Aliment Pharmacol Ther 2011; 34: 344–352
Background Optimal duration of anti-viral therapy in chronic hepatitis B virus (HBV) infection remains unclear.
Aim To investigate factors that could predict relapse after stopping anti-viral agents.
Methods Chronic hepatitis B patients who were treated with anti-viral agents (lamivudine, adefovir, entecavir) and have stopped the treatment were recruited. Anti-viral agents were stopped according to the recommendations of the Asian Pacific Association for the Study of the Liver. Virological relapse was defined as an increase in serum HBV DNA to >1000 copies/mL after discontinuation of treatment.
Results Eighty-four (69 treatment naïve and 15 lamivudine resistant) patients were eligible for this study. Thirty-seven patients developed virological relapse at 4.3 ± 2.9 (range 1–11) months after discontinuation of therapy. The 1-year cumulative probability of virological relapse was 42% and 47% in HBeAg (hepatitis B e antigen)-positive (n = 41) and HBeAg (hepatitis B e antigen)-negative (n = 43) patients, respectively. On multivariate analysis by Cox proportional hazard model, pre-existing lamivudine resistance, delayed suppression of HBV DNA to undetectable level during anti-viral therapy and to a higher HBsAg (hepatitis B surface antigen) level at the end of treatment were associated with virological relapse. Twelve of the 15 (80%) lamivudine resistant patients developed virological relapse. Among the 11 treatment naïve patients who had HBsAg ≤2 log IU/mL at the end of treatment, 1 (9%) of them had virological relapse.
Conclusions Treatment cessation among lamivudine resistant patients is associated with high risk of virological relapse. Serum HBsAg level at the end of treatment and rate of HBV DNA suppression can provide supplementary information to guide the timing of stopping anti-viral drugs.
Chronic hepatitis B is the major cause of liver cirrhosis and hepatocellular carcinoma in China and most parts of Asia.1, 2 Patients who have liver cirrhosis, elevated hepatitis B virus (HBV) DNA and elevated alanine aminotransferase are at risk of developing hepatocellular carcinoma.3–5 Nowadays, peginterferon and oral anti-viral agents are the mainstays of therapy. Meta-analyses of long-term follow-up studies have shown that both interferon and oral anti-viral agents can reduce the risk of hepatocellular carcinoma, cirrhotic complications and mortality related to chronic hepatitis B virus (HBV) infection.6, 7 The therapeutic benefit is most obvious among patients who have histologic response to anti-viral therapy.8
Peginterferon has the benefit of sustained immune control after a finite duration of therapy both in patients with positive and negative hepatitis B e antigen (HBeAg).9–11 It is therefore considered as one of the first line treatments by the regional guidelines despite its adverse effects, inconvenience route of subcutaneous administration and limited efficacy.12–14 On the other hand, oral anti-viral agents can be prescribed as once-daily oral dosing with minimal side effects, and are very effective in viral suppression and normalisation of liver enzymes. However, most patients require long-term therapy and virological relapse is common after premature cessation of therapy.15, 16
Owing to the high cost of the anti-viral drugs, some Asian countries, such as Taiwan and Indonesia, only reimburse the drugs for a finite duration of time.17 In other places such as Hong Kong, the indications of reimbursement for anti-viral agents are very restricted so that only a minority of patients can be benefited.18 For cost-saving and improving patient compliance, it is prudent to investigate on the best timing to stop anti-viral therapy. Clearance of hepatitis B surface antigen (HBsAg) is the ideal endpoint to stop treatment, but its occurrence is usually lower than 5% in 5 years with anti-viral therapy.19, 20 In the recommendation by the Asian Pacific Association for the Study of the Liver (APASL), anti-viral drug is recommended to stop when HBeAg seroconversion has developed for more than 6 months among HBeAg-positive patients.12 For HBeAg-negative patients, the APASL consensus recommended stopping anti-viral treatment when HBV DNA remained undetectable for three separate occasions 6 months apart.12 In a survey among 124 gastroenterologists in 12 Asian-Pacific countries, approximately 60% of them will stop anti-viral treatment according to these criteria.21 Nonetheless, approximately 25% to 50% of the patients may still develop hepatitis relapse after stopping anti-viral therapy even if these recommendations are followed.15, 22–24
In this study, we aimed to investigate the factors that could predict the best timing of stopping anti-viral therapy. A prospective cohort of patients were recruited and followed closely since the time of starting anti-viral treatment, and the timing of stopping treatment was complied with the recommendations of the APASL. The results of our study would provide important supplementary information to the APASL recommendations in the selection of patients and timing of stopping anti-viral agents.
Materials and methods
Chronic hepatitis B patients were recruited from the Department of Infectious Diseases, the First Affiliated Hospital of Guangxi Medical University. Chronic hepatitis B was diagnosed as serum HBsAg-positive for at least 6 months. Treatment naïve patients were treated by anti-viral agents if they had elevated HBV DNA (HBV DNA ≥105 copies/mL for HBeAg-positive and HBV DNA ≥104 copies/mL for HBeAg-negative patients) and abnormal alanine aminotransferase (ALT) for ≥6 months. Chronic hepatitis B patients were treated with lamivudine, adefovir dipivoxil or entecavir according to the joint decision of the patients and the physicians based on the reimbursement policy and the financial affordability of the patients. Patients who presented with lamivudine resistance (virological breakthrough with serum HBV DNA ≥1000 copies/mL and >1 log rise from nadir) were treated with salvage adefovir (either monotherapy or add-on therapy). The baseline visit was taken as the time of starting anti-viral therapy for treatment naïve patients and starting salvage therapy for lamivudine resistant patients. All patients who had regular follow-up with HBV DNA, HBV serological tests and ALT were monitored every month before HBV DNA was suppressed to <1000 copies/mL and then every 3 months thereafter. Only patients who had good compliance to medication and follow-up were analysed in this study. Patients who had co-infection with hepatitis C virus, hepatitis D virus or human immunodeficiency virus were excluded by commercialised serology assays. Patients who suffered from acute hepatitis, decompensated liver cirrhosis, hepatocellular carcinoma and those who had a history of interferon treatment and immunosuppressive therapy were also excluded from this study.
Anti-viral therapy was stopped according to the recommendations by the APASL in 2008.12 In brief, anti-viral agent was stopped in HBeAg-positive patients when HBeAg seroconversion with undetectable HBV DNA was documented on two separate occasions at least 6 months apart. In HBeAg-negative patients, treatment discontinuation would be considered if undetectable HBV DNA has been documented on three separate occasions 6 months apart. After the cessation of anti-viral therapy, HBV DNA, HBV serology tests and ALT were monitored every month within first 6 months and then every 2 months afterwards.
The primary end-point of our analysis was virological relapse, which was defined as an increase in serum HBV DNA to >1000 copies/mL on at least two determinations more than 4 weeks apart after discontinuation of treatment. Sustained response was defined as undetectable HBV DNA (<1000 copies/mL) after stopping anti-viral therapy for ≥6 months till the last follow-up visit.
Hepatitis B virus DNA was measured by real-time polymerase chain reaction (PCR) assay (Shanghai Kehua Bio-engineering Co., Ltd, Shanghai, China) with a lower limit of quantification at 1000 copies/mL. Serum HBsAg, HBeAg and anti-HBe were measured by time-resolved fluorescence immunoassay kits (Sum-Bio Lifescience, Shanghai, China). This HBsAg kit provided a linear range of quantification between 0 and ≥225 ng/mL. Among the 84 patients in this study, 73 patients had serum HBsAg at the end of anti-viral therapy quantified by the Architect chemiluminescent microparticle immunoassay (Abbott Laboratories, Abbott Park, IL, USA). The sensitivity of the Architect assay ranged from 0.05 to 250 IU/mL. Samples with HBsAg titer higher than 250 IU/mL were retested after dilution to 1:500–1000 to bring the reading to the range of the calibration curve.
Statistical tests were performed by spss (version 13.0; Chicago, IL, USA). Categorical variables were compared by Chi-squared test. Continuous variables were compared by Student’s t-test or Mann–Whitney U-test as appropriate. The cumulative virological relapse was determined by the Kaplan–Meier method, and the difference between groups was determined by the log rank test. Cox proportion hazard model was performed to analyse factors associated with virological relapse, and significant factors (P < 0.05) in the univariate analysis were subjected to multivariate analysis to determine independent predictive factors. Serum HBsAg results by Architect assay were logarithmic transformed for analysis. Area under the receiver operating characteristics (ROC) curve was performed to assess the predictive values of HBsAg at the end of therapy for virological relapse. Statistical significance was defined as a P value of less than 0.05. All statistical tests were two-sided.
Patient recruitment started in May 2004 and the last patient follow-up was in October 2010. One hundred and twenty-one consecutive chronic hepatitis B patients stopped anti-viral therapy in the out-patient clinic of the hospital during this study period. After excluding 27 patients who did not meet APASL cessation criteria and 10 patients who were followed up for <6 months with no relapse, 84 patients were eligible for this analysis. The baseline clinical features of the 84 patients in this study are shown in Table 1. Among the 69 treatment naïve patients, 38 (55%) received lamivudine monotherapy, 18 (26%) received adefovir monotherapy and 13 (19%) received entecavir monotherapy. Among the 15 lamivudine resistant patients, 8 (53%) received adefovir monotherapy and 7 (47%) received combination therapy of lamivudine and adefovir. Three patients had sonographic features of liver cirrhosis at baseline, but liver biopsy before stopping treatment confirmed absence of cirrhosis. The remaining 81 patients did not have any clinical or ultrasonic features of liver cirrhosis before starting anti-viral therapy.
|Factors||Total||Virological relapse||Sustained response|
|Number of patients||84||37||47|
|Age (year)||37.0 ± 11.9||39.6 ± 11.6||34.9 ± 11.8|
|Male (%)||56 (67)||25 (45)||31 (55)|
|Alanine aminotransferase (IU/L)||79 (13–607)||75 (13–453)||83 (15–607)|
|HBeAg-positive (%)||41 (49)||17 (42)||24 (59)|
|HBV DNA (log copies/mL)||5.90 ± 1.42||5.90 ± 1.40||5.91 ± 1.50|
|Treatment naïve (%)||69 (82)||25 (36)||44 (64)|
|Lamivudine resistance (%)||15 (18)||12 (80)||3 (20)|
|Time to first undetectable HBV DNA during treatment (month)||3.14 ± 1.9||3.81 ± 2.2||2.62 ± 1.4|
|Level of HBsAg at the end of therapy (ng/mL)||158.7 ± 81.4||183.8 ± 65.0||138.9 ± 88.0|
|Total treatment duration (month)||32.1 ± 7.5||33.1 ± 9.1||31.4 ± 5.9|
|Follow-up period after drugs cessation (month)||8.4 ± 6.0||4.3 ± 2.9||11.6 ± 5.9|
Thirty-seven of the 84 patients developed virological relapse at 4.3 ± 2.9 (range 1–11) months after discontinuation of therapy. The cumulative probability of virological relapse after stopping anti-viral therapy was 35% in 6 months and 44% in 1 year (Figure 1). Among them, 12 patients had ALT elevated to more than two times the upper limit of normal (148.2 ± 39.1 IU/L).
Among the 47 sustained responders, they were followed up for 11.6 ± 5.9 (range 6–26) months. Five patients (4 HBeAg-positive and 1 HBeAg-negative at baseline) cleared HBsAg after treated for 37.6 ± 10.7 (range 27–51) months. All of the patients who had HBsAg clearance were treatment naïve; four were treated with lamivudine and one with adefovir.
HBeAg status and virological relapse
Forty-one HBeAg-positive patients received anti-viral treatment for a total of 31.4 ± 7.0 (range 24–61) months, including a consolidation treatment for 21.5 ± 7.1 (range 12–36) months after HBeAg seroconversion has developed. Seventeen (41%) patients developed virological relapse within 5.3 ± 3.1 (range 1–11) months. The cumulative probability of virological relapse in 6 months was 30% and in 1 year was 42%. The remaining 24 patients had sustained virological response in a follow-up of 12.0 ± 6.3 (range 6–26) months. There was no difference in time to the first HBeAg seroconversion among the relapsers (12.7 ± 7.8 months) vs. the sustained responders (8.1 ± 3.9 months) (P = 0.098). The duration from first HBeAg seroconversion to treatment cessation was 19.9 ± 8.6 (range 12–36) months among relapsers vs. 22.7 ± 5.6 (range 12–33) months among sustained responders (P = 0.13).
Forty-three HBeAg-negative patients received anti-viral treatment for 32.7 ± 7.9 (range 24–59) months. Among 43 HBeAg-negative patients, 20 (47%) developed virological relapse within 3.3 ± 2.3 (range 1–9) months. The cumulative probability of virological relapse in 6 months was 40% and in 1 year was 47%. The remaining 23 patients had sustained virological response in a follow-up of 11.3 ± 5.6 (range 6–26) months. The duration from first undetectable HBV DNA to treatment cessation was 30.7 ± 8.9 (range 23–54) months among relapsers vs. 29.7 ± 6.6 (range 22–48) months among sustained responders (P = 0.67).
Factors associated with virological relapse
There was no difference in the age, gender ratio, HBeAg status, HBV DNA and ALT levels at the time of starting anti-viral therapy, and there was no difference in the treatment duration between the relapsers and the sustained responders (Tables 1 and 2). Patients who developed virological relapse tend to have pre-existing lamivudine resistance, delayed suppression of HBV DNA to undetectable level during anti-viral therapy and to a higher HBsAg level at the end of treatment. All three factors remained as independent factors associated with virological relapse on multivariate analysis.
|Factors||Univariate analysis||Multivariate analysis|
|HR||95% CI||P||HR||95% CI||P|
|Baseline ALT (IU/L)||0.999||0.995–1.002||0.43|
|Baseline positive HBeAg||0.888||0.466–1.692||0.72|
|Baseline HBV DNA (log copies/mL)||0.994||0.788–1.254||0.96|
|Time to first undetectable HBV DNA during treatment (month)||1.290||1.087–1.531||0.004||1.292||1.083–1.542||0.004|
|Level of HBsAg at the end of therapy (ng/mL)||1.005||1.000–1.010||0.043||1.007||1.001–1.012||0.015|
|Total treatment duration (month)||1.024||0.984–1.066||0.25|
Presence of lamivudine resistance
Presence of lamivudine resistance was the most important predictor of virological relapse after stopping anti-viral therapy (hazard ratio 3.650 on multivariate analysis) (Table 2). Of the 15 patients who were resistant to lamivudine, 12 (80%) of them developed virological relapse in spite of meeting the APASL criteria after 15 ± 6.4 (range 12–20) months of therapy. The 6-month and 1-year cumulative probabilities of virological relapse were 73.3% and 80.0% among patients with lamivudine resistance as compared to 26% and 37% among patients who were treatment naïve, respectively (log rank test P < 0.0001) (Figure 2a).
Rate of virological suppression on anti-viral therapy
Patients who had slower virological suppression to PCR undetectable during anti-viral therapy tend to have a higher risk of virological relapse after treatment cessation (Tables 1 and 2). Among the 56 patients who had HBV DNA suppressed to undetectable level in 3 months, 19 (34%) developed virological relapse. On the other hand, among the 28 patients who still had detectable HBV DNA at month 3 of anti-viral treatment, 18 (64%) developed virological relapse after stopping treatment. The 6-month and 1-year cumulative probabilities of virological relapse were 29% and 36% among patients who had HBV DNA undetectable within 3 months vs. 46% and 61% among those who still had detectable HBV DNA at month 3 of anti-viral treatment, respectively (log rank test, P = 0.034) (Figure 2b).
Serum HBsAg level at the end of treatment
Overall, higher serum HBsAg at the end of treatment was associated with higher risk of virological relapse (Figure 2c). Among the 73 patients who had HBsAg quantified by the Architect assay, the HBsAg level was 2.74 ± 1.21 (range −1.15 to 4.28) log IU/mL. The area under ROC curve for serum HBsAg to predict virological relapse was 0.68 (95% confidence interval 0.56, 0.81, P = 0.008). Thirty-five patients had HBsAg ≤3 log IU/mL at the end of treatment and 11 (31%) of them had virological relapse at the last follow-up. The 6-month and 1-year cumulative probabilities of virological relapse among patients who had serum HBsAg ≤3 log IU/mL were 26% and 31%, respectively. Among the 11 patients who had HBsAg ≤2 log IU/mL at the end of treatment, only 1 (9%) of them had virological relapse at the last follow-up. The 6-month and 1-year cumulative probabilities of virological relapse among patients who had serum HBsAg ≤2 log IU/mL were 9% and 9%, respectively.
Among the 11 lamivudine resistant patients with HBsAg measured by the Architect assay, 10 (91%) had virological relapse. A combined algorithm of using HBsAg level at the end of treatment and HBV DNA suppression at month 3 was developed among the 62 treatment naïve patients (Figure 3). Patient who had either end-of-treatment HBsAg ≤2 log IU/mL or HBsAg >2–3 log IU/mL with undetectable HBV DNA at month 3 of treatment had lower risk of virological relapse.
In this cohort of 84 chronic hepatitis B patients stopped anti-viral therapy according to the recommendations by APASL, we have shown a 1-year cumulative probability of virological relapse of 44%. Both HBeAg-positive and HBeAg-negative patients had similar risk of virological relapse. Patients who had lamivudine resistance had the highest risk of relapse of 80% at 1 year. The rate of HBV DNA suppression in the first 3 months of treatment and the serum HBsAg level at the end of treatment could also serve as predictors of relapse.
The durability of HBeAg seroconversion after stopping oral anti-viral agents is known to be lower than that with interferon.16 Nonetheless, among HBeAg-positive patients, all regional guidelines recommend stopping anti-viral treatment 6–12 months after HBeAg seroconversion is achieved.12–14 In a recent report in Korea, among 178 HBeAg-positive patients who stopped lamivudine after achieving a complete response (HBeAg clearance, HBV DNA <140 000 copies/mL and normal ALT for 3–6 months), 16% and 30% of patients had relapse at 1 year and 5 years.25 It was likely an underestimate as relapse was defined as detectable HBV DNA by Digene II assay, which had a lower limit of detection at 140 000 copies/mL. In reports from Hong Kong, Taiwan and the Netherlands, relapse was observed in 48–69% of patients and even additional lamivudine was prescribed for more than 6 months post-HBeAg seroconversion before treatment cessation.22, 23, 26 The high rates of relapse were in line with that reported in the present study (41.5% in 1 year).
There were only a few reports in the literature on the risk of virological relapse after stopping anti-viral therapy in HBeAg-negative patients. In a randomised, placebo-controlled, multicentre study of 2-year lamivudine for HBeAg-negative chronic hepatitis B in China and Hong Kong, 17 of the 26 (65%) patients who had undetectable HBV DNA by PCR at the end of treatment had reappearance of HBV DNA at 6 months after stopping lamivudine.27 In another report from China, among 61 HBeAg-negative patients who had received lamivudine for at least 2 years, relapse (HBV DNA >10 000 copies/mL) occurred in 37% of patients in 1 year and 56% of patients in 5 years.28 Besides younger age, no clinical or virological parameter could predict relapse in this Chinese cohort. None of these reports had clear documentation of the observation of APASL criteria for treatment cessation. In our study, among the 43 patients who received anti-viral treatment for 32.7 months and stopped treatment in accordance to the APASL recommendation, 47% of patients still had virological relapse in 1 year.
A few factors have been identified to predict the risk of relapse after stopping lamivudine in HBeAg-positive patients. Higher pre-treatment HBV DNA was found to associate with higher risk of virological relapse in Taiwan and Hong Kong.22, 23 However, these studies have not assessed the impact of HBV DNA kinetics during anti-viral therapy. In our study, with close monitoring of HBV DNA after commencing anti-viral therapy, we were able to demonstrate the importance of rapid virological suppression on the risk of post-treatment relapse. On the other hand, pre-treatment HBV DNA could not be shown as an important factor to predict virological relapse in the present study. In reports from Taiwan and Korea, additional lamivudine for <8–12 months after HBeAg seroconversion was associated with increased risk of hepatitis relapse after stopping treatment.25, 29, 30 As all of our patients had additional lamivudine for >12 months after HBeAg seroconversion, we could not demonstrate any benefit of extended lamivudine treatment before stopping the drug.31
One novel finding of our study was the high risk of post-treatment relapse among patients suffering from lamivudine resistance. Most studies reported long-term use of adefovir rescue therapy for lamivudine resistance, but little data are available on the attempt of stopping therapy.32–35 It remains uncertain whether or not the cellular immune response to lamivudine resistant HBV is lower than the treatment naïve patients. Based on the current study, long-term anti-viral treatment is needed for lamivudine resistant HBV regardless of the virological response during treatment as the post-treatment relapse rate is too high.
Serum HBsAg quantification has been a hot marker of research due to its relationship with the amount of covalently closed circular DNA inside the liver.36, 37 In natural history studies, lower serum HBsAg levels were associated with improved immune viral clearance.38, 39 In peginterferon-treated patients, a lower on-treatment HBsAg level can predict sustained responders to therapy.40, 41 In studies with telbivudine, a more dramatic HBsAg decline during treatment has been shown to associate with HBsAg clearance and sustained remission of disease.42, 43 In this study, we showed that serum HBsAg levels were lower among patients who did not develop virological relapse after stopping anti-viral therapy. The relapse rate was 17% if HBsAg was ≤3 log IU/mL together with a rapid HBV DNA suppression to undetectable at month 3; and the relapse rate was 9% if HBsAg was ≤2 log IU/mL. These findings were in line with a recent finding in Hong Kong among 53 HBeAg-negative patients who stopped lamivudine treatment, in which HBsAg ≤2 log IU/mL could predict sustained remission of disease.44 On the other hand, the relapse rate among patients who had end-of-treatment HBsAg >2–3 logs but month 3 HBV DNA >1000 copies/mL might be higher than those who had serum HBsAg >3 log IU/mL (Figure 3). Unfortunately, we did not have the baseline sample for measurement of HBsAg level using the Architect assay and the use of HBsAg decline could not be assessed in this study.
Our study has a few limitations. First, owing to the limited resource in Guangxi, we could only use local HBV DNA and HBsAg tests in routine clinical practice. Although the HBV DNA test was not as sensitive as the Taqman assay, our results could demonstrate the importance of rapid on-treatment viral suppression in sustained viral control. We also tried to evaluate the clinical use of serum HBsAg assay by retesting the available serum samples at the end of lamivudine treatment by the Architect assay. Unfortunately, we were unable to retrieve the earlier samples to investigate the effect of HBsAg kinetics during treatment on the risk of relapse. Second, the post-treatment follow-up of our study was relatively short as compared to other studies. Based on previous reports in the Chinese population, most of the virological relapse occurred in the first year post-treatment.22, 23, 28 Nonetheless, some of the late relapses might be missed and the high relapse rate in this report might be an underestimate. Third, HBV genotype was not available in this study. In Southern China, the prevalence of genotype B and C HBV were approximately 50% each.45 So far, there was no suggestion that HBV genotype would affect the risk of post-treatment relapse in previous reports.22, 23, 26
In conclusion, we have demonstrated that the APASL recommendations for stopping anti-viral therapy in both HBeAg-positive and HBeAg-negative chronic hepatitis B are inadequate. More than 40% of patients develop virological relapse after stopping anti-viral drugs based on the APASL guideline. Patients with lamivudine resistant HBV infection has too high risk of post-treatment relapse and should be put on long-term anti-viral therapy. Among treatment naïve patients, serum HBsAg level should be monitored for the best timing of treatment cessation. Nonetheless, we cannot recommend the algorithm as an on-treatment response guide at the early course of treatment, and HBsAg is used as an end-of-treatment biomarker while month 3 HBV DNA can only be looked at retrospectively. The safety of post-treatment virological relapse should be another factor of concern among patients with liver cirrhosis or those presented with severe acute exacerbation of chronic hepatitis B.46 As this study is a retrospective analysis of a limited number of patients with heterogeneous treatment regimes, future prospective, large, multicentre studies are needed to confirm the usefulness of the serum HBsAg monitoring during anti-viral therapy prior to these recommendations can be incorporated into treatment guidelines.
Declaration of personal interests: H. L. Y. Chan is a consultant and advisory board member of Abbott, Bristol-Myers Squibb, F. Hoffmann-La Roche, Merck and Novartis Pharmaceutical. V. W.-S. Wong is an advisory board member of Gilead Sciences and has received honorarium for lectures from Bristol-Myers Squibb and F. Hoffmann-La Roche. Y. Liang handled patient recruitment and follow-up, database management, data analysis and manuscript writing. J. Jiang performed study design, patient recruitment and follow-up, data analysis, manuscript writing and administrative support. M. Su, Z. Liu, W. Guo, X. Huang, R. Xie, S. Ge, J. Hu, Z. Jiang and M. Zhu handled patient recruitment and follow-up. V. W.-S. Wong wrote the manuscript. H. L. Y. Chan handled data analysis and manuscript writing. Declaration of funding interests: This work was supported by Natural Science Foundation of Guangxi (NO.2010GXNSFD013046, 0832117, 0542092), Medical Science Experimental Center Foundation of Guangxi (KFJJ2010-20) and National Key Technologies Research and Development Program of China during the 5-year Plan Period to J. Jiang.