Liver Cancer Biomarkers Ready

Primary liver cancer or hepatocellular carcinoma (HCC) is one of the most common and lethal malignancies in Hong Kong and worldwide. It accounts for approximately half a million annual deaths globally. According to Global Cancer Statistics (2002), HCC ranks sixth as the most prevalent cancers. It accounts for 5.7% of all human cancer cases.

This clinical situation is dismal in developing countries, where most of the new cases arise, and about 55% of all HCC incidences are identified in China, according to CA Cancer J Clin 55 (2005). Although current treatment modalities such as liver transplantation (<5%), local ablative therapies (20%), hepatic resection (20%), transarterial therapies (25%) and systemic therapies (>30%) are available for afflicted patients, there are no definite and effective curative therapies to eliminate this malignant neoplasm as reported in Best Pract Res Clin Gastroenterol 19 (2005) and J Gastroenterol 40 (2005). Thus, a report in CA Cancer J Clin 55 says the prognosis of HCC patients remains pessimistic with the overall five-year survival rate of them remains at approximately 10-20% over these years.

Liver cancer is often asymptomatic at the early and most curable stages and so patients are usually diagnosed at the very advanced stages at presentation when effective treatments are unavailing. Therefore, early detection of HCC is vital for afflicted patients in order for them to receive therapeutic benefits from curative surgery. Today, there is no effective biomarker for early detection of HCC or monitoring its recurrence. The current standard diagnosis of HCC relies on detection of the serum alpha-fetoprotein (AFP) level in at-risk subjects (hepatitis carriers and cirrhosis) followed by hepatic ultrasonography to identify suspicious nodules. Reports in Biomarkers Med 1 (2007), Liver Int 27 (2007) and Oncogene 25 (2006) say the AFP marker is well-received for its low sensitivity (64.8% - 78%) and specificity (50% - 93%), while the diagnostic imaging method is highly operator-dependent and often subjected to high false-negative rate particularly when the tumor nodule is <2 cm in diameter. To this end, reliable and accurate biomarkers are urgently needed to overcome the shortcomings of the current methodology in HCC diagnosis. Our team is at the forefront in the search for HCC biomarkers, and the approach is illustrated in Figure 1.
Clinical analysis
In the biomarker discovery phase, we recruited a Chinese cohort of 120 HCC patients and 20 normal subjects for prioritizing the newly identified putative targets. The ideal markers should be highly expressed in the tumor (T) tissues but not the adjacent non-tumor (NT) or normal (N) liver tissues which should also be present in the patient serum samples. The same set of clinical tissues was analyzed by genomic and proteomic profiling using the cDNA microarray and proteomic 2-dimensiontal gel (2-DE)-based approach respectively. The experimental workflow is summarized in Figure 2. As published in Biomarkers Med 1 (2007) and Oncogene 25 (2006), the resulting data were analyzed statistically to identify differentially expressed genes and proteins. Data-mining algorithms were also employed to exploit information derived from the datasets.

In addition, clinical correlation of each putative biomarker was analyzed with certain defined pathological data in relation to the demographic and prognostic features such as tumor size, differentiation, staging, vascular invasion, recurrence and survival rate. The experimental findings have been confirmed by a separate set of clinical samples using molecular assays quantitative PCR, immunohistochemistry and western blot. It is also imperative to validate the diagnostic utility of a candidate biomarker or biomarker signature in prospective, multicenter clinical trial studies, preferably from different geographical locations.

Data from our laboratories have indicated a panel of genetic and protein biomarkers for HCC, and some of them have shown to denote the onset and development of liver malignancy. For instance, the molecular profiling analysis by cDNA microarray based on about 6,000 genes could clearly distinguish tumorous phenotypes from non-malignant phenotypes in about 200 samples, as published in Mol Biol Cell 13 (2002). In general, the up-regulated gene clusters were largely related to the cell cycle family members, whereas the down-regulated genes to the homeostatic functions and metabolisms. Of particular interest, Clin Cancer Res 10 (2004) reports that over-expression of granulin-epithelin precursor (GEP) gene was associated with tumor invasion and metastasis as demonstrated by both clinical correlation analysis and in vitro assays.

Serum samples from HCC patients also revealed elevation of GEP protein levels, and thus a dual monoclonal antibodies-based sandwich immunoassay for detection of serum GEP has been developed (US Patent filed). Furthermore, two biomarker transcripts, AA454543 (reported in Neoplasia 7 (2005)) and claudin-10 (reported in Clin Cancer Res 11 (2005)), were also detectable in plasma samples by quantitative PCR. These two transcripts are independent prognostic markers for poor prognosis of HCC. These findings were further evidenced by the presence of MXR7 (Glyplican-3, a fetal protein in liver) transcript in plasma samples from HCC patients.

Meanwhile, we have employed the latest cutting-edge technology using SELDI-ToF and mass spectrometry-based proteomic methods to search for specific protein biomarkers in liver cancer. In brief, we have collected sufficient data to identify specific molecules or biomarker signatures that enable us to discriminate between malignant tumor liver tissue and non-malignant liver tissue, detect small-sized HCC tumors (<2cm), and
predict early tumor recurrence after curative surgery.

Two data-mining methods, namely artificial neural network (ANN) and classification and regression tree (CART), have been used to analyze proteomic profiling data and to provide ways and means to discriminate phenotypes of unidentified liver tissues.

In a cross-examination using a separate dataset, the proteomic profiling method in combination with ANN and CART allowed us to detect HCC at (96.97% and 81.82%) sensitivities and (87.88% and 78.79%) specificities respectively. From Biochem Biophys Res Commun 361 (2007), the three principal classifiers were identified as cytochrome b5, heat shock 70 kDa protein 8 isoform 2, and cathepsin B. Furthermore, when comparing tumors with the adjacent non-tumor and normal liver tissues, it was found that expression of certain heat shock proteins (HSP) or chaperones was remarkably elevated the HCC These proteins included HSP27, HSP70, HSP90 and glucose-regulated protein (GRP) 75 and GRP78. Aberrant expression of HSP in HCC was found associated with certain pathologic features (reported in Proteomics 6 (2006)). In this context, mortalin (HSP9A) was shown to be up-regulated in a subgroup of HCC with early tumor recurrence after curative surgery, and associated with vascular invasion and short disease-free survival time (reported in Mol Cell Proteomics (2007)). These HSP markers are useful for prognosis of HCC and for prediction of disease relapse. The ability to detect HCC at its early stage and to predict cancer recurrence could save many lives and enhance the quality of life of cancer patients.

Conclusion
Taken together, the current paradigm is that a panel of biomarkers, either in mRNA transcripts, protein forms, or signature patterns, would be more feasible for future application in the diagnosis of early HCC as well as for better prognostic prediction. With the advent of the latest genomic technology and new proteomic methods such as Multiple Reaction Monitoring (MRM) assay in combination with Stable Isotope Standards with Capture by Anti-Peptide Antibodies (SISCAPA) offering high throughput detection of low-abundance proteins in plasma, we are optimistic that the newly identified biomarkers will soon be ready for prospective, multicenter clinical trial studies.
Figure 1
Figure 2
Figure 3
Figure 4