Isolation of Circulating Tumor Cells in Patients with Hepatocellular Carcinoma Using a Novel Cell Separation Strategy

Hepatocellular carcinoma (HCC) is the fifth most common cancer in the world and the third most common cause of cancer-related deaths (1, 2). The incidence of HCC is highest in South Africa and East Asia, but it is rising gradually in the United States, European countries, and other regions (1, 2). Hepatectomy and orthotopic liver transplantation are currently the only choices to effect a radical cure (3, 4), which is elusive due to high incidence of recurrence. Because only the residual cancer cells outside the liver of a recipient can cause recurrence after liver transplantation in HCC patients, isolated tumor cells must be the source of recurrence (4). As hematogenous spread is the major route of HCC recurrence (5), detection of circulating tumor cells (CTC) undoubtedly has important clinical significance in predicting recurrence and monitoring therapeutic efficiency in HCC patients.

Immunomagnetic bead separation based on antibodies for tumor cell surface antigens is currently the standard method employed to detect CTCs in the clinical setting. Epithelial cell adhesion molecule (EpCAM) is an epithelial cell–specific adhesion molecule that is widely expressed on the surface of epithelial cells and epithelial-derived tumor cells (6). The CellSearch system, based on EpCAM antibody–coated magnetic beads, has been approved by the U.S. Food and Drug Administration for detection of CTCs in breast cancer, colon cancer, and prostate cancer (7). Although HCC cells are epithelial cells, only about 35% of HCC cases express EpCAM (8, 9). Therefore, the CellSearch system is not appropriate for detection of CTCs in HCC patients. Given that there are currently no antibodies specific for HCC cell surface antigens, few reports have been published on the isolation of CTCs in HCC patients (10–12). α-Fetoprotein (AFP), a well-known marker for HCC, is expressed in the cytoplasm of HCC cells. AFP mRNA in the peripheral blood detected by reverse transcriptase-PCR (RT-PCR) has been reported to be useful as a prognostic predictor of HCC (13–15). However, about 30% to 40% of HCC patients are AFP negative (16, 17), and AFP mRNA was reported to be detected even in patients with chronic hepatitis B or cirrhosis (15, 18). Moreover, the RT-PCR–based assays cannot accurately quantify the number of CTCs or obtain intact CTCs for further investigation. In summary, it is imperative to establish a sensitive and specific method to detect CTCs in HCC patients.

The asialoglycoprotein receptor (ASGPR) is a transmembrane protein exclusively expressed on the surface of hepatocytes, which can bind and internalize molecules with terminal galactose and N-acetylgalactosamine residues (19, 20). By taking advantage of this characteristic, liver-targeting systems for drugs and genes have been developed using glycosylated macromolecules as vehicles that can be selectively recognized by ASGPR (21–23). Here we describe the development and validation of a sensitive and specific system to magnetically separate CTCs in HCC patients, mediated by the interaction of the ASGPR with its ligand. In the system, HCC cells were bound by biotinylated asialofetuin, an ASGPR ligand, and subsequently labeled by antibiotin antibody–coated magnetic beads, followed by magnetic separation. The separated HCC cells were then identified by immunofluorescence staining by using the hepatocyte-specific antibody Hep Par 1. Given that normal hepatocytes do not circulate, unless they become tumorous, any of the cells detected by the system are circulating HCC cells. This system solves the inherent problems (i.e., problems with specificity and sensitivity) of previously reported methods (13–15, 18). To evaluate its clinical impact, we have applied this system to patients with HCC and correlated results with the clinical parameters.

From June to December 2009, peripheral blood samples were collected from 85 HCC patients, 37 patients with benign liver diseases (16 patients with cirrhosis, 4 patients with chronic hepatitis B, 6 patients with acute hepatitis A, 8 patients with hepatic hemangioma, and 3 patients with liver cysts), 20 healthy volunteers, and 14 patients with miscellaneous advanced cancers other than HCC (2 testicular, 3 thyroid, 3 colon, 2 kidney, and 4 metastatic breast cancer). The clinical characteristics of HCC patients are summarized in Table 1. HCC was histologically diagnosed using surgically obtained specimens in 63 subjects. The other HCC patients were diagnosed on the basis of clinical data, computed tomographic signs of HCC, and elevated AFP levels. HCC patients were classified according to 2 criteria: the sixth edition of International Union Against Cancer (UICC) tumor-node-metastasis (TNM) staging system (24) and the Milan criteria (25). Fourteen patients with non-HCC cancers were histologically diagnosed using surgically obtained specimens. Five milliliters of blood was drawn from each subject and collected in BD Vacutainer tubes containing sodium heparin (Becton Dickinson). The samples were stored at 4°C and processed within 6 hours after collection. From August to October 2010, another 18 HCC patients were enrolled for FISH analysis of CTCs. Written informed consent was obtained from each patient, and the study protocol conformed to the ethical guidelines of the 1975 Declaration of Helsinki as reflected in a priori approval by the Institutional Review Board at Eastern Hepatobiliary Surgery Hospital (Shanghai, China).

Human hepatoma cell lines HepG2 and Hep3B, human breast cancer cell line MCF-7, and human kidney cancer cell line A498 were purchased from American Type Culture Collection. HepG2, Hep3B, and MCF-7 cells were cultured in Dulbecco's modified Eagle medium (Invitrogen) supplemented with 10% FBS (Invitrogen). A498 cells were cultured in Eagle's minimum essential medium (Invitrogen) supplemented with 10% FBS. Cells were maintained at 37°C in a humidified atmosphere containing 5% CO₂ and harvested with trypsin before use. The cell suspensions were used only when their viability as assessed by trypan blue exclusion exceeded 90%.

Asialofetuin (Sigma–Aldrich) was biotinylated using Sulfo-NHS-LC-Biotin (Pierce) according to the manufacturer's instructions.

For each staining, a total of 4 × 10⁵ cells were incubated with biotinylated asialofetuin at 37°C for 45 minutes, followed by incubation with antibiotin–fluorescein isothiocyanate (FITC; Miltenyi Biotec GmbH) at 4°C for 30 minutes in the dark. Then cells were resuspended in 250 μL of staining buffer (Becton Dickinson). Two washes with the staining buffer were carried out between each step. Flow cytometric analysis was carried out on FACSCalibur (Becton Dickinson), and the obtained data were analyzed with CellQuest software (Becton Dickinson).

A total of 4 × 10⁵ HepG2 cells were incubated with biotinylated asialofetuin, together with 50-fold molar excess of asialofetuin or 10 mmol/L EDTA (Sigma–Aldrich) at 37°C for 45 minutes, followed by incubation with antibiotin–FITC at 4°C for 30 minutes in the dark. Then, cells were resuspended in 250 μL of staining buffer. Two washes with the staining buffer were carried out between each step. Flow cytometric analysis was carried out as described earlier.

Mononuclear cells and tumor cells were isolated from the whole blood samples by density gradient (Ficoll-Paque PLUS; GE Healthcare) according to the manufacturer's instructions. Briefly, 5 mL of whole blood mixed with an equal volume of balanced salt solution (NaCl solution, 0.14 mol/L) was carefully layered on 8 mL Ficoll-Paque PLUS and centrifuged at 400 × g for 30 minutes at 18°C in a 50-mL centrifuge tube (Corning Inc.). Then, the mononuclear cell layer was collected and washed twice with balanced salt solution.

For magnetic labeling, cells were incubated with biotinylated asialofetuin for 45 minutes at 37°C. After washing with dilution buffer (PBS containing 50 units of sodium heparin per mL), cells were incubated with antibiotin microbeads (Miltenyi Biotec GmbH) for 15 minutes at 8°C. Then, cells were washed once and resuspended in 1 mL of dilution buffer. Magnetically labeled cells were isolated over the AutoMACS Pro Separator (Miltenyi Biotec GmbH) with “posseld2” program referring to the user manual. The positive fraction was spun down on a polylysine-coated slide using a cytocentrifuge (Wescor Inc.) and air-dried at 37°C for 30 minutes. The slide was fixed for 15 minutes in 4% formaldehyde and investigated immediately by immunoflourescence staining or stored at −20°C until further processing.

To identify HCC cells on slides, a mouse anti-human monoclonal antibody Hep Par 1 (Dako) for hepatocytes and a rat anti-human CD45 monoclonal antibody (Santa Cruz Biotechnology Inc.) for hematologic cells were used. Immunofluorescence staining was carried out according to the manufacturer's instructions. Briefly, slides were rinsed with washing buffer (PBS with 0.1% Tween 20). After blocking nonspecific binding sites with 5% bovine serum albumin in PBS with 0.1% Tween 20 for 30 minutes, slides were incubated for 1 hour with Hep Par 1 and rat anti-human CD45 monoclonal antibody at 37°C in a humidified chamber. A Cy5 rabbit anti-mouse IgG (Pierce) and an Alexa Fluor 488 rabbit anti-rat IgG (Invitrogen) served as secondary antibodies. After 4′,6-diamidino-2-phenylindole (DAPI; Pierce) staining, the slides were mounted in antifade solution (Vector Laboratory).

An Olympus BX61 upright microscope with an automated Prior ProScan stage (Prior Scientific) was used to image the slides. The slides were scanned automatically in a 1 × 1-mm² grid format with the programmable stage and Qcapture Pro software (version 5.1; Media Cybernetics). Captured images (at 400 × total magnification) were carefully examined and the objects that met preset criteria were counted. Color, brightness, and morphometric characteristics such as cell size, shape, and nuclear size were considered in identifying potential CTCs and excluding nonspecific cells. Cells that stained positive for Hep Par 1 and DAPI, negative for CD45, and met the morphologic characteristics consistent with malignant cells, including large cellular size, high nuclear to cytoplasmic ratio, and visible nucleoli, were scored as CTCs. Cell counts were expressed as the number of cells per 5 mL of blood.

This assay was conducted partly as previously described (26). Briefly, for recovery, sensitivity, and linearity experiments, five 5-mL aliquots of peripheral blood collected into BD Vacutainer heparin tubes from each of 5 healthy volunteers were prepared. The 5 aliquots from each volunteer were spiked with various numbers of Hep3B cells to produce separate samples with approximately 10, 30, 90, 270, or 810 cells per 5 mL of blood. For specificity assay, five 5-mL aliquots of blood were collected from another healthy volunteer, and 100 MCF-7 cells were spiked into each aliquot. The 30 samples were then processed and analyzed according to the CTC separation and identification protocols described earlier. The actual number of spiked cells was determined by a COULTER LH 750 Hematology analyzer (Beckman Coulter). All procedures were carried out by a single operator.

Pretreatment and denaturation of slides were executed as previously described (27). Chromosome enumeration probe (CEP) 17 (SpectrumGreen) and locus-specific probes for HER-2 (SpectrumRed) and TP53 (SpectrumOrange) were provided by Vysis (Downers Grove). Hybridization and posthybridization washes were carried out according to manufacturer's instructions. After DAPI (Pierce) staining, the slides were mounted in antifade solution (Vector Laboratory) and examined using a Zeiss Axioplan 2 epifluorescence microscope controlled by Isis imaging software (MetaSystems GmbH). The leukocytes on the slides were used as internal controls for hybridization efficiency and specificity. Signal counts were recorded for all probes, and the ratio of HER-2 (red) and TP53 (orange) signals to chromosome 17 pericentromeric signal (green) was assessed. For HER-2, the ratio equal to or greater than 2 was scored as amplified (28), and for TP53, the ratio less than 1 was scored as deletion (29). Abnormal signal patterns were accepted as true genomic changes if present in at least 3 cells.

RNA was extracted from isolated cells with the TRIzol Plus RNA Purification Kit (Invitrogen) according to the protocol supplied and resuspended in 1 μL of diethylpyrocarbonate-treated water for each milliliter of peripheral blood processed. Six microliters of RNA solution was denatured with 10 units of RNasin (RNAguard; Pharmacia Biotech) and 40 ng of hexamer random primers (GE Healthcare Bio-Sciences Corp.) in a 10-μL final volume at 65°C for 5 minutes and then reverse-transcribed in a 20-μL final volume, after the addition of 10 more units of RNasin, 1 × buffer supplied with the enzyme, 40 mmol/L of 4 deoxynucleotides, and 20 units of Moloney murine leukemia virus reverse transcriptase (Invitrogen) at 37°C for 60 minutes. Nested PCR for amplification of AFP cDNA was conducted as previously described (13). To evaluate the amplification of RNA, a 626-bp fragment of the human β-actin cDNA was amplified as previously reported (30). After separation through 6% polyacrylamide gels, all PCR products were visualized by ethidium bromide staining.

Categorical data displayed in a contingency table with 2 columns and 2 rows were analyzed using Fisher's exact test. For the continuous data, the Mann–Whitney test was used for comparison between 2 groups. Spearman rank correlation analysis was used for nonparametric correlation analysis. All statistical analyses were carried out with the SPSS statistical software package (SPSS/PC+; SPSS Inc.). A 2-sided P < 0.05 was considered statistically significant.

Flow cytometry was used to analyze the specificity of the interaction between biotinylated asialofetuin and ASGPR. It is known that ASGPR is highly expressed in the human hepatoma cell lines HepG2 and Hep3B (22, 31, 32), and in agreement, biotinylated asialofetuin showed high level binding in these cells (Fig. 1A), indicating that asialofetuin had been biotinylated and the biotinylated asialofetuin could interact with ASGPR. On the other hand, the human breast cancer cell line MCF-7 and the human kidney cancer cell line A498, which do not express ASGPR (32, 33), did not bind to biotinylated asialofetuin (Fig. 1A). Binding inhibition assay showed that binding of biotinylated asialofetuin to ASGPR could be suppressed by excessive amounts of asialofetuin or EDTA (Fig. 1B). These results suggested the specificity of the interaction between biotinylated asialofetuin and ASGPR-expressing cells.

Different numbers of Hep3B cells were spiked into blood, and recovery was measured using our system. The spiked number of Hep3B cells (i.e., 10, 30, 90, 270, or 810) was plotted against the number of Hep3B cells detected in the samples (Fig. 2). The results are summarized in Table 2. Linear regression of the number of detected tumor cells versus the number of spiked tumor cells yielded a slope of 0.61 (95% CI: 0.59–0.63), an intercept of 3.5 (95% CI: −3.9 to 10.8), and a correlation coefficient (R²) of 0.995. As expected, with the reduction of spiked cells, the coefficient of variation increased as the number of cells spiked decreased, ranging from 5.6% (810-cell spike) to 22.8% (10-cell spike). The average recovery of Hep3B cells was 61% or more at each spiking level. In the samples spiked with 10 cells, none had less than 5 tumor cells detected. No tumor cells were detected in any samples spiked with human breast cancer cell line MCF-7 (100 cells spiked).

No CTCs were detected in any of the blood samples from the 20 healthy volunteers. In the samples from 16 patients with cirrhosis, 4 patients with chronic hepatitis B, and 6 patients with acute hepatitis A, no CTCs were detected. In the samples from 14 patients with non-HCC cancers, including 2 testicular, 3 thyroid, 3 colon, 2 kidney, and 4 metastatic breast cancers, no CTCs were detected. Among 11 patients with benign hepatic space-occupying lesions (8 patients with hepatic hemangioma and 3 patients with liver cysts), none had CTCs detected before surgery and all had Hep Par 1–positive cells detected 2 and 8 days after surgery. Interestingly, the number of Hep Par 1–positive cells detected 8 days after surgery significantly decreased compared with that 2 days after surgery, and no Hep Par 1–positive cells were detected 12 days after surgery (data not shown). These results indicate that liver resection can induce release of liver cells into the peripheral blood circulation and the nontumorous cells will disappear within about 2 weeks, presumably owning to anoikis (34).

CTCs were measured in blood samples from 85 HCC patients at various clinical stages. The characteristics of CTCs isolated from samples of HCC patients include larger cell size with intact nuclei and high nucleus-to-cytoplasm ratios, DAPI and Hep Par 1 positive and CD45 negative (Fig. 3A–E). If Hep Par 1 was replaced by a pan-cytokeratin–specific antibody CK3-6H5 (Miltenyi Biotec GmbH), cells stained positive for cytokeratin and DAPI and negative for CD45 were scored as CTCs (Fig. 3F–J). CTCs from the blood samples of AFP-positive (>200 ng/mL) HCC patients were positive for anti-AFP antibody (Invitrogen) staining (Fig. 3K–O).

In Table 3, the number and percentage of patients with 1 or more (CTC positive), 1 to 5, 5 to 30, and 30 or more CTCs in 5 mL of blood are provided. The mean and median of CTC number of those patients with 1 or more CTCs are also shown in the table. CTCs were detected in the blood samples from 69 of 85 (81%) patients with HCC. The number of CTCs detected ranged from 0 to 125 per 5 mL, with an average of 19 ± 24 (mean ± SD). The positivity rate of CTCs in patients with portal vein tumor thrombus (92%) was significantly higher than that in patients without portal vein tumor thrombus (67%; P < 0.001), and the number of CTCs detected in patients with portal vein tumor thrombus (30 ± 27) was significantly higher than that in patients without portal vein tumor thrombus (4 ± 6; P < 0.001). The positivity rate of CTCs in patients beyond Milan criteria (91%) was significantly higher than that in patients within Milan criteria (69%; P = 0.009), and the number of CTCs detected in patients beyond Milan criteria (27 ± 27) was significantly higher than that in patients within Milan criteria (6 ± 9; P < 0.001). The positivity rate of CTCs in patients with Edmondson–Steiner grade III or IV (85%) was significantly higher than that in patients within grade I or II (59%; P = 0.036). Also, the number of CTCs detected in patients with Edmondson–Steiner grade III or IV (17 ± 22) was significantly higher than that in patients with grade I or II (8 ± 12; P = 0.030). Spearman correlation analysis showed that there was a high correlation between the positivity rate of CTCs and tumor size (r = 0.255, P = 0.019) and between the number of CTCs detected and tumor size (r = 0.329, P = 0.002). The positivity rate of CTCs was highly correlated with TNM staging from 66% in stage I to 100% in stage IV (r = 0.323, P = 0.003), and also the number of CTCs detected was highly correlated with TNM staging from 3 ± 4 in stage I to 67 ± 35 in stage IV (r = 0.709, P < 0.001). Neither the positivity rate of CTCs nor the number of CTCs detected was correlated with age, sex, etiology, Child–Pugh class, or serum AFP level (Table 4).

From August to October 2010, another 18 HCC patients were enrolled for FISH analysis of CTCs, which were isolated and enumerated using our system, and samples from 11 patients with CTC counts 5 or more were analyzed by FISH using probes for HER-2 and TP53 genes, and centromere sequence for chromosome 17, which contains HER-2 and TP53. The results are shown in Table 5. HER-2 amplification was detected in 2 patients, both with CTC counts of more than 50. TP53 gene deletion was observed in 6 patients, of whom 1 had HER-2 amplification. Biallelic deletion of TP53 was found in 2 patients, both with chromosome 17 gain. Examples with HER-2 amplification and/or TP53 deletion are shown in Figure 4. No abnormality was detected in samples from 4 patients.

AFP mRNA was not detected in any cases without HCC, except in 1 of 16 (6%) cases of cirrhosis, whereas it was detected in 26 of 85 (31%) HCC patients. However, none of the 4 cases with serum AFP level of less than 20 ng/mL and 2 of 13 cases with serum AFP level of 20 to 100 ng/mL were judged to be positive. Correlations between positivity rate of AFP mRNA and clinical variables of HCC patients are shown in Table 6. The positivity was significantly associated with the existence of HCC (P = 0.023) and the existence of portal vein tumor thrombus (P = 0.039), but not with tumor size, differentiation status, or the disease extent as classified by the TNM classification or the Milan criteria.

ASGPR is abundantly and exclusively expressed on the surface of hepatocytes and HCC cells (31–33). Hep Par 1 is a human hepatocyte–specific antibody, which recognizes antigens localized in mitochondria and has been used to identify liver-derived cells including normal hepatocytes and HCC cells (35). We provide a unique isolation and enumeration system for CTCs by using ASGPR-based magnetic separation and Hep Par 1–based immunoidentification. Recovery of Hep3B cells spiked into 5 mL of blood by using this system was greater than 61% at each spiking level. In addition, when as few as 10 Hep3B cells were spiked into 5 mL of blood, no fewer than 5 cells were detected in each sample, indicating the high sensitivity of the system. In all blood samples spiked with cells of the human breast cancer cell line MCF-7, no samples had tumor cells detected. No CTCs were detected in blood samples from healthy volunteers or from patients with cirrhosis, chronic hepatitis B, and acute hepatitis A, indicating that the results were not affected by the liver disease background. It is reported that ASGPR is also expressed at low levels in extrahepatic cells such as Neisseria gonorrhoeae–infected urethral epithelial cells and proximal tubular epithelial cells (36, 37). However, these cells usually do not appear in the blood of HCC patients. Even if there are trace amounts of non–liver-derived ASGPR-expressing cells in circulation, they cannot be detected because the Hep Par 1 antibody used for detection can recognize only liver-derived cells. In fact, we examined peripheral blood samples from patients with advanced cancers other than HCC and detected no Hep Par 1–positive cells. These results suggest excellent specificity of our system.

We examined peripheral blood samples from 85 patients with HCC at various stages with this system. The HCC patients were classified according to the widely accepted sixth edition of UICC TNM staging system (24) and the Milan criteria (25). Results showed that CTCs could be detected in most HCC patients (>80%), even in patients at early stage or in patients with tumor size of less than 2 cm. Both the positivity rate and the number of detected CTCs were positively correlated with the disease extent as classified by the TNM classification. For more than a decade, the Milan criteria have been most widely used for the selection of candidates for liver transplantation (38). In patients who do not meet the Milan criteria, both the positivity rate and the number of detected CTCs were significantly higher than those in patients who meet the Milan criteria, suggesting that detection of CTCs may have potential applications in selecting HCC patients for liver transplantation. Notably, we found that both the positivity rate and the number of CTCs were significantly correlated with the differentiation status as defined by Edmondson–Steiner grading (39), an authorized and extensively used histologic classification (40). It was previously thought that portal vein tumor thrombus would mainly cause intrahepatic metastases (41, 42). In our study, however, both the positivity rate and the number of detected CTCs in patients with portal vein tumor thrombus were higher than those in patients without portal vein tumor thrombus, which suggest that portal vein tumor thrombus may also be a source of systemic spread of CTCs. The correlation between CTC prevalence and disease extent also suggests the specificity of our system. Up to now, we have carried out a minimum 10-month follow-up of 7 HCC patients who received orthotopic liver transplantation. In 3 CTC-negative cases, no recurrence was observed within 1 year after surgery, whereas in 4 CTC-positive cases, 3 developed recurrences within 6 months after surgery (data not shown).

TP53 gene functions as a tumor suppressor and plays a central role in regulating transcriptional activation of crucial growth regulatory genes that control cell-cycle progression and cell division (43). TP53 is frequently inactivated in various types of malignant tumors, including HCC (43). HER-2 (neu or ErbB-2) gene encodes a tyrosine kinase receptor, which belongs to the epidermal growth factor receptor family (44). HER-2 amplification has been reported in breast cancer, and a monoclonal antibody trastuzumab, has become an important therapeutic option for patients with HER-2 amplification (44). However, reports on the status of HER-2 in HCC are conflicting (45–48). By FISH, we showed a high frequency of deletion for TP53. Amplification of HER-2 was observed in 2 patients with higher CTC counts among the 11 patients analyzed. Such amplifications are uncommon in primary tumors of HCC (45, 46), whereas some studies show frequent HER-2 overexpression in HCC (47, 48). The frequency of HER-2 amplification of CTCs in HCC needs to be investigated with more patients so as to clarify whether it could be a candidate gene for targeted therapy in HCC. In addition, we found that 2 patients had chromosome 17 gain in apparent triploid background, both with biallelic deletion of TP53. This finding is in accordance with the known hypothesis that TP53 is important for the maintenance of chromosomal stability (43).

To date, CTCs in HCC patients have been mainly evaluated by detecting AFP mRNA in peripheral blood. As in the cases with portal vein tumor thrombus in present study, detection of AFP mRNA could provide useful information. However, AFP mRNA was reported to be detected in the peripheral blood of patients with chronic hepatitis B or cirrhosis (15, 18). Indeed, AFP mRNA was detected in 1 of 16 (6%) cases with cirrhosis in our study. Besides, the positivity rates of AFP mRNA were only 31% (26 of 85) in all 85 HCC patients and 12% (2 of 17) in 17 HCC patients with serum AFP level of less than 100 ng/mL, although the detection procedure was the same as that used in previous report (13). In contrast, the corresponding positivity rates of CTCs detected using our system were up to 81% (69 of 85) and 76% (13 of 17), respectively.

It is noteworthy that our study showed that in patients with benign hepatic space-occupying lesions, liver resection could cause release of hepatocytes into the blood circulation, which is in agreement with previous reports (49). More importantly, we found that the nontumorous cells would disappear from circulation within 2 weeks, which should be taken into consideration when using this system to evaluate CTCs in patients undergoing liver resection so as to minimize false-positive results caused by normal hepatocytes released into blood.

In summary, we have established a highly sensitive and specific system for CTC detection in HCC patients, which provides an opportunity for enumeration and biological properties characterization of individual isolated CTCs. It is likely clinically useful in diagnosis and monitoring of HCC and may have a role in clinical decision making.

No potential conflicts of interest were disclosed.

We thank Wei-Wei Zhang and Ye-Fang Gong for technical assistance.

This work was supported by grants from the National High-Tech Research and Development Program of China (2007AA02Z461), China National Key Projects for Infectious Disease (2008ZX10002-021), and the National Natural Science Foundation of China (30672002, 30772513, and 30801342).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.