Published OnlineFirst December 16, 2016; DOI: 10.1158/1078-0432.CCR-16-1500 Personalized Medicine and Imaging Clinical Cancer Research A Glycomics-Based Test Predicts the Development of Hepatocellular Carcinoma in Cirrhosis Xavier Verhelst1,2, Dieter Vanderschaeghe3,4, Laurent Castera 5,6, Tom Raes3,4, Anja Geerts1,2, Claire Francoz5,6, Roos Colman7,Francois¸ Durand5,6, Nico Callewaert3,4, and Hans Van Vlierberghe1,2 Abstract Purpose: Cirrhosis is a major risk factor for the development of who did not. For patients with a baseline GlycoCirrhoTest exceed- hepatocellular carcinoma but remains underdiagnosed in the ing 0.2, the HR for hepatocellular carcinoma development over compensated stage. Fibrosis progression and cirrhosis are asso- the entire study (Cox regression) was 5.1 [95% confidence interval ciated with changes in blood serum glycomic profiles. Previously, (CI), 2.2–11.7; P < 0.001], and the HR for hepatocellular carci- the serum glycomics-based GlycoCirrhoTest was shown to iden- noma development within 7 years was 12.1 (95% CI, 2.8–51.6; tify 50% to 70% of compensated cirrhosis cases in chronic liver P ¼ 0.01) based on the cut-off value optimized in the same cohort. disease cohorts, at >90% specificity. This study assessed GlycoCir- An absolute increase in GlycoCirrhoTest of 0.2 was associated rhoTest for the risk of hepatocellular carcinoma development in with an HR of 10.29 (95% CI, 3.37–31.43; P < 0.001) for compensated cirrhosis. developing hepatocellular carcinoma. In comparison, the HR Experimental Design: Serum glycomics were analyzed in sera for the development of hepatocellular carcinoma within 7 years of 133 patients, with compensated cirrhosis collected between for AFP levels above the optimal cutoff in this study (5.75 ng/mL) 1995 and 2005 in a surveillance protocol for hepatocellular was 4.65 (95% CI, 1.59–13.61). carcinoma using an optimized glycomic technology on a DNA Conclusions: This prognostic study suggests that GlycoCir- sequencer. rhoTest is a serum biomarker that identifies compensated Results: Baseline GlycoCirrhoTest values were significantly cirrhotic patients at risk for developing hepatocellular carcino- increased in patients who developed hepatocellular carcinoma ma. Screening strategies could be guided by a positive test on after a median follow-up of 6.4 years as compared with patients GlycoCirrhoTest. Clin Cancer Res; 23(11); 2750–8. Ó2016 AACR. Introduction to accurately identify the main risk groups. In this regard, it is well established that liver cirrhosis is the most important risk Hepatocellular carcinoma represents up to 85% of the pri- factor for hepatocellular carcinoma development. Indeed, in mary liver cancer burden (1). In recent years, an increasing most prevalent etiologies of chronic liver disease, the vast number of biomarkers have been proposed for the diagnosis of majority of hepatocellular carcinoma cases originate on a hepatocellular carcinoma (2). Other markers have been pro- background of cirrhosis (4–7), likely because hepatocellular posed to better assess the prognosis of the outcome of hepa- cell proliferation in the inflammatory context of cirrhotic tocellular carcinoma (3). However, to increase the effectiveness nodules provides a strongly enlarged pool of dividing hepato- of screening aimed at detecting hepatocellular carcinoma at the cytes in which mutagenesis can result in tumor formation. In early stage that is amenable to curative therapy, it is important patients with compensated cirrhosis, the annual incidence of hepatocellular carcinoma ranges from 1% to 8% (8). European 1Department of Hepatology and Gastroenterology, Ghent University Hospital, Association for the Study of the Liver (EASL) and American Ghent, Belgium. 2Laboratory of Hepatology, Ghent University, Ghent, Belgium. Association for the Study of Liver Diseases (AASLD) guidelines 3Center for Medical Biotechnology, VIB, Ghent, Belgium. 4Department of Bio- advocate systematic ultrasound (US)-based screening for hepa- chemistry and Microbiology, Ghent University, Ghent, Belgium. 5Hepatology and tocellular carcinoma in any patient with cirrhosis on the basis ^ 6 Liver Intensive Care Unit, Hopital Beaujon, Clichy, France. INSERM U773, Centre of ultrasonography every 6 months (9, 10). The aim of screen- 7 de Recherche Biomedicale Bichat Beaujon CRB3, Paris, France. Department of ing is to detect small tumors with more chance of curative Public Health, Biostatistics Unit, University of Ghent, Ghent, Belgium. therapy (11). This screening strategy in cirrhotic patients Note: Current address for D. Vanderschaeghe: NLO bvba, Technologiepark 19, showed a reduction in hepatocellular carcinoma mortality rates Zwijnaarde 9052, Belgium. (12) and it is cost-effective (13). Unfortunately, the proportion Corresponding Author: Xavier Verhelst, Ghent University Hospital, De Pintelaan of cirrhosis patients who do have screening remains low. For 185, Ghent B-9000, Belgium. Phone: 329-332-2371; Fax: 329-332-4984; E-mail: instance,inaNorthAmericancohort,lessthan20%patients [email protected] with hepatocellular carcinoma reported to have received regu- doi: 10.1158/1078-0432.CCR-16-1500 lar screening before diagnosis (14). An important reason for Ó2016 American Association for Cancer Research. this is that the compensated stage of liver cirrhosis remains 2750 Clin Cancer Res; 23(11) June 1, 2017 Downloaded from clincancerres.aacrjournals.org on September 29, 2021. © 2017 American Association for Cancer Research. Published OnlineFirst December 16, 2016; DOI: 10.1158/1078-0432.CCR-16-1500 GlycoCirrhoTest for Prediction of Hepatocellular Carcinoma Translational Relevance This prospective study demonstrates that a simple serum blood test based on the analysis of serum glycomics can discern patients with compensated cirrhosis with a high risk for development of hepatocellular carcinoma from patients with a low hepatocellular carcinoma risk. Screening strategies for patients with compensated cirrhosis could be guided by a positive test on GlycoCirrhoTest. This is the first biomarker that allows for powerful risk stratification in this population. Interestingly, this biomarker is supported by a strong patho- physiologic rationale. Furthermore, this technique for glyco- mic assessment can be easily implemented on routine capil- lary electrophoresis–based analyzers, widely available in rou- tine hospital laboratory facilities. underdetected. The current diagnosis of compensated cirrhosis is through liver biopsy in chronic liver disease patients. How- ever, biopsy is unsuited for regular patient monitoring, and a reliable and specific noninvasive biomarker that identifies the cirrhosis-characteristic hepatocyte proliferation that predis- poses to hepatocellular carcinoma development could fill this gap. We have previously shown that the GlycoCirrhoTest, a "gly- comics" biomarker based on profiling of the N-glycans from the total serum protein using capillary electrophoresis (CE), could distinguish chronic liver disease patients with compen- sated cirrhosis from those with earlier stages of fibrosis. Fur- Figure 1. thermore, GlycoCirrhoTest has been optimized for use in The glycomic analysis and GlycoCirrhoTest. A, The structures of the N-glycan peaks in the total serum of a cirrhotic patient as obtained using CE yields 13 clinical laboratories, using high-throughput DNA sequencers peaks. From left to right: Peak 1 is an agalacto, core-alpha-1,6-fucosylated or CE-based analyzers (15, 16), including those that are in use biantennary (NGA2F), peak 2 is an agalacto, core-alpha-1,6-fucosylated in clinical chemistry for routine serum protein electrophoresis bisecting biantennary (NGA2FB), peak 3 and peak 4 are single agalacto, core- (unpublished results). alpha-1,6-fucosylated biantennary structures (NG1A2F), peak 5 is the The GlycoCirrhoTest profile of patients with cirrhosis is bigalacto biantennary glycan NA2, peak 6 is the bigalacto, core-alpha-1,6- characterized by an increase in the proportion of bisecting fucosylated biantennary glycan NA2F, peak 7 is the bigalacto, core-alpha-1,6- – fucosylated bisecting biantennary glycan NA2FB, peak 8 is the triantennary N-acetylglucosamine (GlcNAc) containing N-glycans and a glycan NA3, peak 9 is the branching alpha-1,3-fucosylated triantennary glycan decrease in the proportion of triantennary N-glycans on gly- NA3Fb, peak 9 is the core-alpha-1,6-fucosylated triantennary glycan NA3Fc, coproteins in serum (Fig. 1). The enzyme N-acetylglucosami- peak 10 is the branching alpha-1,3-fucosylated and core alpha-1,6-fucosylated nyltransferase III (GnT-III) catalyzes the addition of a bisecting triantennary glycan NA3Fbc, peak 11 is a tetra-antennary (NA4), and peak 12 is GlcNAc residue in b1,4 linkage to the b-linked mannose of the a branching alpha-1,3-fucosylated tetra-antennary (NA4Fb) glycan. The b trimannosyl core structure of N-linked oligosaccharides of symbols used in the structural formulas are as follows: square, -linked GlcNAc; yellow circle, beta-linked galactose; triangle indicates alpha/beta-1,3/ glycoproteins, using UDP-GlcNAc as a donor substrate (17). 6-linked fucose; green circle, a/b-linked mannose. B, The GlycoCirrhoTest This leads to the formation of the defining sugar moiety of the profile of patients with cirrhosis is characterized by an increase in the relative GlycoCirrhoTest (Fig. 1). Neither bisecting GlcNAc residues nor expression of NA2FB, a bisecting N-acetylglucosamine containing N-glycan, GnT-III activity are detectable in a rat model in nonnodular and a decrease