A baseline model including quantitative anti-HBc to predict response of peginterferon in HBeAg-positive chronic hepatitis B patients

Abstract

Background

Few models to predict antiviral response of peginterferon were used in hepatitis B e antigen (HBeAg)-positive chronic hepatitis B patients and the prediction efficacy was unsatisfied. Quantitative antibody to hepatitis B core antigen (anti-HBc) is a new predictor of treatment response. We aimed to develop a new model to identify HBeAg-positive Chinese patients who were more likely to respond to peginterferon.

Methods

Data from 140 peginterferon recipients with HBeAg-positive were applied with generalized additive models and multiple logistic regression analysis to develop a baseline scoring system to predict serological response (SR: HBeAg loss and HBeAg seroconversion 24 weeks post-treatment) and combined response (CR: SR plus serum HBV DNA levels <2000 IU/mL 24 weeks post-treatment).

Results

Anti-HBc levels, alanine aminotransferase ratio, and HBeAg were retained in the final model. The new model scored from 0 to 3. Among patients with scores of 0, 1, or ≥2, SR was achieved in 6.45% (2/31), 13.21% (7/51), and 55.36% (31/56), respectively, and CR in 3.23% (1/31), 9.43% (5/53), and 25.00% (14/56), respectively. Our model has a higher AUROC for SR comparing to Chan’s (Z = 2.77 > 1.96, p < 0.05) and Lampertico’s (Z = 2.06 > 1.96, p < 0.05) model. The negative predictive value for SR and CR were both 100% in patients with score 0 and hepatitis B surface antigen ≥20,000 IU/mL at week 12.

Conclusions

Patients with higher scores at baseline were more likely to respond to peginterferon. This new model may predict the treatment response.

Keywords

antibody to hepatitis B core antigen peginterferon serological response combined response

Introduction

Hepatitis B virus (HBV) remains an important global public health problem with a devastating cause of morbidity and mortality.¹ Antiviral therapy can effectively reduce the incidence of HBV-related complications and improve the quality of life.^2–4 Peginterferon (Peg-IFN) has both antiviral and immunomodulatory activity, which has already been routinely used in clinic. Compared with nucleos(t)ide analogs (NAs), Peg-IFN has the greatest advantage of inducing immune control and maintaining long-term therapy efficacy. However, its response rate is lower and the side effects are significantly higher compared to NAs.^5–7 Therefore, how to screen out the patients with good response to Peg-IFN is a big puzzle for clinical doctors.

Indicators to predict antiviral response of Peg-IFN have been used in some studies, which are experimental use and difficult to apply in clinical practice.^8,9 Other prediction methods mainly focused on establishing models including the levels of HBV surface antigen (HBsAg) or hepatitis B e antigen (HBeAg), alanine aminotransferase (ALT), sex, and age. Those models are easy to use, but the actual prediction efficacy is not satisfied.^10–13 Antibody to hepatitis B core antigen (anti-HBc) is a novel use of an old test indicator of HBV infection which can be used to estimate the risk of HBV reactivation in patients undergoing immunosuppression.¹⁴ However, its clinical use is limited due to the lack of quantitative detection methods. Based on the application of double antigen enzyme-linked immunosorbent assay and standard quantitative technology, quantitative assays of measuring anti-HBc levels become useful.¹⁵ Quantitative anti-HBc could be used to distinguish the different disease stages of chronic hepatitis B (CHB) infection.^16,17 More importantly, it is a novel biomarker for predicting treatment response in HBeAg-positive patients receiving NAs¹⁸ and Peg-IFN.¹⁹ In patients with HBsAg clearance after Peg-IFN treatment, the levels of anti-HBc at the end of treatment can also predict the recurrence for those patients.²⁰

Sonneveld et al. found that response rates were low in CHB patients with HBsAg level >20,000 IU/mL after the monotherapy of Peg-IFN. The HBsAg levels at week 12 or 24 have been a stopping rule for Peg-IFN therapy.²¹ We aimed to establish a prediction model including anti-HBc levels, which can better screen out patients suitable for Peg-IFN therapy. Meanwhile, we also compared the two existing models^10,11 with the new one we established in order to select a better model for clinical service. Then we used the stopping rule of Peg-IFN²¹ treatment to confirm our new prediction model.

Methods

Patients

140 of the 205 patients confirming the inclusion criteria were enrolled in the study (Supplementary figures). All the patients received Peg-IFN therapy alone. And the total cycle was 72 weeks, including 48 weeks of Peg-IFN treatment period and 24 weeks of follow-up period after treatment. Serum samples were collected every 12 weeks.

The inclusion criteria were as follows: (1) aged between 18 to 65 years old; (2) HBsAg-positive for more than 6 months; (3) HBeAg-positive; (4) had two episodes of elevated serum ALT levels (>1.5 times the upper limit of the normal range) within 6 months before randomization, and the normal range of serum ALT levels was defined as ≤ 40 IU/L; and (5) HBV DNA level > 10⁵ IU/mL.

The exclusion criteria were as follows: (1) received antiviral or immunosuppressive therapy within 6 months; (2) co-infection with hepatitis A, hepatitis C, hepatitis D, hepatitis E, or human immunodeficiency virus; (3) other acquired or inherited causes of liver disease; (4) neutrophil count <1.0 × 10⁹/L or platelet count <50 × 10⁹/L; (5) decompensated liver disease; (6) immunologically mediated disease; (7) pregnancy; and (8) alcoholism.

The definition of response: Serological response (SR) is defined as HBeAg loss and HBeAg seroconversion 24 weeks post-treatment. Combined response (CR) is defined as SR plus serum HBV DNA levels <2000 IU/mL 24 weeks post-treatment. NSR means no SR, and NCR means no CR.

The details of two published prediction models of Peg-IFN on CHB patients: (1) Chan et al.’s model¹⁰ (Supplemetary Table) will be called Chan’s mode and (2) Lampertico et al.’s model¹¹ (Supplementary Table) will be called Lampertico’s model.

Laboratory measurements

Serum samples collected at weeks 0, 12, 24, 48, and 72 were analyzed. The serum anti-HBc levels were measured by a newly developed double antigen enzyme-linked immunosorbent assay (Wantai, Beijing, China) that was calibrated using immunoassay.¹⁵ The HBsAg, HBeAg, and antibodies to hepatitis B e antigen (anti-HBe) were detected with the Architect assay (Abbott Laboratories, IL, USA). Serum HBV DNA was measured using COBAS AmpliPrep/COBAS TaqMan (Roche Diagnostics, Basel, Switzerland). The HBV genotype was assessed by sequencing.

Statistical analysis

SPSS software version 20.0 (SPSS Inc., Chicago, IL, USA) was used to perform statistical analysis. Continuous variables were presented as mean ± standard deviation (SD) for normally distributed data or median (interquartile range) for non-normally distributed data. Categorical variables were presented as counts and percentages. Areas under the receiver operating characteristic curve (AUROC) were applied to discriminate the model. Comparison between two different prediction models was carried out using a formula as follows: Z = A₁-A₂/√SE₁²+SE₂²-2rSE₁SE₂, Z > 1.96, p < 0.05; Z > 2.58, p < 0.01.²²

Model establishment

First, generalized additive models were used to detect the relationship between continuous variables and the two response indexes (SR and CR) in order to find the appropriate cut-off values for use in the logistic regression analysis. Second, associations between the baseline continuous indexes and treatment response were explored by univariate logistic regression analyses, aiming to find the optimal cut-off values. Continuous variables were classified into two different groups according to the above optimal cut-off values. Categorical variables were compared using χ2 test or Fisher’s exact test as appropriate. Third, multiple logistic regression analyses were performed using a backward elimination process. Baseline factors associated with SR or CR with a p < 0.50 were included in the multiple logistic regression analyses. p < 0.1 was required in the backward selection procedure for a baseline factor to remain in the final model. The stability of the selected model was evaluated by bootstrap resampling.²³ A baseline variable was considered to be a reliable predictor if it was selected in at least ≥50% of 500 bootstrapped samples with replacement. AUROC and the Hosmer–Lemeshow tests were applied to assess the goodness of fit of the final model.

The article involved two different response indexes, and the SR model–related baseline variables were considered to be the primary model. The scoring system was derived by assigning points according to the method of Sullivan et al.²⁴ The reference category for each predictive factor was assigned to a score of 0. Points were then assigned according to the size of the regression coefficient in the final model.

Results

Baseline characteristics

140 (68.29%) patients were included in the study and 65 patients were excluded (Supplemetary Figure). The baseline characteristics of patients are shown in Table 1. 40 (28.57%) patients achieved SR and 20 (14.29%) patients experienced CR in the 140 patients recruited.

Table 1.

Baseline characteristic of patients stratified by the follow-up endpoint.

Characteristic	Total (n = 140)	SR (n = 40)	NSR (n = 100)	p value	CR (n = 20)	NCR (n = 120)	p value
Age, years, mean (±SD)	26.75 (6.99)	25.33 (6.72)	27.32 (7.04)	0.054	23.80 (4.37)	27.24 (7.23)	0.050
Male sex, n (%)	103 (73.57)	27 (67.50)	76 (76.00)	0.303	14 (70.00)	89 (74.17)	0.696
HBV genotype, n (%)
B	51 (36.43)	16 (40.00)	35 (35.00)	0.652	7 (35.00)	44 (36.67)	0.886
B/C
C	89 (63.57)	24 (60.00)	65 (65.00)		13 (65.00)	76 (63.33)
ALT ratio × ULN, mean (±SD)	4.66 (3.56)	6.13 (5.02)	4.07 (2.61)	<0.001	5.50 (2.12)	4.52 (3.76)	0.006
HBsAg, log10 IU/mL, mean (±SD)	4.07 (0.60)	3.95 (0.50)	4.12 (0.62)	0.027	3.85 (0.59)	4.11 (0.59)	0.027
HBeAg, log10 S/CO, mean (±SD)	2.76 (0.55)	2.45 (0.75)	2.88 (0.38)	<0.001	2.60 (0.68)	2.78 (0.52)	0.099
HBV DNA, log10 IU/mL, mean (±SD)	7.64 (0.85)	7.50 (0.64)	7.70 (0.92)	0.116	7.30 (0.57)	7.70 (0.88)	0.017
Anti-HBc, log10 IU/mL, mean (±SD)	4.26 (0.67)	4.58 (0.50)	4.13 (0.69)	<0.001	4.61 (0.51)	4.18 (0.69)	0.016

Serological response (SR) is defined as HBeAg loss and HBeAg seroconversion 24 weeks post-treatment; NSR means no SR.

Combined response (CR) is defined as SR plus serum HBV DNA levels <2000 IU/mL 24 weeks post-treatment; NCR means no CR.

ALT ratio = alanine aminotransferase level divided by the upper limit of normal (ULN) for the local laboratory.

HBV, hepatitis B virus; ALT, alanine aminotransferase; HBsAg, hepatitis B surface antigen; HBeAg, hepatitis B e antigen; Anti-HBc, antibodies to hepatitis B core antigen; SD, standard deviation; ULN, upper limit of normal.

Compared to NSR patients, SR patients had significantly higher anti-HBc levels (4.58 ± 0.50 log₁₀ IU/mL vs.4.13 ± 0.69 log₁₀ IU/mL, p < 0.001) and ALT ratio (6.13 ± 5.02 × ULN vs. 4.07 ± 2.61 × ULN, p < 0.001) but significantly lower HBsAg (3.95 ± 0.50 log₁₀ IU/mL vs. 4.12 ± 0.62 log₁₀ IU/mL, p = 0.027) and HBeAg S/CO levels (2.45 ± 0.75 log₁₀ S/CO vs. 2.88 ± 0.38 log₁₀ S/CO, p < 001). Compared to no NCR patients, CR patients had significantly higher anti-HBc levels (4.61 ± 0.51 log₁₀ IU/mL vs. 4.18 ± 0.69 log₁₀ IU/mL, p = 0.016) and ALT ratios (5.50 ± 2.12 × ULN vs. 4.52 ± 3.76 × ULN, p = 0.006) but significantly lower HBV DNA levels (7.30 ± 0.57 log₁₀ IU/mL vs. 7.70 ± 0.88 log₁₀ IU/mL, p = 0.017) and HBsAg levels (3.85 ± 0.59 log₁₀ IU/mL vs. 4.11 ± 0.59 log₁₀ IU/mL, p = 0.027).

Among 51 genotype B patients, 16 (31.37%) patients achieved SR and 7 (13.73%) patients achieved CR. Among 89 genotype C patients, 24 (26.97%) patients achieved SR and 13 (14.61%) patients achieved CR. There is no significant difference between SR and NSR patients in HBV genotype (p = 0.652), so as well between CR and NCR patients (p = 0.886).

Generalized additive model and univariate logistic regression analysis to identify baseline predictive characteristics

The results of generalized additive models showed that the response rates of SR and CR were gradually increased with the decrease of HBsAg levels, HBeAg S/CO, HBV DNA levels, and age. SR and CR rates showed an obvious increase with the increasing anti-HBc levels and ALT ratio (Supplementary figure). Based on the generalized additive models and univariate logistic regression analyses, the cut-off values were as follows: age <30 and ≥30 years, anti-HBc levels <30,000 and ≥30,000 IU/mL, HBsAg levels <7500 and ≥7500 IU/mL, HBeAg levels <800 and ≥800 S/CO, HBV DNA levels <7.70 and ≥7.70 log₁₀ IU/mL, and ALT <4 and ≥4 ratio × ULN. The univariate analysis results are shown in Table 2.

Table 2.

Univariate logistic regression analysis of factors associated with SR and CR 24 weeks post‐treatment.

Baseline characteristic	Regression coefficient	Odds ratio (95% CI)	p value^a	p value (effect)^b
SR 24 weeks post-treatment
Age <30 vs ≥30 years	0.203	1.225 (0.531–2.827)	0.634	0.634
Sex, female versus male	0.422	1.525 (0.682–3.411)	0.305	0.305
Genotype, B versus C	0.214	0.808 (0.380–1.717)	0.579	0.579
Anti-HBc < 30000 vs ≥30000 IU/mL	1.406	4.080 (1.885–8.832)	<0.0001	<0.0001
HBsAg <7500 vs ≥ 7500 IU/mL	0.744	2.104 (0.983–4.502)	0.055	0.055
HBeAg <800 vs ≥ 800 S/CO	1.455	4.286 (1.975–9.301)	<0.0001	<0.0001
HBV DNA <7.70 vs ≥ 7.70 log₁₀ IU/mL	0.566	1.761 (0.836–3.709)	0.137	0.137
ALT < 4 vs ≥ 4 ratio × ULN	1.466	4.333 (1.818–10.327)	0.001	0.001
CR 24 weeks post-treatment
Age <30 vs ≥ 30 years	0.783	2.189 (0.841–5.697)	0.109	0.109
Sex, female versus male	0.207	1.230 (0.435–3.481)	0.696	0.696
Genotype, B versus C	0.072	1.075 (0.399–2.297)	0.886	0.886
Anti-HBc < 30000 vs ≥ 30000 IU/mL	1.389	4.012 (0.885–18.186)	0.072	0.072
HBsAg <7500 vs ≥ 7500 IU/mL	0.847	2.333 (0.894–6.091)	0.084	0.084
HBeAg <800 vs ≥ 800 S/CO	0.346	1.413 (0.543–3.676)	0.478	0.478
HBV DNA <7.70 vs ≥ 7.70 log10 IU/mL	1.587	4.889 (1.543–15.489)	0.007	0.007
ALT < 4 vs ≥ 4 ratio × ULN	1.635	5.127 (1.427–18.414)	0.012	0.012

Serological response (SR) is defined as HBeAg loss and HBeAg seroconversion 24 weeks post-treatment.

Combined response (CR) is defined as SR plus serum HBV DNA levels <2000 IU/mL 24 weeks post-treatment.

HBV, hepatitis B virus; HBsAg, hepatitis B surface antigen; HBeAg, hepatitis B e antigen; Anti-HBc, antibodies to hepatitis B core antigen.

ULN, upper limit of normal; ALT ratio = alanine aminotransferase level divided by the upper limit of normal (ULN) for the local laboratory.

^a Wald chi‐square test for each regression coefficient.

^b Wald chi‐square test for each variable.

Multiple logistic regression analysis of selected baseline characteristics with SR and CR

Six baseline variables (sex, anti-HBc, HBsAg, HBeAg, HBV DNA, and ALT) associated with SR and CR in the univariate analysis with p < 0.50 were included in the multiple logistic regression analysis. Three factors (ALT ratio, HBeAg S/CO, and anti-HBc levels) were retained in the final model (Table 3). The SR model–related baseline variables were considered to be the primary model. The AUROC was 0.782 and the Hosmer–Lemeshow test (p = 0.249 > 0.15) also indicated sufficient discrimination and goodness of fit.

Table 3.

Multiple logistic regression analysis of factors associated with SR and CR 24 weeks post‐treatment.

Baseline characteristic	Regression coefficient	Odds ratio (95% CI)	p value^a	p value (effect)^b	Selected in bootstrap models
SR 24 weeks post-treatment
Anti-HBc < 30000 vs ≥ 30000 IU/mL	1.070	2.914 (1.264–6.721)	0.012	0.012	81%
HBeAg <800 vs ≥ 800 S/CO	1.201	3.324 (1.444–7.653)	0.005	0.005	89%
ALT < 4 vs ≥ 4 ratio × ULN	1.251	3.492 (1.391–8.766)	0.008	0.008	84%
		Hosmer–Lemeshow p value = 0.249
CR 24 weeks post-treatment
Anti-HBc < 30000 vs ≥ 30000 IU/mL	1.304	2.173 (0.627–3.737)	0.009	0.009	82%
HBV DNA <7.70 vs ≥ 7.70 log₁₀ IU/mL	1.307	3.694 (1.126–12.121)	0.031	0.03	68%
ALT < 4 vs ≥ 4 ratio × ULN	1.413	4.110 (1.104–15.297)	0.035	0.035	62%
		Hosmer–Lemeshow p value = 0.813

Serological response (SR) is defined as HBeAg loss and HBeAg seroconversion 24 weeks post-treatment.

Combined response (CR) is defined as SR plus serum HBV DNA levels <2000 IU/mL 24 weeks post-treatment.

HBV, hepatitis B virus; Anti-HBc, antibodies to hepatitis B core antigen.

ALT ratio = alanine aminotransferase level divided by the upper limit of normal (ULN) for the local laboratory.

^a Wald chi‐square test for each regression coefficient.

^b Wald chi‐square test for each variable.

The scoring system was derived from three baseline indicators (anti-HBc levels, ALT ratio, and HBeAg S/CO) retained in the final model. The points of indicators were based on the size of regression coefficient²⁴ (Table 4). The total scores of patients ranged from 0 to 3, and higher scores always predicated a higher response possibility (Figure 1). In the 140 patients included in our study, the distribution of baseline prediction score was 0 (n = 31, 22.14%), 1 (n = 53, 37.86%), 2 (n = 34, 24.29%), and 3 (n = 22, 15.71%), respectively.

Table 4.

Scoring system for predictive baseline characteristics.

Baseline characteristic		Score
Anti-HBc, IU/mL	<30000	0
Anti-HBc, IU/mL	≥30000	1
HBeAg, S/CO	≥800	0
HBeAg, S/CO	<800	1
ALT ratio × ULN	<4	0
ALT ratio × ULN	≥4	1

Anti-HBc, antibodies to hepatitis B core antigen; HBeAg, hepatitis B e antigen.

ALT ratio = alanine aminotransferase level divided by the upper limit of normal (ULN) for the local laboratory.

Figure 1.

SR and CR response rates 24 weeks post-treatment by our baseline new predictive score. SR = HBeAg loss and HBeAg seroconversion 24 weeks post-treatment. CR = SR plus HBV DNA levels <2000 IU/mL 24 weeks post-treatment.

Response rates according to baseline prediction scores

The overall SR and CR rates were 28.57% (40/140) and 14.29% (20/140), respectively, and both the response rates gradually increased with increasing baseline prediction scores. According to the response rates of different scores, we classified the patients into three groups: 0 point was defined as low response, 1 point as moderate response, and ≥2 points as high response. Among 31 patients in the low response group, only two patients (6.45%) achieved SR and one patient (3.23%) achieved CR. The negative predictive value (NPV) in low response group was 93.55% with SR and 96.77% with CR. Among 53 patients in moderate response group, seven patients (13.21%) experienced SR and five patients (9.43%) experienced CR. Among 56 patients in high response group, 31 patients (55.36%) achieved SR and 14 patients (25.00%) achieved CR. The rates of SR and CR in high response group were significantly higher than the overall response rates (Figure 1).

Comparison of AUROC between the new prediction model and two published prediction models

For all the 140 enrolled patients, Chan’s model¹⁰ had 0.627 (p = 0.019) AUROC curve for SR and Lampertico’s¹¹ model had 0.669 (p = 0.002) AUROC curve for SR. However, the new prediction model in our study had 0.782 (p<0.001) AUROC curve for SR (Supplementary figure). Compared with the AUROC of Chan’s model,¹⁰ our new prediction model had a higher AUROC (Z = 2.77 > 1.96, p < 0.05). Similarly, our new prediction model also had a higher AUROC (Z = 2.06 > 1.96, p < 0.05) comparing with the AUROC of Lampertico’s model¹¹ (Figure 2).

Figure 2.

The comparison of AUROC between our new prediction model and the two published prediction models. AUROC = Areas under the receiver operating characteristic curve; NS = not significant; ∗ p<0.05.

The new baseline prediction score combined with week 12 or 24 HBsAg levels to predict SR and CR

123 patients (87.86%) had HBsAg levels <20,000 IU/mL at week 12, while the rest 17 patients (12.14%) had HBsAg levels ≥20,000 IU/mL. Among the 123 patients with HBsAg levels <20,000 IU/mL, 38 (30.89%) and 20 (16.26%) individuals achieved SR and CR, respectively. Among 53 patients with HBsAg levels <20,000 IU/mL and prediction scores of 2–3, 30 individuals (56.60%) achieved SR and 15 (28.30%) achieved CR. Among the 17 patients with HBsAg levels ≥20,000 IU/mL, two individuals (11.76%) achieved SR but none achieved CR. Among 10 patients with HBsAg levels ≥20,000 IU/mL and prediction scores of 0, no individual achieved SR or CR. The NPV for SR and CR were both 100% in patients with HBsAg levels ≥20,000 IU/mL at week 12 and prediction scores of 0.

The association between response rates and the combination of baseline prediction score with HBsAg level at week 24 was similar (data not shown). The NPV for SR and CR were both 100% in patients with HBsAg levels ≥20,000 IU/mL at week 24 and prediction scores of 0.

Discussion

Our study showed that the new prediction model was useful to identify HBeAg-positive CHB patients who were more likely to respond to Peg-IFN therapy. Meanwhile, a significantly higher NPV (>90%) was observed in non-responsive patients. All the 140 patients enrolled were CHB patients of genotype B or C, and the response rates of the two genotypes were similar. Since almost all CHB patients in China are genotype B or C, genotyping tests might not be necessary for Chinese HBeAg-positive CHB patients receiving Peg-IFN therapy.

Chan H et al. found that a pre-treatment scoring system using readily available baseline characteristics (sex, age, ALT ratio, HBsAg levels, and HBV DNA levels) identified HBeAg-positive Asian patients likely to experience sustained HBeAg seroconversion after treatment with Peg-IFN.¹⁰ Lampertico P et al. also found that an easy-to-use baseline scoring system would allow clinicians to identify patients with a low or high chance of achieving a durable post-treatment response to Peg-IFN.¹¹ These two models were also carried out for the 140 patients involved in our study, and the results were used to verify their efficacy in practice. Unfortunately, their AUROC were both <0.700, and the actual potency was unsatisfied. There were two prediction models containing semi-quantitative anti-HBc levels, which revealed that the baseline level of anti-HBc might be a good indicator for predicting SR of interferon.^12,13 A novel diagnostic immunoassay testing procedure for anti-HBc using homogeneous purified full-length hepatitis B core antigen (HBcAg) capsids obtained from Escherichia coli could quantify the levels of anti-HBc in serum.¹⁵ Based on quantitation of anti-HBc, a new prediction model in our study was established by combining baseline anti-HBc levels with other two baseline indicators. The final scores of the selected indexes were as follows: anti-HBc levels <30,000 (0) or ≥30,000 IU/ml (1); HBeAg levels <800 (0) or ≥800 S/CO (1); and ALT <4 (0) or ≥4 ratio × ULN (1).

The new scoring system ranged from 0 to 3, with higher scores predicting a higher likelihood of experiencing SR and CR. According to the response rates of different scores, we classified patients into three groups. Patients with baseline prediction scores 2 to 3 (40% of the total patients) had the highest probability of achieving SR (55.36%) and CR (25.00%). The probability of achieving SR (13.21%) and CR (9.43%) in patients with baseline prediction score 1 (38% of all patients) was similar to that of the general populations. Patients with baseline prediction score 0 (22% of all patients) had the lowest likelihood of achieving SR and CR (both<10%). Compared to the two previously published models, the AUROC of our new model was significant higher. It showed that the new model had more advantages in identifying patients between responsive and non-responsive individuals, which indicated that our model could be more superior for clinical use.

HBsAg levels at weeks 12 and 24 had already been a stopping rule for HBeAg-positive CHB patients receiving Peg-IFN therapy alone.^6,21 The NPV for SR as well as CR was 100% in patients with HBsAg levels ≥20,000 IU/mL and prediction scores of 0 at weeks 12 and 24. So our new model enables us to master the withdrawal indications more accurately at week 12 and adjust the treatment strategy in time to avoid receiving useless treatment or leading to potential adverse events.

The purpose of our article is to provide scientific basis for patient selection of Peg-IFN therapy and adjustment during treatment. Compared with the published models, our prediction model could significantly increase the PPV and NPV of SR and CR. For patients with prediction scores ≥2, they were suitable candidate to receive Peg-IFN therapy. But for patients with prediction score 0, Peg-IFN was not routinely recommended. For patients who had already received Peg-IFN treatment, HBsAg levels at week 12 combined with baseline prediction score 0 could be a useful stopping rule. The most advantage of our article is that all patients were treated with Peg-IFN alone. However, our research still has certain limitations. The sample size was not large enough, and further expansion of sample size in the future would lead to more convincing results.

In conclusion, our new baseline scoring system reveals that CHB patients with higher baseline prediction scores usually have a higher likelihood of experiencing SR and CR. It is helpful for clinicians to identify patients with a low or high possibility of achieving a durable treatment response to Peg-IFN treatment in CHB patients.

Supplemental Material

sj-pdf-1-avt-10.1177_13596535211059895 – Supplemental Material for A baseline model including quantitative anti-HBc to predict response of peginterferon in HBeAg-positive chronic hepatitis B patients

Supplemental Material, sj-pdf-1-avt-10.1177_13596535211059895 for A baseline model including quantitative anti-HBc to predict response of peginterferon in HBeAg-positive chronic hepatitis B patients by Yu-Qing Fang, Xiao-Yan Xu, Feng-Qin Hou and Wei Jia in Antiviral Therapy

Footnotes

Acknowledgments

We would like to thank Liu-Wei Song (National Institute of Diagnostics and Vaccine Development in Infectious Diseases, Xiamen, Fujian, China) for his technical support of measuring anti-HBc.

Author contributions

All authors contributed to the study conception and design. Material preparation, data collection, and analysis were performed by Wei Jia, Xiao-Yan Xu, and Feng-Qin Hou. The first draft of the manuscript was written by Yu-Qing Fang and all authors commented on previous versions of the manuscript. All authors read and approved the final manuscript.

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

Availability of data and material

Data are available within the article or its supplementary materials.

Informed consent

Informed consent was obtained from all individual participants included in the study. All patents provided written informed consent for the scientific use of their clinical data, sample. Patients signed informed consent regarding publishing their data.

Ethics approval

The study has been approved by the local ethics committee of Peking University First Hospital before experiment was started and that has been conducted in accordance with the principles set forth in the Helsinki Declaration and its later amendments.

ORCID iD

Yu-Qing Fang

Supplemental material

Supplemental material for this article is available online.

References

Ott

Stevens

Groeger

, et al. Global epidemiology of hepatitis B virus infection: new estimates of age-specific HBsAg seroprevalence and endemicity. Vaccine 2012; 30: 2212–2219.

European Association for the Study of the Liver, European Organisation for Research and Treatment of Cancer . EASL-EORTC clinical practice guidelines: management of hepatocellular carcinoma. J Hepatology 2012; 56: 908–943.

Zhuang

Zhang

Liu

, et al. Significant impairment of health-related quality of life in mainland Chinese patients with chronic hepatitis B: a cross-sectional survey with pair-matched healthy controls. Health Qual Life Outcomes 2014; 12: 101.

T-H

Chen

C-Y

, et al. Four-year entecavir therapy reduces hepatocellular carcinoma, cirrhotic events and mortality in chronic hepatitis B patients. Liver Int 2016; 36: 1755–1764.

Terrault

Bzowej

Chang

K-M

, et al. AASLD guidelines for treatment of chronic hepatitis B. Hepatology 2016; 63: 261–283.

European Association for the Study of the Liver . EASL 2017 Clinical Practice Guidelines on the management of hepatitis B virus infection. J Hepatology 2017; 67: 370–398.

Terrault

Lok

ASF

McMahon

, et al. Update on prevention, diagnosis, and treatment of chronic hepatitis B: AASLD 2018 hepatitis B guidance. Hepatology 2018; 67: 1560–1599.

Chuaypen

Posuwan

Payungporn

, et al. Serum hepatitis B core-related antigen as a treatment predictor of Pegylated interferon in patients with HBeAg-positive chronic hepatitis B. Liver Int 2016; 36: 827–836.

Deng

Liu

, et al. Impaired expression and function of TLR8 in chronic HBV infection and its association with treatment responses during peg-IFN-α-2a antiviral therapy. Clin Res Hepatol Gastroenterol 2017; 41: 386–398.

10.

Chan

HLY

Messinger

Papatheodoridis

, et al. A baseline tool for predicting response to Peginterferon alfa-2a in HBeAg-positive patients with chronic hepatitis B. Aliment Pharmacol Ther 2018; 48: 547–555.

11.

Lampertico

Messinger

Oladipupo

, et al. An easy-to-use baseline scoring system to predict response to Peginterferon alfa-2a in patients with chronic hepatitis B in resource-limited settings. Antivir Ther 2018; 23: 655–663.

12.

Wang

C.-T.

Zhang

Sun

, et al. Models for predicting hepatitis B e antigen seroconversion in response to interferon-α in chronic hepatitis B patients. World J Gastroenterol 2015; 21: 5668–5676.

13.

Zhu

Wang

Zhang

, et al. Prediction model for sustained hepatitis B e antigen seroconversion to Peginterferon alfa-2a in chronic hepatitis B. J Gastroenterol Hepatol 2016; 31: 1963–1970.

14.

Nishida

Matsubara

Yakushijin

, et al. Prediction and clinical implications of HBV reactivation in lymphoma patients with resolved HBV infection: focus on anti-HBs and anti-HBc antibody titers. Hepatol Int 2019; 13: 407–415.

15.

Yuan

Huang

, et al. Novel double-antigen sandwich immunoassay for human hepatitis B core antibody. Clin Vaccin Immunol 2010; 17: 464–469.

16.

Jia

Song

L-W

Fang

Y-Q

, et al. Antibody to Hepatitis B Core Antigen Levels in the Natural History of Chronic Hepatitis B. Medicine 2014; 93: e322.

17.

Song

L-W

Liu

P-G

Liu

C-J

, et al. Quantitative hepatitis B core antibody levels in the natural history of hepatitis B virus infection. Clin Microbiol Infect 2015; 21: 197–203.

18.

J-H

Song

L-W

, et al. Baseline hepatitis B core antibody predicts treatment response in chronic hepatitis B patients receiving long-term entecavir. J Viral Hepat 2017; 24: 148–154.

19.

Hou

F-Q

Song

L-W

Yuan

, et al. Quantitative hepatitis B core antibody level is a new predictor for treatment response in HBeAg-positive chronic hepatitis B patients receiving Peginterferon. Theranostics 2015; 5: 218–226.

20.

Wang

Lin

, et al. Quantitative of serum hepatitis B core antibody is a potential predictor of recurrence after interferon-induced hepatitis B surface antigen clearance. J Microbiology, Immunology, Infection = Wei Mian Yu gan Ran Za Zhi 2019; S1684-1182: 30145–30148.

21.

Sonneveld

Hansen

Piratvisuth

, et al. Response-guided Peginterferon therapy in hepatitis B e antigen-positive chronic hepatitis B using serum hepatitis B surface antigen levels. Hepatology 2013; 58: 872–880.

22.

Hanley

McNeil

. A method of comparing the areas under receiver operating characteristic curves derived from the same cases. Radiology 1983; 148: 839–843.

23.

Steyerberg

. Clinical Prediction Models. A Practical Approach to Development, Validation, and Updating. New York: Springer-Verlag, 2009, p. 105.

24.

Sullivan

Massaro

D'AgostinoSr

. Presentation of multivariate data for clinical use: the Framingham Study risk score functions. Stat Medicine 2004; 23: 1631–1660.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.49 MB