LC-MS Based Detection and Quantification of N-glycans in Human Serum Samples Tsung-Heng Tsai¹, Minkun Wang¹, Cristina Di Poto¹, Yi Zhao¹, Yunli Hu², Shiyue Zhou², Yehia Mechref², Habtom W. Ressom¹ ¹Lombardi Comprehensive Cancer Center, Georgetown University, Washington, DC; ²Department of Chemistry and Biochemistry, Texas Tech University, Lubbock, TX Study Population The participants in this study consist of 89 subjects (40 HCC cases and 49 cirrhotic controls) from Egypt (Table I) and 94 subjects (48 HCC cases and 46 cirrhotic controls) from the US (Table II). LC-MS Data Acquisition Thermo Scientific LTQ Orbitrap Velos mass spectrometer coupled to the Ultimate Dionex 3000 HPLC system (Nano LC 350 nL/min) Five MS/MS scans per MS scan on positive mode HCC (N=48)CIRR (N=46)p-value AgeMean (SD)60.2 (6.0)58.9 (7.1) GenderMale (%)77.1%73.9% Ethnicity (%) Caucasian50.0%63.0% African American 33.3%26.1% Others16.7%10.9% HCV Serology HCV Ab (+)68.8%41.3% HCV RNA (+)62.5%39.1% HBV Serology Anti-HBV (+)45.8%26.1% HBsAg (+) 8.3% 2.2% MELD Mean (SD)11.3 (4.1)17.3 (16.1) MELD ≤ %10.9% AFPMedian (IQR)38.8 (91.1)4.5 (11.85) HCC Stage Stage I54.2% Stage II22.9% Stage III 6.3% Unknown16.7% Table I: Characteristics of the Egyptian cohortTable II: Characteristics of the US cohort Objective The objective of this study is to identify candidate N-glycan biomarkers by comparing the levels of permethylated N-glycans in sera of hepatocellular carcinoma (HCC) patients with those of cirrhotic patients from two cohorts (Egypt and US) using LC-MS. Study Design Each cohort analyzed in four batches (n≈24): E1/E2/E3/E4 in the Egyptian cohort; U1/U2/U3/U4 in the US cohort. Balanced assignment of cases (HCC) and controls (CIRR) into each batch in terms of age, race, gender, smoking, alcohol, and BMI. Hepatocellular Carcinoma Hepatocellular carcinoma (HCC) is a significant worldwide health problem with as many as 600,000 new cases diagnosed each year. Primary liver cancer is the fifth most common cancer worldwide and the third most common cause of cancer mortality. Patients with cirrhosis have an annual risk of 1-2% for developing HCC. Malignant conversion of cirrhosis to HCC is often diagnosed at a late stage. Glycoproteins as new HCC markers include AFP-L3, GP73, etc. LC-MS Data Preprocessing Deisotoping (DeconTools) [Jaitly et al., BMC Bioinformatics 2009]: RAW data → Ion list Peak detection (In-house algorithm): Ion list → Peak list (1) Find the ion (deisotoped) with highest intensity (2) Record its mass (), scan number and charge (3) Based on the desired precision, define the mass range [−∆, +∆] (4) Link ions within the mass range in adjacent scans Peak matching (SIMA) [Voss et al., Bioinformatics 2011]: Peak list → consensus list Statistical Analysis p-value < 0.05/4 FC same direction Pre-processed LC-MS data t-test on each batch Significant peaks in each batch Overlapping between batches ANOVA test on common peaks among 4 batches Group comparison within each batch Δm/z: 20 ppm Δrt: 50 s Δm/z: 20 ppm Δrt: 50 s FC same direction p-value < 0.05 Wilcoxon test on each batch Log-transformation Table III: List of candidate N-glycan biomarkers Candidate N-Glycan Biomarkers 14 candidate markers are identified with putative structure. Seven of them were previously reported in the literature (Table III). A preliminary stratified analysis on the US cohort revealed the presence of four markers specific to Caucasians, six to African Americans and an additional one to both races. PNGase F Digestion Solid-Phase Extraction Solid-Phase Permethylation LC-MS/MS Proteins & Glycoproteins Proteins & N-Glycans N-Glycans Reduced N-Glycans Permethylated N-Glycans LC-MS/MS Data Reducing N-Glycans Summary Permethylation of glycans enzymatically removed from serum proteins allows relative quantification of hundreds of oligosaccharides. The proposed workflow design allows identification of reliable candidate biomarkers for HCC. Future Work Clustering of related ions with different charge states and adduct types Stratified analysis to evaluate the effect of age, gender, race, viral infection, alcoholic cirrhosis, etc. Multivariate analysis to identify a panel of biomarkers Quantitation of candidates from this study by SRM on TSQ Vantage Acknowledgements This work was supported by NIH Grants R01CA and R01GM HCC (N=40)CIRR (N=49)p-value AgeMean(SD)53.2 (3.9)53.8 (7.6) GenderMale (%)77.5%67.3% HCV Serology HCV Ab (+)100.0% HBV Serology HBsAg (+)0.0%6.1% MELD Mean (SD)18.6 (7.7)18.9 (7.1) MELD ≤ %12.2% AFP Median (IQR) (1244.3) HCC Stage Stage I72.5% Stage II15.0% Stage III5.0% Unknown7.5% Overlapping between batches Structure CohortAdductm/zzRTBatchFCLiterature [4,3,2,0,0] EgyptRe[M+2H] E1↓ 1.85 [Goldman et al., Clin Cancer Res 2009] [Liu et al., Hepatology 2007] [Tanabe et al., Biochem Biophys Res Commun 2008] E3↓ E4↓ 1.80 USRe[M+2H] U3↓ U4↓ 2.47 [4,3,0,1,0] Egypt Re[M+H] E1↓ 2.77 [Liu et al., 2007] [Tanabe et al., 2008] E4↓ 1.54 Re[M+NH 4 ] E1↓ 2.76 Re[M+2H] E1↓ E4↓ 1.38 Re[M+H+NH 4 ] E1↓ 2.75 Re[M+2NH 4 ] E1↓ 2.61 Re[M+3H] E4↓ 1.44 [4,3,0,0,0] Egypt Re[M+H+NH 4 ] E1↓ 2.68 [Tanabe et al., 2008] E3↓ 1.75 Re[M+2H] E1↓ 2.17 USRe[M+2H] U4↓ 3.84 [5,3,0,1,0] Egypt Re[M+2H] E1↓ 2.74 [Liu et al., 2007] [Tanabe et al., 2008] E4↓ 1.69 Re[M+H+NH 4 ] E1↓ 2.49 Re[M+3H] E1↓ 2.46 US Re[M+2H] U3↓ 1.34 Re[M+2H] U4↓ 2.19 [5,3,1,0,1] Egypt Re[M+2H] E1↓ 2.41 [Goldman et al., 2009] [Ressom et al., J Proteome Res 2008] E4↓ 1.76 Re[M+H+NH 4 ] E4↓ 1.52 Re[M+3H] E1↓ E4↓ 1.74 US Re[M+2H] U3↓ U4↓ 2.67 Re[M+3H] U3↓ U4↓ 2.56 [5,3,3,1,3] USRe[M+2H] U2↓ 2.06 [Tang et al., J Proteome Res 2010] U3↓ U4↓ 1.33 [4,3,1,1,0] US Re[M+H+NH 4 ] U1↑ 1.58 [Liu et al., 2007] [Ressom et al. 2008] [Tanabe et al., 2008] [Tang et al., 2010] Re[M+2NH 4 ] U1↑ U4↑ 1.89