Serum Protein N-Glycans Profiling for the ... - ACS Publications

The aim was to evaluate the use of blood serum N-glycan fingerprinting as a tool for differential diagnosis of nonalcoholic steatohepatitis from steat...
1 downloads 0 Views 603KB Size
Serum Protein N-Glycans Profiling for the Discovery of Potential Biomarkers for Nonalcoholic Steatohepatitis Cuiying Chen,*,†,‡,# Hemda Schmilovitz-Weiss,|,⊥,# Xue-en Liu,†,‡,§ Orit Pappo,∇ Marisa Halpern,O Jaqueline Sulkes,[ Marius Braun,| Maya Cohen,¶ Nir Barak,| Ran Tur-Kaspa,| Valerie Vanhooren,†,‡ Hans Van Vlierberghe,+ Claude Libert,†,‡ Roland Contreras,†,‡ and Ziv Ben-Ari| Department for Molecular Biomedical Research, VIB, Ghent, Belgium, Department of Molecular Biology, Ghent University, Ghent, Belgium, Department of Microbiology, Peking University Health Science Center, Beijing, People’s Republic of China, Liver Institute and Department of Medicine D, Beilinson Hospital, Gastroenterology Unit, Hasharon-Golda Campus, Rabin Medical Center, Histopathology Department, Beilinson Hospital, Hasharon-Golda Campus, Epidemiology Unit, Radiology Department, Beilinson Hospital, Rabin Medical Center, Petah-Tiqwa, and Sackler School of Medicine, Tel-Aviv University, Tel-Aviv, Israel, and Department of Gastroenterology and Hepatology, Ghent University Hospital, Ghent, Belgium Received August 19, 2008

The hepatic histology in nonalcoholic fatty liver disease can vary from isolated hepatic steatosis to steatohepatitis can progress to cirrhosis and liver-related death. The aim was to evaluate the use of blood serum N-glycan fingerprinting as a tool for differential diagnosis of nonalcoholic steatohepatitis from steatosis. A group of 47 patients with NAFLD was diagnosed by clinical laboratory analysis and ultrasonography, and was studied histologically using the Brunt’s scoring system. The control group included 13 healthy individuals. N-glycan profiles of serum proteins were determined by DNA sequencerbased carbohydrate analytical profiling. We have found that the concentrations of two glycans (NGA2F and NA2) and their logarithm ratio of NGA2F versus NA2 (named GlycoNashTest) were associated with the degree of NASH-related fibrosis, but had no correlation with the grade of inflammation nor steatosis severity. When used to screen NAFLD patients, GlycoNashTest could identify advanced NASHrelated fibrosis (F3-F4) with the diagnosis sensitivity of 89.5% and specificity of 71.4%. The serum N-glycan profile is a promising noninvasive method for detecting NASH or NASH-related fibrosis in NAFLD patients, which could be a valuable supplement to other markers currently used in diagnosis of NASH. Keywords: NASH • NAFLD • liver fibrosis • noninvasive • fatty liver • N-glycan • glycomics

1. Introduction Nonalcoholic steatohepatitis (NASH) is a significant form of chronic liver disease in adults and children.1 Nonalcoholic fatty liver disease (NAFLD) describes a range of conditions that affect people who drink little or no alcohol but suffer from all of the classic histological lesions of alcoholic liver diseases.2 Although asymptomatic in most case, NAFLD can progress to cirrhosis, * To whom correspondence should be addressed. E-mail, [email protected] or [email protected]; tel, +32 93313702; fax, +32 93313609. † Department for Molecular Biomedical Research, VIB. ‡ Department of Molecular Biology, Ghent University. # C.C. and H.S.-W. contributed equally to this work. | Liver Institute and Department of Medicine D, Beilinson Hospital. ⊥ Gastroenterology Unit, Hasharon-Golda Campus, Rabin Medical Center. § Peking University Health Science Center. ∇ Histopathology Department, Beilinson Hospital. O Hasharon-Golda Campus. [ Epidemiology Unit. ¶ Radiology Department, Beilinson Hospital, Rabin Medical Center and Tel-Aviv University. + Ghent University Hospital. 10.1021/pr800656e CCC: $40.75

 2009 American Chemical Society

which causes progressive to hepatocellular carcinoma. The mildest type is simple fatty liver (steatosis or NAFL), an accumulation of fat within the liver that usually does not cause liver damage. A potentially more serious type, nonalcoholic steatohepatitis (NASH) with or without fibrosis and cirrhosis, is associated with liver-damaging inflammation.3,4 NASH has become a serious public health problem, because the incidence of NASH has risen with the increase of overweight and obesity world widely.5 Many studies have revealed the significant correlation with obesity and insulin resistance, and between hepatic steatosis, cardiovascular disease and increased intimamedia thickness.6,7 Previous epidemiologic evidence has suggested that about 20% of adults have NAFLD and 3-4% of adults have NASH, with a third of those or 1% of adults having fibrosis and thus being at risk for progressing to cirrhosis.8 The only way to distinguish NASH from other forms of fatty liver disease is with a liver biopsy, which remains a gold standard for the evaluation of liver histology.9 Unfortunately, because it is invasive, it is associated with discomfort, pain and some risk. It is therefore not suitable for screening of a high Journal of Proteome Research 2009, 8, 463–470 463 Published on Web 01/13/2009

research articles

Chen et al.

Table 1. Characteristics of the Patients with NAFLD case group, case number NAFL (n ) 9) NASH (n ) 38)

gender, female, n (%)

20 (52.6%)

age, year BMI (kg/m2) ALT (IU/L) AST (IU/L) Alkaline phosphatase (IU/L) GGT (IU/L) total bilirubin (mg/dl) total serum protein (g/dL) Albumin (g/dL) Globulin (g/dL) Cholesterol (mg/dL) Triglycerides (mg/dL) HDL (mg/dL) LDH (mg/dL)

1 (11.1%)

8 (61.5%)

mean

(SD

mean

(SD

mean

(SD

52.58 29.78 86.24 59.13 97.73 93.61 1.11 7.40 4.13 3.25 184.64 132.31 45.03 109.80

12.01 4.18 68.41 36.46 40.76 92.42 1.00 0.61 0.62 0.52 40.29 60.73 12.51 37.47

46.44 28.24 75.44 48.56 129.78 231.11 1.04 7.66 4.51 3.11 198.71 162.71 60.29 91.75

10.63 2.38 51.63 30.16 60.88 167.56 0.52 0.53 0.35 0.45 31.68 56.87 46.51 37.65

57.23 25.05 19.62 21.23 69.77 25.92 0.68 7.42 4.41 3.02 202.85 112.23 54.92 139.85

9.34 2.22 4.72 6.06 20.44 28.11 0.25 0.60 0.26 0.45 28.48 31.99 7.14 31.49

risk population and evaluating every potential NAFLD patient. The limitations of a liver biopsy prompted researchers to develop new laboratory tests and imaging techniques for diagnosis of NAFLD. Radiological imaging of the liver by sonography, computed tomography (CT) and magnetic resonance imaging (MRI), used either singly or in combination, have an adequate threshold for detection of fatty infiltration of the liver, but none of them can distinguish steatohepatitis from steatosis.10,11 Another noninvasive, ultrasound-based approach is to estimate hepatic fibrosis by assessing elasticity using an elastographic device called FibroScan.12 However, it still needs to be evaluated in large-scale studies in patients with NAFLD. The measurement of serum ALT levels has been commonly used as a screening test to diagnose presumed NAFLD on a population-wide basis.10 However, this test’s specificity, sensitivity, and predictive value leave room for improvement.13 For example, serum ALT levels may be completely normal in patients with advanced steatohepatitis or even with cirrhosis. Moreover, the degree of ALT elevation does not correlate with the extent of hepatic damage. Given the increasing importance of NAFLD, it is critical that researchers identify clinical factors that can be used to reliably predict which patients have benign steatosis and which are at risk for hepatocellular injury and progressive fibrosis. A number of reports have isolated clinical predictors on the initial diagnostic biopsy. Among these, age greater than 40-50 years, obesity, type 2 diabetes mellitus and hypertriglyceridemia are the most reliable. Model based on clinical and laboratory test have been developed as alternative to liver biopsy in the Poynard’s group, a noninvasive biochemical marker (NashTest) for the prediction of NASH in patients with nonalcoholic fatty liver disease.14 The NashTest was developed using algorithms that combine 13 parameters. However, it is not known which factors could clinically differentiate NASH from simple steatosis in patients with NAFLD. Other improved serological markers, whether used alone or with others, are needed to distinguish NASH from steatosis. Our present study aimed to evaluate the use of blood serum N-glycan fingerprinting as a tool for differential diagnosis of nonalcoholic steatohepatitis from steatosis. 464

control (n ) 13)

Journal of Proteome Research • Vol. 8, No. 2, 2009

2. Patients and Methods 2.1. Patients. Forty-seven patients with clinical, ultrasonographic and histologic findings compatible with a diagnosis of NAFLD were enrolled in this study.11 Nineteen percent of the patients were in the stage of NAFL (n ) 9) and 81.9% of the patients had developed NASH (n ) 38). Routine clinical, laboratory, radiographic and histological examinations were used to exclude patients with a history of ethanol consumption (>20 g/daily) or other concomitant causes of liver disease, such as viral hepatitis B or C, autoimmune hepatitis and metabolic or hereditary liver disease. The control group consisted of 13 healthy individuals matched for age and gender, and with normal liver function tests and liver ultrasound examination. The study protocol was approved and supervised by the Scientific Committee of the Rabin Medical Center (Petah-Tiqva, Israel), and all subjects gave their written informed consent to participate in the study. 2.2. Laboratory Tests. Clinical and laboratory data were collected on the day the liver biopsy was taken. A complete medical history was obtained from all patients and control individuals, and they were subjected to physical examination. Weight, height and waist circumference were measured, and body mass index (BMI) was calculated as weight /height2 (kg/ m2). Diabetes mellitus was defined using the American Diabetes Association criteria. Laboratory evaluation included complete blood count; fasting blood glucose; total-, LDL-, and HDLcholesterol; and total triglycerides. Liver function was evaluated by measuring serum levels of enzymes and other substances produced by the liver, namely, alanine aminotransferase (ALT), aspartate aminotransferase (AST), total bilirubin, albumin, total serum protein, alkaline phosphatase and γ-glutamyltransferase (GGT). The main clinical and biological data of the patients are summarized in Table 1. Mean age was 51.4 ( 11.9 years; 55.3% were female. Mean BMI was 29.5 ( 3.9 kg/m2 with 61.7% of the patients >28 kg/m2. Triglyceride level was more than 150 mg/mL in 32.6%. Type 2 diabetes mellitus had been diagnosed in 19.4% of the patients. The level of ALT, the ALT/AST ratio, alkaline phosphatase and GGT were similar in NASH and NAFL patients.

research articles

N-Glycan Marker for Diagnosis of NASH Table 2. Histological Characteristics According to Brunt’s Scores NASH (%)

0 1 2 3 4 0-0.5 1 2 3 0 1 2 3

NAFL (%)

Fibrosis Stage 4.3% 21.3% 17.0% 17.0% 21.3%

14.9% 4.3% 0 0 0

Steatosis Severity 2.3% 31.0% 28.5% 19.0%

9.5% 7.1% 2.1% 0

Inflammation Severity 0 31% 35.7% 14.3%

2.3% 11.9% 2.3% 2.3%

2.3. Clinical Stage and NAFLD Stage. The histological findings are shown in Table 2. Liver biopsies were examined by one liver pathologist who was unaware of the patient’s clinical and laboratory data. The diagnosis of NAFLD is based on two criteria: (1) establishing the presence of a fatty liver or steatohepatitis, and (2) excluding ethanol as an etiological factor.15 NASH from other forms of fatty liver disease was diagnosed as steatosis with lobular inflammation and either ballooning of hepatocytes or abnormal (stage 1-4) fibrosis. The degree of fibrosis was staged according to the 4-point scale using the Brunt’ scoring system:16 stage 0 ) normal connective tissue; stage 1 ) perivenular/pericellular fibrosis in zone 3; stage 2 ) perivenular or pericellular fibrosis confined to zone 3 plus periportal fibrosis; stage 3 ) bridging or septal fibrosis; and stage 4 ) cirrhosis. The degree of steatosis was assessed on a scale of 0-3: 0 ) hepatocellular steatosis less than 10%; 1) mild (10-33% of hepatocytes affected); 2 ) moderate (33-66% of hepatocytes affected); and 3 ) severe (>66% of hepatocytes affected). Severity of inflammation was graded on a scale of 0-3; 0 ) none; 1 ) mild; 2 ) moderate; and 3 ) severe. The presence of Mallory’s hyaline and ballooning of hepatocytes was also recorded. 2.4. Processing Blood Samples for Protein N-Glycome Analysis. The N-glycans attached to the proteins in 2 µL of serum were released, labeled, and analyzed as described.17 Labeled N-glycans were analyzed by DNA Sequencer Assisted Fluorophore Assisted Carbohydrate Electrophoresis (DSAFACE) technology, using a capillary electrophoresis (CE)-based ABI3130 sequencer. Data were analyzed with the GeneMapper v3.7 software (Applied Biosystems, Foster City, CA). We measured the heights of the peaks that were detected in all the samples, obtained a numerical description of the profiles, and analyzed these data with SPSS 12.0 statistical software. 2.5. Statistical Analysis. Statistical analyses were performed with SPSS for Windows software (SPSS, Chicago, IL). Results are presented as means ( SD. All reported p-values are twotailed, using a t test for independent samples. Pearson’s coefficients of correlation (with 95% confidence intervals and their associated probability, p) were used to evaluate the relationships between parameters. The Receiver Operating Characteristics (ROC) curve was used as an index of accuracy; values close to 1.0 indicate high diagnostic accuracy.

Figure 1. A typical desialylated N-glycan profile from total serum protein. The structures of the N-glycan peaks are shown below the panel. Peak 1 is an agalacto, core-R-1,6-fucosylated biantennary glycan (NGA2F), peak 2 is an agalacto, core-R-1,6-fucosylated bisecting biantennary (NGA2FB), peak 3 and peak 4 are a single agalacto, core-R-1,6-fucosylated biantennaries (NG1A2F), peak 5 is a bigalacto, biantennary glycan (NA2), peak 6 is a bigalacto, core-R-1,6-fucosylated biantennary (NA2F), peak 7 is a bigalacto, core-R-1,6-fucosylated bisecting biantennary (NA2FB), peak 8 is a triantennary (NA3), peak 9 is a branching R-1,3fucosylated triantennary (NA3Fb). The symbols used in the structural formulas are: square indicates β-linked N-acetylglucosamine (GlcNAc); open circle indicates β-linked galactose; triangle indicates R/β-1,3/6-linked fucose; filled circle indicates R/βlinked mannose.

3. Results 3.1. Altered N-Glycan Profiles in NASH but Not in NAFL. We used DSA-FACE to determine the N-glycome profiles of desialylated sera obtained from NAFLD patients with NASH (n ) 38) and NAFL (n ) 9). We also analyzed serum from agematched healthy donors (n ) 13). The glycan fingerprint (Figure 1) is similar to what we previously reported.17,18 The N-glycans detected in serum was represented as peaks and the relative concentration was quantified by normalizing its height to the sum of the heights of all peaks in the profile. The individual N-glycan structure was identified previously as shown in the Figure 117,18 and was confirmed in this study using hydrolyzing exoglycosidases (data not shown). However, the putative structures in Figure 1 might contain groups of isomers or very small amount of other structures that comigrate with the major peak. We previously reported that the concentration of the core fucosylated bisecting biantennary (represented as peak 7) and triantennary (represented as peak 8) glycans were altered in liver cirrhosis patients with HCV infection19 and the branch fucosylated triantennary glycan (represented as peak 9) was increased in the liver of HCC patients with HBV infection.17 However, in this study, we found no alteration of peak 7, peak 8 or peak 9 in the NAFLD (Supporting Information Figure 1). The GlycoCirrhoTest and GlycoHCCTest were generated based on the three glycans using the same method but defining different sets of peaks.17,19 In this study of a cohort of NAFLD, GlycoCirrhoTest and GlycoHCCTest markers had no correlation with NASH, underlining their specificity for detecting cirrhosis and HCC. To specifically diagnose NASH in people with fatty liver, we focused on identifying glycan structures whose abundance Journal of Proteome Research • Vol. 8, No. 2, 2009 465

research articles

Chen et al.

Figure 3. Receiver operating characteristic (ROC) curve for prediction of clinically significant for detection of NASH in the NAFLD group using the values of peak 1, peak 5 and GlycoNashTest. Area under the curves (AUC) show diagnosis power of peak 1 (0.74 ( 0.09), peak 5 (0.29 ( 0.01) and GlycoNashTest (0.74 ( 0.08). Table 3. The Pearson Correlation Coefficient between Glycan Markers and Brunt’s Scores peak 1

peak 5

grade Pearson Correlation -0.067 0.028 inflammation Sig. (2-tailed) 0.671 0.86 steatosis Pearson Correlation 0.005 -0.203 severity Sig. (2-tailed) 0.975 0.196 fibrosis stage Pearson Correlation 0.488a -0.576a Sig. (2-tailed) 0.001 0.0001 a

Figure 2. Trends in derived diagnostic variables for the detection of NASH in NAFLD patients. The vertical axis represents the glycan values of peak 1, peak 5 and GlycoNashTest. Glycan value of peak 1 increased in NASH patients (A), whereas peak 5 decreased in NASH patients (B). GlycoNashTest was significantly higher in the NASH group than in the NAFL and control groups (C). Error bars represent 95% confidence interval for the means.

would change in NASH but not in NAFL. We found that peak 1 (NGA2F) was significantly higher in the NASH group than in the NAFL and control groups (p < 0.025), whereas peak 5 (NA2) was lower (p < 0.055) (Figure 2). The other peaks remained 466

Journal of Proteome Research • Vol. 8, No. 2, 2009

GlycoNashTest

-0.056 0.724 0.111 0.485 0.562a 0.0001

Correlation is significant at the 0.01 level (2-tailed).

unchanged in NASH patients compared to NAFL and controls (data not shown). The log(peak1/peak5) ratio was significantly elevated in NASH patients (p < 0.025) compared to NAFL patients and healthy controls (Figure 2). We therefore named log(peak1/peak5) as GlycoNashTest, in parallel to the “GlycoFibroTest, GlycoCirrhoTest and GlycoHCCTest”. ROC curve analysis showed that peak 1, peak 5 and the GlycoNashTest could identify NASH patients among NAFLD patients with an accuracy of 70-74.3% ( 2.5% (Figure 3). According to the ROC cutoff, for example, the GlycoNashTest could distinguish NASH from NAFL with a specificity of 90% and a sensitivity of 45%. Moreover, both total agalactose (the sum of peaks 1, 2, 3 and 4) and total core-1,6-fucose (the sum of peaks 1, 2, 3, 4, 6 and 7) were significantly higher (p < 0.05) in NASH than in NAFL and the controls. Considering the remarkable increase of peak 1 and the tendency to an increase of peak 2 in NASH patients, we believe that the alteration of total agalactose and core-1,6 fucose reflect the alterations of peak 1 and peak 2, but not the other peaks. 3.2. Glycan Alterations Are Associated with Fibrosis Stage, but Not with Degree of Steatosis and Severity of Inflammation. The correlation of N-glycans with the stage of fibrosis, grade of seatosis and inflammation severity was analyzed in the NAFLD patients using the Peason Correlation test. As shown in Table 3, the abundance of peak 1 was positively associated with fibrosis stage (r ) 0.488, p < 0.001), whereas peak 5 was negatively associated (r ) -0.576, p