Proteomics of Hepatocellular Carcinoma: Serum Vimentin As a

Feb 2, 2010 - Dr. John Luk, Departments of Pharmacology and Surgery, National ... Small hepatocellular carcinomas (HCCs) can be effectively cured by ...
0 downloads 5 Views 2MB Size
Proteomics of Hepatocellular Carcinoma: Serum Vimentin As a Surrogate Marker for Small Tumors (e2 cm) Stella Sun,† Ronnie T. P. Poon,† Nikki P. Lee,† Chun Yeung,† K. L. Chan,† Irene O. L. Ng,‡ Philip J. R. Day,§ and John M. Luk*,†,| Department of Surgery, University of Hong Kong, Pokfulam, Hong Kong, China, Department of Surgery and Department of Pharmacology, National University Health System, National University of Singapore, 117597 Singapore, Department of Pathology, University of Hong Kong, Queen Mary Hospital, Pokfulam, Hong Kong, China, and Interdisciplinary Molecular Medicine, The Manchester Interdisciplinary Biocentre, University of Manchester, Manchester, United Kingdom Received November 25, 2009

Small hepatocellular carcinomas (HCCs) can be effectively cured by surgery with good clinical outcomes. However, the conventional AFP marker is ineffective in detecting small tumors. Here we employed a proteomic profiling approach to identify a candidate marker for HCC (e2 cm) in tumor tissues and then evaluate its clinical feasibility in patients’ sera. The study was divided into 2 phases. (i) Biomarker discovery: we collected 76 frozen liver tissues (40 HCC and 36 controls) for proteomics profiling. Candidate protein markers were identified by MALDI-TOF/TOF and confirmed by immunoblot and qPCR. (ii) Clinical evaluation: Selected biomarker was tested by ELISA for sensitivity and specificity using serum samples from a separate cohort of 152 subjects (88 HCC and 64 controls). Vimentin was found significantly overexpressed in HCC, in particular the small-size subgroup (e2 cm) with p < 0.01. When tested in the serum samples, vimentin level was significantly higher in small tumors than the nonneoplastic controls (AUC ) 0.69 and p < 0.01). Further analysis suggested that elevated circulating vimentin level could detect small HCC at 40.91% sensitivity and 87.50% specificity. Moreover, vimentin was found to be superior to serum AFP assayed at different cut-offs in detecting small tumors. When combined with AFP, the detection sensitivity and specificity could be further enhanced to 58.77 and 98.15%, respectively. In conclusion, serum vimentin is a potential surrogate marker, either alone or in combination with AFP, for detection of small HCCs. Keywords: AFP • biomarker • HCC • tumor size • tumor detection • cancer screening

Introduction Despite the development of clinical therapeutic strategies, patients with hepatocellular carcinoma (HCC) have gained only marginal improvement in survival during the past two decades. The estimated incidence of new cases is more than 500 000 per year, causing significantly high death rates globally.1 Among the cases of HCC, the incidence is particularly high in China and Eastern Asia (over 20 cases/100 000 population),2 where hepatitis B virus (HBV) infections are endemic, and patients with chronic hepatitis are prompted to develop cirrhosis and HCC.3 The lethality of HCC is due mainly to the advanced stage of the malignancy at diagnosis, owing to the limitation of current surveillance programs and the absence of effective treatment for advanced HCC.4 Early detection of HCC in high* To whom correspondence should be addressed. Dr. John Luk, Departments of Pharmacology and Surgery, National University of Singapore, NUHS Clinical Research Centre, MD11 #05-09, 10 Medical Drive, 117597 Singapore. Fax: 65-68737690. E-mail: [email protected]. † Department of Surgery, University of Hong Kong. ‡ Department of Pathology, University of Hong Kong. § University of Manchester. | National University of Singapore. 10.1021/pr901085z

 2010 American Chemical Society

risk subjects (hepatitis and cirrhosis) could improve their clinical outcomes5 and, indeed, many lives can be saved by effective treatments such as hepatectomy and local radiofrequency ablation therapies when the tumors are still small.6 However, the detection rate for small HCC is extremely low and varies in studies,7,8 largely due to the asymptomatic nature of HCC and the current ineffective detection tools. A solitary small (e2 cm) HCC is regarded as the earliest stage of tumor development.9 Treatment of small HCC by surgical resection can achieve a low recurrence rate and long survival time. The overall survival rates at 3, 5, and 10 years were reported as 83, 69, and 36%, respectively, and the corresponding disease-free survival rates were 63, 41, and 10%.10 It has also been shown recently that ablation of small HCC e 2 cm can achieve favorable long-term results.11 Today, the current practice for HCC screening includes hepatic ultrasonography and measurement of serum alphafetoprotein (AFP) at 6-12 month intervals but neither of these is adequate for detecting very small (e2 cm) HCC tumors. The sensitivity (Sen: 39-65%) and specificity (Spe: 76-94%) of AFP are limited to the overall HCCs12 and also vary significantly13 owing to benign nodules and non-neoplastic liver conditions. Journal of Proteome Research 2010, 9, 1923–1930 1923 Published on Web 02/02/2010

research articles

Sun et al.

Table 1. Demographic and Clinicopathological Characteristics of Patients in HCC Cancer Biomarker Study training set (for 2-DE analysis)

validation set (for ELISA test)

n ) 76 clinical variables

Male:Female Age (yr)a HBsAg(() SGOT (u/L)a SGPT (u/L)a Total bilirubin (µmol/L)a Child Pugh score Grade A Grade B Grade C Tumor characteristics AFP (ng/mL)a Venous infiltrationb Absent Present Tumor stage (AJCC) Stage I Stage II Stages IIIA&B Edmonson gradeb,c Grades I-II Grades III Grade IV Tumor size (cm) e2 >2 No of tumor nodules Undiffused (1-2 nodules) Diffused a

Mean ( SD.

b

controls (n ) 36)

n ) 152 HCC (n ) 40)

HCC (n ) 88)

34:2 50.3 ( 10.1 20:16 44.0 ( 18.8 40.0 ( 30.5 12.8 ( 3.3

40:0 52.3 ( 8.6 40:0 72.3 ( 46.4 67.2 ( 65.4 14.8 ( 18.2

48:16 41.5 ( 11.8 40:24 47.8 ( 72.1 48.5 ( 55.9 12.7 ( 13.1

72:16 56.1 ( 11.3 81:7 60.7 ( 44.9 58.7 ( 44.7 28.4 ( 87.8

N/A

(100%)

N/A

80 (91%) 5 (6%) 3 (3%)

192.1 ( 208.6

50 007.16 ( 22 116.78

122.30 ( 270.38

25 304.3 ( 139 709.41

21 (53%) 19 (47%)

38 (60%) 25 (40%)

17 (43%) 10 (25%) 13 (32%)

48 (55%) 20 (23%) 19 (22%)

11 (31%) 17 (49%) 7 (20%)

14 (22%) 44 (69%) 6 (9%)

10 (25%) 30 (75%)

44 (50%) 44 (50%)

36 (90%) 4 (10%)

67 (84%) 13 (16%)

Partial data was not available, and statistics was based on available data. c Tumor cell differentiation.

In this study, we employed proteomic profiling analysis of HCC tumors ranging from 1.5 to 15 cm, to identify protein markers that can effectively detect small HCCs. We further evaluated the clinical feasibility in serum samples by ELISA in a separate cohort of HCC patients.

Materials and Methods Patient Selection. Patients included in this study were Hong Kong Chinese (mean age: 52.3 ( 8.6 years) with chronic HBV infection (Table 1). We collected 60 tissues samples from 20 patients with liver cirrhosis and 40 with HCC during surgical resection; 16 normal liver tissues from residual grafts of liver donors were included for comparison. The 10 recruited patients with small (e2 cm) HCC tumors had a Child Pugh liver function score of grade A and stage I/II of new American Joint Committee on Cancer (AJCC) staging. The other 30 HCC patients with tumor sizes of >2 cm were recruited with no particular selection criteria. The tumor size was defined by contrast computed tomography scan or magnetic resonance imaging. The 20 cirrhotic liver tissues were diagnosed histologically, with activity ranging from mild to severe. All resected tissues were collected between 1998 and 2006 at Queen Mary Hospital, Pokfulam, Hong Kong. Prior to proteomic and molecular analyses, the pathological diagnosis for all tissues was confirmed histologically. A separate cohort of HCC patients (n ) 88) and non-neoplastic control subjects (n ) 64) from the same hospital was used to validate the biomarker performance in serum samples by ELISA (Table 1). This study was vetted and 1924

controls (n ) 64)

Journal of Proteome Research • Vol. 9, No. 4, 2010

approved by the Institutional Ethics Committee, and all the tissues and serum were collected with informed consent from patients. Cell Lines. Liver cell lines (MIHA, Huh7, HepG2, PLC/PRF/ 5) were obtained and detailed in previous studies.14 Cells were cultured in complete medium [Dulbecco’s modified Eagle’s medium (DMEM) supplemented with 10% fetal bovine serum (FBS)] at 37 °C in a 5% CO2 incubator. Proteomic 2-DE Analysis of HCC Tissues. Methods have been previously described.15,16 In brief, approximately 25 µg of tissue lysates were subjected to isoelectric focusing using the IPGphor system (GE Biosciences, Buckinghamshire, England) and electrophoresed in 12.5% precast Ettan DALT Gel (GE Biosciences) for second-dimension separation. After silver staining, the gel images were captured with a GS-800 Calibrated Densitometer (Bio-Rad, Hercules, CA) and analyzed using PDQuest 8.0 (Bio-Rad) according to the manufacture guidelines. Spot intensities were exported for normalization using “Total quantity in valid spots” method17 and reference markers (β actin and Hsp 60) that have recently been identified as internal housekeeping reference controls across different silverstained 2DE gels.18 Statistical analysis with using one-way ANOVA test was used to compare significancy in normal, cirrhosis, and tumorous samples. Spots with significant (p < 0.05) expression difference were chosen for further mass spectrometry analysis. MALDI-TOF/TOF. After in-gel trypsin digestion (MS grade, Promega, Madison, WI), the peptide extracts were purified by

research articles

Proteomics of Hepatocellular Carcinoma 16

ZipTip (Millipore, Bedford, MA) as previously described. MS/ MS spectra data were generated by ABI 4800 Proteomics Analyzer MALDI-TOF/TOF (Applied Biosystems, CA). A maximum of 5 most intense tandem mass spectra were collected for MS/MS in positive ion mode and of collision energy 2 kV. Monoisotopic peaks were automatically determined and fragment ion masses were compared with the NCBI database for protein identification using the MASCOT version 2.1 (Matrix Science). Western Blot Analysis. Protein extracts (25 µg) were separated on 10% polyacrylamide SDS gels and electroblotted on nitrocellulose membranes as previously described.19 After blocking with 5% nonfat milk in TBS-T (20 mM Tris, 137 mM NaCl, 0.1% Tween-20, pH 7.6), the membrane was incubated with mouse monoclonal antivimentin antibody (CHEMICON International, MA) at 1:3000 dilution at 4 °C overnight, followed by peroxidase-conjugated goat antimouse secondary antibody (Invitrogen-Zymed laboratories, South San Francisco, CA). Mouse antihuman β-actin antibody (Santa Cruz Biotechnology, CA) at 1:1000 dilution was included as a loading reference control. Densitometry data were analyzed by Quantity One (Bio-Rad) for protein quantification of vimentin normalized with β-actin. Real-Time qPCR. Primers and probe sequences were designed using the Universal Probe Library Assay Design Centre (http://www.roche-applied-science.com). After RNA purification, the first-strand cDNA synthesis was performed using the Superscript first strand synthesis system (Invitrogen) followed by PCR amplification using the Roche LightCycler 480 (Roche Diagnostic, UK). Detailed procedures were described in previous study.14 Standard curves were generated from five repeated 10-fold serial dilutions of oligo-template for quantification. Data were reported as copies numbers of transcript per ng of cDNA. Serum Vimentin ELISA. Circulating vimentin in blood sera was measured by indirect ELISA, using a commercially available antiepitope antibody against vimentin (V9, Santa Cruz Biotechnology) and standard antigen control of recombinant fulllength vimentin protein. In brief, vimentin cDNA in the pET43.1b (+) expression vector (Invitrogen) was induced by 1 mM isopropylthiogalactopyranoside (IPTG) at 37 °C overnight. The protein lysates were obtained by sonication in lysis buffer [50 mM Tris-HCl, 300 mM NaCl, 10 mM Imidazole], followed by ultracentrifugation and HiTrap affinity chromatography (GE healthcare). Protein purity and concentration were determined by SDS-PAGE and NanoDrop (Thermo Electronics, TX), respectively. A sigmoidal dose-dependent titration curve was then generated from various concentrations (10, 70, 140, 281, 526, and 1125 ng/mL) of the purified recombinant vimentin protein, and the serum level of circulating vimentin in each individual sample was estimated using the equation (y ) 0.0003x + 0.0173, r2 ) 0.9977) based on the standard curve. The precision values of the intra-assay and interassay were also determined as the coefficient of variance (CV). The mean intra-CV values were 7.9% (628.2 ng/mL) and 10.1% (217.7 ng/mL), whereas the inter-CV values were 9% (596.4 ng/mL) and 10.6% (200.7 ng/ m), with a minimum detection sensitivity of 81 ng/mL. Serum samples were coated in duplicates on a 96-well ELISA plate (Corning Incorporated, NY) for overnight incubation at 4 °C. After blocking with 5% BSA, antivimentin V9 antibody at 2.5 µg/mL was added and incubated for 2 h at 37 °C. After 3 washes with TBS/T, secondary peroxidase-conjugated goat antimouse IgG antibody was added and incubated at 37 °C for 2 h. The color was developed by incubation with ABTS

substrate (Invitrogen) at room temperature for 20 min, and the optical density at 415 nm was measured by an ELISA Plate reader (Bio-Rad). Vimentin (VIM) Gene Cloning and Transient Transfection in HCC Cells. The whole cDNA coding region of vimentin (Accession no. NM_003380, 1401bp) was cloned into the pCR 2.1-TOPO vector (Invitrogen) and subcloned into the pcDNA3.1A expression vector (Invitrogen) by Hind III and EcoRV digestions. After bidirectional DNA sequencing, pcDNA3.1A empty vector (control) and pcDNA-VIM were transiently transfected into the PLC/PRF/5 cells using Lipofectamine 2000 reagent (Invitrogen) according to the manufacturer’s protocols. Proteins were collected consecutively for 5 days and the transfection efficacy were examined by Western blot analysis. Subsequently, the culture supernatants from PLC/PRF/5 cells transfected with the empty vector control and pcDNA-VIM were harvested for measurement of soluble vimentin level by indirect ELISA assay. Statistical Analysis. All 2-DE data were log-transformed for parametric analyses using both SPSS for Windows (version 16.0, SPSS, Chicago, IL) and Prism 4 software (GraphPad Inc., San Diego, CA). Student’s t-test or the Mann-Whitney test was used when appropriate to determine whether a difference existed between two groups. One-way ANOVA analysis or the KruskalWallis test was used for comparison of more than 2 groups. The choice of a parametric or nonparametric test was based on the distribution of data. P values of less than 0.05 were considered statistically significant. Continuous data were presented as the mean ( SD. Receiver operating characteristic (ROC) curves were performed to determine the area under the curve (AUC), and a cutoff value was selected for optimal sensitivity and specificity.

Results Proteomic Identification of Vimentin Overexpressed in HCC. By 2-DE profiling in the biomarker discovery phase, a total of 14 protein spots that showed at least 2-fold difference (p < 0.05) among the HCC, cirrhotic, and normal liver samples were identified (Supplementary Table 1, Supporting Information). Among the proteins identified as up-regulated using mass spectrometry, SSP 1615 -vimentin was ideally found to be consistently overexpressed in more than 72 folds in HCCs compared to cirrhotic and normal liver tissues. Considering early detection of HCC in high risk subjects (hepatitis and cirrhosis) could improve patients’ clinical outcomes and prolong survival preferably when the tumors are small (2 cm or less). We then motivated to distinguish any of the identified biomarker with expression related to small size HCCs. Intriguingly, vimentin was found to express significantly more strongly (p < 0.01) in small (e2 cm) HCCs than others MS identified proteins. Moreover, significant overexpression (p < 0.05) of this protein was maintained in larger size (>2 cm) HCC tumors suggested that expression of vimentin may be related to HCC development (Figure 1A). Finally, the high protein score and CI% (protein score > 80 and protein CI% > 99%), sequence coverage percentages and the number of peptide matches to the predicted protein further reinforced the isolation of the protein for detailed validation (Supplementary Table 2, Supporting Information). Verification of Vimentin Overexpression in All Size HCC Tumors. To confirm the results from 2-DE analysis, additional independent samples from cirrhosis, small HCCs, HCCs > 2 cm and their respective nontumorous tissues were examined. Western blot analysis showed higher concentrations of vimenJournal of Proteome Research • Vol. 9, No. 4, 2010 1925

research articles

Sun et al.

Figure 1. Overexpression of vimentin in small HCCs. (A) Proteomic identification of vimentin overexpressed in HCC (one-way ANOVA in different sizes of HCC tumors versus normal and cirrhosis controls; * P < 0.05, ** P < 0.01.) (B) Scatter diagram shows the densitometry data from Western blot analysis. Differences are expressed as the ratio of vimentin in different disease stages to actin density. (C) Transcript levels of vimentin showed overexpression using qPCR. Data was presented in normalized copies per nanogram of RNA. Statistical analysis was performed by student-t-test in HCC tumors of different sizes relative to non-neoplastic control *P < 0.05, **P < 0.01. Table 2. Concentrations of Serum Vimentin and AFP Levels in Different Disease Stagesa serum vimentin (ng/mL)

serum AFP (ng/mL)

diagnosis (n ) 152)

mean ( SD

median (interquartile range)

mean ( SD

median (interquartile range)

Controls (n ) 64) HCC e2 cm (n ) 44) HCC >2 cm (n ) 44)

176.0 ( 61.3 271.5 ( 165.9 231.3 ( 132.3

163.0 (128.3-212.5) 202.5 (160.0-328.5) ** 179.0 (150.0-264.0) *

122.3 ( 270.4 227.8 ( 573.2 50380.8 ( 195484.6

11 (5.0-24.0) 43 (9.0-225.0) 233 (57.5-233.0) **

a Statistical analysis was performed using Mann-Whitney test and student-t-test (after data log-transformed) by comparing non-neoplastic conditions with HCCs of different sizes. *p < 0.05, **p < 0.01.

tin protein in both HCCs e 2 cm (n ) 10) and those >2 cm (n ) 10) (Figure 1B) compared with the grouped nontumorous control (cirrhotic and adjacent nontumorous tissues) (n ) 10). Densitometric analysis showed more vimentin in the small (0.702; mean ratio relative to actin) and larger HCC (0.706) than in the nontumorous liver control (0.059) (p < 0.05). RNA transcripts from the same set of samples were analyzed using qPCR (Figure 1C). In line with the 2-DE data, small HCCs contained more vimentin mRNA (mean: 2300 normalized copies per nanogram) than the non-neoplastic control (mean: 763.9 normalized copies per nanogram) (p < 0.05, Student’s t test). The vimentin transcript was also present significantly at higher levels in HCCs larger than 2 cm (mean: 2535 normalized copies per nanogram) and data were less dispersed than in the small tumors. Vimentin as a Potential Serological Marker for Small HCC. Initially, a pilot study was performed to test whether sera from HCC patients contain soluble vimentin. Immunoprecipitation was conducted to enrich and captured vimentin antigen from human HCC sera followed by Western blot analysis. Two groups of pooled serum samples from 10 HCC patients (5 patients /group) were tested to be vimentin positive (data not shown), suggesting that it might serve as a serological marker for HCCs. Subsequently, the presence of vimentin in serum was tested in another cohort of 152 individuals using an indirect ELISA. The concentrations of serum vimentin and AFP in different disease stages are summarized in Table 2. There was a significant difference (**, p < 0.01,*, p < 0.05) in the level of circulating vimentin between non-neoplastic controls and HCC of all sizes, whereas for serum AFP a significant difference (**, p < 0.01) was seen only between those with non-neoplastic liver conditions and HCCs of >2 cm. In addition, a cutoff value for soluble vimentin was chosen with 1926

Journal of Proteome Research • Vol. 9, No. 4, 2010

the highest Youden’s index which gave the optimal sensitivity and specificity. Circulating vimentin assayed at a cutoff level of 245 ng/mL detected small HCCs with sensitivity of 40.91% (95% CI: 26.34% to 56.75%). At the same cutoff value, circulating vimentin has a specificity of 87.5% (95% CI: 76.85% to 94.45%) in separating non-neoplastic subjects from those with HCCs of small size. The positive likelihood ratio of using cutoff 245 ng/mL is 3.27 and Youden’s index is 0.28. Soluble Vimentin Is Superior to Serum AFP for Detection of Small Size HCCs. Figure 2A demonstrates the expression levels of soluble vimentin and serum AFP in respect to the HCC tumor sizes (cm). Patients with small size tumors perceptibly contained more vimentin (cutoff 245 ng/mL) than those with larger size tumors suggested that its expression was associated with small size tumors. In the same cohort of patients, AFP was less sensitive to detect small HCC tumors (cutoff 100 ng/mL calculated from our study cohort), and its expression increased corresponding to the tumor sizes. ROC analysis also showed that soluble vimentin (AUC: 0.69) is superior to serum AFP (AUC: 0.62) in separating non-neoplastic individuals from those with small HCCs (Figure 2B). The results are consistent with the previous reports that AFP is insensitive to detect small size HCCs. To further compare the diagnostic performance of circulating vimentin with serum AFP, more defined statistical analyses were performed (Table 3). Overall, circulating vimentin concentration (g245 ng/mL) appeared to be a superior single marker with higher sensitivity (40.91%) than serum AFP (cutoffs: 100 ng/mL, Sen ) 30.23% and 400 ng/mL, Sen ) 16.28%) in detecting small HCCs. Moreover, at the same cutoff value, circulating vimentin had a higher specificity (87.5%) than serum AFP (85.19%) in separating cirrhosis/normal liver from small HCCs. Although AFP g 20 ng/mL demonstrated a higher

research articles

Proteomics of Hepatocellular Carcinoma

Figure 2. Elevated vimentin levels in small tumor sizes of HCC. (A) Schematic scatter diagram shows distribution of serum AFP and vimentin in respect to the size of largest tumor-length (sltl-cm) in HCC. The y-axis is presented in logarithmic base 10. (B) Comparison of the diagnostic performance of serum AFP and vimentin in detection of small HCCs by receiver operating characteristic (ROC). Area under the ROC curve (AUC) is a measure of the overall performance of sensitivity and specificity. Table 3. Predictive Performance of Circulating Vimentin and AFP As Biomarkers for Detection of Small HCCs non-neoplastic controls (n ) 64)vs HCCs e2 cm (n ) 44) statistical parameters

vimentin (g245 ng/mL)

AFP (g20 ng/mL)

AFP (g100 ng/mL)

AFP (g400 ng/mL)

vimentin (g245 ng/mL) + AFP (g100 ng/mL)a

Sensitivity, SEN Specificity, SPE False positive rate, FPR False negative rate, FNR Accuracy, AC Youden’s index Positive likelihood ratio, LR+ Negative likelihood ratio, LRPositive Predictive Value, PPV+ Negative Predictive Value, NPV-

40.91% 87.50% 12.50% 59.09% 68.52% 0.28 3.27 0.68 69% 68%

54.55% 59.26% 40.74% 45.45% 56.34% 0.13 1.34 0.77 69% 44%

30.23% 85.19% 14.81% 69.77% 51.43% 0.15 2.04 0.82 76% 43%

16.28% 85.19% 14.81% 83.72% 42.86% 0.02 1.10 0.98 64% 39%

58.77% 98.15% 1.85% 41.23% 82% 0.57 31.77 0.420 96% 78%

a

Combined diagnostic performance was performed by parallel (Sensitivity) and serial (Specificity) tests.

sensitivity than vimentin, its specificity was comparatively low (59.26%), therefore Youden’s index may exemplify a better approach as it is one way to attempt summarizing test accuracy into a single numeric value. Other statistical indexes also suggested that circulating vimentin was superior to serum AFP in detecting HCC e 2 cm. Recognizing the limitations of individual tests, we performed parallel and serial diagnostic tests to assess the ability of

vimentin in conjunction with serum AFP at optimized cut-offs in detection of small HCCs. AFP g 100 ng/mL is one of the cut-offs regularly used in clinical analysis and has been chosen because of its high predictive performance in our study. In combination, the predictive performance of vimentin and serum AFP was remarkably enhanced and yielded sensitivity and specificity of 58.77 and 98.15%, respectively (Table 3). In fact, these two biomarkers when used in combination at other Journal of Proteome Research • Vol. 9, No. 4, 2010 1927

research articles cutoff values (e.g., vimentin 245 ng/mL and AFP 20 ng/mL: Sen ) 73.14%, Spe ) 94.91%; vimentin 150 ng/mL and AFP 100 ng/mL: Sen ) 85.73%, Spe ) 90.51%), could possibly further increased the diagnostic accuracy for detection of small HCCs. Soluble Vimentin in HCC Cell Culture Supernatants. Twohundred microliters of cell culture supernatants were collected from different liver cells (MIHA, PLC/PRF/5, HepG2 and Huh7) after 4 days of culture and vimentin level was measured by indirect ELISA. Soluble vimentin was detected in the culture supernatant of Huh-7 with vimentin concentration at 311 ng/ mL but not in MIHA (immortalized normal liver cell) and HepG2 (Hepatoma). Intriguingly, vimentin in the culture supernatants are comparable to the endogenous expression of vimentin where MIHA, PLC/PRF/5 and HepG2 have no endogenous expression of vimentin but Huh-7. Thus, we would like to investigate whether secreted vimentin is dependent on the intracellular vimentin protein. After transiently transfected vimentin into the PLC/PRF/5 cells, culture supernatants were collected at days 3 and 5 for ELISA analysis. Soluble vimentin was markedly increased in the culture supernatant at day 3 post-transfection (2601 ng/mL) when compared to the vector control (below the minimum detection sensitivity). The soluble vimentin was also present at higher levels in the culture supernatant at day 5 post-transfection (1617.7 ng/mL) but not to the same degree as in day 3. The expression level reflected the transfection efficiency of vimentin in PLC/PRF/5 cells.

Discussion Herein, through the analysis of protein expression profiles in 76 liver tissues, we have identified a set of proteins that distinguishes both small (e2 cm) and large HCCs from highrisk cirrhotic subjects. So far, no study has reported a similar investigation of size-defined HCC tumors using 2-DE. From this set, we chose vimentin for further assessment and demonstrated a positive correlation between vimentin protein and mRNA abundance in different sizes of HCC tumors. Validation of the presence of high levels of vimentin in small HCCs confirmed its clinical usefulness for the diagnosis of early HCC. The conventional biomarker for detection of HCC is AFP but its specificity and sensitivity are limited. Various glycated isoforms of AFP have been shown to be superior to total AFP as a marker of HCC, for example, Lens culinaris agglutininreactive AFP and des-γ-carboxyprothrombin, but these are still poor at detecting small (