroup IHC analysis identified several varied clinicopathological correlations with the panel of selected molecular markers (Ki-67, p53, MSI, CK19, GPC3 and S100A6). DISCUSSION In conclusion, our overall survival rate was comparable to the international literature and MVI was the main independent predictor of poor prognosis in our patients. Though our preoperative MVI prediction model did not achieve a clinically applicable AUC, decreased expression of protein m/z 9,961 significantly correlated with the presence of MVI in the proteomics subgroup. This protein should be identified as it could play an important role in liver carcinogenesis. Finally, the clinicopathological correlations in the IHC subgroup were of varied value and additional work could be considered with larger sample sizes.