Performance evaluation of HBsAg by Lumipulse HBsAg-HQ: The agreement with HBsAg by Architect HBsAg-QT and the effectiveness in predicting liver tissue pathological states of chronic hepatitis B patients

Results. In both HBeAg-positive and HBeAg-negative patients, HBsAg-HQ was significantly positively correlated with HBsAg-QT (r = 0.913 and r = 0.959, respectively), the overall disagreement rates between HBsAg-HQ and HBsAg-QT were 2.72% (4/147) and 4.69% (6/128), respectively. In HBeAg-positive patients, the area under the ROC curve (AUC) of HBsAg-HQ and HBsAg-QT for predicting the grade ≥G3 (0.686 and 0.684, respectively) and stage ≥S4 (0.739 and 0.745, respectively) were the greatest compared with other pathological states; the optimal cutoffs of HBsAg-HQ and HBsAg-QT for predicting the grade ≥G3 were <2.244 × 107 mIU/mL and <3.589 × 107 mIU/mL, and those for predicting the stage ≥S4 were 7.328 × 106 mIU/mL and <6.194 × 106 mIU/mL, respectively.


Introduction
The detection of serum hepatitis B surface antigen (HBsAg) is the leading hallmark for screening and diagnosing hepatitis B virus (HBV) infection.In the past 10 years, serum HBsAg quantification assays have been developing rapidly.−8 Furthermore, some evidence has reinforced its value in predicting drug efficacy and evaluating the prognosis in patients with chronic hepatitis B (CHB). 9,10ecently, a novel high-sensitivity linearized HBsAg quantification assay, Lumipulse HBsAg-HQ (Fujirebio Inc., Tokyo, Japan), was developed. 11The original studies on this assay indicated that the detection of serum HBsAg by Lumipulse HBsAg-HQ (HBsAg-HQ) could potentially assist in diagnosing occult hepatitis B infection 12−15 and showed consistent results with the qualitative and quantitative detection of serum HBsAg by Architect HBsAg-QT (HBsAg-QT) and HB-sAg by Elecsys HBsAg II (HBsAg-E II). 16,17However, the correlation between the serum HBsAg-HQ levels and the natural history and the liver tissue pathological states of chronic HBV infection have not yet been adequately investigated. 18

Objectives
The objective of this study is to further evaluate the agreement of serum HBsAg-HQ with HBsAg-QT and comparatively investigate the effectiveness of serum HBsAg-HQ and HBsAg-QT in predicting the liver tissue pathological states of CHB patients.

Patients
A total of 275 Chinese patients with chronic HBV infection who were hospitalized in the Shanghai Public Health Clinical Center of Fudan University (Shanghai, China) between August 2012 and July 2015 were prospectively enrolled, among whom 147 and 128 patients were hepatitis B e antigen (HBeAg)-positive and HBeAg-negative, respectively.The diagnoses of all the patients were in accordance with the standards set forth in the Asian-Pacific clinical practice guidelines for the management of hepatitis B (2015 update). 19Patients with other forms of viral hepatitis, drug-induced liver injuries, hereditary liver diseases, schistosomiasis japonica infection, autoimmune diseases, endocrine and metabolic diseases, and blood system diseases were excluded.Patients who had been treated with interferon alpha, nucleosi(t)des, steroids, or licorices were also excluded.
All patients provided written consent prior to a liver biopsy, and all clinical investigations were conducted according to the principles expressed in the 1995 Declaration of Helsinki.

Histological assessment
Ultrasound-assisted liver biopsies were performed using a 1-second liver biopsy needle (16G).The biopsies were collected immediately after the procedure and transferred into plastic tubes for freezing.One pathologist, who was blinded to all biochemical, serologic and virological parameters, was assigned to review all biopsy specimens.A biopsy length of at least 10 mm was required for inclusion in this study.The pathological diagnosis of liver tissues was performed independently by 1 experienced pathologist.The pathological diagnosis referred to the Scheuer standard, 20 in which a grade is used to describe the intensity of the necro-inflammatory activity, and a stage is used as a measure of fibrosis and architectural alteration.The grades include 5 levels, G0−G4, and the stages include 5 levels, S0−S4.

Laboratory assays
Serum samples used for measurements were taken within 1 week before and 1 week after liver biopsy and stored at -40°C.Serum HBsAg-HQ and HBsAg-QT were quantified using chemiluminescence enzyme immunoassay (CLEIA) in a Lumipulse G1200 automated analyzer (Fujirebio Inc., Tokyo, Japan) and chemiluminescence microparticle immunoassay (CMIA) in an Abbott Architect I2000 automated analyzer (Abbott Laboratories, Chicago, USA), respectively.The HB-sAg-HQ reagents were provided by Fujirebio Inc., and the HBsAg-QT reagents were purchased from Abbott Laboratories.The linear detection range of the HBsAg-HQ was from 5 to 150 000 mIU/mL.If the serum exceeded the upper detection limit, it was diluted 1,000 times and remeasured.Similarly, the linear detection range of the HBsAg-QT was 50 to 250 000 mIU/mL, and if the serum exceeded the upper detection limit, it was diluted 500 times and remeasured.
The serum HBeAg was measured using CMIA Abbott Architect I2000 automated analyzer, with a lower detection limit of 1 S/CO.The serum HBV DNA was measured using Bio-Rad Icycler PCR System (Bio-Rad Laboratories, Inc., USA), and the polymerase chain reaction (PCR) kits were obtained from Qiagen Shenzhen Co. Ltd. (Shenzhen, China).The linear detection range of HBV DNA was 5 × 10 2 IU/mL to 5 × 10 7 IU/mL.

Statistical analysis
Statistical analyses were performed using MedCalc v. 15.1 (MedCalc Software, Broekstraat, Mariakerke, Belgium).A paired-samples t-test was used to compare the differences between the serum HBsAg-HQ and HBsAg-QT.Pearson correlation, linear regression analysis and Bland-Altman plots were used to evaluate the agreement between the HBsAg-HQ and HBsAg-QT quantitation.Spearman's correlation analysis was used to analyze the correlation of the serum HBsAg-HQ and HBsAg-QT levels with the liver tissue pathological grade and stage.The receiver operating characteristic (ROC) curve was used to assess the effectiveness of serum HBsAg-HQ and HBsAg-QT for predicting the different liver tissue pathological states.The paired-samples DeLong non-parametric test was used to compare the differences in the area under the ROC curve (AUC) between the serum HBsAg-HQ and HBsAg-QT for predicting the same liver tissue pathological states.A 2-sided p-value of less than 0.05 was considered to be significant.

Clinical characteristics of the patients
There was no significant difference in the male-to-female ratio (p = 0.490) between the HBeAg-positive and HBeAg-negative patients; however, there was a significant difference in the average age (p = 0.000) between the HBeAg-positive and HBeAg-negative patients.The difference in alanine transaminase (ALT) (p = 0.947) between the HBeAg-positive and HBeAg-negative patients was not statistically significant.The differences in the serum HBsAg-HQ, HBsAg-QT and HBV DNA (all p = 0.000) between the HBeAg-positive and HBeAg-negative patients were all statistically significant.There was a significant difference (p = 0.046) in the frequency of the different pathological grades, but there was no significant difference (p = 0.469) in the frequency of the different pathological stages between the HBeAg-positive and HBeAg-negative patients (Table 1).

Comparison between HBsAg-HQ and HBsAg-QT levels
Regardless of serum HBeAg state and HBsAg levels, there was no significant difference between the serum HBsAg-HQ and HBsAg-QT (p = 0.691).Grouping the patients according to serum HBeAg state and HBsAg levels showed that there were no significant differences between serum HBsAg-HQ and HBsAg-QT in patients with both HBeAg-positive (p = 0.853) and HBeAg-negative (p = 0.647) and in patients with both higher HBsAg levels (HBsAg-QT ≥100 000 mIU/mL) (p = 0.942) and lower HBsAg levels (HBsAg-QT <100 000 mIU/mL) (p = 0.089) (Table 2).

Performance of HBsAg-HQ and HBsAg-QT in predicting pathological states
In HBeAg-positive patients, all AUCs of the serum HB-sAg-HQ and HBsAg-QT for predicting the pathological grades ≥G2 and ≥G3, and stages ≥S2, ≥S3 and ≥S4 were significantly greater than the area under the diagonal reference line (all p < 0.01).Of these, only the AUCs of serum HBsAg-HQ and HBsAg-QT for predicting pathological stage ≥S3 and ≥S4 were >0.70 (Table 4, Fig. 5).In the HBeAg-negative patients, all the AUCs of serum HBsAg-HQ and HBsAg-QT for predicting pathological grades ≥G2 and ≥G3, and stages ≥S2, ≥S3 and ≥S4 were not significantly greater than the area under diagonal reference line (all p > 0.05) (Table 4).
In HBeAg-positive patients there were no significant differences for predicting all the same pathological states (all p > 0.05) between the AUCs of the serum HBsAg-HQ and HBsAg-QT (Table 4).In the HBeAg-negative patients, there were no significant differences for predicting the same pathological states except for predicting the pathological stage ≥S4 between the AUCs of serum HBsAg-HQ and HB-sAg-QT (p = 0.046 for predicting the pathological stage ≥S4, p > 0.05 for predicting the other pathological states) (Table 4).

Discussion
Choi et al. confirmed that the qualitative results of serum HBsAg-HQ highly agreed with those of the serum HBsAg-QT, with a concordance rate of 99.8% (k = 1.00; 95% CI, 0.99−1.00).Of the 315 HBsAg-QT-positive samples, 314 were HBsAg-HQ-positive.Of the 685 HBsAg-QT-negative samples, 684 were HBsAg-HQ-negative. 16Recently, Yang et al. evaluated and compared the agreement of the qualitative and quantitative results of serum HBsAg-HQ with HB-sAg-QT and HBsAg-EII. 17The results showed that of the 2,043 samples tested, 1,844 samples gave negative results for both HBsAg-HQ and HBsAg-QT, and 172 samples yielded positive results for both HBsAg-HQ and HBsAg-QT.Of the remaining 27 samples that had inconsistent HBsAg results, 3 were HBsAg-QT-positive and 24 were HBsAg--HQ-positive.Among these, none of the 3 HBsAg-QT--positive samples were subsequently confirmed to be positive.Twenty of the 24 HBsAg-HQ-positive samples were confirmed to be positive, 3 were confirmed to be negative and 1 gave an indeterminate result and therefore was excluded from the specificity calculation for HBsAg-HQ.The specificity was 99.84% for HBsAg-HQ and 99.84% for HBsAg-QT.Yang et al. also found that, of the 112 tested samples, HBsAg-HQ displayed an excellent correlation with both HBsAg-QT and HBsAg-EII (r = 0.985 and r = 0.990); the Bland-Altman analyses demonstrated that, compared to HBsAg-QT and HBsAg-EII, HBsAg-HQ had an upward bias of 0.19 log 10 IU/mL with a 95% LOA of -0.01 to 0.39 log 10 mIU/mL and 0.07 log 10 IU/mL with a 95% LOA of -0.12 to 0.25 log 10 mIU/mL, respectively. 17n this study, the serum HBsAg-HQ was significantly correlated with HBsAg-QT.The Bland-Altman analyses showed that, compared to the serum HBsAg-QT, the HBsAg-HQ had a slight bias of 0.04 log 10 mIU/mL with a 95% LOA of -0.61 to 0.69 log 10 mIU/mL; the overall disagreement rate was 3.64%.Further analyses of the grouping according to the HBeAg state (HBeAg-positive and HBeAg-negative) and the HBsAg levels (higher HBsAg levels (HBsAg-QT ≥100 000 mIU/mL) and lower HBsAg levels (HBsAg-QT < 100 000 mIU/mL)) also showed similar results.This further demonstrated that the serum HBsAg-HQ levels were highly correlated and highly agreed with HBsAg-QT.
Seto et al. comparatively investigated the changes in the serum HBsAg-HQ and HBsAg-EII in different phases in the natural history of chronic HBV infection. 18−23 Unexpectedly, the serum HBsAg-HQ levels were significantly higher than the HBsAg-EII levels in either the immune tolerance or the activation phase of the HBeAg-positive patients.However, the serum HBsAg-HQ levels were similar to the HBsAg-EII levels in either the immune escape or control phase of the HBeAg-negative patients.However, the reason why the serum HBsAg-HQ levels in HBeAg-positive patients were significantly higher than HBsAg-E levels is not clear.The investigators speculated that this could possibly be due to the enhanced detection of minor viral populations with "a" determinant mutations in patients with higher viral loads.However, Yang et al. reported that the S gene mutations within the "a" determinant, such as T126A, T126S, Q129H, Q129R, T140S, and G145E, did not affect the correlation and agreement observed between the serum HBsAg-HQ and HBsAg-QT and HBsAg-EII. 17n this study, regardless of the serum HBeAg state and HBsAg levels, the difference between the serum HBsAg-HQ and HBsAg-QT levels was not significant.Further analyses of the grouping according to the HBeAg state and HBsAg levels also showed similar results.Our results agreed with the study by Yang et al. but failed to find a serum HBsAg-HQ level that was not significantly higher than HBsAg-QT. 175][26][27][28][29] However, the predicable optimal pathological state of serum HBsAg and HBVDNA in the HBeAgpositive patients was not consistent between the different studies, 7,[24][25][26][27] although for the HBV DNA in the HBeAgnegative patients, there was an agreement in the pathological grade ≥G2 and stage ≥S2. 28,29The results of this study were consistent with the findings by Cheng et al. and Jia  7,[24][25][26][27] Furthermore, we did not observe significant differences in the AUCs for predicting the same liver tissue pathological states between the serum HBsAg-HQ and HBsAg-QT in the HBeAg-positive patients.This study further demonstrated that, in the HBeAg-positive patients, serum HBsAs for predicting the liver tissue pathological states was valuable.Importantly, the efficacies of serum HBsAg-HQ for predicting the liver tissue pathological states were highly consistent with those of HBsAg-QT.
The results of this study showed that, in HBeAg-positive patients, the optimal cut-off of serum HBsAg-HQ for predicting a pathological grade ≥G3 was < 7.351 log 10 mIU/mL (2.244 × 10 7 mIU/mL), with a difference of -0.204 log 10 mIU/mL (-0.625 mIU/mL) from HBsAg-QT.The corresponding sensitivity, specificity, and positive and negative predictive values were 87.18%, 43.52%, 35.8%, and 90.4%, respectively.The optimal cut-off of serum HBsAg-HQ for predicting a pathological stage ≥S4 was < 6.865 log 10 mIU/mL (7.328 × 10 6 mIU/mL), with a difference of 0.073 log 10 mIU/mL (1.183 mIU/mL) from HBsAg-QT.The corresponding sensitivity, specificity, and positive and negative predictive values were 81.25%, 64.35%, 38.8%, and 92.5%, respectively.This suggests that, in HBeAg-positive patients, the optimal cut-offs of serum HBsAg-HQ for predicting pathological grade ≥G3 and stage ≥S4 should be highly consistent with those of HBsAg-QT, and serum HBsAg-HQ and HBsAg-QT should be very valuable for predicting pathological grade ≥G3 and stage ≥S4.
In conclusion, this study further evaluated the agreement between the serum HBsAg-HQ and HBsAg-QT levels and comparatively investigated the effectiveness of serum HBsAg-HQ and HBsAg-QT in predicting the liver tissue pathological states of CHB.The results showed that the serum HBsAg-HQ was highly correlated and agreed with the HBsAg-QT in both the HBeAg-positive and HBeAgnegative patients, regardless of whether they presented with higher or lower HBsAg levels.Furthermore, serum HBsAg-HQ and HBsAg-QT had good predictive efficacy on the pathological grade ≥G3 and stage ≥S4 in HBeAgpositive patients, but did not have predictive efficacy on the pathological states in HBeAg-negative patients.

Fig. 1 .
Fig. 1.Correlations of the serum HBsAg-HQ and HBsAg-QT with serum ALT in HBeAg-positive (A, B) and HBeAg-negative (C, D) patients

Fig. 2 .
Fig. 2. Correlations of the serum HBsAg-HQ and HBsAg-QT with serum HBV DNA in HBeAg-positive (A, B) and HBeAg-negative (C, D) patients

Fig. 4 .
Fig. 4. Correlation of the serum HBsAg-HQ and HBsAg-QT levels with the liver tissue pathological grade and stage The measurement units of HBsAg-HQ and HBsAg-QT were both log 10 mIU/mL.The box-plots represent the serum distributions of HBsAg-HQ and HBsAg-QT in different pathological grades and stages in the HBeAg-positive (A, B) and HBeAg-negative patients (C, D), in which the top and bottom of the box indicates the upper and lower quartile, and the horizontal line in the box indicates the median.The upper and lower horizontal lines outside the box indicate the 95% quintile, and the circle indicates the extreme value.

Fig. 5 .
Fig. 5. ROC curves of serum HBsAg-HQ and HBsAg-QT for predicting a pathological grade ≥G3 and stage ≥S4 in HBeAg-positive patients

Table 1 .
Clinical characteristics of the patients according to their serum HBeAg state # ULN − upper limit of normal = 40 IU/L; * HBeAg-positive vs HBeAg-negative; a c 2 test; b independent samples t-test.

Table 2 .
Comparison between the serum HBsAg-HQ and HBsAg-QT levels

Table 3 .
Pearson correlation and agreement between the serum HBsAg-HQ and HBsAg-QT

Table 4 .
AUCs of serum HBsAg-HQ and HBsAg-QT for predicting the different liver tissue pathological states

Table 5 .
Optimal cut-offs of serum HBsAg-HQ and HBsAg-QT for predicting different pathological states and the corresponding diagnostic parameters in HBeAg-positive patients