WO2009117666A1

WO2009117666A1 - Glycan markers of hepatocellular carcinoma

Info

Publication number: WO2009117666A1
Application number: PCT/US2009/037818
Authority: WO
Inventors: Milos V. Novotny; Yehia S. Mechref; Radoslav Goldman; Habtom W. Ressom
Original assignee: Indiana University Research And Technology Corporation; Georgetown University
Priority date: 2008-03-21
Filing date: 2009-03-20
Publication date: 2009-09-24

Abstract

Selective and specific biomarkers for early detection and progression of cancer are disclosed. The present invention describes the quantitation of selected sets of serum proteins and N-glycans as useful biomarkers of hepatocellular carcinoma.

Description

GLYCAN MARKERS OF HEPATOCELLULAR CARCINOMA

CROSS REFERENCE TO RELATED APPLICATIONS

This application claims priority under 35 USC § 119(e) to US Provisional Application Serial No. 61/038,491, filed March 21, 2008, the disclosure of which is incorporated herein by reference.

BACKGROUND

Hepatocellular carcinoma (HCC) is a major worldwide health problem and a cancer with increasing incidence in the United States. Currently available systemic therapies demonstrate a modest response rate and have not been shown to improve survival in patients with HCC. A complete surgical resection and liver transplant are at present the only curative treatment options; however, many patients suffer from advanced unresectable disease not amenable to definitive local therapies (Schwartz & Ham, Curr. Treat. Options. Gastroenterol, 2003, 6(6) 465-472; Lopez & Marrero, Curr. Opin. Gastroenterol, 2004, 20(3) 248-253). The slow development and late detection of HCC suggest that the identification of biomarkers of disease progression and early detection are needed.

The current diagnosis of HCC relies on clinical information, liver imaging, and measurement of serum alpha- fetoprotein (AFP). The reported sensitivity (41-65%) and specificity (80-94%) of AFP are not effective for early diagnosis (Gupta et al., Ann Intern. Med., 2003, 139(1) 46-50). Analysis of protein glycosylation in blood has been pursued as an indicator of liver pathology because of the major role of the liver in the homeostasis of blood glycoproteins (Turner, Clin. Chim. Acta., 1992, 208(3) 149-171; Lee et al., Science, 2002, 295(5561) 1898-1901).

An alternative to the analysis of glycoproteins in blood is the analysis of protein-associated glycans. N-glycans are a diverse group of carbohydrate molecules that share a common core asparagine linkage and can be enzymatically released from serum glycoproteins. Permethylated N-glycan structures can readily be measured by matrix-assisted laser desorption/ionization time -of- flight (MALDI-TOF) mass spectrometry (MS). As disclosed herein, the quantitative measurement of changes in the levels of selected N-glycans is useful as a sensitive and specific biomarker for the diagnosis and early detection of HCC.

SUMMARY OF INVENTION

A selective and specific biomarker for cancer has been developed based on the analysis of serum N-glycan levels. In one illustrative embodiment, the relative concentrations of a plurality of selected serum N-glycans are analyzed as biomarkers for the early detection hepatocellular carcinoma and monitoring of cancer progression and responsiveness to therapy.

In accordance with one embodiment, the method for detecting early stage hepatocellular carcinoma comprises the steps of analyzing a biological sample obtained from a patient to determine the relative concentration of a glycan selected from the group consisting of

Glycan 2 (Mass: 1981.6):

Glycan 3 (Mass: 2069.7):

Glycan 5 (Mass: 2472.9):

Glycan 6 (Mass: 2929.9):

= N-acetylglucosamine; W = mannose; \ = galactose;

L-S. = fucose; and ^^ = N-acetylneuraminic acid. The concnetration of the detected glycan is the compared to concentrations of the glycan found in cancer free individuals. A statistically significant difference between the glycan concentrations detected in the biological sample relative to those in the general population of cancer free individuals indicates the presence of hepatocellular carcinoma in said patient. In one embodiment the patient is a human and the biological sample represents blood or a blood derivative such as plasma or serum. In one embodiment the method comprises determining the relative concentration of 1-3 glycans selected from the group consisting of

ι

Glycan 1 (Mass: 3241.9):

Glycan 5 (Mass: 2472.9):

Glycan 6 (Mass: 2929.9):

wherein : I — I = N-acetylglucosamine; W = mannose; ^ = galactose;

/A = fucose; and — = N-acetylneuraminic acid. -A-

In another illustrative embodiment, there is provided a method of identifying, selecting and measuring a plurality of peptides or N-glycans for use as biomarkers of hepatocellular carcinoma.

BRIEF DESCRIPTION OF THE DRAWINGS

The detailed description particularly refers to the accompanying figures.

Figs. 1A-1F represent Box plots of six glycans comparing HCC cases (n=73), Chronic Liver Disease (CLD) controls (n=52), and population controls (CTRL; n=77). More particularly, Fig. IA represents detected glycan 2 (1981.6 Da); Fig. IB represents glycan 3 (2069.7 Da); Fig. 1C represents detected glycan 4 (2185.8 Da); Fig. ID represents glycan 5 (2472.9 Da); Fig. IE represents detected glycan 6 (2929.9 Da); Fig. IF represents glycan 1 (3241.9 Da).

Fig. 2 represents a ROC curve for three glycans comparing a blinded validation set of HCC cases (n=47) and CLD controls (n=27).

Figs. 3A-3C represent data presented as Box plots comparing Glycan 1 (Fig. 3C: 3241.9 Da); Glycan 5 (Fig. 3A: 2472.9 Da); and Glycan 6 (Fig. 3B: 2929.9 Da) in fibrosis (n=22), cirrhosis (n=25), and early HCC (n=18).

Fig. 4 represents an example of biomarker selection using PSO-SVM.

Fig. 5 is a schematic representation of the peptide selection scheme.

Figs. 6A-6F represent data presented as Box plots of the relative concentrations of selected peptides comparing HCC and population controls. More particularly, the data present in Fig. 6A represents glycans with approximate mass of about 1863.4-1871.3 Da (MALDI A), Fig. 6B represents glycans with approximate mass of about 933.6-938.2 Da (MALDI B), Fig. 6C represents glycans with approximate mass of about 2528.7-2535.5 Da (MALDI C), Fig. 6D represents glycans with approximate mass of about 1737.1-1744.6 Da (MALDI D), Fig. 6E represents glycans with approximate mass of about 1379.0-1381.2 Da (MALDI E), and Fig. 6F represents glycans with approximate mass of about 4085.6-4097.9 Da (MALDI F). Fig. 7 represents ROC analysis of six individual marker candidates and for a combined SVM classifier. Each curve is based on spectra from a blinded validation set of samples (53 HCC, 47 control).

Fig. 8A & 8B represent data presented as a Box plot showing the peptide levels of MALDI A (Fig 8A) & MALDI B (Fig. 8B) in progression of HCV to HCC.

Fig. 9A-9B represents data obtained from an ELISA of secreted liver proteins and present in Box plot format.

Figs. 10A- 1OF represent data presented as Box plots showing the relative amounts of glycan detected. More particularly, Fig. 1OA represents a glycan having a mass of 2569 Da; Fig. 1OB represents a glycan having a mass of 1799.8 Da ; Fig. 1OC represents glycan 1 (3241.9 Da); Fig. ID represents glycan 2 (1981.6 Da); Fig. IE represents a glycan having a mass of 2149.9 Da; Fig. IF represents a glycan having a mass of 1543.7 Da)

Fig. 11 represents a ROC analysis of three individual marker candidates and for a combined SVM classifier. Each curve is based on a spectra from a blinded validation set of samples (53 HCC, 72 controls including 25 with CLD).

Figs. 12A-12C represent data presented as Box plots showing the relative amounts of three glycans having a mass of 1543.7 Da (Fig. 12A), 1799.8 Da (Fig. 12B) and 3241.9 Da (Fig. 12C) detected in serum of fibrosis, cirrhosis, and HCC patients.

DETAILED DESCRIPTION

DEFINITIONS

In describing and claiming the invention, the following terminology will be used in accordance with the definitions set forth below.

As used herein the term "patient" or without further designation is intended to encompass any warm blooded vertebrate domesticated animal (including for example, but not limited to livestock, horses, cats, dogs and other pets) and humans.

As used herein the term "glycan" without further designation is intended to encompass any branched oligosaccharide (typically containing 5 to 20 sugar monomers). An "N-glycan" represents a glycan that is found naturally linked to the amide nitrogen of an amino acid side chain.

As used herein the term "serum N-glycans" refers to N-glycans that have been recovered from a patient's serum. An "enzymatically released serum N- glycan" is an N-glycan that is recovered from a serum sample after enzymatically treating the sample to release and recover glycans.

EMBODIMENTS

While the invention has been illustrated and described in detail in the drawings and the foregoing description, such an illustration and description is to be considered exemplary and not restrictive in character, it being understood that only illustrative embodiments have been shown and described and that all modifications, equivalents, and alternatives within the spirit of the invention are desired to be protected.

As disclosed herein branched oligosaccharides are used as markers for cancer. More particularly, biological samples recovered from patients are analyzed to detect the presence, and relative concentrations, of glycoprotein glycans, wherein statistically significant deviation from population levels of one or more glycans is a diagnostic indicator of cancer, including for example hepatocellular carcinoma. In accordance with one embodiment serum level glycans, optionally cleaved from their corresponding glycoproteins, can be used as an early stage diagnostic of hepatocellular carcinoma. In accordance with one embodiment serum proteins are purified, the glycoproteins are enzymatically cleaved to release the glycan moieties, and the relative concentration of glycans in a patient's biological sample are determined, using standard techniques known to those skilled in the art, including for example mass spectrometry analysis.

In accordance with one embodiment the glycoproteins are cleaved with an enzyme (e.g., PNGase F) to release N-glycans, and the relative concentrations of N-glycans are measured as a diagnostic indicator of hepatocellular carcinoma. In accordance with one embodiment the detected concentrations of N-glycans in the patient's sample are compared to concentrations of N-glycans in non-cancerous populations wherein a statistical difference in relative N-glycan concentration for certain select N-glycans is diagnostic of hepatocellular carcinoma.

In accordance with one embodiment an N-glycan selected from those disclosed in Table 2 of the present specification is selected for use as a diagnostic marker for cancer. Selective and specific biomarkers for cancer have been developed based on the analysis of serum N-glycan levels. Changes in N-glycans enzymatically released from serum glycoproteins were analyzed as candidate markers for the detection of HCC. In accordance with one embodiment serum N-glycans are measured using mass spectrometry analysis, including for example, MALDI-TOF quantification of glycans enzymatically detached from serum proteins.

In accordance with one illustrative embodiment, a set of six N-glycans with the following characteristics are selected as biomarkers for the early detection and progression of hepatocellular carcinoma.

Glycan 1 (Mass: 3241.9):

Glycan 2 (Mass: 1981.6):

Glycan 3 (Mass: 2069.7):

Glycan 6 (Mass: 2929.9):

wherein : I — I = N-acetylglucosamine; -^ = mannose ,; > I^== galactose; /A == ffuuccoossee;: aanndd ¹ ^^^ = N-acetylneuraminic acid.

In one embodiment the relative concentration two or more enzymatically cleaved serum N-glycans are measured in a patient's biological sample to diagnose a patient with early stage hepatocellular carcinoma. In a further embodiment three or more enzymatically cleaved serum N-glycans are measured in a patient's biological sample to diagnose a patient with early stage hepatocellular carcinoma. In one embodiment a patient's biological sample is analyzed to determine the relative concentration of one to three N-glycans selected from the group consisting of

ι

Glycan 1 (Mass: 3241.9):

Glycan 5 (Mass: 2472.9):

Glycan 6 (Mass: 2929.9):

= N-acetylglucosamine; w = mannose; U"" = galactose;

L\ = fucose; and ^^ = N-acetylneuraminic acid. In one embodiment the relative concentration three to six N-glycans are measured in a patients biological sample to diagnose a patient with early stage hepatocellular carcinoma. In accordance with one embodiment a method is provided for monitoring the effectiveness of an anti-cancer therapy, including the monitoring of the efficacy of an anti-hepatocellular carcinoma therapy. The method comprises the steps of obtaining a biological sample (e.g., a serum sample) from the patient diagnosed with hepatocellular carcinoma prior to starting the anti-cancer therapy and at least one biological sample after initiating an anti-cancer therapy. In one embodiment multiple samples are taken, at regular intervals after initiating the anti-cancer therapy. The samples are then analyzed to determine the relative concentration of preselected N- glycans. In one embodiment each of the serum samples are screened for one or more glycans selected from the group consisting of Glycans 1-6 as disclosed herein.

In accordance with one embodiment N-glycans present in a biological sample are identified by mass spectrographic analysis. Briefly, the proteins are purified from other sample components and the glycans are enzymatically released (using PNGase F, for example) from serum glycoproteins. The free N-glycans are then purified using standard techniques (e.g., C 18, active charcoal) and the purified glycans subjected to solid phase permethylation. Permethylated N-glycan structures can then be readily measured by matrix-assisted laser desorption/ionization time-of- flight (MALDI-TOF) mass spectrometry (MS), or other analytical techniques known to those skilled in the art.

EXAMPLE 1

Hepatocellular carcinoma (HCC) cases and controls were enrolled in collaboration with the National Cancer Institute of Cairo University, Egypt, from 2000 to 2002, as described previously (Ezzat et al., IntJHyg. Environ. Health, 2005, 208(5) 329-339). Briefly, adults with newly diagnosed HCC, aged 17 and older, without a previous history of cancer, were eligible for the study. Diagnosis of HCC was confirmed by pathology, cytology, imaging (CT, ultrasound), and serum AFP levels. Controls were recruited from the orthopedic department of Kasr El Aini Faculty of Medicine, Cairo University (Ezzat et al., IntJHyg. Environ. Health, 2005, 208(5) 329-339). The patients with chronic liver disease (CLD), fibrosis and cirrhosis, were recruited from Ain Shams University Specialized Hospital and Tropical Medicine Research Institute, Cairo, Egypt, during the same period. The diagnosis of liver disease in this group of controls was confirmed by ultrasound- guided liver biopsy. The patients negative for hepatitis B viral (HBV) infection but positive for hepatitis C virus (HCV) RNA and with AFP less than 100 ng/ml were selected for the study. All participants signed informed consent, provided a blood sample, and answered a questionnaire with demographic information, personal habits, medical history, and occupational history. The study protocol was approved by the institutional review committee and conformed to the ethical guidelines of the 1975 Helsinki Declaration.

Blood samples were collected by a trained phlebotomist each day around 10 a.m. and processed within a few hours according to a standardized protocol. Aliquots of sera were frozen immediately after collection and kept at about -80 ⁰C until analysis; all mass spectrometric measurements were performed on twice- thawed sera. Each patient's hepatitis B and C viral infection status was assessed by enzyme immunoassay for anti-HCV, anti-HBC, and hepatitis B surface antigen (HBsAg), and by PCR for HCV RNA (Ezzat et al, IntJHyg. Environ. Health, 2005, 208(5) 329-339; Abdel-Hamin et al., J. Hum. Virol, 1997, 1(1) 58-65). The characteristics of this population are summarized in Table 1, which shows increased markers of viral infections (HCV RNA, anti HCV, and anti-HBV) in cancer cases (see also Table 5) (Ezzat et al., IntJHyg. Environ. Health, 2005, 208(5) 329-339). We obtained stage information on 51 cases, with 18 cases classified as early (Stage I and II) and 33 cases as advanced (Stage III and IV) according to the AJCC staging system (AJCC Cancer Staging Manual, 6^th Edition, American College of Surgeons, 2002, Philadelphia, Lippincott-Raven); for the remaining cases, the available information was not sufficient to assign the stage. Table 1. Demographic variables and viral infections of Patient Population

Glycomic Analysis

Release of N -gly cans from glycoproteins

Human serum samples were reduced and alkylated as described previously (Kyselova et al, J. Proteome. Res., 2007, 6(5) 1822-1832; Kang et al, Rapid Commun. Mass Spectrom., 2005, 19(23) 3421-3428). Briefly, a 10-μl aliquot of serum was added to 150 μl of 25 mM ammonium bicarbonate and 2.5 μl of 200 mM DTT prior to incubation at about 60 ⁰C for about 45 min. A 10-μl aliquot of 200 mM IAA was added and allowed to react at room temperature for about 1 h in the dark. Subsequently, a 2.5-μl aliquot of DTT was added to react with the excess IAA. The reaction mixture was diluted with 100 μl of ammonium bicarbonate to adjust the pH to about 7.5 - 8.0 for the enzymatic release of N-glycans using PNGase F. Next, a 5 mU aliquot of PNGase F was added to the mixture prior to incubation overnight (about 18-22 h) at about 37 ⁰C.

Solid-phase extraction of N-glycans

The volume of enzymatically released glycans was adjusted to 1 ml with deionized water and applied to a C18 Sep-Pak® cartridge (Waters, Milford, MA), which was preconditioned with ethanol and deionized water as described previously (Kang et al., Rapid Commun. Mass Spectrom., 2005, 19(23) 3421-3428). The reaction mixture was circulated through the cartridge about 5 -times to retain peptides and 0-linked glycopeptides. Glycans were present in the pass-through and the 0.25 ml deionized water washes. The combined eluents were then passed over activated charcoal microcolumns (Harvard Apparatus, Holliston, MA) preconditioned with 1 ml of ACN and 1 ml aqueous solution of about 0.1% trifluroacetic acid (TFA). The microcolumn was washed with 1 ml of about 0.1% TFA and samples were eluted with 1-ml of 50% aqueous ACN with 0.1% TFA. The purified N-glycans were evaporated to dryness using vacuum CentriVap Concentrator (Labconco Corporation, Kansas City, MO) prior to solid-phase permethytion.

Solid-phase permethylation

The permethylation was carried out as described in a recent report (Kang et al, Rapid Commun. Mass Spectrom., 2005, 19(23) 3421-3428). Tubes, nuts and ferrules from Upchurch Scientific (Oak Harbor, WA) were used to assemble the sodium hydroxide capillary reactor. Sodium hydroxide powder was suspended in ACN and packed into Peek tubes (1 mm i.d.; Polymicro Technologies, Phoenix, AZ) using a 100-μl syringe from Hamilton (Reno, NV) and a syringe pump from KD Scientific, Inc. (Holliston, MA). The sodium hydroxide reactor was conditioned with 60 μl of dimethyl sulfoxide (DMSO) at about a 5 μl/min flow rate. Purified N- glycans were resuspended in a 50-μl aliquot of DMSO with 0.3 μl of water and 22 μl methyl iodide. This permethylation procedure has been shown to minimize oxidative degradation and peeling reactions and to eliminate excessive clean-up. Sample was infused through the reactor at about 2 μl/min and washed with 230 μl ACN at about 5 μl/min. All eluents were combined while permethylated N-glycans were extracted using 200 μl chloroform and washed 3 times with 200 μl of water prior to drying.

MALDI-TOF MS instrumentation

Permethylated glycans were resuspended in 2 μl of (50:50) methanol: water solution. A 0.5-μl aliquot of the sample was spotted on a MALDI plate and mixed with an equal volume of DHB-matrix (10 mg DHB in 1 ml of (50:50) methanol: water containing about 1 mM sodium acetate to promote formation of sodium adducts in MALDI-MS). The MALDI plate was dried under vacuum to ensure uniform crystallization. Mass spectra were acquired using an Applied Biosystems 4800 MALDI TOF/TOF Analyzer (Applied Biosystems Inc., Framingham, MA) equipped with a Nd:YAG 355-nm laser, as described previously (Kyselova et al, J. Proteome. Res., 2007, 6(5) 1822-1832). MALDI-spectra were recorded in the positive-ion mode, since permethylation eliminates the negative charge normally associated with sialylated glycans (Mechref & Novotny, Anal. Chem., 1998, 70(3) 455-463).

Data Processing and Analysis

Raw spectra were exported as text files for further analysis. Each spectrum consisted of approximately 121,000 m/z values with the corresponding intensities in the mass range of about 1,500-5,500 Da. Analyses were carried out in MATLAB (MathWorks, Natick, MA) and SAS (SAS Inc., Cary, NC) software packages; overlay of spectra was created in ClinProTools (Bruker Daltonics, Billerica, MA). MALDI-TOF mass spectra were processed as described previously (Ressom et al., Bioinformatics, 2007, 23(5) 619-626). Briefly, the dimension of each spectrum was reduced to 13,030 bins (100 ppm step). Baseline-corrected spectra were normalized by dividing each spectrum by its total ion current. A total of 203 spectra were analyzed. One spectrum was eliminated due to truncated acquisition. Peak identification was carried out on a randomly selected training set of 74 samples (25 HCC, 24 controls, 25 controls with CLD). After scaling the peak intensities to an overall maximum intensity of 100, local maximum peaks above a specified threshold were identified and nearby peaks within 300 ppm mass were coalesced into a single window to account for drift in m/z location.

This procedure identified 85 peak-containing windows; the maximum intensity in each window was used as the variable of interest. The threshold intensity for peak identification was set so that isotopic clusters were represented by a single peak. The isotopic cluster at 1543-1547 Da was the only cluster resolved by the procedure to three individual peaks. This cluster was grouped to one variable prior to all analyses. This resulted in a final comparison of 83 features (glycan intensities). Logistic regression models were used to determine association of the glycans and covariates including HCV and HBV viral infections (independent variables) with HCC status (dependent variable). Glycan intensities were dichotomized for the regression analyses of HCC by the median of the appropriate control group (population or CLD controls).

For determination of prediction accuracy and construction of receiver operating characteristic (ROC) curves, the 74 spectra randomly selected for window definition were used as a training dataset (25 HCC, 24 control, 25 controls with CLD) and the remaining 128 spectra served as a blinded validation set (48 HCC, 53 control, 27 controls with CLD). Spectra in the blinded validation set were processed using the same criteria described above for the 74 training samples. We performed two sets of analyses, one comparing intensity of all 83 m/z windows, second using only 48 glycans with assigned structure. A hybrid algorithm that interfaces ant colony optimization with support vector machine (ACO-SVM) was used to select six mass windows (N-glycan peaks) for classification of HCC based on comparison of HCC cases (n=25) and controls (n=24) from the training dataset as described previously (Ressom et al., Bioinformatics, 2007, 23(5) 619-626). Briefly, the algorithm was run 100 times to select 5 peaks at a time. Each run consisted of 500 iterations. A 4-fold cross-validation was used to estimate the classification accuracy. A frequency plot was used to select m/z windows from the 100 runs that occurred more than 50% of the time. An SVM classifier was used to classify the blinded spectra using the selected peaks. Sensitivity and specificity of the marker-candidates were evaluated on the blinded validation dataset including the set of controls with CLD.

RESULTS

MALDI-TOF mass-spectrometric analysis of permethylated N- glycans, which were enzymatically detached from serum glycoproteins, allowed relative quantification of 83 features. Serum samples from a total of 202 participants were analyzed. Glycan analysis (Kyselova et al., J. Proteome. Res., 2007, 6(5) 1822- 1832; Kang et al., Rapid Commun. Mass Spectrom., 2005, 19(23) 3421-3428) and spectral processing (Ressom et al., Bioinformatics, 2007, 23(5) 619-626) was carried out, as described previously, with minor modifications. Peak selections were adjusted so that isotopic clusters resolved by the high-resolution reflectron acquisition were represented by only one glycan peak. A total of 74 spectra were used from HCC cases (n=25), population controls (n=24), and controls with CLD (n=25) to define parameters for spectral processing and to identify glycan peaks. Comparison of average spectra in HCC cases (n=73) and controls (n=77) showed marked differences in glycan abundance (Fig. 1). Analysis of the 83 peak intensities by t-test showed significant differences (p<0.01) in the abundance of 57 glycans; we chose p<0.01 to adjust for the multiple comparisons. Details of the analysis are presented in Table 2, together with a description of the known glycan structures. A graphical overview of the spectra in the mass range of 1.5 - 5.5 kDa is provided in Figure 5.

Table 2. Analysis of Glycan Abundance in HCC Cases and Population Controls

Compound structures for Table 2:

Compound A:

Compound B:

Compound C:

Compound D:

Compound E:

Compound F:

Compound G:

Compound H:

Compound I:

Compound J:

Compound K:

Compound L:

-O -O

Compound M: ⁰

Compound N:

Compound O:

Compound P:

Compound Q:

Compound R:

Compound S:

Compound T:

Compound U:

Compound V:

Compound W:

Compound X:

Compound Y:

Compound Z:

Compound AA:

Compound BB:

Compound CC:

Compound DD:

Compound EE:

Compound FF:

Compound GG:

Compound HH:

Compound II:

Compound JJ:

Compound KK:

Compound LL:

Compound MM:

Compound NN:

Compound 00:

Compound PP:

Compound QQ:

Compound RR:

Compound SS:

Compound TT:

Compound UU:

Structural composition of 48 of the 83 N-glycans was determined by a combination of enzymatic sequencing and tandem MS as described previously (Mechref & Novotny, Anal. Chem., 1998, 70(3) 455-463; Mechref et al, Anal. Chem., 2003, 75(18) 4895-4903). To select candidate markers for detection of HCC, we focused on the glycans with known structure to allow a robust validation of our selection in other laboratories. We selected six of the 48 N-glycans as candidate markers for classification of HCC by ACO-SVM computational methods, as described previously (Ressom et al., Bioinformatics, 2007, 23(5) 619-626). These six glycans were selected with greater than 50% frequency in 100 repeats of the ACO- SVM algorithm carried out with 25 HCC and 24 population control spectra. Association of the glycans and covariates (age, gender, HCV and HBV viral infections, and smoking) with HCC (dependent variable) was analyzed by univariate logistic regression using 73 HCC and 77 control spectra (Table 5). Glycan intensities were dichotomized by the median value in population controls; the analysis of glycans as continuous variables did not substantially affect the outcome. The analysis showed that viral infections and five of the six selected glycans are associated with HCC. The association of the sixth glycan with HCC bordered on significance and became significant after adjustment for HCV infection.

To evaluate association of these N-glycans with viral infections, we analyzed association of each of the six N-glycans (independent variable) and viral infections (dependent variable) in the population controls (32% positive for HCV antibodies and 52% positive for HBV antibodies, see Table 1). The structure of the six selected N-glycans is as follows:

Glycan 1

Glycan 2

Glycan 3

Glycan 4

Glycan 5

ycan 6

= galactose; --A fucose; and O = N-acetylneuraminic acid.

None of the selected N-glycans were associated with the presence of HBV antibodies; glycans 2 and 4 were associated with HCV antibodies (see Table 6). Next, we used seven multivariate regression models to evaluate the association of each glycan (independent variable) with HCC (dependent variable), following an adjustment for matching variable age, gender and for HCV infection (Table 3). All six N-glycans were associated with HCC following the adjustment. We did not include HCV RNA in the regression models because it is correlated with anti HCV (correlation coefficient =0.823). Table 3. Association of Glycans with HCC

Next, we examined the six glycans in comparison to the CLD control group, which is a clinically relevant group for the detection of HCC. This group of participants had a biopsy-confirmed fibrosis (n=22) or cirrhosis (n=25); 5 remaining CLD controls did not have sufficient clinical information. Box plots of the intensities of the six N-glycans in HCC cases (n=73), population controls (n=77), and controls with CLD (n=52) are presented in Figure 2. Glycans 1 and 3 decreases from population controls to HCC, all the remaining N-glycans increase in intensity. Multivariate logistic regression comparing HCC cases (n=73) with CLD controls (n=52) showed that glycans 1, 5, and 6 remain significantly associated with HCC following adjustment for age and gender (Table 4). We did not adjust for viral infection in this analysis because all participants in the CLD group were selected to be HBV-negative and HCV-positive. Since all CLD and 80% HCC cases carry HCV viral infection, the result strongly suggests that the observed change in N-glycan concentration is associated with HCC. Descriptive statistics for the six glycans in population controls (n=77), CLD controls (n=52), and HCC (n=73) are presented in Table 6 and Table 7. Table 4. Comparison of HCC with Chronic Liver Disease (CLD) Controls

Multivariate logistic regression controlled for age and gender; 73 HCC cases and 52 CLD controls.

Tabic 5. Association of Glvcans and Covariates with HCC

Univariate regression analysis of the association of glycans and covariates with HCC; population controls (n=77) and HCC cases (n=73).

Table 6. Association of HCV and HBV Antibodies Λvith GIyeans in tlie Population

Controls.

Association of the antibodies to HCV and HBV with glycans in the population controls (n=77); 25 anti-HCV positive (32%), 42 anti-HBV positive (54%).

^'Table 7, Descriptive Statistics of Six Glycans Evaluated as Candidate Markers for the Detection of HCC in Cases, Population Controls, and CLD Controls.

Six glycans evaluated as candidate markers for the detection of HCC in cases (n=73), population controls (n=77), and CLD controls (n=52).

Fig. 2 shows ROC curves of the three individual glycans that were different compared to CLD controls and their combination in a blinded, independent validation set of HCC cases (n=48) and CLD controls (n=27). The area under the receiver operating characteristic (AuROC) for individual glycans ranged from 89- 93%, while the combined classifier has a sensitivity of 92% and specificity of 96% in a blinded validation set of 47 HCC cases and 27 CLD controls. Glycan 1 is a triantennary complex glycan that decreases in HCC. Glycan 5 is a bisecting glycan and glycan 6 is a core fucosylated biantennary glycan; both of these glycans increase in HCC. This is consistent with the general trends of changes observed in our study as discussed below.

In accordance with another illustrative embodiment three N-glycans with approximately the following characteristics are selected for utility as biomarkers for the early detection and progression of hepatocellular carcinoma.

Glycan 1 (Mass: 3241.9):

Glycan 5 (Mass: 2472.9):

Glycan 6 (Mass: 2929.9):

wherein : I — I = N-acetylglucosamine; ^- = mannose; . >

= galactose; l\ = fucose; and ^^ = N-acetylneuraminic acid.

EXAMPLE 2

We extracted disease stage information from the clinical records for 51 HCC cases and 46 CLD controls. We were not able to obtain sufficient clinical information for the remaining patients. The progression from fibrosis to cirrhosis and early cancer, stage 1 and 2 disease (AJCC Cancer Staging Manual, 6^th Edition, American College of Surgeons, 2002, Philadelphia, Lippincott-Raven), is presented in Fig. 3. Seventeen of the 18 early cancers and 24 out of 25 cirrhotic controls were classified correctly by the three glycans. Because AFP was used as a selection criterion for the CLD group (<100 ng/ml included), we could not compare the prediction accuracy directly to AFP. However, 30% of the cases had AFP <200 ng/ml; of these 22 HCC cases 18 were correctly classified by the three glycans.

In this study, we compared HCC cases with two groups of controls, a population control group, and a hospital-based control group with a biopsy-verified CLD. Overall, the differences in glycosylation between HCC cases and population controls are substantial with 57 of 83 N-glycans, as differentiated by t-test at p<0.01. We selected three of 48 glycans with known structure as candidate markers for the detection of HCC on a CLD background. The selection is based on prediction accuracy and facilitates the choice of the glycans for disease classification (Ressom et al, Bioinformatics, 2007, 23(5) 619-626). The following precautions were used to limit the possibility of false discoveries (Ransohoff, Nat. Rev. Cancer, 2005, 5(2) 142-149; Ransohoff, Nat. Rev. Cancer, 2004, 4(4) 309-314). The cases and controls are a representative sample of the Egyptian population with a substantial proportion of controls carrying HCV infection; it is not a "convenience sample" (Ezzat et al., IntJ Hyg. Environ. Health, 2005, 208(5) 329-339). A standardized sample-collection and processing protocol was used to minimize variability; the analytical methods were optimized to a mean CV of approximately 10% (Kang et al., Rapid Commun. Mass Spectrom., 2005, 19(23) 3421-3428). Glycan quantification/selection followed established procedures with evaluation of prediction accuracy of marker-candidates on a blinded validation set of samples (Ressom et al., Bioinformatics, 2007, 23(5) 619-626). In addition, sugar composition of the three glycans selected as candidate- markers for detection of HCC was determined.

Grouping of the glycans with determined structural composition into high-mannose, hybrid, and complex structure types showed some general trends. The complex structures represented the majority of the observed glycans (n=28), with 19 of the complex structures core fucosylated. We detected a relatively high number of bisecting glycans (n=9), with 5 of them core-fucosylated. We also detected 5 high- mannose structures and four hybrid glycans one of which was core fucosylated. As described previously, core fucosylation is a frequent event in liver disease (Turner, Clin. Chim. Acta, 1992, 208(3) 149-171; Block et al, Proc. Natl. Acad. Sci. USA, 2005, 102(3) 779-784); 9 of the core-fucosylated complex structures increased in HCC, as compared to the controls, including CLD. In contrast, 8 of 9 complex structures without core fucosylation decreased in HCC. All four hybrid structures increased in CLD compared to controls but only the core fucosylated hybrid glycan were higher in HCC compared to CLD. Seven of 9 bisecting glycans (with or without core fucosylation) increased in CLD and 4 of 9 further increased in HCC. Similarly, 3 of 5 high-mannose structures increased in CLD, and 4 of 5 further increased in HCC. Sialylation results are mixed with 15 of the 32 sialylated glycans higher in cancer compared to control, 9 higher in control compared to cancer, and 8 unchanged.

We present selection of three glycans as candidate markers for the detection of HCC in patients with CLD. Each of the three glycans has an AuROC between 0.89-0.93 for the detection of HCC compared to CLD (Fig. 2) and the markers detect early cancer (Fig. 3). The individual marker candidates have good prediction accuracy in this population, in general, comparable to AFP (Marrero, Clin. Liver Dis., 2005, 9(2) 235-251, vι). The combination of the three glycans performs with sensitivity of 90% and specificity of 96% in the independent validation set of 48 HCC and 27 CLD samples.

In accordance with another illustrative embodiment of the invention a set of six peptides with approximate mass of about 1863.4-1871.3 Da (MALDI A), about 933.6-938.2 Da (MALDI B), about 2528.7-2535.5 Da (MALDI C), about 1737.1-1744.6 Da (MALDI D), about 1379.0-1381.2 Da (MALDI E), and about 4085.6-4097.9 Da (MALDI F) are selected for their utility as biomarkers for the early detection and progression of hepatocellular carcinoma. The FDA has approved glycoproteins as markers of cancer for purposes of diagnosis, staging, monitoring, screening, and selection of therapy for certain cancers (Table 8). Table 8. FDA Approved Glycoproteins as Cancer Markers.

Ludwig, JA and Weinstein, JN (2005), Nat Rev Cancer, 5, 845 Glycoproteins as new HCC markers: AFP-L3, GPC3, GP73

EXAMPLE 3

Using a similar process and sample set of the population as described in Example 1 , we demonstrated the utility of a set of six peptides as biomarkers for the detection of hepatocellular carcinoma. The characteristics of this sample set are summarized in Table 9, which shows increased markers of viral infections (HCV RNA, anti HCV, and anti-HBV) in cancer cases. Comparison of average spectra in HCC cases (n=73) and controls (n=77) showed differences in peptide abundance. Table 9. Association of HCV with HCC.

Presence ot HCV vfnis in chrome infection (HCV RNA], HCV antstødies (ante HCV). HBV surface antigen {HBsAg_? and HBV anybodies (ants HBV) was tested by RT-PCR and ELISA.

Chrome liver disease ξCLO) controls {n=5Q} are drrrhαss (25) ara3 fiLaosis (25) patterns ail HCV RNA positive and HBs.Ag negative with derailed clinical inTonnation

Data Processing and Analysis

Spectra were binned (136,000 m/z to 20,896 bins) and baseline corrected. Peaks were defined as slope change with a greater than a selected minimal intensity. Peaks were compared across all spectra and grouped in a window if within 0.03% relative mass. Maximum intensity in a window was normalized by total ion current and adjusted to an overall scale of 100%. The dataset was split into a training (25 HCC, 25 control) and a blinded validation (53 HCC, 47 control) set. The training set was used to select a useful combination of peaks for a support vector machine (SVM) classifier of HCC. Selection was based on a particle swarm optimization (PSO) or an ant colony optimization (ACO) algorithm. The blinded validation set was used for ROC analysis and for the evaluation of prediction accuracy.

An example of the results of using the PSO-SVM algorithm is shown in Fig. 4. A frequency plot was used to select the m/z windows for peptide classification from 600 runs of the PSO-SVM algorithm (Fig. 5). Six peptides were selected based on their relative abundance in HCC and controls. A box plot of the intensity of six selected peptides is shown in Fig. 6. ROC curves were generated for the six individual peptide marker candidates and a combined SVM classifier based on spectra from a blinded validation set of samples (53 HCC, 47 control (Fig.7)). The prediction accuracy for individual peptides (MALDI A, MALDI B, MALDI C, MALDI D, MALDI E, and MALDI F) is provided in Table 10. Individually, the sensitivity and specificity of each peptide is comparable to alpha fetoprotein, with sensitivities ranging from 50-96% and specificities ranging from 36-91%. An SVM classifier consisting of six markers, however, has a favorable sensitivity of 100% and specificity of 91%. The relative abundance of two peptides (MALDI A, 1865 Da and MALDI B, 935 Da) was compared among controls and patients with chronic liver disease, and early or late stage HCC (Fig. 8).

Table 10: Sensitivity and Specificity of individual peptides

Structural composition of the peptides in serum was determined by a combination of enzymatic sequencing and tandem MS using standard techniques. One hundred and nine peptides corresponding to 24 proteins were identified, including:

1. Kininogen (8)

2. inter-alpha-trypsin inhibitor heavy chain H4 precursor (10)

3. Fibrinogen alpha (31)

4. Prothrombin (6)

5. Coagulation faction XIII A chain (3)

6. Alpha-2-HS glycoprotein (5)

7. Apolipoprotein A-IV (5)

8. Apolipoprotein C-III (3)

9. Fibrinogen beta (7)

10. Complement C3 (5)

11. Complement C4 (3)

12. Clusterin precursor ( 1 )

13. Apolipoprotein E precursor ( 1 ) 14. Apolipoprotein A-I precursor ( 1 )

15. Gelsolin precursor ( 1 )

16. Zyxin 2 (7)

17. Thymosin beta-4 (2)

18. Plexin domain-containing protein 2 precursor ( 1 )

19. Corticotropin-lipotropin ( 1 )

20. Secretory granule proteoglycan core protein (1)

21. Dermicidin precursor ( 1 )

22. Alpha- 1 -antitrypsin ( 1 )

23. LIM and SH3 domain protein 1 (LASP-I) (1), etc.

Three hepatically secreted proteins involved in proteolysis and coagulation, SERPINC 1 (antithrombin III), LACI (lipoprotein associated circulating factor) and CPB2 (carboxypeptidase B2), were found to be differentially abundant in HCC. The relative amount of each hepatically secreted protein was determined by ELISA and compared between HCC cases and controls (Fig. 9). Quantification of the LMW serum fraction was also performed by isotope dilution and TripleQuad.

In accordance with another illustrative embodiment of the invention a set of 3 N-glycans are selected for their utility as biomarkers for the early detection and progression of hepatocellular carcinoma.

EXAMPLE 4

Methods for sample collection, N-glycan extraction, permethylation, MALDI-TOF, and data analysis were similar to as described in Example 1. Overlays of the MALDI-TOF spectra (1.5-5.5 kDa) in HCC cases (n=78) and controls (n=72) were prepared and spectra of glycans were processed as described for peptides in Example 3. Six glycans were selected by ACO-SVM analysis in HCC cases (n=78), population controls (n=72) and chronic liver disease controls (n=50) and compared for relative intensity in each group (Fig. 10). The ROC curves for three individual glycan marker candidates, with a mass of about 1799.8 Da, about 3241.9 Da, and about 1543.7 Da, as well as the combined SVM classifier are shown in Fig. 11. Serum levels of the three selected glycans were determined for patients with fibrosis, cirrhosis, and HCC (Fig. 12). An SVM classifier using the three glycans was found to have 92% sensitivity and 96% specificity in separating HCC (n=53) from CLD controls (n=25) in a blinded validation set. Glycans identified included compounds CC, DD, EE, FF, GG, HH, II, JJ, KK,

LL, MM, NN, the structure of which was previously disclosed in Table 2. In addition the following Glycan were also detected:

Glycan VV (mass: 3013.1):

Glycan XX (mass: 3271.6):

Glycan structures were grouped into complex/hybrid, high-mannose, bisected, and fucosylated structures which showed some general trends. Of the 48 known glycan structures, 32 were complex/hybrid glycans, 9 were bisected glycans, and 5 were high mannose glycans. Twenty of the complex/hybrid glycans and 5 of the bisected glycans were core fucosylated. Ten of the 20 core fucosylated complex/hybrid glycans increased in HCC compared to the controls, including CLD. Eight of the 12 complex/hybrid structures without core fucosylation decreased in HCC. Seven of the 9 bisected glycans, with or without core fucosylation, increased in CLD without much further increase in HCC. Three of the 5 high-mannose glycans increased in CLD, 4 of 5 further increased in HCC compared to CLD.

Claims

WHAT IS CLAIMED IS:

1. A method for detecting early stage hepatocellular carcinoma in a test patient, said method comprising the steps of analyzing a biological sample obtained from said test patient for the presence of a glycan selected from the group consisting of

Glycan 1 :

Glycan 2:

Glycan 3 :

Glycan 5 :

Glycan 6:

wherein : I — I = N-acetylglucosamine; ^ = mannose; I^ = galactose;

LS. = fucose; and v^¹ = N-acetylneuraminic acid; comparing the relative concentration of the glycan detected in said biological sample to concentrations of the glycan found in cancer free patients, wherein a statistically significant difference between the glycan concentration in said biological sample relative to those in cancer free patients indicates the presence of hepatocellular carcinoma in said test patient.

2. The method of claim 1 wherein said cancer free patients and test patient are humans.

3. The method of claim 2 wherein the biological sample is blood sample.

4. The method of claim 3 wherein the blood sample is a serum sample.

5. The method of claim 4 wherein said analyzing step comprises isolating serum proteins from said serum sample; enzymatically releasing glycans from their corresponding glycoproteins; and detecting said glycans by mass spectrometry analysis.

6. The method of claim 1 wherein the glycan detected is selected from the group consisting of

Glycan 1 :

Glycan 5 :

Glycan 6:

wherein : I — I = N-acetylglucosamine; ^ = mannose; I^ = galactose;

Z-A = fucose; and ^^ = N-acetylneuraminic acid.

7. The method of claim 1 wherein the relative concentration of 2- 6 different glycans are determined in the biological sample of said test patient and compared to concentrations of the corresponding glycans found in cancer free patients.

8. The method of claim 7 wherein the relative concentration of each of glycan 1 , glycan 5 and glycan 6 are determined in the biological sample of said test patient and compared to concentrations of the corresponding glycans found in cancer free patients.

9. A composition comprising an isolated N-glycan selected from the group consisting of

Glycan 1 :

Glycan 2:

Glycan 3 :

Glycan 5 :

Glycan 6:

wherein : I — I = N-acetylglucosamine; ¹^ = mannose; . > .

galactose;

= N-acetylneuraminic acid.

10. The composition of claim 9 comprising 2 or more isolated enzymatically cleaved N-glycans selected from the group consisting of

Glycan 1 :

Glycan 5 :

Glycan 6:

= N-acetylglucosamine; ^ = mannose; ^

galactose; l\ = fucose; and ^^ = N-acetylneuraminic acid.

11. A method for the selection of biomarkers of cancer comprising quantitation of a plurality N-glycans in a patient's serum samples and identifying those N-glycans that statistically differ in concentration relative to the corresponding glycan concentration in non-cancer patients.

12. A method for detecting early stage hepatocellular carcinoma in a test patient, said method comprising the steps of analyzing a biological sample obtained from said test patient for the relative concentration of an N-glycan; comparing the relative concentration of the N-glycan detected in said biological sample to concentrations of the N-glycan found in cancer free patients, wherein a statistically significant difference between the glycan concentration in said biological sample relative to those in cancer free patients is diagnostic for hepatocellular carcinoma.