Hepatocellular carcinoma (HCC) has a high tendency for recurrence after radical treatment. Apart from tumor and liver function parameters, little is known about the role of hepatitis B virus (HBV) factors in the recurrence of HCC. The objective of this study was to identify the potential relation between viral load and HCC recurrence in patients undergoing transarterial chemolipiodolization.
This was a cohort study of 62 consecutive patients who had HBV-related HCC and achieved complete necrosis with transarterial chemolipiodolization. Risk factors, including viral load for posttreatment recurrence, were analyzed.
Overall, 32 of 62 patients (51.6%) developed a recurrence during the study period (7.2–37.5 months). Multivariate analysis established Child-Pugh Class B (P = .014), multiple tumors (P = .013), and high viral load (HBV DNA levels >105 copies/mL) at complete necrosis (P = .001) as independent risk factors for recurrence. On both univariate and multivariate analyses, high viral load at the time of complete necrosis was identified as the strongest factor for recurrence; moreover, its statistically significant effects still were observed even when conducting the analyses separately for both local recurrence (P = .018) and distant recurrence (P = .009).
Hepatocellular carcinoma (HCC) is a major health problem in regions like Korea, where the hepatitis B virus (HBV) is endemic. Although surgical resection or liver transplantation are the only potentially curative options for patients with HCC, only a few patients are eligible for these treatments because of various limitations, including tumor multiplicity, underlying poor liver function, and organ shortage. Several locoregional therapies have been proposed for patients with unresectable and nontransplantable HCC in whom the curative options are precluded. Among these, transarterial chemotherapy with or without embolization is used commonly as part of a palliative treatment for unresectable HCC, providing objective response rates that range from 12% to 55%.1 This approach reportedly yielded a complete necrosis rate of 39.6% to 64% when it was applied to small HCCs.2, 3 However, even in the presence of an initial remission, only a few patients remain in long-term remission after intra-arterial chemotherapy, with an overall recurrence rate of around 50%.4
The results from a recent analysis of patients who underwent surgical resection suggested that HBV DNA levels are correlated positively with the risk of HCC recurrence.5 This issue is intriguing but, to our knowledge, never has been investigated in patients undergoing transarterial chemotherapy. With respect to patients who have HBV-related HCC, a special consideration needs to be given to the potential increase in HBV viral load during chemotherapy; this is based on recent observations that HBV replication often is reactivated during the course of either transarterial6–8 or systemic therapy.9 In fact, the question of whether or not the increase in viral load may promote a recurrence of HCC has yet to be answered in the treatment of HCC, particularly in the transarterial setting, because this mode of treatment indeed is one of the most used approaches in the treatment of unresectable HCC.
To address this issue, we conducted a cohort study with the objective of identifying potential correlations between the HBV viral load and HCC recurrence in patients undergoing transarterial chemolipiodolization. From a risk-factor analysis, this study presents evidence of an independent role of HBV viral levels in the development of both local and distant recurrence.
MATERIALS AND METHODS
Between March 2003 and February 2005, a total of 168 patients who were diagnosed with HBV-related HCC at our liver unit underwent transarterial chemolipiodolization as a primary treatment for HCC. All patients were positive for serum hepatitis B surface antigen (HBsAg) for at least 6 months before their diagnosis of HCC and were negative for antibodies to hepatitis C and human immunodeficiency virus. The selection criteria for transarterial chemolipiodolization included serum bilirubin level <3 mg/dL; albumin level >2.8 g/dL; prothrombin time >50%; and the absence of extrahepatic metastasis, main portal vein thrombosis, and underlying cardiac or renal diseases. Chemolipiodolization also was received by patients who either refused resection or were ineligible for radiofrequency ablation or percutaneous ethanol injection therapy because of tumor location adjacent to bowel or gall bladder or because of inaccessibility by ultrasonography. Treatment was received only after informed consent was obtained from each patient.
In total, 62 of 168 patients who underwent chemolipiodolization achieved complete necrosis and were entered consecutively into the study. Among these, 19 patients previously had been enrolled in the pre-emptive lamivudine arm of a randomized study, as described elsewhere.7 In addition, 5 other patients received lamivudine, which was started before or during transarterial treatment with clinical indication. Thus, 24 of 62 patients (38.7%) in the current study received lamivudine therapy. No other antiviral therapy except lamivudine was given during the study period.
The chemolipidolization regimen that we used was a combination of intra-arterial epirubicin (50 mg/m2) and/or cisplatin (60 mg/m2) in a mixture of lipiodol (5–10 mL) without gelfoam embolization. The procedure was repeated at 1- to 2-month intervals. The dosage of chemotherapeutic agents and treatment interval were modified as described in our previous report.10 The transarterial chemolipiodolization was continued without any limit on the number of cycles until the achievement of complete necrosis. All patients were tested at baseline for HBsAg, antibody to HBsAg, hepatitis B e antigen (HBeAg), antibody to HBeAg (anti-HBe) (Abbott Laboratories, Abbott Park, Ill), serum aminotransferase levels, albumin, bilirubin, prothrombin time, α-fetoprotein (AFP), and HBV DNA (VERSANT 3.0, branched DNA assay; Bayer HealthCare LLC, Tarrytown, NY; detection range, from 2 × 103 copies/mL to 1 × 108 copies/mL). Complete blood cell counts and liver biochemical tests were performed before and during up to 1 week after each cycle of chemolipiodolization. Tests for serum HBV DNA, HBeAg/anti-HBe, and AFP were repeated at each visit for a scheduled course of chemolipiodolization or more frequently if necessary. For patients who achieved complete necrosis, the treatment was stopped; then, an abdominal computed tomography (CT) scan was obtained and serum AFP levels were checked every 2 to 3 months over the follow-up period.
The diagnosis of HCC was made histologically, or it was based on an elevated serum AFP level (>400 ng/mL) with a typical radiologic feature, or there were consistent findings from at least 2 separate imaging techniques, including a CT scan, magnetic resonance image, or angiography. Complete necrosis was defined as the radiologic disappearance of both a contrast-enhanced area on a dynamic CT scan and tumor staining on angiography together with normalization of AFP levels on 2 consecutive occasions. Recurrence was considered when there was a reappearance of viable tumor area on follow-up radiologic examinations. The enhanced or vascular staining area located within or adjacent to the treated tumor was defined as a local recurrence, and the presence of a new lesion developing at a different site from the previous tumor area was defined as a distant recurrence.
Data were expressed as the mean ± standard deviation or median (range). The cumulative rate of recurrence after complete necrosis was estimated using the Kaplan-Meier method, and the difference was determined by the log-rank test. Variables that had an association (P < .20) with recurrence in the univariate analysis also were included in a multivariate Cox regression analysis to determine the independent contribution of each variable to the recurrence after complete necrosis. A P value <0.05 was considered significant. The analysis software used was the Statistical Package for Social Science (version 11.5; SPSS Inc., Chicago, Ill).
Demographic and baseline characteristics of the enrolled patients are shown in Table 1. In total, 62 patients (40 men and 22 women) were included in the study, and the mean patient age was 56.2 ± 8.8 years. In total, 265 cycles of chemolipiodolization were performed to achieve complete necrosis with a median of 4 cycles per patient (range, 3–8 cycles per patient). The mean interval between treatment cycles was 5.8 ± 1.3 weeks (range, 4.1–9.2 weeks). Patients were followed for a median of 16.6 months (range, 7.2–37.5 months) after achieving complete necrosis.
Table 1. Baseline Characteristics of All Patients With Hepatitis B Virus-related Hepatocellular Carcinoma Who Achieved Complete Necrosis With Transarterial Chemolipiodolization
Characteristics, n = 62
HBeAg indicates hepatitis B e antigen; HBV, hepatitis B virus; ALT, alanine aminotransferase; AFP,α-fetoprotein.
56.2 ± 8.8
Sex, men:women, no.
HBeAg status, positive:negative, no.
HBV DNA level: Median (range), ×103 copies/mL
ALT: Mean ± SD, IU/L
52.7 ± 33.1
Albumin: Mean ± SD, g/dL
3.7 ± 0.5
Prothrombin time: Mean ± SD, %
81.3 ± 18.8
Total bilirubin: Mean ± SD, mg/dL
0.98 ± 0.55
Child-Pugh classification, A:B
AFP: Median (range), ng/mL
Tumor size: Mean ± SD, cm
3.1 ± 1.3
No. of tumors, single:multiple
Tumor type, nodular:infiltrative, no.
Overall HCC Recurrence and Its Risk Factors
Overall, 32 of 62 patients (51.6%) had a recurrence of HCC during the study period. The estimated recurrence rates at 6 months, 12 months, 18 months, and 24 months after complete necrosis were 3.8%, 29.6%, 56.9%, and 64.1%, respectively. Of the 32 patients who developed a recurrence, 13 patients (40.6%) had a local recurrence, and the remaining 19 patients (59.4%) had a distant recurrence.
To evaluate the factors that affected recurrence after complete necrosis, the 11 variables of interest shown in Table 2 were included in the analysis. In the univariate analysis, Child-Pugh Class B (P = .013), multiple tumors (P = .010), and high serum HBV DNA levels at complete necrosis (P = .008) were associated with recurrence after achieving complete necrosis. Figure 1 shows serum HBV DNA levels at complete necrosis in patients with and without recurrence, indicating a tendency toward higher HBV DNA levels in patients with recurrence than in patients without recurrence (P = .087). Consistently, multivariate analysis also indicated that Child-Pugh Class B (odds ratio [OR], 3.81; 95% confidence interval [95% CI], 1.52–9.52; P = .014), multiple tumors (OR, 2.83; 95%CI, 1.25–6.43; P = .013), and serum HBV DNA levels >105 copies/mL at complete necrosis (OR, 3.77; 95%CI, 1.70–8.38; P = .001) were independent risk factors for recurrence (Table 3). On both univariate and multivariate analyses, the serum HBV DNA level at the time of complete necrosis was identified as the strongest factor for recurrence after complete necrosis. Patients who had HBV DNA levels >105 copies/mL at the time of complete necrosis experienced more frequent episodes of posttreatment cancer recurrence after complete necrosis than patients who had HBV DNA levels ≤105 copies/mL (P = .005; log-rank test) (Fig. 2).
Table 2. Univariate Analysis of Factors Related to Recurrence After Complete Necrosis With Chemolipiodolization in Patients With Hepatitis B Virus-related Hepatocellular Carcinoma
All patients (n = 62)
Median RFS, mo
RFS indicates recurrence-free survival; HBeAg, hepatitis B e antigen; HBV, hepatitis B virus; AFP, α-fetoprotein.
Age: ≤55 y vs >55 y
16.8 vs 17.3
Sex: Men vs women
17.3 vs 23.7
HBeAg status: Postive vs negative
15.2 vs 19.5
Baseline HBV DNA: ≤105 copies/mL vs >105 copies/mL
21.4 vs 16.4
Child-Pugh Class A vs Class B
19.5 vs 10.1
AFP ≤100 ng/mL vs >100 ng/mL
18.8 vs 16.4
Tumor size ≤3 cm vs >3 cm
17.3 vs 15.2
No. of tumors: Single vs multiple
21.4 vs 10.9
Tumor type: Nodular vs infiltrative
16.8 vs 15.5
Lamivudine during transarterial therapy: Treated vs untreated
27.5 vs 15.2
HBV DNA levels at complete necrosi: ≤105 copies/mL vs >105 copies/mL
27.5 vs 14.7
Table 3. Multivariate Analysis of Factors Related to Recurrence After Complete Necrosis With Chemolipiodolization in Patients With Hepatitis B Virus-related Hepatocellular Carcinoma
All patients, n = 62
OR (95% CI)
OR indicates odds ratio; 95% CI, 95% confidence interval; HBeAg, hepatitis B e antigen; HBV, hepatitis B virus.
No. of tumors
HBV DNA levels at complete necrosis
With respect to the use of lamivudine, 10 of 24 patients (41.7%) who received lamivudine and 22 of 38 patients (57.9%) who did not receive lamivudine developed a recurrence during follow-up. At the time of complete necrosis, the median serum HBV DNA levels were 13,940 copies/mL (range, from <2 × 103 copies/mL to 3.9 × 107 copies/mL) in patients who received lamivudine and 2,084,000 copies/mL (range, from <2 × 103 copies/mL to >108 copies/mL) in patients who did not receive lamivudine (P = .017). Overall, the patients who received lamivudine tended to have a lower rate of recurrence after complete necrosis compared with patients who did not receive lamivudine (P = .089) (Table 2). However, the observed effect of lamivudine was less evident in the multivariate analysis (Table 3).
Because this study included patients who received lamivudine therapy, which may have affected the interpretation of the current results, we also examined the effect of viral loads on HCC recurrence in the 38 patients who did not receive lamivudine therapy. Consequently, HBV DNA levels >105 copies/mL at complete necrosis (P = .046), multiple nodules (P = .053), and Child-Pugh Class B (P = .061) were associated significantly or marginally with recurrence in the univariate analysis. On multivariate analysis, only Child Class B (OR, 3.40; 95%CI, 1.15–10.04; P = .027) and HBV DNA levels >105 copies/mL at complete necrosis (OR, 3.87; 95%CI, 1.21–12.37; P = .022) finally remained independently correlated with recurrence.
Viral Loads and Location of Recurrence
To determine whether or not the impact of viral load on HCC recurrence would vary according to the locational pattern of recurrence, overall recurrence was categorized into the 2 groups: local recurrence and distant recurrence. Table 4 shows that a high viral load (HBV DNA levels >105 copies/mL) at the time of complete necrosis was identified as the most important independent factor for both local recurrence (OR, 5.24; 95%CI, 1.33–20.59; P = .018) and distant recurrence (OR, 4.11; 95%CI, 1.43–11.81; P = .009). The other risk factors for recurrence included younger age and Child-Pugh Class B for local recurrence and Child-Pugh Class B and tumor multiplicity for distant recurrence (Table 4).
Table 4. Multivariate Analyses of Factors Related to Local and Distant Recurrence
OR (95% CI)
OR (95% CI)
OR indicates odds ratio; 95% CI, confidence interval; AFP, α-fetoprotein; HBV, hepatitis B virus.
Age ≤55 y
AFP >100 ng/mL
Child-Pugh Class B
Tumor size >3 cm
Multiple no. of tumors
Lack of lamivudine therapy
HBV DNA >105 copies/mL at complete necrosis
Viral Loads and Necroinflammatory Activities
To investigate the effect of viral load on necroinflammatory activities, laboratory data were compared between patients with HBV DNA levels >105 copies/mL and ≤105 copies/mL at the time of complete necrosis. The results indicated that albumin levels and prothrombin time were significantly worse in patients who had high viremia (>105 copies/mL) than in patients with low levels of virus. Aminotransferase levels also were marginally higher in high viremic patients (Table 5). However, none of these biochemical variables was associated significantly with HCC recurrence in the univariate or multivariate analyses (data not shown).
Table 5. Comparison of Laboratory Data From Patients According to Viral Load at the Time of Complete Necrosis
Patients with HBV DNA
>105 Copies/mL, n = 37
≤105 Copies/mL, n = 25
HBV indicates hepatitis B virus; HBeAg, hepatitis B e antigen; SD, standard deviation; AST, aspartate aminotransferase; ALT, alanine aminotransferase.
HBeAg seropositivity: No. of patients (%)
HBV DNA, ×103 copies/mL
Mean ± SD
24,476.9 ± 35,209.7
6.6 ± 11.7
74.0 ± 83.8
43.1 ± 23.7
65.8 ± 65.1
41.1 ± 19.2
0.97 ± 0.59
0.86 ± 0.77
3.54 ± 0.57
3.82 ± 0.40
Prothrombin time, %
78.8 ± 16.1
88.8 ± 17.6
Although many surgical and nonsurgical options have been developed for the treatment of HCC, the prognosis for these patients largely remains dismal. Even in those who receive radical therapies, posttreatment recurrence remains a medical challenge. Several factors have been reported to be associated with an increased risk of recurrent HCC after surgical resection or local ablation therapies, including tumor characteristics, such as multiplicity, size, AFP levels, portal invasion, and surgical tumor findings; and liver functional parameters, such as albumin levels, indocyanine green retention rates, and Child-Pugh class.5, 11, 12 Consistently, the 2 retrospective studies that analyzed patients undergoing transarterial treatment indicated previously that only tumor characteristics were risk factors for tumor recurrence after complete necrosis.4, 13 Those studies, however, recruited heterogeneous individuals with deferent etiologies of the disease and failed to validate viral factors of HBV as a possible risk factor for recurrence. With the inclusion of a consecutive HBV-related cohort, our study is the first to our knowledge that explores this issue in a transarterial setting.
The current study focused primarily on the correlation between HBV viral load and cancer recurrence after complete necrosis obtained by transarterial chemolipiodolization. According to our statistical analysis, the study results highlight that high HBV viral load is one of the most important risk factors for a recurrence of HCC after complete necrosis. Moreover, its statistically significant effects still were observed even when the analyses were conducted separately for both local and distant recurrences. Presumably, local recurrence occurs from an incompletely treated tumor, and distant recurrence probably is caused by multicentric carcinogenesis. In view of this, the current observation that the viral load independently affected both local and distant recurrences does not appear to be treatment-specific but, rather, suggests strong evidence of the true impact of HBV viral load on a recurrence of HCC in the setting of transarterial chemotherapy.
With regard to an optimal cut-off value for an HBV DNA level that is predictive of recurrent HCC, our data indicated a strongly positive correlation between an HBV DNA level >105 copies/mL at complete necrosis and the increased risk of posttreatment recurrence. However, it should be noted that the cut-off value of 104 copies/mL of HBV DNA also was associated significantly with recurrence in our analysis (P = .012; data not shown), although its effect was not as significant as that observed with a level of 105 copies/mL of HBV DNA. Further studies are needed to determine a true cut-off level; however, the rationale for decreasing viral loads may be apparent, because the risk of recurrence undoubtedly remained reduced in patients who had low viremia.
It is noteworthy that, in the current analysis, the influence of viral load on recurrence was identified as significant only for serum HBV DNA levels at the time of complete necrosis and not for baseline HBV DNA levels. It is possible that the lack of a significant effect with a baseline viral load may be attributable to a highly dynamic alteration in the HBV DNA levels during chemotherapy. Indeed, we reported previously that HBV viral loads often can increase during transarterial chemolipiodolization in patients with HBV-related HCC, whereas such events were extremely scarce during other treatment modalities.6 Thus, in the current analysis, the effect of baseline HBV DNA level on HCC recurrence may become less evident with the subsequent increasing or fluctuating levels of serum HBV DNA during repetitive courses of chemolipiodolization, whereas the viral status at complete necrosis (which is expected to remain virtually unchanged without further treatment) may be associated more closely with posttreatment tumor recurrence.
Given the strong association between HBV viral load and cancer recurrence, it is anticipated that the implementation of strategies for antiviral use that results in a durable suppression of HBV replication ultimately will lead to a reduction in the recurrence of HCC. However, our data failed to demonstrate an independent contribution of lamivudine in lowering the risk of HCC recurrence, although there was a trend in favor of the use of lamivudine. The lack of a significant benefit of lamivudine may have been related in part to the small number of patients analyzed and to an unsatisfactory response in a subset of the patients who received lamivudine. The issue of the ultimate benefit of antiviral therapy in this setting has yet to be confirmed from larger randomized studies.
Although the precise mechanism by which HBV DNA induces hepatocarcinogenesis is unclear, it is possible that an increased viral load and subsequent active viral replication may contribute to the carcinogenic process, probably caused by the augmented, direct oncogenic potential of HBV and the accompanying necroinflammatory process resulting from the increased viral load.14–16 This postulation is supported practically by a growing body of evidence from large-scale prospective studies, which have indicated a very strong correlation between viral load and the risk of developing HCC.17, 18 In addition, several investigations have suggested an association between active hepatitis and the postoperative recurrence of HCC.5, 19 Consistently, we observed that biochemical profiles indicative of active inflammation in our data were worse in patients who had high viremia than in patients who had low viremia, further supporting the theory of the potential carcinogenic process through active inflammation associated with high viremia.
One limitation of this study is the lack of information on viral factors other than viral load. From previous epidemiologic studies, there is growing evidence in support of a pathogenic role of HBV genotypes and molecular variants in the development of HCC. Most studies on this issue have suggested genotype C or B and core promoter/precore variants as factors associated with an increased risk of HCC.20, 21 However, in Korea, almost all chronic HBV carriers are infected with genotype C HBV, which frequently has been associated with the presence of variants in the core promoter region,22, 23 and we believe that this type of HBV represents the majority of the viral genomes in our study cohort.
In conclusion, among individuals undergoing transarterial chemolipiodolization, a high serum HBV DNA level at the time of complete necrosis was associated independently with an increased risk of HCC recurrence, irrespective of the locational pattern of its recurrence. Because this cancer has a high tendency for recurrence even after curative treatments, it is very important to explore effective ways to increase disease-free survival in the patients who achieve an optimal response. Apart from other tumor and liver function parameters that potentially may be associated with recurrence, viral loads are modifiable with antiviral therapy. In this regard, we expect that the current findings will provide an equitable basis for conducting future prospective studies that test the applicability of antiviral therapy in reducing the risk of HCC recurrence in the setting of transarterial treatment for HBV-related HCC.