Determination of Prognosis in Hepatocellular Carcinoma: The Role of Adaptive Molecular Mechanisms to Microenvironmental Factors

Primary liver cancer or hepatocellular carcinoma (HCC) is the sixth most common cancer worldwide. When the diagnosis of HCC has been made, the treatment and prognosis depend on the stage of the disease. Current staging systems for HCC take into account the size of thetumor, the number of tumors, the presence or absence of macroscopic vascular invasion and distant metastases. Other features such as histology and molecular markers are currently not used in daily clinical practice. However, the gross characteristics of the tumor give only crude estimates of the prognosis, and for the individual patient it remains difficult to accurately predict the prognosis.Furthermore, the response to therapy is difficult to predict based on thesefactors. Apparently tumor size, tumor number and vascular invasion do notalways represent the aggressiveness of the tumor. Different other factors playa role in the aggressive behavior of a tumor, such as the microenvironment. Oneof the microenvironmental factors that plays a pivotal role in solid tumors ishypoxia. Different study groups have tried to identify molecularmarkers or predictive gene sets to better understand the tumor biology andbetter predict prognosis. But whole genome research faces some methodologicaldifficulties, for example studies are frequently underpowered and microarraysmeasure a considerable amount of noise. Furthermore, the results of one studyare not transposable to another. Better study design and biostatisticalanalysis could improve the performance of whole genome research.We used a mechanism driven analysis, based on the hypothesisthat chronic hypoxia is present during the outgrowth of HCC and induces anaggressive gene expression pattern. First, in an experimental setting wecharacterized a cell culture model for chronic hypoxia and assessed thedifferential gene expression in these conditions. Next, we analyzed theprognostic role of the selected chronic hypoxia related genes in differentindependent data sets of HCC patients. We identified a small 7-gene set withstrong prognostic relevance. The 7-gene set separated patients with good andpoor prognostic characteristics. And more importantly, by developing a hypoxiascore based on the expression of these 7 genes, we could predict both survivaland early recurrence in a retrospective data set. Our method has severaladvantages. By selecting only the genes with a differential expression underchronic hypoxia, the number of genes tested is more in relation to the numberof patients studied. And further, by testing the hypoxia related genes indifferent data sets, results do not longer depend on microarray technique andpatient selection. In the future, this small prognostic gene set should betested in a prospective manner, and if possible in formalin fixed paraffinembedded (FFPE) tissue. Moreover, our analytical sequence could also be used inother (gastrointestinal) malignancies. Since chronic hypoxia is of prognostic importance, a pathwayanalysis was performed in our in vitromodel of chronic hypoxia. The transforming growth factor-beta (TGF-ß) signalingand the peroxisome proliferator-activated receptor alpha (PPARα) and retinoid Xreceptor alpha (RXRα) signaling were identified as important pathways in theresponse to prolonged hypoxia. These results can be relevant for futureresearch to find new therapeutic targets. The last part of this research concerned the effects oflong-term sorafenib exposure in hepatocellular carcinoma cell lines. Sorafenibis the first targeted therapy with a significant benefit in patients withadvanced stage disease. Nevertheless, all patients under sorafenib eventuallyshow progressive disease. We showed that long-term in vitro exposure of sorafenib to HCC cell lines can induceresistance, and in HepG2 cell this is accompanied by anepithelial-to-mesenchymal transition. Cancer cells with a mesenchymal phenotypeare more aggressive, since these cells are very motile and can migrate todistant sites. Furthermore, we demonstrated that withdrawal of sorafenib canlead to rebound growth. This finding is important for clinical practice wherepatients who develop side-effects temporarily stop treatment or have dosereductions. Our model can be used for new drug development.

Publication year:2011