Serum profile of low molecular weight fucosylated glycoproteins for early diagnosis of hepatocellular carcinoma
- Authors:
- Published online on: June 11, 2020 https://doi.org/10.3892/ol.2020.11727
- Pages: 1597-1606
-
Copyright: © Yao et al. This is an open access article distributed under the terms of Creative Commons Attribution License.
Abstract
Introduction
As one of the most common malignant tumors in China, new cases of hepatocellular carcinoma (HCC) and associated deaths accounted for ~50% of the global total in 2012 (1). The cause of HCC in 80% of Chinese patients can be attributed to hepatitis B virus (HBV) infection (2), which may further increase the mortality rate of HCC and decrease the 5-year survival rate after surgical treatment (3). Early diagnosis is effective to improve therapeutic effects (4). At present, the diagnosis of HCC is mainly dependent on clinical data, imaging tests of the liver and measurement of serum α-fetoprotein (AFP) levels. Although AFP is the most widely used serum marker for diagnosing and monitoring HCC, >40% of patients with HCC may exhibit normal AFP levels, particularly in the early stages (5,6). Due to the low sensitivity of measuring AFP levels, false-positive results constitute another problem, as benign liver tumors may also cause a rise in the AFP serum levels. Therefore, the sensitivity and specificity of AFP in HCC diagnosis is questionable (7).
AFP can specifically bind to Lens culinaris agglutinin (LCA), referred to as AFP-L3, and this has been reported to be a specific marker for diagnosing HCC (8). Since LCA can further bind to fucose, AFP-L3 is also referred to as fucosylated AFP (9). In 2005, the United States Food and Drug Administration officially approved AFP-L3 as a marker for diagnosing HCC (10).
However, the application of AFP or AFP-L3 alone cannot meet the clinical requirements of patients with AFP-negative HCC. To date, research in patients with HCC with abnormal serum AFP levels has revealed that fucosylation may not be restricted to AFP, and a previous study confirmed its presence in other proteins (11). Although the molecular mechanism of abnormal fucosylation in HCC has not been fully clarified, previous research identified multiple proteins with differential fucosylation, including the Golgi protein 73 (12), haptoglobin (13), α-1-acid glycoprotein (14) and kininogen (15), which may be used as potential markers for HCC diagnosis. However, none of these markers could be applied as widely as AFP-L3 in clinical practice.
Abnormal fucosylation of different glycoproteins provides a novel insight into the diagnosis of HCC, but no effective method has been presented to compare and analyze all fucosylated glycoproteins in patients with HCC. In our previous study (16), matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (MALDI-TOF MS) was used to capture serum-associated fucosylated glycoproteins using LCA-coated magnetic beads. By using MALDI-TOF MS, a salivary protein fingerprint model was established to assist the diagnosis of HCC. Since MALDI-TOF MS is mainly utilized to detect proteins <10 kD, our previous study used trypsin to digest all proteins captured by the magnetic beads. However, since protein fractionation required complex operation and blinded test results were unsatisfactory, the aforementioned model may not be appropriate for use in clinical practice.
The present study was designed to fragment all captured proteins and to only detect those that were <10 kD, including peptides produced by the natural degradation of some proteins. Therefore, the present study aimed to provide a novel and more appropriate method for use in clinical practice, and discussed its potential in HCC diagnosis.
Patients and methods
Patients
Serum samples (n=425) from patients (n=339) and healthy controls (n=86) were collected at Fuzhou General Hospital of Nanjing Command (Fuzhou, China) between November 2007 and December 2017. The present study was approved by the Ethics Committee of Fuzhou General Hospital of Nanjing Military Command, and written informed consent was obtained from all participants. Demographic and clinical data were obtained, and a blood sample was collected from each study subject. The HBV infection status was assessed using a Hepatitis B Virus Surface Antigen Diagnostic kit (enzyme-linked immunoassay) (Livzon Pharmaceutical Group Inc.; http://en.livzon.com.cn/). Hepatitis C virus (HCV) infection status was assessed using a Hepatitis C Virus Antibody Diagnostic kit (enzyme-linked immunoassay) (Livzon Pharmaceutical Group Inc.). HBV-DNA was detected using a Hepatitis B Virus Nucleic Acid Detection kit (Sun Yat-Sen University Daan Gene Co., Ltd.). (PCR-fluorescent probe method). These kits were used according to the manufacturer's protocols.
Four groups of patients were enrolled in the present study (Table I). The first group (n=86) included subjects with no history of liver disease and normal liver biochemistry, no risk factors for viral hepatitis and an alcohol consumption <40 g/week. The second group (n=86) consisted of patients with chronic hepatitis B (CHB) who were positive for HBsAg, hepatitis B e-antigen and HBV DNA, and negative for HCsAg and abnormal liver biochemistry. The third group (n=97) consisted of patients with HBV-infected cirrhosis (HBV-cirrhosis). Diagnosis of cirrhosis was based on liver histology or on clinical, laboratory and imaging evidence of hepatic decompensation or portal hypertension (17). Patients with cirrhosis without a suspected or malignant nodule were assessed by B-ultrasound (US). If the serum AFP level was elevated, CT or magnetic resonance imaging scans of the liver within 3 months before enrollment and 6 months after enrollment had to exhibit no liver mass. The fourth group (n=141) included patients with HBV-infected HCC (HBV-HCC). The diagnosis of HCC was carried out by histopathology (18). Tumor staging was determined via the United Network of Organ Sharing that uses the modified TNM staging system for HCC (19).
All patients were randomly divided into two cohorts. The training cohort was composed of 26 healthy volunteers, 21 patients with CHB, 32 patients with HBV-cirrhosis and 32 patients with HBV-HCC. The blinded test cohort was composed of 60, 65, 65 and 109 patients, respectively. Demographic data and etiology of liver diseases of all patients and healthy volunteers are presented in Table II.
In conjunction with an ongoing cohort study, prediagnostic sera from 9 patients with HBV-cirrhosis who developed HBV-HCC within 1 year of US screening and 6 patients with HBV-cirrhosis who remained free of HBV-HCC for the subsequent 3 years were also obtained. These subjects constituted the third analysis cohort (Table III).
Preparation of blood samples
Blood from patients was collected from an elbow vein 3–5 days before surgery in glass tubes without additive (BD Vacutainer®; BD Diagnostics) and was allowed to clot at room temperature for 40 min. Serum was separated by centrifugation at 912 × g for 15 min at room temperature, immediately split into 200-µl aliquots and frozen at −80°C until analysis. The time between collection and frozen storage was <60 min.
The processing, collection and storage protocols were the same for all individuals. Each sample used for proteomic profiling had not been thawed more than once. Blood samples from patients were drawn before commencement of treatment.
Serum protein fractionation
Serum samples were thawed and purified using a LCA magnetic beads kit (Bruker Corporation). A total of 10 µl serum was mixed with 2 µl beads and the samples were purified by binding, washing and elution according to the manufacturer's protocol. Each incubation (at room temperature) step lasted 1 min. Elution was carried out with 10 µl elution buffer and the purified material was 8-fold diluted with the elution solution prior to MS analysis.
MALDI-TOF MS
For MALDI-TOF MS analysis, 1 µl of the aforementioned diluted purified sample was mixed with 0.5 µl matrix solution (0.4 mg/ml α-cyano-4-hydroxycinnamic acid in ethanol:acetone 2:1) and allowed to dry onto the MALDI sample plate (600-µm AnchorChip™; Bruker Daltonics; Bruker Corporation). Laser desorption was targeted randomly on the sample plate and samples were measured using an Autoflex II MALDI-TOF mass spectrometer (Bruker Daltonics; Bruker Corporation) operated in positive ion linear (reflection) mode. Ionization was achieved by irradiation with a 50 Hz nitrogen laser (λ=337 nm). Spectra were the mean of 100 ionizations with fixed laser power in linear geometry mode, and mass maps were obtained in reflectron mode. The spectra were calibrated externally with a mixture of protein/peptide standards in the range of 1,000-10,000 Da (Bruker Daltonics; Bruker Corporation). Three MALDI preparations (MALDI spots) were measured from each sample. For each MALDI spot, 400 spectra were acquired (50 laser shots at eight different spot positions). The spectra from all samples (training cohort) were imported into the CLINPROT™ software (Bruker Daltonics; Bruker Corporation) for spectra processing, model building, model recognition and internal model validation. CLINPROT™ software is composed of data acquisition software FlexControl 2.2, viewing software FlexAnalysis 3.0 and analysis software ClinProTools 2.1. The spectra were processed in the following order: i) Spectra normalization to total ion current; ii) spectra recalibration using prominent peaks; iii) baseline subtraction, peak smoothing (Savitsky-Golay algorithm) and peak detection; and iv) calculation of peak areas for each spectrum. Peak detection was performed using S/N≥5 and peak areas were calculated using a zero level integration type.
All MALDI-TOF MS spectra were analyzed using flexAnalysis™ version 3.0 to detect the peak intensities of interest and CLINPROT™ software (both Bruker Daltonics; Bruker Corporation) to compile the peaks across the spectra obtained from all samples of the training set. This analysis allowed for discrimination between HCC and control samples. The Supervised Neural Network (SNN) contained within this software suite was used to select clusters of signals for the model to discriminate between the two populations. A leave-20%-out cross-validation was calculated to avoid over-fitting of pattern recognitions. The spectra from all samples of the blinded test cohort were processed according to the same method. The spectra data of two sets (the training cohort and the blinded test cohort) were used for internal and external model validation.
Sample processing and MALDI analysis
Each serum sample from the training cohort was analyzed using LCA magnetic beads enriched with fucosylated glycoproteins. The MALDI-TOF approach was applied to exhibit spectral peaks in the 1,000-10,000 m/z range, as well as the effect of pre-processing and normalization. Each serum sample was tested in duplicate. ClinprotTools v2.1 (Bruker Daltonics; Bruker Corporation) was used to statistically analyze the differences in peak positions and intensities.
Blinding
The analysts were blinded to the origin of the samples. Initially, the established model was used to distinguish HCC samples from all serum samples. Subsequently, healthy samples were excluded, and the analysts attempted to distinguish HCC samples from samples with HBV-associated diseases. Finally, the analysts attempted to exclude CHB, and distinguish HCC from all cirrhosis samples. At the same time, in the third analysis cohort, the analysts attempted to distinguish HCC from all cirrhosis samples.
Nano-liquid chromatography (LC)-electrospray ionization (ESI)-tandem mass spectrometry (MS/MS) analysis
All fucosylated proteins in serum samples purified and enriched by LCA magnetic beads in the eluent were directly used for Nano LC-ESI-MS/MS analysis. The enriched proteins dried using vacuum centrifugation (at room temperature, 2,000 × g for ~3 h), and re-suspended in the solution containing 5% ACN and 0.1% formic acid. Peptides were separated by nano-LC and analyzed using a mass spectrometer. The experiments were performed on an LC-20AD system (Shimadzu Corporation) connected to an LTQ Orbitrap mass spectrometer (Thermo Fisher Scientific, Inc.) equipped with an online nanoelectrospray ion source (Bruker-Michrom, Inc.). In total, 5 µl peptide sample was loaded onto the trap-column with a flow rate of 60 µl/min, eluted with a gradient of 5–45% solvent B (95% ACN in 0.1% formic acid) over 60 min and then injected into the mass spectrometer at a constant column-tip flow rate of ~500 nl/min. The electrospray voltage of 2.2 kV vs. the inlet of the mass spectrometer was used. The Orbitrap mass spectrometer was operated in the data-dependent mode to switch automatically between MS and MS/MS acquisition. Survey full-scan MS spectra were acquired in Orbitrap with a mass resolution of 60,000 at 400 m/z followed by eight sequential MS/MS. The AGC target was set to 1,000,000, and the maximum injection time was 300 ms. MS/MS acquisition was performed in Orbitrap with the resolution was 15,000 at 400 m/z. The intensity threshold was 50,000, and the maximum injection time was 100 ms. The AGC target was set to 100,000, and the isolation window was 2.0 m/z. Ions with charge states 2+, 3+ and 4+ were fragmented by collisional induced dissociation with a normalized collision energy of 30%. In all cases, one microscan was recorded using dynamic exclusion of 30 sec (20).
Database searching
Mass spectra were searched against the human International Protein Index (IPI) database (ftp://ftp.ebi.ac.uk/pub/databases/IPI/last_release/old/HUMAN/ipi.HUMAN.v3.35.fasta.gz.) (IPI human v3.35 fasta with 68,348 entries) using the Bioworks software (v3.3.1; Thermo Fisher Scientific, Inc.) based on the Sequest algorithm. The search parameters included the following: i) Precursor ion mass tolerance <10 ppm; ii) fragment ion mass tolerance <1 Da; and iii) digestion mode was unspecific. The corresponding reversed sequence database was used to generate score criteria that yielded an estimated false positive rate of 1% (precision of 0.99). To minimize false positives, all output results were combined together using a homemade software to generate score criteria: The cross-correlation scores (Xcorr) of matches were >2.81, 3.22 and 3.41 for charged state 2, 3 and 4 peptide ions, respectively. To obtain reliable protein identification, only peptides with a ΔCn score >0.1 were used, and the ranks of the primary scores were 1, and the posterior error probability (PEP) <0.001.
Statistical analysis
The processed spectra were analyzed using an unpaired Student's t-test and Wilcoxon tests. All the results were expressed as mean ± standard deviation (SD) and P<0.05 was considered to indicate a statistically significant difference.
Results
MALDI analysis
Spectra of serum samples from healthy controls and patients with CHB, HBV-cirrhosis and HBV-HCC were imported into the CLINPROT™ software to calculate the average spectrum (Fig. 1). A total of 48 peaks exhibited differential expression (P<0.05; data not shown), and Table IV lists 14 peaks of those with significant difference (P<0.002).
 | Table IV.Mass spectral characteristics of proteins with differential expression (P<0.002) among four cohorts. |
To assess the classification efficiency, the mean ± SD of the mass of the 14 peak values was calculated. Among them, 2,660.05 and 7,640.48 Da demonstrated the most significant differential expression (P<0.0001; Fig. 2 presents their average peak values and relative concentrations in the four groups). Notably, the peak values of 2,660.05 Da in healthy controls and patients with CHB were similar, and the intensity level was increased in patients with HBV-cirrhosis and was increased even further in patients with HBV-HCC. The current data indicated that the degree of fucosylation may increase with the development of HCC. On the other hand, the peak value of 7,640.48 Da was the highest in healthy controls, suggesting a decreasing degree of fucosylation during disease progression. Overall, the two peptides appeared to have discriminatory potential (Fig. 3).
Analysis of classification models and blind testing
Peak values obtained from the training cohort were processed using the SNN algorithm in CLINPROT™ software to establish classification models for cross-validation. Using this method, three models were established to distinguish HCC from all serum samples (HBV-HCC vs. HBV-cirrhosis + CHB + healthy controls), from HBV-associated chronic liver diseases (HBV-HCC vs. HBV-cirrhosis + CHB) and from HBV-associated cirrhosis (HBV-HCC vs. HBV-cirrhosis). Table V presents the peak values, constituting three classification models.
By using serum samples collected from the blinded test cohort, the accuracy of the established classification models was examined. According to the present results, the models had a sensitivity and specificity of 74.31 and 76.32%, respectively, to distinguish HCC from all serum samples, 81.65 and 83.08%, respectively, to distinguish HCC from HBV-associated cirrhosis and chronic hepatitis B, and 88.99 and 84.62%, respectively, to distinguish HCC from HBV-associated cirrhosis (Table VI). When these models were combined with AFP measurement (AFP >20 ng/ml), the sensitivity and specificity were elevated to 80.73 and 87.37%, 87.16 and 90.00%, and 92.66 and 93.84%, respectively (Table VI).
| Table VI.Blind test results of HCC diagnostic classification models and the results of joint AFP detection. |
In addition, serum samples from 9 patients with cirrhosis 1 year before HCC diagnosis and from 6 patients with cirrhosis exhibiting no signs of HCC for 3 years were collected. The HBV-HCC vs. HBV-cirrhosis model was applied to analyze their data; 7/9 (77.78%) patients with HCC were classified into the HCC group, and all 6 patients without HCC were classified into the cirrhosis group (Table VII). The present results suggest that the aforementioned classification models may be useful for early HCC diagnosis.
Identification of candidate biomarkers
Two peptides (2,660.05 and 2,368.23 Da) were included in the peak values with differential expression (Table IV) and in the peak values constituting the classification models (Table V). Therefore, nano-LC-ESI-MS/MS analysis was performed for their isolation and identification. This revealed that the 2,660.05 Da peptide was isoform 1 of inter-α-trypsin inhibitor heavy chain 4 precursor (ITIH4), while the 2,368.23 Da peptide was thymosin β-4-like protein 3 (TMSL3; Fig. 4).
Discussion
In our previous study (16), an LCA-based fucosylated glycoprotein spectrum was established for differentiating HCC diagnosis; however, the ultimate goal is to develop a novel method that is more useful in clinical practice. MALDI-TOF MS could detect peak values with a mass-to-charge ratio between 1,000 and 10,000 m/z, which are peptides with a molecular weight <10 kD. Previously, trypsin was used to digest all fucosylated glycoproteins that were captured by LCA-coated beads (16); however, this complex operation, together with the unsatisfactory results in blinded validation, resulted in this model being less appropriate for clinical application. In the present study, the captured proteins were used directly for MALDI-TOF MS. Therefore, only proteins <10 kD, including peptides produced from natural degradation of some proteins, were measured. In comparison with our previous study, the model established in the present study presented some advantages: i) It identified more differentially expressed peak values (while in the previous study 89 peak values with differential expression were detected and only 36 were of statistical significance, the present study identified 48/64 detected peak values to be statistically significant); ii) it exhibited a stronger predictive ability (data not shown); and iii) it obtained improved blinded test results (the sensitivity and specificity of the HBV-HCC vs. HBV-cirrhosis model were 81.00 and 82.00%, respectively, in our previous study, while they were increased to 88.99 and 84.62%, respectively, in the present study). The main reason for these differences may be due to our previous study analyzing the peptides after protein digestion, while the present peptides included not only glycopeptides, but also other peptides. Furthermore, the majority of these peptides had a molecular weight <5,000 Da, indicating that numerous peptide segments with the same or similar size may mask the differential expression of glycopeptides (21).
The sample size for blind testing was increased to 299 in the present study, thus the results were more reliable. Additionally, the HBV-HCC vs. HBV-cirrhosis classification model was used to analyze serum samples collected from patients with cirrhosis 1 year before being diagnosed with HCC. The model identified 7/9 (77.78%) patients as HCC, suggesting that the model is able to identify HCC prior to diagnostic imaging. Indeed, when HCC is diagnosed by imaging, it is often too late to miss the optimal treatment time. The potential of the HBV-HCC vs. HBV-cirrhosis model established in the present study may require further examination by expanding the sample size for blinded testing, as well as investigating differential peak values involved in the model. However, the ability of the present model to diagnose HCC early suggests that it may be able to assist in the clinical staging of HCC and cirrhosis. Since ClinprotTools v2.1 is mainly suitable for analysis between two groups of samples, the simultaneous analysis of multiple groups of samples requires improvement, which limits the development of studies on classification models and clinical stages of HCC and cirrhosis. Therefore, upon setting up the experimental group, patients with liver cancer were not divided into subgroups according to their stage, and patients with cirrhosis were not graded for liver function tests.
The present results indicate that the protein peaks of the three diagnostic classification models in Table V were not the same, and that they were also different from the protein peaks listed in Table IV. There are two main reasons for this discrepancy. The first reason is that the model established in the present study used the SNN algorithm. The most important consideration of this algorithm is the ability to classify the combination of protein peaks, not the classification ability of a single protein peak (22). Therefore, when establishing a diagnostic classification model, it is important not to select the protein peaks that have the most significant differences, but to consider which protein peaks are most powerful when combined. This is why the differential protein peaks used in the classification of the HCC group and the different control groups in Table V were different. The second reason is that Table IV lists only the 14 protein peaks with the most significant differences when the analysis software analyzed four groups of samples simultaneously (P<0.002). However, a total of 48 proteins were identified as differentially expressed (P<0.05) and only 14 protein peaks with the most significant differences (P<0.002) were presented due to space limitations, which does not indicate that these 14 protein peaks were more important than others when building the classification diagnosis model. For the purpose of the present study, only some examples of proteins with different expression patterns in the occurrence and development of HCC were presented (Fig. 2). Additionally, an example was provided to illustrate the ability of two differential protein peaks to classify four groups of samples (Fig. 3). The present study tried to establish a classification model to diagnose four groups of samples simultaneously, but the results were unsatisfactory (data not shown). The three diagnostic classification models in Table V were used to distinguish between the HCC group and the different control groups, not to distinguish among the four groups of samples simultaneously, hence why they are different from the protein peaks listed in Table IV.
Despite these differences, the 2,660.05 and 2,368.23 Da peak values were identified in both calculations (Tables IV and V). Therefore, these peptides were purified and identified as ITIH4 and TMSL3. Since ITIH4, TMSL3 and their respective antibodies are not sold commercially, and our laboratory does not have the technology for recombinant protein expression, follow-up experiments, such as western blotting, investigating these two proteins were not performed in the present study. Future research should focus on indicating whether these two proteins may be used as novel biomarkers for HCC diagnosis.
Notably, our previous study (16) reported defucosylation of some glycoproteins during the development and progression of HCC. Although this finding was inconsistent with the majority of existing studies (9–14), it was re-observed in the present study. The 7,640.48 Da peptide had the highest expression in healthy controls, suggesting that the degree of fucosylation may decrease during HCC progression. Therefore, fucosylation of serum glycoproteins is a complex and variable process during HCC development, and its mechanism requires more in-depth investigation in the future.
In conclusion, the present study focused on fucosylated glycoproteins with a molecular weight <10 kD and established a classification model of low-molecular-weight fucosylated glycoproteins for early diagnosis of HCC. The proposed diagnostic model underwent preliminary clinical validation and exhibited great potential in the early diagnosis of HCC. However, the present study was only a single-center study. Additionally, cases were not divided into subgroups, data for comparison with other types of cancer were lacking and there were various limitations for its practical application in the clinic. In the future, the number of samples should be increased, cases should be subdivided, and multi-center studies should be conducted to make the current model more appropriate for use in clinical practice.
Acknowledgements
Not applicable.
Funding
The present study was supported by the Natural Science Foundation of Fujian (grant nos. 2017J01326 and 2017J01219), Science and Technology Plan Project of Longhai (grant no. 201811) and the Key Project of Science and Technology of Fujian (grant no. 2012Y0058).
Availability of data and materials
The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.
Authors' contributions
JL and MC conceived and supervised the present study. JL and KW designed the experiments. WY, YJ and ZH performed the experiments. HY and SH collected the blood samples. WY and YH analyzed the data. WY and YJ wrote the manuscript. MC and JL revised the manuscript. All authors reviewed the results and all authors read and approved the final manuscript.
Ethics approval and consent to participate
The present study was reviewed and approved by the Ethics Committee of Fuzhou General Hospital of Nanjing Command (Fuzhou, China), and written informed consent was obtained from all participants.
Patient consent for publication
Not applicable.
Competing interests
The authors declare that they have no competing interests.
Glossary
Abbreviations
Abbreviations:
|
MALDI-TOF MS |
matrix-assisted laser desorption/ionization time-of-flight mass spectrometry |
|
AFP |
α-fetoprotein |
|
ITIH4 |
inter-α-trypsin inhibitor heavy chain 4 precursor |
|
TMSL3 |
thymosin β-4-like protein 3 |
|
HCC |
hepatocellular carcinoma |
|
HBV |
hepatitis B virus |
|
LCA |
Lens culinaris agglutinin |
|
CHB |
chronic hepatitis B |
|
US |
ultrasound |
|
CT |
computed tomography |
|
SNN |
Supervised Neural Network |
References
|
Torre LA, Bray F, Siegel RL, Ferlay J, Lortet-Tieulent J and Jemal A: Global cancer statistics, 2012. CA Cancer J Clin. 65:87–108. 2015. View Article : Google Scholar : PubMed/NCBI | |
|
Marrero JA: Hepatocellular carcinoma. Curr Opin Gastroenterol. 22:248–253. 2006. View Article : Google Scholar : PubMed/NCBI | |
|
El-Serag HB, Mason AC and Key C: Trends in survival of patients with hepatocellular carcinoma between 1977 and 1996 in the United States. Hepatology. 33:62–65. 2001. View Article : Google Scholar : PubMed/NCBI | |
|
Hoofnagle JH and di Bisceglie AM: The treatment of chronic viral hepatitis. N Engl J Med. 336:347–356. 1997. View Article : Google Scholar : PubMed/NCBI | |
|
Di Bisceglie AM and Hoofnagle JH: Elevations in serum alpha-fetoprotein levels in patients with chronic hepatitis B. Cancer. 64:2117–2120. 1989. View Article : Google Scholar : PubMed/NCBI | |
|
Sherman M, Peltekian KM and Lee C: Screening for hepatocellular carcinoma in chronic carriers of hepatitis B virus: Incidence and prevalence of hepatocellular carcinoma in a North American urban population. Hepatology. 22:432–438. 1995. View Article : Google Scholar : PubMed/NCBI | |
|
Buamah PK, Gibb I, Bates G and Ward AM: Serum alpha fetoprotein heterogeneity as a means of differentiating between primary hepatocellular carcinoma and hepatic secondaries. Clin Chim Acta. 139:313–316. 1984. View Article : Google Scholar : PubMed/NCBI | |
|
Taketa K, Okada S, Win N, Hlaing NK and Wind KM: Evaluation of tumor markers for the detection of hepatocellular carcinoma in Yangon General Hospital, Myanmar. Acta Med Okayama. 56:317–320. 2002.PubMed/NCBI | |
|
Mita Y, Aoyagi Y, Suda T and Asakura H: Plasma fucosyltransferase activity in patients with hepatocellular carcinoma, with special reference to correlation with fucosylated species of alpha-fetoprotein. J Hepatol. 32:946–954. 2000. View Article : Google Scholar : PubMed/NCBI | |
|
Mehta A and Block TM: Fucosylated glycoproteins as markers of liver disease. Dis Markers. 25:259–265. 2008. View Article : Google Scholar : PubMed/NCBI | |
|
Comunale MA, Lowman M, Long RE, Krakover J, Philip R, Seeholzer S, Evans AA, Hann HW, Block TM and Mehta AS: Proteomic analysis of serum associated fucosylated glycoproteins in the development of primary hepatocellular carcinoma. J Proteome Res. 5:308–315. 2006. View Article : Google Scholar : PubMed/NCBI | |
|
Jiang K, Shang S, Li W, Guo K, Qin X, Zhang S and Liu Y: Multiple lectin assays for detecting glyco-alteration of serum GP73 in liver diseases. Glycoconj J. 32:657–664. 2015. View Article : Google Scholar : PubMed/NCBI | |
|
Shang S, Li W, Qin X, Zhang S and Liu Y: Aided diagnosis of hepatocellular carcinoma using serum fucosylated haptoglobin ratios. J Cancer. 8:887–893. 2017. View Article : Google Scholar : PubMed/NCBI | |
|
Tanabe K, Kitagawa K, Kojima N and Iijima S: Multifucosylated alpha-1-acid glycoprotein as a novel marker for hepatocellular carcinoma. J Proteome Res. 15:2935–2944. 2016. View Article : Google Scholar : PubMed/NCBI | |
|
Wang M, Shen J, Herrera H, Singal A, Swindell C, Renquan L and Mehta A: Biomarker analysis of fucosylated kininogen through depletion of lectin reactive heterophilic antibodies in hepatocellular carcinoma. J Immunol Methods. 462:59–64. 2018. View Article : Google Scholar : PubMed/NCBI | |
|
Liao J, Zhang R, Qian H, Cao L, Zhang Y, Xu W, Li J, Wu M and Yin Z: Serum profiling based on fucosylated glycoproteins for differentiating between chronic hepatitis B and hepatocellular carcinoma. Biochem Biophys Res Commun. 420:308–314. 2012. View Article : Google Scholar : PubMed/NCBI | |
|
Marrero JA, Fontana RJ, Fu S, Conjeevaram HS, Su GL and Lok AS: Alcohol, tobacco and obesity are synergistic risk factors for hepatocellular carcinoma. J Hepatol. 42:218–224. 2005. View Article : Google Scholar : PubMed/NCBI | |
|
Bruix J, Sherman M, Llovet JM, Beaugrand M, Lencioni R, Burroughs AK, Christensen E, Pagliaro L, Colombo M and Rodés J; EASL Panel of Experts on HCC, : Clinical management of hepatocellular carcinoma. Conclusions of the Barcelona-2000 EASL conference. European Association for the Study of the Liver. J Hepatol. 35:421–430. 2001. View Article : Google Scholar : PubMed/NCBI | |
|
Faria SC, Szklaruk J, Kaseb AO, Hassabo HM and Elsayes KM: TNM/Okuda/Barcelona/UNOS/CLIP International Multidisciplinary Classification of hepatocellular carcinoma: Concepts, perspectives, and radiologic implications. Abdom Imaging. 39:1070–1087. 2014. View Article : Google Scholar : PubMed/NCBI | |
|
Cao J, Hu Y, Shen C, Yao J, Wei L, Yang F, Nie A, Wang H, Shen H, Liu Y, et al: Nanozeolite-driven approach for enrichment of secretory proteins in human hepatocellular carcinoma cells. Proteomics. 9:4881–4888. 2009. View Article : Google Scholar : PubMed/NCBI | |
|
Sparbier K, Asperger A, Resemann A, Kessler I, Koch S, Wenzel T, Stein G, Vorwerg L, Suckau D and Kostrzewa M: Analysis of glycoproteins in human serum by means of glycospecific magnetic bead separation and LC-MALDI-TOF/TOF analysis with automated glycopeptide detection. J Biomol Tech. 18:252–258. 2007.PubMed/NCBI | |
|
Baggerly KA, Morris JS, Wang J, Gold D, Xiao LC and Coombes KR: A comprehensive approach to the analysis of matrix-assisted laser desorption/ionization-time of flight proteomics spectra from serum samples. Proteomics. 3:1667–1672. 2003. View Article : Google Scholar : PubMed/NCBI |
