Tissue biomarkers as predictors of outcome and selection of transplant candidates with hepatocellular carcinoma

AFP, alpha-fetoprotein; Ang2, angiopoietin 2; HCC, hepatocellular carcinoma; LT, liver transplantation; miRNA, microRNA; mRNA, messenger RNA; REMARK, Reporting Recommendations for Tumor Marker Prognostic Studies; VEGF, vascular endothelial growth factor. Hepatocellular carcinoma (HCC) is a common cause of cancer deaths worldwide, and its annual incidence is rising. Liver transplantation (LT) is an accepted curative treatment for patients with tumors satisfying the Milan criteria (a single tumor ≤ 5 cm in diameter or up to 3 tumors with individual diameters ≤ 3 cm and no macrovascular invasion). These criteria predict an overall 5-year survival rate of 70% after LT.1 Since the introduction of the Milan criteria, subsequent studies have explored the expansion of transplant recipient selection to include individuals with tumors exceeding the Milan criteria.2 A recent study demonstrated an acceptable overall 5-year survival rate (71.2%) for patients who underwent transplantation for tumors that were beyond the Milan criteria but satisfied the up-to-7 rule (7 is the sum of the size of the largest tumor in centimeters and the number of tumors) in the absence of microvascular invasion.3 This approach is based on the best data available for understanding tumor behavior after LT, but it is still based on pathological data. The tumor size and the tumor number cannot be used to define subclasses of patients with better biology and better outcomes, so biomarkers are expected to be a major step forward in this setting during the next decade. Numerous molecular pathways involved in the pathogenesis of HCC have been identified. These include activation pathways that are involved in angiogenesis [vascular endothelial growth factor (VEGF)], in cell proliferation and survival (epidermal growth factor, insulin-like growth factor, and hepatocyte growth factor/Met), and in cell differentiation and proliferation (Wnt/β-catenin and hedgehog signaling). The activation of VEGF,4 Serine/threonine protein kinase Akt (Bombyx mori) [AKT],5 and met proto-oncogene (hepatocyte growth factor receptor) [MET]6 has been correlated with an aggressive phenotype and a poor prognosis after liver resection. Similarly, several gene signatures have been used to predict the outcomes of patients with HCC.7 Gene expression profiling with formalin-fixed, paraffin-embedded tissue samples from HCC resection specimens has been described and validated for the prediction of survival outcomes for patients after resection for HCC.8 This profiling technique offers the ability to perform retrospective studies with stored histological specimens. In addition, it potentially offers a practical clinical application through the ability to perform gene profiling with common formalin-fixed biopsy specimens rather than frozen tissue. Role of tissue biomarkers in the diagnosis of HCC. Role of biomarkers in the prediction of prognosis (ie, the use of gene signatures or tissue biomarkers to predict a patient's prognosis and thus aid in the extension of the Milan criteria for HCC). Role of biomarkers in the prediction of the response to molecular-targeted therapies. The diagnosis of HCC is based on pathological or noninvasive criteria.9 The pathological differentiation of dysplastic nodules (particularly high-grade nodules) from very early HCC is sometimes difficult, especially with a cirrhotic background. Few studies have tested the accuracy of the molecular diagnosis of early HCC in this setting. For instance, gene signatures have allowed molecular demarcations between low-grade dysplastic nodules, high-grade dysplastic nodules, and early HCC in both Asian10 and Western patients.11 More specifically, a 3-gene signature (including glypican 3, lymphatic vessel endothelial hyaluronan receptor 1, and survivin) has been reported to distinguish early HCC (<2 cm) from dysplastic nodules with an accuracy of approximately 90%.12 Nonetheless, this signature has not yet been externally validated. More recently, an immunohistochemistry study found the expression of glypican 3, heat shock protein 70, and glutamine synthetase to be useful in detecting well-differentiated HCC in biopsy samples,13 and this is currently being considered for HCC management guidelines.9 Patients who develop HCC with cirrhosis and undergo resection have a high rate of recurrence (approximately 70% at 5 years).2, 14 A molecular assessment of the prognosis could determine which patients with HCC would benefit from adjuvant therapy after resection or radio frequency ablation (2 curative treatments with a high risk of relapse). Moreover, it could be used to refine the group of patients who should undergo transplantation for HCC beyond the Milan criteria. Whether the risk of tumor seeding counterbalances the advantages of tissue-based molecular profiling is still an area of discussion. In a recent meta-analysis, the risk of tumor seeding after liver biopsy was 2.7% with a median time of 17 months between biopsy and seeding.15 These data also include large tumors, so the risk of complications with small, early tumors is expected to be significantly lower and thus acceptable. Biomarkers predicting a patient's prognosis or response to therapy are crucial in modern oncology. Novel prognostic biomarkers enabling tumor classification, disease state monitoring, or both could advance our efforts to realize the potential of personalized medicine in cancer.16 Besides reports on AFP levels and outcomes,17-19 recent studies have correlated various types of markers, such as gene expression, microRNAs (miRNAs), and methylation changes, with the survival of HCC patients; this topic has been reviewed elsewhere20 (see Table 1). Five markers or signatures (epithelial cell adhesion molecule [EPCAM signature], which is a hepatic stem cell marker in tumor tissue21, 22; the G3 proliferation subclass23; the expression status of the miR-26 miRNA precursor24; and 2 prognostic gene signatures in nontumor hepatic tissue8, 25) have emerged as more consistent ones. Finally, both VEGF and Ang2 were shown to have independent prognostic value in a large cohort of patients with advanced HCC.26 Although these results support the possibility of using these genetic and molecular markers as prognostic biomarkers for patients with HCC, they require external validation before they can be included in staging systems and/or incorporated into clinical management guidelines. The fractional allelic imbalance, which is used to measure chromosomal instability, has been associated with outcomes for patients with HCC and with recurrence after LT; this observation requires attention in future studies.27, 28 Similarly, data about CD90+ circulating cells may lead to a tractable supply of tissue for molecular characterization, but this is still under investigation.29 In this era of limited organ availability, better predictors of HCC recurrence are needed for selecting appropriate LT candidates whose tumors exceed the Milan criteria. The identification of a subgroup of patients whose tumors are beyond the Milan criteria but who have a favorably low risk of recurrence after transplantation offers a potential cure to those who would otherwise be excluded according to current organ allocation policies. Whether any of the aforementioned biomarkers or gene signatures can be used to identify those patients with better biological profiles needs to be elucidated in molecular studies addressing this point. Only a small study has addressed this question in a specific manner, and it found that chromosomal instability (measured with the fractional allelic imbalance) independently predicted which patients beyond the Milan criteria had a low risk of recurrence.27 Similarly, preliminary reports describing surrogates of microvascular invasion (the main predictor of HCC recurrence after LT) require independent validation in the setting of transplantation.30 Biomarkers for treatment responses are still rarities in oncology; only a few have made their way into routine clinical use. Well-defined biomarkers are believed to characterize an oncogenic addiction loop (the proposed mechanism by which a tumor cell becomes largely reliant on a single activated oncogene31) and define particular tumor subtypes that respond to specific molecular-targeted therapies. Examples of oncogenic addiction include an amplification of human epidermal growth factor receptor 2 in patients with breast cancer responding to trastuzumab,32 mutations in epidermal growth factor receptor that distinguish patients with non–small cell lung cancer responding to erlotinib,33 and v-kit Hardy-Zuckerman 4 feline sarcoma viral oncogene homolog (c-KIT)–positive gastrointestinal stromal tumors responding to the multikinase inhibitor imatinib.34 In addition, wild-type v-Ki-ras2 Kirsten rat sarcoma viral oncogene homolog (KRAS) has recently emerged as a marker of a response to cetuximab and panitumumab in patients with colorectal cancer, although the mechanism is entirely different and involves the downstream regulation of epidermal growth factor receptor signaling.35 Moreover, a new step in personalized medicine has been achieved recently with the development of a specific inhibitor of mutated V600E v-raf murine sarcoma viral oncogene homolog B1 (BRAF); this inhibitor has shown impressive clinical efficiency with few adverse events in a recent phase 2 study of melanoma.36 In the future, therefore, mapping the genetic alterations of tumors before the treatment or after treatment failure will improve the clinical care of patients with cancer.37 The use of biomarkers for HCC is somewhat more complex because HCC is a very heterogeneous disease for which oncogenic addiction loops have yet to be characterized. Initial approaches for defining a molecular classification have not yet been linked to specific treatment responses.38, 39 So far, only 1 small molecule, sorafenib, has been shown to improve the survival of HCC patients.40 Sorafenib is a multikinase inhibitor that targets a number of kinases; these kinases include VEGF receptors 2 and 3, platelet-derived growth factor receptor β, c-KIT, Ret proto-oncogene (RET), fms-related tyrosine kinase 3, and Raf kinase, efector of Ras (RAF).41 Isolated reports have described the use of sorafenib in the adjuvant setting after LT. In a companion biomarker study of the pivotal Sorafenib HCC Assessment Randomized Protocol trial, 10 serum markers and 1 tissue marker were tested, but none of them succeeded in identifying subclasses of responders.26 Nonetheless, the fast development of new biotherapies and the growing number of clinical trials for HCC are expected to lead to the use of the molecular features of tumors in defining types of treatment. In this setting, we have to reevaluate the utility of tumor biopsy for easy access to tissue and its frequency. Novel molecular data may change our approach to the diagnosis, staging, and prognosis of HCC in this decade. For prognosis assessments, recently reported prognostic gene signatures and miRNAs may be added to staging systems to complement clinical variables once they have been externally validated by independent studies. These advances in our understanding of HCC ultimately need to be transferred to clinical practice as daily tools for selecting management and treatment methods. Moreover, treatment response predictors will emerge along with novel drugs for the treatment of HCC. Positive results with sorafenib40 have opened a new era in HCC research. Future trends in drug development will pivot on the accurate assessment of genetic traits associated with human diseases on an individual basis (ie, personalized medicine). For HCC, the identification of these singularities will allow maximization of the therapeutic response through the selection of the best drug for the ideal candidate.