We enrolled 328 HBV hepatitis patients who had been diagnosed with HCC based on a resected specimen and whose medical records and formalin-fixed paraffin blocks of tumor tissue were available from the archives of the Department of Pathology at Seoul National University Hospital (SNUH) from 1998 to 2004. We excluded co-infected hepatitis C virus hepatitis patients and patients who had neither serologic nor clinical evidence of HBV infection. The matched non-neoplastic hepatic parenchyma was available for 155 of the 328 patients. Clinical information, such as age, sex, surgical procedure, underlying etiology of liver disease, preoperative serum α-fetoprotein (AFP, μg/mL), preoperative treatment, and postoperative tumor recurrence, was collected from the medical records. Serological results of HBsAg, anti-HBs (HBsAb), IgG anti-HBc (HBcAb), hepatitis B virus e antigen (HBeAg), and anti-HBe (HBeAb) were based on the most recent preoperative tests from medical records. Depending on the state of the serum viral marker, liver function test, and clinical symptom of hepatic failure or portal hypertension (e.g., hypoalbuminemia, prolonged prothrombin time, ascites, hyperbilirubinemia, or hepatic encephalopathy), “asymptomatic carriers” had positive serum viral markers, but a normal liver function test and no clinical symptom of hepatic failure, while “noncirrhotic” patients had positive serum viral marker with abnormal liver function test, but no symptom of hepatic failure. Patients with symptom of hepatic failure were assigned to the cirrhotic patient group. Disease recurrence was defined as newly appearing lesions diagnosed by radiologic examinations such as ultrasonography and X-ray computed tomography, or based on serum tumor markers such as AFP, after an operation. Pathologic information, such as tumor size, number of tumors, gross type (vaguely nodular, expanding nodular, nodular with perinodal extension, and multinodular confluent), angioinvasion, large vessel invasion, Edmondson-Steiner (ES) nuclear grade, histologic pattern of the tumor, cellular type of tumor cells (hepatic vs. non-hepatic including giant, pleomorphic, spindle, and clear cell types), and extent of tumor invasion, was collected from pathology reports and review of the slides. Criteria for pathologic T stage (pT) followed the liver tumor staging of the American Joint Committee on Cancer, seventh edition [13]. Clinicopathologic parameters were assessed according to the general rules for examining primary liver cancer [14].

Of the 328 patients, 283 were male and 45 were female (M:F ratio of 6.3:1), with a median age of 55 years (range, 25 to 80 years). The mean size of a tumor was 5.44 cm (range, 0.8 to 24.0 cm). Most patients (92.1%, 302/328) had undergone partial hepatectomy, such as right or left lobectomy, caudate lobectomy, or segmentectomy, whereas the remaining patients (7.9%, 26/328) had undergone total hepatectomy for transplantation. Follow-up periods ranged from 0 to 161 months (median, 51 months). This study was approved by the Institutional Review Board of Seoul National University Hospital (H-1011-046-339).

Hematoxylin and eosin slides were reviewed, and one representative formalin-fixed paraffin-embedded (FFPE) archival block was selected for each case. Each core tissue biopsy (2 mm in diameter) was taken from individual FFPE blocks (donor blocks) and arranged in recipient paraffin blocks (tissue array blocks) using a trephine. Immunohistochemical studies were performed on 13 array blocks containing 328 HCC samples and on six array blocks containing 155 samples of HCC-adjacent non-neoplastic tissue as a healthy control (Superbiochips Laboratories, Seoul, Korea). Each tissue microarray had four cores of normal liver, normal bile duct, and normal gastrointestinal tract mucosa as a negative control.

Sections (4 μm) were stained for HBcAg (1:800, hepatitis B core antigen rabbit polyclonal antibody, Cat. No. B0586, DAKO, Copenhagen, Denmark), HBsAg (1:200, HBsAg mouse monoclonal antibody, clone number S1-210, Cell Marque, Rocklin, CA, USA), and HBx (1:100, HBx mouse monoclonal antibody, clone number 3F6-G10, Thermo, Rockford, IL, USA) after antigen retrieval using a microwave and pH 6.0 citrate buffer. The slides were stained according to methods specified in the Ultravision LP kit (Lab Vision, Fremont, CA, USA) or the Envision kit (Dako, Glostrup, Denmark) using the Bond polymer Refine Detection kit (Leica, Wetzlar, Germany). Unequivocal cytoplasmic staining for HBsAg in more than 5% of tumor cells was considered positive, and unequivocal nuclear and cytoplasmic staining for HBcAg in more than 5% of tumor cells was considered positive as described in a previous study [4]. For HBx, both cytoplasmic and nuclear staining were observed in tumorous and nontumorous liver tissue; thus, we separately assessed the cytoplasmic and nuclear expression of HBx. The intensity of cytoplasmic staining was graded as weak, moderate, or strong by two pathologist (H.Y.Jung and K.-B.Lee), and a grade higher than moderate was considered as positive criteria. Nuclear staining was assessed by image analysis of digitally scanned slides (Nuclear V9 algorithm, ScanScope, Aperio Technologies, Vista, CA, USA). The intensity of nuclear staining was graded as 1+ (intensity range, 230 to 210), 2+ (intensity range, 210 to 188), or 3+ (intensity range, 188 to 162), and more than 5% of tumor cells with more than 2+ intensity were considered as positive criteria.

Comparative analysis of HBsAg, HBcAg, and HBx expression with clinicopathologic parameters was assessed using the chisqaure test. Survival analysis was performed using the Kaplan-Meier method. The results were considered statistically significant when p-values were <.05. All calculations were performed using the PASW statistics ver. 18.0 (SPSS Inc., Chicago, IL, USA).

The representative immunohistochemical stainings and expression rates of HBsAg, HBcAg, and HBx in the nucleus and cytoplasm are summarized in Table 1 and Fig. 1, respectively. Among the 328 patients with HBV, the positive rate for HBsAg was significantly lower in HCC than in nontumorous tissue (7.3% vs 57.4%, respectively; p<.001), and the positive rate for nuclear HBx was higher in HCC than in nontumorous tissue (52.1% [171/328] vs 30.3% [47/155], respectively; p<.001). Cytoplasmic HBx expression was less frequent in HCC samples than in nontumorous tissue (43.3% [142/328] vs 81.3% [126/155], respectively; p<.001).

The clinicopathologic features and expression of HBsAg, HBcAg, and HBx are summarized in Table 2. HBsAg-expressing HCCs had lower ES nuclear grade and lower pT stage than HCCs without HBsAg expression (p<.001 and p=.013, respectively). HBcAg was more frequently expressed in well-differentiated HCCs (lower ES grade) and in HCCs without preoperative treatment, such as transarterial chemoembolization, radiofrequency ablation, or percutaneous ethanol injection, as well as in patients who had lower serum AFP (p=.018, p=.037, and p=.015, respectively). Cytoplasmic expression of HBx was associated with similar features: lower ES nuclear grade, smaller tumor size, less frequent angioinvasion, lower pT stage, and a lower serum AFP level compared to HCCs not expressing cytoplasmic HBx (p<.05) (Table 2). Compared to HBsAg and HBcAg, nuclear HBx was more frequently expressed in HCCs with higher ES grade and with large-vessel invasion (p=.008, and p=.017, respectively). Although nuclear HBx was more frequently expressed in HCCs with high ES nuclear grade, the histologic pattern of HCCs with nuclear HBx expression was mostly trabecular and retained the hepatic cell type (p=.002 and p=.086, respectively) (Table 2). Age and sex were not different between HBV protein positive groups and negative groups (data not shown).

We set out to test whether the serum level of hepatitis B markers or the expression profile of nontumorous liver tissue showed any correlation with the expression of HBV proteins in HCCs. All patients with HBsAg-positive HCCs were HBsAg positive in serum. Patients with positive serum HBeAb results showed lower positive rates of HBsAg in HCCs than did patients with negative serum HBeAb (10.8% [7/65] vs 1.8% [2/108], p=.027). Expression of HBcAg and expression of nuclear HBx in HCCs were not significantly concordant with the serum status of HBsAg, HBsAb, HBcAb, HBeAg, and HBeAb. As for the clinical stage of hepatitis, nuclear HBx was more frequently expressed in HCCs of noncirrhotic chronic hepatitis patients compared to asymptomatic carriers or cirrhotic patients (noncirrhotic vs asymptomatic vs cirrhosis; 75% [15/20] vs 35% [13/37] vs 43.9% [54/123], p=.023).

To test whether the HBsAg, HBcAg, or HBx status of tumors influences cancer progression, we analyzed DFS time after operation in the 328 patients with HBV-associated HCC using Kaplan-Meier analysis. We could not determine any correlation between DFS time and the expression status of HBsAg, HBcAg, and HBx in tumors and nontumorous tissue or the serum levels of hepatitis markers (data not shown). The size and multiplicity of the tumor, angioinvasion, ES nuclear grade, and the extent of tumor invasion were statistically significant prognostic factors for tumor recurrence in the entire sample of 328 patients (data not shown). The low frequency of positive HBsAg in HCCs and well-differentiated histology of HBsAg-expressing HCCs may be a confounder of the above-mentioned negative results in the entire group, so we stratified the study group and analyzed progression-free survival in 99 patients with early-stage (pT1) and well-differentiated HCC (ES nuclear grade 1 and 2); these patients were a relatively low-risk group for HCC recurrence. In 48.5% (48/99) of the patients, the tumor recurred and median DFS time was 84 months, ranging from 2 to 153 months. As shown in Table 3, nuclear HBx in tumors was a statistically significant prognostic factor (p=.015), whereas the presence of HBsAg in a tumor showed a tendency for poor prognosis of early-stage HCCs, but did not reach statistical significance (p=.095). Cytoplasmic expression of HBx in tumor and nontumorous tissue was not associated with tumor recurrence. In nontumorous liver tissue, HBsAg-, HBcAg-, and HBx-positive patients had a relatively lower median DFS, but this effect was statistically insignificant (log-rank p>.05) (Table 3). The presence of HBsAg in the tumor, serum, and nontumorous tissue was correlated with shorter DFS and early recurrence compared to HBsAg-negative patients, as shown in Fig. 2, but this effect was not statistically significant (Table 3). Survival analysis for the high-risk group (advanced stages, pT2–4) or higher nuclear grade (ES nuclear grade 3 and 4) showed no significant difference in DFS according to the status of HBV-associated proteins in serum, tumor, and nontumor tissues (data not shown).