Hepatitis B virus surface antigen levels: A guide to sustained response to peginterferon alfa-2a in HBeAg-negative chronic hepatitis B
Hepatology April 2009
"On-treatment quantification of HBsAg in patients with HBeAg-negative chronic hepatitis B treated with peginterferon alfa-2a may help identify those likely to be cured by this therapy and optimize treatment strategies......HBsAg clearance represents the best possible and closest to cure outcome of antiviral therapy in patients with CHB, but is realistic almost exclusively among patients receiving interferon-based regimens, which are recommended as a first-line therapeutic approach. Quantitative on-therapy measurement of serum HBsAg may be useful to identify patients who are most likely to benefit from interferon-based therapy and may help tailor and optimize treatment duration. Further studies to determine the optimal time points to measure HBsAg level are therefore warranted."
Maurizia Rossana Brunetto 1 *, Francesco Moriconi 1, Ferruccio Bonino 2, George K. K. Lau 3, Patrizia Farci 4, Cihan Yurdaydin 5, Teerha Piratvisuth 6, Kangxian Luo 7, Yuming Wang 8, Stephanos Hadziyannis 9, Eva Wolf 10, Philip McCloud 11, Richard Batrla 12, Patrick Marcellin 13
1UO Epatologia, Azienda Ospedaliero Universitaria Pisana, Pisa, Italy
2Scientific Direction, Foundation IRCCS Policlinico of Milan and University of Pisa, Italy
3Department of Medicine, Queen Mary Hospital, Hong Kong, China
4Dipartimento di Scienze Mediche, Universita di Cagliari, Monserrato, Italy
5Department of Gastroenterology, University of Ankara, Turkey
6Department of Internal Medicine, Songklanagarind Hospital, Prince of Songkla University, Hat Yai, Thailand
7Department of Infectious Diseases, Nanfang Hospital, Guangzhou, China
8Infectious Disease Department, Xinan Hospital, Chongqing, China
9Department of Medicine and Hepatology, Henry Dunant Hospital, Athens, Greece
10MUC Research, Munich, Germany
11Roche, Dee Why, Australia
12Roche, Basel, Switzerland
13Hopital Beaujon, University of Paris, Clichy, France
See Editorial on Page 1063
Potential conflict of interest: Dr. Brunetto is on the speakers' bureau of Roche. Dr. Bonino is a consultant for, advises, and is on the speakers' bureau of Abbott, Bristol-Myers Squibb, Gilead, and Novartis. He is also in the speakers' bureau of Schering-Plough. Dr. Lau is on the speakers' bureau of and received grants from Roche and Novartis. Dr. Piratvisuth advises and is on the speakers' bureau of Roche. He advises Novartis and Schering-Plough and is on the speaker's bureau of GlaxoSmith Kline. Dr. Hadziyannis is on the speakers' bureau of and received grants from Roche. He advises and received grants from Gilead. He advises Novartis and Bristol-Myers Squibb. Dr. McCloud owns stock in Roche.
We investigated the relationship between hepatitis B virus surface antigen (HBsAg) serum level decline and posttreatment response in patients with hepatitis B e antigen (HBeAg)-negative chronic hepatitis B from a large multinational study of pegylated interferon alfa-2a (peginterferon alfa-2a), with or without lamivudine, versus lamivudine alone. Serum HBsAg was quantified using the Architect assay (Abbott Diagnostics) at pretreatment, end of treatment (week 48), and 6 months after the end of treatment (week 72) in sera from 386 of the 537 patients who participated in the multinational study (peginterferon alfa-2a, 127; peginterferon alfa-2a plus lamivudine, 137; lamivudine monotherapy, 122). Pretreatment HBsAg levels varied according to genotype, with the highest levels present in patients infected with genotypes A (median, 4.11 log10 IU/mL) and D (median, 3.85 log10 IU/mL). Significant on-treatment decline in HBsAg was observed during treatment with peginterferon alfa-2a (alone or combined with lamivudine; mean decline at week 48, -0.71 and -0.67 log10 IU/mL, respectively, P < 0.001), but not during treatment with lamivudine alone (-0.02 log10 IU/mL). Significantly more patients treated with peginterferon alfa-2a (21%) or peginterferon alfa-2a plus lamivudine (17%) achieved HBsAg levels <100 IU/mL at the end of treatment compared with lamivudine (1%) (both P < 0.001 versus lamivudine). End-of-treatment HBsAg level correlated strongly with HBV DNA suppression to < /=400 copies/mL 6 months posttreatment. An HBsAg level <10 IU/mL at week 48 and on-treatment decline >1 log10 IU/mL were significantly associated with sustained HBsAg clearance 3 years after treatment (both P < 0.0001). __Conclusion: On-treatment quantification of HBsAg in patients with HBeAg-negative chronic hepatitis B treated with peginterferon alfa-2a may help identify those likely to be cured by this therapy and optimize treatment strategies.
Chronic hepatitis B is a global health problem accounting for 1 million deaths each year. Age-adjusted death rates are 3 to 3.6 times higher in carriers of hepatitis B virus (HBV) surface antigen (HBsAg) than in persons without HBV infection. The aim of treatment for patients with chronic hepatitis B (CHB) is to decrease progression of liver disease to cirrhosis and hepatocellular carcinoma, with the ultimate aim of improving survival. This can be pursued by maintaining constant inhibition of viral replication through a long-term treatment with nucleos(t)ide analogs or by inducing, through the combined antiviral and immunomodulatory effects of interferon, a sustained immune response. Maintenance of the inactive carrier state (HBV DNA <2,000 IU/mL and normal alanine aminotransferase [ALT]) after treatment discontinuation has been associated with good prognosis in both HBeAg-positive and HBeAg-negative CHB.[3-7] Nevertheless, HBsAg seroconversion, characterized by the loss of serum HBsAg and development of anti-HBs antibodies, is the hallmark of a successful immunological response to HBV infection and the closest outcome to clinical cure. It is associated with favorable long-term clinical outcomes, including reduced incidence of cirrhosis and hepatocellular carcinoma and longer survival.[8-12] HBsAg seroconversion can be achieved after interferon-based therapy as demonstrated by conventional interferon[5-7] and, more recently, by peginterferon alfa-2a and peginterferon alfa-2b in patients with HBeAg-positive disease and by peginterferon alfa-2a in patients with HBeAg-negative disease.
Interferon-based therapy leads to an off-therapy response in approximately one third of patients, and these patients have an increasing likelihood of achieving HBsAg clearance during long-term follow up. In contrast, HBsAg clearance is rarely achieved in patients treated with nucleos(t)ide analogs, especially in those with HBeAg-negative disease where rates are typically 0% with 1 year of therapy. Nucleos(t)ide analogs need to be given long term, because they suppress HBV DNA only during therapy. Although quantification of HBV DNA during treatment is appropriate for determining response to nucleos(t)ide analogs, it is not as suitable for monitoring the mechanisms responsible for the achievement of sustained antiviral response to interferon-based therapy in chronic hepatitis B patients.
A recent retrospective analysis suggested that quantitative determination of HBsAg level in patients treated with interferon may provide an insight into the likelihood of eventual HBsAg seroconversion. We hypothesized that the analysis of response to interferon-based therapy using the additional marker of HBsAg quantification might improve our understanding of which patients are most likely to achieve a complete posttreatment response. We performed a comparative analysis of HBsAg levels in patients with HBeAg-negative chronic hepatitis B who were treated as part of a large randomized multinational study of peginterferon alfa-2a, with or without lamivudine, and lamivudine alone. The main objectives of this study were (1) to investigate pretreatment levels of HBsAg and their possible correlation with other baseline factors including HBV genotype, (2) to quantify changes in HBsAg level during different treatment regimens, and (3) to determine if on-treatment HBsAg decline is correlated with response 6 months posttreatment and subsequent HBsAg clearance up to 3 years posttreatment.
ALT, alanine aminotransferase; CART, classification and regression tree; CHB, chronic hepatitis B; HBeAg, hepatitis B e antigen; HBsAg, hepatitis B virus surface antigen; HBV, hepatitis B virus; peginterferon alfa-2a, pegylated interferon alfa-2a.
Our analyses provide an insight into levels of HBsAg in patients with HBeAg-negative chronic hepatitis B, a direct comparison of the effects of different treatment regimens upon them and how end-of-treatment HBsAg levels relate to sustained posttreatment response. Considerable variation in HBsAg level is observed within the patient population, and levels appear to be affected by HBV genotype, with the highest levels in those patients infected with HBV genotypes A and B. There was some degree of correlation of pretreatment HBsAg with HBV DNA level, lower HBsAg levels being associated with lower levels of HBV DNA.
Serum HBsAg levels declined significantly during 48 weeks of treatment with peginterferon alfa-2a ± lamivudine, but not with lamivudine alone (P < 0.001). The decline in HBsAg was similar in the two peginterferon alfa-2a-containing regimens with approximately 10% of patients achieving HBsAg levels <10 IU/mL at end of treatment. These patients had the highest rates of HBV DNA suppression to below 400 copies/mL 6 months after treatment. This suggests that the HBsAg level reached at the end of treatment could be used as an indicator of anticipated sustained response in peginterferon alfa-2a-treated patients. Previous studies have shown that patients who achieve sustained virological posttreatment response to interferon therapy have an increasing chance of achieving subsequent clearance of HBsAg, which is considered the closest outcome to clinical cure in patients with chronic hepatitis B.
Of the 23 patients in the long-term follow-up study who had achieved an end-of-treatment HBsAg level <10 IU/mL, 52% had cleared HBsAg by 3 years after treatment. In contrast, there was no significant association between suppression of HBV DNA to <400 copies/mL at end of treatment and sustained HBsAg clearance 3 years after treatment. HBsAg level may therefore be a more appropriate marker of response to peginterferon than HBV DNA level.
Despite lowering HBV DNA levels through its antiviral effects, lamivudine had little effect on HBsAg levels, as shown by Manesis et al. Thus, antiviral suppression alone is clearly not sufficient to reduce HBsAg levels. Interferon has not only antiviral activity that suppresses viral replication, but also immunomodulatory effects which can prevent infection of previously uninfected hepatocytes and also result in clearance of infected hepatocytes. It is the combined effect of HBV DNA suppression and immunological enhancement that may help tip the balance to allow the host to gain immunological control and thereby a sustained off-therapy response and eventual HBsAg clearance. Therefore, the reduction of HBsAg serum levels during peginterferon alfa-2a therapy may be considered the hallmark for achievement of the immune competence that is needed for persistent control of chronic HBV infection. Accordingly, the modeling of HBV dynamics during therapy showed a similar or even greater impact on the clearance of the infected cell by peginterferon alfa-2a compared with lamivudine, even in the absence of the high degree of HBV DNA inhibition induced by the nucleos(t)ide analogs. Other oral antivirals, including entecavir - despite being more potent inhibitors of HBV replication - have not so far induced HBsAg clearance rates comparable to those achieved by peginterferon alfa-2a in patients with HBeAg-negative disease, despite treatment durations of 2 to 5 years.[22-24]
The potential value of quantifying serum HBsAg to predict long-term response to conventional interferon was suggested previously in patients with HBeAg-positive CHB. A significant reduction in serum HBsAg levels was seen in patients who responded to interferon (as determined by HBeAg seroconversion), but not in patients without HBeAg seroconversion (P< 0.001).
The on-treatment reduction in HBsAg afforded by peginterferon alfa-2a and its association with sustained response suggests that its monitoring could be beneficial in several ways. It may be possible, for example, to identify early during the course of therapy which patients are achieving HBsAg decline and thus most likely to respond. For example, patients treated with peginterferon alfa-2a who achieved an end-of-treatment HBsAg level <10 IU/mL had a very high chance (88%) of achieving suppression of HBV DNA to 400 copies/mL 6 months after treatment. Conversely, given the fact that not all patients are responsive to peginterferon, lack of HBsAg decline could be used as a potential indicator of nonresponse and thus be used to consider switching a patient to an alternative treatment approach.
We found a significant association between on-treatment decline in serum HBsAg of >1 or >2 log10 IU/mL and HBsAg clearance 3 years after treatment. On-treatment decline of HBsAg <1.87 log10 IU/mL or an absolute HBsAg level of >380 IU/mL were negative predictors of sustained HBsAg clearance 3 years after treatment. Similarly on-treatment decline of HBsAg <0.46 log10 IU/mL or an absolute HBsAg level of >19 IU/mL could be used as a negative predictor for HBV DNA suppression to <400 copies/mL 3 years after treatment, with negative predictive values of 95% and 92%, respectively. Such a negative prediction could be taken into account when deciding whether a patient with a large decline in HBsAg levels after 12 months of treatment might stop treatment or whether a patient without significant HBsAg decline might benefit from switching to an alternative treatment strategy. Conversely, prolongation of treatment might be indicated in patients with a slow decline in HBsAg levels. In fact, extending interferon treatment duration was shown to benefit some patients with HBeAg-negative CHB, and a recent pilot study has investigated extending treatment duration with peginterferon alfa-2a. This small study of 13 patients suggests that prolonging treatment with peginterferon alfa-2a to 60 weeks may be a viable way to increase response rates; quantification of HBsAg demonstrated a >90% decrease in HBsAg levels in five of 13 patients after 60 weeks of treatment. The authors concluded that further clinical studies in which peginterferon alfa-2a is given for even longer than 60 weeks are warranted. Although the optimal duration of interferon-based treatment in patients with HBeAg-negative CHB remains unresolved, it is generally regarded as being longer than that for the HBeAg-positive form of the disease. An on-treatment marker predicting sustained off-treatment response may be used to determine the optimal treatment duration of peginterferon in patients with HBeAg-negative CHB and, therefore, enable the individualization of therapy and optimization of response.
In conclusion, HBsAg clearance represents the best possible and closest to cure outcome of antiviral therapy in patients with CHB, but is realistic almost exclusively among patients receiving interferon-based regimens, which are recommended as a first-line therapeutic approach. Quantitative on-therapy measurement of serum HBsAg may be useful to identify patients who are most likely to benefit from interferon-based therapy and may help tailor and optimize treatment duration. Further studies to determine the optimal time points to measure HBsAg level are therefore warranted.
Pretreatment HBsAg Level and Correlation with Other Baseline Factors.
Patient demographic and laboratory characteristics were comparable across the three treatment arms (Table 1). The mean HBsAg level was 3.4 log10 IU/mL (2,500 IU/mL; range, 0.78-4.89 IU/mL); 21.5% of patients had an HBsAg level <3 log10 IU/mL (1,000 IU/mL), 63.2% had levels 3 and <4 log10 IU/mL (1,000-10,000 IU/mL), and 15.3% had levels > /=4 log10 IU/mL (10,000 IU/mL).
Pretreatment HBsAg showed a positive correlation with baseline HBV DNA level (r = 0.38), but there was no apparent correlation with baseline ALT level (r = 0.09) or age (r = -0.14). Median HBsAg levels were similar in males (3.39 log10 IU/mL, n = 326) and females (3.42 log10 IU/mL, n = 60). Variation in pretreatment HBsAg levels was observed between genotypes (Fig. 1), with the highest median levels in patients infected with HBV genotypes A (median, 4.11 log10 IU/mL) and D (median, 3.85 log10 IU/mL).
HBsAg and HBV DNA Decline According to Treatment Regimen.
The magnitude of HBsAg decline from pretreatment to week 48 was approximately 30 times higher with peginterferon alfa-2a (alone or combination with lamivudine) versus lamivudine alone (mean declines of -0.71 and -0.67 log10 IU/mL, respectively, versus -0.02 log10 IU/mL in patients treated with lamivudine monotherapy; P <0.01) (Fig. 2A). The pattern of decline in HBsAg and HBV DNA levels from pretreatment to weeks 48 and 72 according to treatment group is provided in Fig. 2B. HBV DNA level declined considerably in all three treatment arms, with the greatest decline at the end of treatment in the peginterferon alfa-2a plus lamivudine arm.
Baseline Factors Associated with Decline in HBsAg to End of Treatment (Week 48) and 6 Months Posttreatment (Week 72) in Patients Treated with Peginterferon alfa-2a ± Lamivudine.
We analyzed whether baseline factors might influence the degree of decline of HBsAg observed in patients treated with a peginterferon alfa-2a-containing regimen. Patients aged <25 years showed a greater on-treatment decline in HBsAg level (>-0.8 log10 IU/mL mean decline) than patients aged >55 years (<-0.4 log10 IU/mL mean decline), whereas there was no apparent pattern in on-treatment decline in HBsAg level according to baseline ALT level.
Analysis of on-treatment HBsAg decline according to genotype showed that the mean decline was most pronounced in patients infected with genotypes A and B (approximately -0.8 log10 IU/mL; 95% confidence intervals, -1.58 to -0.02 and -1.03 to -0.50, respectively) and least pronounced in patients with genotype D (-0.3 log10IU/mL; 95% confidence interval, -0.58 to -0.06). The distribution of HBsAg levels over time in patients infected with HBV genotypes A to D is shown in Fig. 3. Median levels of HBsAg decreased from pretreatment to end of treatment in all genotypes, with less rebound evident between end of treatment and 6 months posttreatment in patients infected with HBV genotypes A and B.
The only baseline factors identified by multivariate analysis that were significantly associated with decline in HBsAg to end of treatment were baseline HBV DNA level (P = 0.003) and age (P = 0.04). Baseline HBV DNA level was significantly associated with reduction in HBsAg level from baseline to 6 months posttreatment (P = 0.005). No significant association was found between decline in HBsAg from end of treatment or 6 months posttreatment and treatment regimen (peginterferon alfa-2a versus peginterferon alfa-2a plus lamivudine), sex, HBV genotype, body mass index, baseline ALT, and baseline HBsAg level.
Association Between Serum HBsAg Level at End of Treatment and Response to Therapy 6 Months Posttreatment.
Overall, the mean decline in HBsAg from pretreatment levels to week 48 was greater for those who achieved a virological response than for those who did not (Fig. 4). The mean HBsAg decline was -1.077 log10 for the HBV DNA < /=400 copies/mL endpoint responders versus -0.263 for nonresponders (P < 0.001). This association held true only for the subgroup of patients treated with peginterferon alfa-2a-based therapy. Among patients treated with lamivudine who achieved an HBV DNA level < /=400 copies/mL at week 72, the mean HBsAg decline was negligible (-0.003 log10 IU/mL) and lower than that seen in patients who did not achieve suppression of HBV DNA to < /=400 copies/mL (-0.022 log10 IU/mL).
Significantly more patients treated with peginterferon alfa-2a ± lamivudine achieved a substantial on-treatment decline in HBsAg compared with those treated with lamivudine only (Fig. 5A); HBsAg levels < /=100 IU/mL at the end of treatment were observed in 21% and 17% of patients treated with peginterferon alfa-2a and peginterferon alfa-2a plus lamivudine compared with only 1% of those treated with lamivudine alone (P < 0.001 for peginterferon alfa-2a ± lamivudine versus lamivudine).
We performed a detailed analysis of end-of-treatment HBsAg levels and response 6 months posttreatment according to treatment group. There was a strong association between level of HBsAg at end of treatment and suppression of HBV DNA to < /=400 copies/mL, in particular for patients treated with a peginterferon alfa-2a-containing regimen. The association was less apparent for ALT response (Fig. 5B). A very high proportion of peginterferon alfa-2a-treated patients with end-of-treatment HBsAg levels < /=10 IU/mL had HBV DNA suppressed to < /=400 copies/mL 6 months posttreatment (88% and 73% for peginterferon alfa-2a and peginterferon alfa-2a plus lamivudine, respectively).
Association Between Serum HBsAg Level at End of Treatment and Response to Therapy 3 Years After Treatment with Peginterferon alfa-2a.
In total, 198 peginterferon alfa-2a-treated patients with available data on quantitative HBsAg entered the long-term observational follow-up study, and we analyzed the relationship between end-of-treatment HBsAg level at subsequent HBsAg clearance 3 years after treatment. Of the 198 patients, 16 (8%) had cleared HBsAg by the 3-year posttreatment analysis. Of the 23 patients with an HBsAg level < /=10 IU/mL, 12 (52%) had cleared HBsAg 3 years after treatment compared with only four (2.3%) of the 171 patients with an HBsAg level >10 IU/mL at the end of treatment (P < 0.0001; relative risk, 22.8; 95% confidence interval, 8-649) (Table 2). Similarly, an on-treatment decline of HBsAg of >1.1 log10 IU/mL was significantly associated with HBsAg clearance (P < 0.0001; relative risk, 14.6; 95% confidence interval, 5.5-38.5).
Multivariate logistic regression analysis using selected baseline and on-treatment factors identified a decline in HBsAg levels from pretreatment to end of treatment as the only factor significantly associated with HBsAg clearance 3 years after treatment. Baseline parameters (age, HBsAg level, HBV DNA level, ALT level) or treatment with peginterferon alfa-2a versus peginterferon alfa-2a plus lamivudine were not associated with HBsAg clearance 3 years after treatment.
Given the association between HBsAg level at week 48 and HBsAg clearance 3 years after treatment, we aimed to identify cutoff levels of HBsAg at week 48 that could be used to provide the best predictive value of response. CART analyses were performed for 3-year posttreatment response including HBsAg clearance or HBV DNA suppression to 400 copies/mL. For response defined as HBsAg clearance 3 years after treatment, the best predictive values were an absolute HBsAg level of < /=380 IU/mL at week 48 or a change in HBsAg level from pretreatment to week 48 of >1.87 log10 IU/mL (Table 3). Of the 64 patients with an end-of-treatment HBsAg level < /=380 IU/mL, 16 (25%) had HBsAg clearance by 3 years after treatment compared with none of the 134 patients who had an end-of-treatment HBsAg level >380 IU/mL. Of the 27 patients with a change in HBsAg level from pretreatment to week 48 of >1.87 log10 IU/mL, 12 patients (44%) had sustained HBsAg clearance 3 years after treatment. Of the 171 patients with a decline in HBsAg level from pretreatment to week 48 of <1..87 log10 IU/mL, only four (2%) had HBsAg clearance 3 years after treatment.
The cutoff level associated with reduction in HBV DNA level to < /=400 copies/mL 3 years after treatment was HBsAg 19 IU/mL at week 48 and an on-treatment reduction in HBsAg of >0.46 log10 IU/mL. Of 27 patients with HBsAg 19 IU/mL at week 48, 17 (63%) had HBV DNA < /=400 copies/mL 3 years after treatment, whereas only 14 of 171 patients (8%) with HBsAg >19 IU/mL at week 48 reached this endpoint. Eighty-two patients had a reduction in HBsAg of >0.46 log10 from baseline to week 48; of these, 25 (30%) had HBV DNA < /=400 copies/mL 3 years after treatment.
Patients and Methods
This was a retrospective analysis of HBsAg levels in sera from a total of 386 of 537 patients with HBeAg-negative CHB who participated in a randomized, placebo-controlled study conducted at 54 sites in 13 countries. The selection criterion for inclusion in the present study was availability of stored sera for quantification of HBsAg level pretreatment, at end of treatment (week 48), and 6 months after end of treatment (week 72).. Sera for the remaining patients were not available. The study was conducted according to the guidelines of the Declaration of Helsinki and with the principles of Good Clinical Practice and was approved by local ethics committees. All patients gave written informed consent. All patients were positive for HBsAg and negative for anti-HBs antibody pretreatment.. Treatment comprised 180 g peginterferon alfa-2a once weekly plus oral placebo once daily (n = 127), 180 ug peginterferon alfa-2a once weekly plus 100 mg lamivudine once daily (n = 137), or 100 mg lamivudine once daily (n = 122) for 48 weeks. Results 6 months posttreatment have been published previously. For the purpose of this analysis, response 6 months posttreatment was defined as HBV DNA suppression to < /=400 copies/mL or biochemical (ALT <30 U/L). Response parameters assessed 3 years after the end of therapy were HBV DNA <400 copies/mL and HBsAg clearance.
Pretreatment HBsAg levels for all 386 patients were analyzed according to the baseline factors HBV DNA level, ALT, HBV genotype, and host factors (age, sex). Decline in HBsAg and HBV DNA over time (pretreatment, end of treatment, and 6 months posttreatment) was determined for all 386 patients according to treatment group. Stratification of decline in HBsAg levels over time according to HBV genotype was performed for the 264 patients treated with peginterferon alfa-2a ± lamivudine.
HBsAg level at end of treatment was analyzed with respect to HBV DNA and ALT response 6 months posttreatment for all 386 patients stratified according to treatment group. The relationship between end-of-treatment HBsAg level and HBsAg clearance 3 years posttreatment was investigated in a total of 198 patients treated with peginterferon alfa-2a ± lamivudine who entered a long-term observational rollover study. Patients with missing data for HBsAg clearance at 3 years posttreatment were considered to be nonresponders.
End-of-treatment HBsAg levels associated with a 3-year posttreatment response were identified using classification and regression tree (CART) analyses. Multiple logistic regression analyses were performed to identify potential baseline or on-treatment parameters associated with sustained HBsAg clearance.
HBsAg was quantified using the Architect HBsAg assay (Abbott Laboratories, Abbott Park, IL; dynamic range, 0.05-250.0 IU/mL) after 1:100 dilution. Samples with HBsAg levels >250.0 IU/mL at 1:100 dilution were retested at a final dilution of 1:1,000. Samples with HBsAg levels <0.05 IU/mL at 1:100 dilution were retested undiluted. HBV DNA was measured using the Amplicor HBV test (Roche Diagnostics, Basel, Switzerland; range, 400-200,000 copies/mL); samples with viremia >200,000 copies/mL were retested after 1:100 dilution according to the manufacturer's recommendations. Serum ALT was measured at the time of sampling. Normal ALT was considered to be 1x the upper limit of normal (30 IU/mL).
Multiple logistic regression analysis was used to investigate baseline factors other than treatment with peginterferon alfa-2a that might be associated with HBsAg reduction (as a continuous variable) from baseline to end of treatment (week 48) and from baseline to 6 months posttreatment (week 72) for the 198 patients treated with peginterferon alfa-2a ± lamivudine. Factors included in the analyses were: treatment with peginterferon alfa-2a versus peginterferon alfa-2a plus lamivudine, sex, genotype, and - as continuous variables - age, body mass index, baseline ALT, baseline HBV DNA, and baseline HBsAg level.
For the subset of 65 patients (treated with peginterferon alfa-2a ± lamivudine) with HBsAg quantification and available 3-year follow-up data, multiple logistic regression analysis was used to investigate factors that might be associated with HBsAg clearance 3 years after end of treatment.. Variables included in the analyses were baseline factors (treatment with peginterferon alfa-2a versus peginterferon alfa-2a plus lamivudine, age, baseline ALT, baseline HBV DNA) and week 48 variables (change in HBsAg level from baseline, absolute HBsAg level, HBV DNA level).
CART analyses were performed using CART software to investigate negative and positive predictive values as well as the sensitivity and specificity of end-of-treatment HBsAg levels for predicting sustained response to peginterferon alfa-2a 3 years after end of treatment. Test data were generated using 10-fold cross-validation; the tree models were then fitted to all the data excluding each subsample in turn. The misclassification cost was calculated for each of the 10 trees by validating each in their respective subsamples.