The Lumipulse G HBsAg-Quant assay for screening and quantification of the hepatitis B surface antigen

Journal of Virological Methods 228 (2016) 39–47

Contents lists available at ScienceDirect

Journal of Virological Methods journal homepage: www.elsevier.com/locate/jviromet

The Lumipulse G HBsAg-Quant assay for screening and quantification of the hepatitis B surface antigen Ruifeng Yang 1 , Guangjun Song 2 , Wenli Guan 3 , Qian Wang 3 , Yan Liu 3 , Lai Wei ∗ Peking University People’s Hospital, Peking University Hepatology Institute, Beijing Key Laboratory of Hepatitis C and Immunotherapy for Liver Diseases, Beijing, China

a b s t r a c t Article history: Received 9 October 2015 Received in revised form 9 November 2015 Accepted 9 November 2015 Available online 23 November 2015 Keywords: Hepatitis B virus HBsAg Quantitative Performance evaluation Diagnosis

Qualitative HBsAg assay is used to screen HBV infection for decades. The utility of quantitative assay is also rejuvenated recently. We aimed to evaluate and compare the performance of a novel ultra-sensitive and quantitative assay, the Lumipulse assay, with the Architect and Elecsys assays. As screening methods, specificity was compared using 2043 consecutive clinical routine samples. As quantitative assays, precision and accuracy were assessed. Sera from 112 treatment-naïve chronic hepatitis B patients, four patients undergoing antiviral therapy and one patient with acute infection were tested to compare the correlations. Samples with concurrent HBsAg/anti-HBs were also quantified. The Lumipulse assay precisely quantified ultra-low level of HBsAg (0.004 IU/mL). It identified additional 0.98% (20/2043) clinical samples with trance amount of HBsAg. Three assays displayed excellent linear correlations irrespective of genotypes and S-gene mutations (R2 > 0.95, P < 0.0001), while minor quantitative biases existed. The Lumipulse assay did not yield higher HBsAg concentrations in samples with concomitant anti-HBs. Compared with other assays, the Lumipulse assay is sensitive and specific for detecting HBsAg. The interpretation of the extremely low-level results, however, is challenging. Quantitative HBsAg results by different assays are highly correlated, but they should be interpreted interchangeably only after conversion to eliminate the biases. © 2015 Elsevier B.V. All rights reserved.

1. Background Infection of hepatitis B virus (HBV) remains a severe threat to the public health worldwide, especially in the Asia-pacific region (Chen et al., 2000). For example, the prevalence of hepatitis B surface antigen (HBsAg) is 7.18% in China; accordingly there are ca.

Abbreviations: HBV, hepatitis B virus; HBsAg, hepatitis B surface antigen; HBeAg, hepatitis B e antigen; anti-HBc, antibodies against hepatitis B core antigen; antiHBs, antibodies against hepatitis B surface antigen; NAT, nucleic acid testing; CHB, chronic hepatitis B; HCC, hepatocellular carcinoma; OBI, occult HBV infection; PEGIFN, peginterferon; NA, nucleos(t)ide analogue; ETV, entecavir; cccDNA, covalently closed circular DNA; CV, coefficient of variation. ∗ Corresponding author at: Peking University People’s Hospital, Peking University Hepatology Institute, Xizhimen South Street, Beijing 100044, China. Tel.: +86 10 8832 5566; fax: +86 10 8832 5723. E-mail addresses: [email protected] (R. Yang), [email protected] (G. Song), [email protected] (W. Guan), [email protected] (Q. Wang), helen [email protected] (Y. Liu), [email protected] (L. Wei). 1 Tel.: +86 10 8832 5727. 2 Tel.: +86 10 8832 5734. 3 Tel.: +86 10 8832 5720. http://dx.doi.org/10.1016/j.jviromet.2015.11.016 0166-0934/© 2015 Elsevier B.V. All rights reserved.

93 million HBV carriers, among whom 30 million are patients with chronic hepatitis B (CHB) (Liang et al., 2009; Lu and Zhuang, 2009). CHB patients are at high risk of progressing to cirrhosis, liver failure and hepatocellular carcinoma (HCC) (Fattovich et al., 2008). As a hallmark of HBV infection, serum qualitative HBsAg assay has been used for screening and diagnosis of HBV infection for decades. It still plays an important role even in the era of nucleic acid testing (NAT), especially for the resource-limited countries with a high prevalence of HBV infection (Allain and Cox, 2011). Recent studies also indicate the clinical significance of the quantitative HBsAg assay. For example, HBsAg levels vary in different phases of the natural history of HBV infection (Tseng and Kao, 2013). Baseline HBsAg < 1000 IU/mL and HBsAg annual decrease >0.3 log10 IU/mL were associated with HBsAg seroclearance. HBsAg ≥ 1000 IU/mL and HBV DNA ≥ 200 IU/mL could be combined to identify patients with risk of switching from inactive carrier to reactivation status (Martinot-Peignoux et al., 2013). The cumulative risk for cirrhosis and HCC rises with the increase of serum HBsAg levels (Lee et al., 2013). On the other hand, baseline or on-treatment HBsAg level can be potentially applied to predict the treatment outcome. Our previous study indicated that serum HBsAg < 1500 IU/mL at week 12 and <2890 IU/mL at week 24 had

CMIA, chemiluminescent microparticle immunoassay; ECLIA, electrochemiluminescence immunoassay; CLEIA, chemiluminescent enzyme immunoassay; mAb, monoclonal antibody; AMPPD, 3-(2 -spiroadamantane)-4-methoxy4-(3 -phosphoryloxy)phenyl-1,2-dioxetane disodium salt

1:100, 1:200 or 1:1000 with NaCl and Tris bufffer 0.005–150 IU/mL (0.005 IU/mL) 29 min 100 ␮L Lumipulse G1200 Fujirebio Lumipulse G HBsAg-Quant

Sandwich principle, two capture mAbs and two detection mAbs

CLEIA (AMPPD)

Yes, to disrupt viral particles and dissociate HBsAg from HBsAg-anti-HBs complexes

18 min None ECLIA (ruthenium)

50 ␮L

0.05–130 IU/mL (0.05 IU/mL)

1:500 with recalcified negative human plasma 1:400 with buffered negative human serum 0.05–250 IU/mL (0.05 IU/mL) 29 min 75 ␮L None

Sandwich principle, capture mAbs and polyclonal detection antibodies Sandwich principle, two capture mAbs and a mixture of mAbs and polyclonal antibodies

CMIA (acridinium)

Assay duration

Molecular E170

where C is the HBsAg concentration (mIU/mL) for the Lumipulse assay, and sample/cutoff ratio (S/Co) for the Architect assay; A is the control reagent; B is the reagent containing neutralizing antibodies.

Roche HBsAg II Quant

CA − CB . CA

Architect i2000SR

Inhibition rate (%) =

Abbott Architect HBsAg

The presence of serum HBsAg was confirmed using the HBsAg confirmatory kits. HBsAg detected using the Lumipulse HBsAg assay was confirmed by the Lumipulse HBsAg confirmatory assay (Fujirebio, Tokyo, Japan), and the HBsAg detected using the Architect HBsAg assay was confirmed by the Architect HBsAg Confirmatory V.1 assay (Abbott, Wiesbaden, Germany). Polyclonal neutralizing antibodies against HBsAg were used to bind the immunodominant epitopes and thereby block the binding sites for the capture antibodies used in the HBsAg assays. Briefly, the sample with repeatedly positive HBsAg results was split as two aliquots. Reagent containing the neutralizing antibodies and the control reagent were added in parallel. HBsAg concentrations were measured after thorough mixing and incubation, and then the inhibition rate was calculated.

Table 1 Characteristics of the three HBsAg quantitative assays.

2.2. HBsAg confirmatory tests

Pre-treatment

The principle and procedure of the Lumipulse G HBsAg-Quant assay (Fujirebio, Tokyo, Japan) was described previously (Shinkai et al., 2013). Samples with results exceeding the upper limit was retested through 100-, 200- or 1000-fold dilution. HBsAg quantification using the Architect HBsAg assay (Abbott Ireland, Sligo, Ireland) and Elecsys HBsAg II Quant assay (Roche Diagnostics, Mannheim, Germany) was performed according to the manufacturer’s instruction. The characteristics of the assays are summarized in Table 1.

Technology (tracer)

2.1. HBsAg assays

Principle

The study was approved through the Peking University People’s Hospital Ethical Committees.

Analyzer

2. Materials and methods

Reaction sample volume

Linear range (analytical sensitivity)

On-board dilution

Traceability (NIBSC code)

negative predictive values >90% for HBV e antigen (HBeAg) seroconversion at week 48 in patients treated with peginterferon (PEG-IFN) (Ma et al., 2010). The absence of HBsAg decline during PEG-IFN treatment helps identify the non-responders and determine the early stopping rule during PEG-IFN treatment (Rijckborst et al., 2012). Although the decline of HBsAg during the nucleos (t) ide analogue (NA) treatment seemed not so drastic (Zoulim et al., 2015), baseline HBsAg <1000 IU/mL and on-treatment HBsAg reduction >0.166 log IU/mL per year were optimal cutoff levels for predicting the long-term HBsAg loss (Seto et al., 2013). Compared with HBV DNA testing, HBsAg assay is simple, cheap and provide results rapidly. There have been several commercial quantitative assays available (Burdino et al., 2014; Lee et al., 2012; Sonneveld et al., 2011). A novel HBsAg quantitative assay, the Lumipulse G HBsAg-Quant assay, has been recently developed (Matsubara et al., 2009). It is based on a new chemiluminescent enzyme immunoassay technology, claiming an ultra-sensitivity of 0.005 IU/mL. The assay is performed automatically on the Lumipulse G1200 immunoanalyzer (Choi et al., 2013). Trace amount of serum HBsAg could be detected even after the seroconversion (Shinkai et al., 2013). In this study, we evaluated and compared the performance of the Lumipulse G HBsAg-Quant with two commonly used assays, the Architect HBsAg and the Elecsys HBsAg II Quant assays in various clinical settings.

WHO first International standard, subtype ad (80/549) WHO second international standard, subtype adw2, genotype A (00/588) WHO second international standard, subtype adw2, genotype (00/588)

R. Yang et al. / Journal of Virological Methods 228 (2016) 39–47

Assay

40

R. Yang et al. / Journal of Virological Methods 228 (2016) 39–47

Inhibition rate more than 50% indicated the existence of HBsAg. The confirmatory result should be regarded as indeterminate if the CA value is below the limit of detection (LOD). 2.3. Specificity comparison using clinical routine samples During June and July 2015, a total of 2043 consecutive serum samples were submitted to our Institute for routine pre-surgery testing. HBsAg was detected using the Lumipulse and Architect assays in parallel. The sample with a positive HBsAg result was retested in duplicate after a 10,000 × g centrifugation. If either retest was reactive, the sample was considered positive for HBsAg. Only samples with inconsistent results were subjected to the confirmatory testing. Specificity was calculated using the following formula: Specificity = number of true-negative samples/(number of true-negative specimens + number of false-positive specimens). The Elecsys assay was not included in comparison because it was not designed for screening purpose. 2.4. Precision evaluation Low and high-positive controls were used to evaluate the within-run and within-laboratory precisions (Lee et al., 2012). To make the low-positive control, HBsAg positive serum samples were pooled and the concentration was adjusted approximately to the 2 folds of LOD, i.e. 0.10 IU/mL for the Architect and Elecsys assays and 0.01 IU/mL for the Lumipulse assay through dilution using the HBsAg negative sera. Concentration of the high-positive control was approximately 120 IU/mL.

41

2.8. Statistical analysis Statistical analysis was performed and results illustrated using GraphPad Prism 6 (GraphPad, San Diego, CA) and MedCalc 13 (MedCalc Software bvba, Ostend, Belgium). The correlation and difference were analyzed using Pearson’s correlation coefficients and Bland–Altman plots. Slopes and intercepts of the linear correlation curves were compared. Statistical significance was defined as P < 0.05. 3. Results 3.1. Clinical specificity comparison between the Lumipulse and Architect assays for detecting routine samples Of the 2043 samples tested, 1844 samples gave negative HBsAg results and 172 samples yielded positive results using both assays; the remaining 27 samples had inconsistent HBsAg results (Table 2). Three samples were determined as positive by the Architect assay and 24 samples were determined as positive by the Lumipulse assay. None of the three Architect positive samples were subsequently confirmed as positive. Twenty of the 24 Lumipulse positive samples were confirmed as positive; three samples were confirmed as negative; and the remaining one sample gave an indeterminate result and therefore it was excluded from the specificity calculation for the Lumipulse assay. The specificity was 99.838% for the Lumipulse assay and 99.840% for the Architect assay. HBV DNA was measured in the 20 samples with confirmed HBsAg results, among which six samples yielded positive HBV DNA results, ranging from 32 IU/mL to 600 IU/mL.

2.5. Accuracy evaluation 3.2. Precision comparison An HBsAg reference panel (03/262, NIBSC, UK) was used to assess the accuracy of the assays, containing a series of 4-fold dilutions of the 2nd International Standard for HBsAg (00/588, NIBSC, UK), with HBsAg concentrations ranging from 8.25 to 0.064 IU/mL. Four further dilutions (0.032, 0.016, 0.008 and 0.004 IU/mL) were also prepared to evaluate the linearity of the Lumipulse assay in extremely low level of HBsAg. Each level of the diluted standard was tested in triplicate. 2.6. Samples from HBV infected individuals for the comparisons of the quantitative results Serum samples from 112 treatment-naïve HBeAg positive CHB patients enrolled in our previous study were used to evaluate the correlations of the HBsAg results quantified by these assays (Song et al., 2014). To compare the utility of the assays in monitoring the kinetics of HBsAg, serial sera from four CHB patients undergoing PEG-IFN or entecavir (ETV) treatment and a patient with acute HBV infection were quantified. Finally, 15 samples with concurrent HBsAg (≥0.05 IU/mL) and anti-HBs (≥10 IU/L) measured by the Abbott Architect assays were retested using the Lumipulse and Elecsys assays, and these samples carried positive anti-HBc antibody. 2.7. Other laboratory measurements HBV genotype and S-gene mutations were determined by direct sequencing, and mutations were confirmed using the clonal sequencing method (Yang et al., 2010). HBV DNA level was measured using the Realtime HBV assay with a limit of detection of 15 IU/mL (Abbott molecular, Des Plaines, IL, USA). HBeAg was quantified using the Elecsys HBeAg assay (Roche Diagnostics, Mannheim, Germany) (Ma et al., 2010).

The precision data are listed in Table 3. 3.3. Accuracy evaluation using the HBsAg international standard panel Three HBsAg assays yielded excellent regression curves when quantifying the serially diluted HBsAg standards (Fig. 1A). The Architect assay slightly overestimated the HBsAg level whereas the Elecsys and Lumipulse assay slightly underestimated it. Moreover, the Lumipulse assay could precisely and accurately quantify the further serially diluted standards containing very low level of HBsAg. At the level of 0.004 IU/ml diluted standard, the coefficient of variation (CV) of the Lumipulse quantitative results was 3.77% (Fig. 1B). 3.4. Correlations of the quantitative HBsAg results tested by the three assays In 112 samples from treatment-naïve CHB patients, 66 were identified as genotype B, 45 identified as genotype C, and one sample identified as a recombination of genotypes C and D (C/D). High HBsAg concentrations were present in these samples, ranging from approximately 102 to 106 IU/mL. The Lumipulse HBsAg assay displayed an excellent correlation with either the Architect or the Elecsys assay when quantifying HBsAg; HBsAg quantification results from samples with HBV genotypes B, C or C/D shared the same correlation curve (Fig. 2A andC). Bland–Altman plot demonstrated that HBsAg concentration measured by the Lumipulse assay was slightly lower than that by the Architect (mean 0.19 log10 IU/mL) and that by the Elecsys assay (mean 0.07 log10 IU/mL) independent of the magnitude of the measurement (P < 0.0001, Fig. 2B and D).

42

R. Yang et al. / Journal of Virological Methods 228 (2016) 39–47

Table 2 Clinical characteristics of the patients with a positive HBsAg result determined only by the Architect or by the Lumipulse assay (n = 27). Age (y), gender

Diagnosis

Lumipulse HBsAg (mIU/mL)

Architect HBsAg (IU/mL)

Anti-HBs (IU/L)

Anti-HBe

Anti-HBc

Confirmatory test

HBV DNA (IU/mL)

1 2 3

35, F <1, M 74, F

10 39.1 323.4

0 0.03 0

0 0.07 23.7

(−) (−) (−)

(−) (−) (+)

(+) (+) (−)

(−) (−) ND

4 5 6 7 8 9 10 11 12 13 14 15 16 17

65, F 70, M 77, M 80, M 35, F 33, F <1, M 32, F <1, M 87, M 28, F 24, F 76, M 29, F

29.8 6.5 15 8.2 11.9 5 5.7 32.8 8.1 9.9 6 284.4 89 5.9

0.04 0.01 0.01 0.01 0 0 0 0 0.01 0 0.01 0 0 0

35.2 4.41 1.55 8.51 5.47 0 18.46 0 1 2.13 8.2 15.49 77.46 89.34

(+) (−) (+) (+) (−) (−) (−) (−) (−) (+) (−) (−) (−) (−)

(+) (+) (+) (+) (−) (−) (−) (−) (−) (+) (−) (−) (+) (−)

(+) (+) (+) (+) (−) (+) Indeterminate (+) (+) (+) (+) (−) (+) (+)

85 (−) (−) 600 ND (−) ND 79 (−) 32 (−) ND (−) (−)

18 19 20 21 22 23 24 25

68, M 26, M 55, F 30, F 31, F 35, M 43, F 29, F

19.9 8.1 13.3 8.3 6.4 8.6 10.4 0.8

0.01 0.02 0.01 0 0.01 0.02 0 0.06

0.47 39.26 ND 36.46 ND ND ND >1000

(+) (+) (+) (+) (−) (−) (−) (+)

(+) (+) (+) (+) (−) (−) (−) (+)

(+) (+) (+) (+) (+) (+) (+) (−)

(−) 254 (−) (−) (−) 361 (−) ND

26 27

25, F 32, F

Prenatal examination Neonatal jaundice Blepharitis and diabetes mellitus Lung cancer Cardiac carcinoma Parathyroid hyperplasis Hepatectomy for HCC prenatal examination Prenatal examination Anemia Periodontitis Premature infant Gastric cancer Physical examination Prenatal examination Lymphocytic leukemia Anti-HBs testing after vaccination Renal dysfunction Stem cell transplantation Physical examination Prenatal examination Prenatal examination Dental caries Physical examination Intrauterine foreign body Prenatal examination Prenatal examination

1.4 0.6

0.12 0.16

0 6.82

(−) (−)

(−) (−)

(−) (−)

ND ND

Sample

ND: not done. Table 3 Precisions of the HBsAg assays.

3.5. HBsAg quantification from samples with HBV S-gene mutations

Assay

Mean (IU/mL)

Within-run precision CV (%)

Within-laboratory precision CV (%)

Lumipulse G HBsAg-Quant

0.01 0.09 118.40

3.6 1.2 1.6

6.6 8.6 2.5

Elecsys HBsAg II Quant

0.09 122.66

3.2 2.7

8.2 6.6

Architect HBsAg

0.11 130.82

7.8 4.8

9.1 7.7

Eight out of 122 samples showed a single HBV S-gene mutation, which included T126A (3), T126S (1), Q129H (1), Q129R (1), T140S (1) and G145E (1). Proportion of the mutant population ranged from 48.28% to 97.44%. The mutations did not interfere with the correlations among these assays (P = 0.77, Fig. 3). 3.6. Correlation of quantitative HBsAg with HBV DNA and HBeAg levels in treatment-naïve CHB patients Serum HBsAg level measured using the Lumipulse, Architect or Elecsys assay was highly correlated with the viral load (R ≥ 0.70, P < 0.0001) and moderately correlated with the HBeAg

Fig. 1. (A) HBsAg concentrations quantified by the three assays against the target values of the serially diluted HBsAg international standard. (B) HBsAg concentrations measured by the Lumipulse HBsAg assay against the extremely low levels of HBsAg of the diluted international standard. Each level of the diluted standard were tested in triplicate. The tested HBsAg results are expressed as mean ± standard deviation (SD) IU/mL. Some error bars are too short to be indicated.

R. Yang et al. / Journal of Virological Methods 228 (2016) 39–47

43

Fig. 2. Correlation and the Bland–Altman plot of the HBsAg level measured by the Lumipulse and Architect assays (A and B) and by the Lumipulse and Elecsys assays (C and D). In the Bland–Altman plot, the solid line represents the mean of difference between the assays; the dashed line illustrates the 95% confidence interval of the mean difference.

level (R2 ≥ 0.24, P < 0.0001, Fig. 4A–C). Moreover, HBV DNA was also correlated moderately with the HBeAg level (R2 = 0.17, P < 0.0001, Fig. 4D). 3.7. HBsAg dynamics monitored by the three assays during PEG-IFN and ETV treatment and in acute infection phase The three assays yielded similar patterns of HBsAg fluctuations during the antiviral treatment. Two patients achieved HBeAg seroconversion under PEG-IFN or ETV treatment, with serum HBsAg drop of 0.40–0.46 log10 IU/mL and 0.30–0.48 log10 IU/mL tested by the assays, respectively (Fig. 5A and C). Two other patients failed to respond to the treatment, with increased HBsAg levels of 0.21–0.33 log10 IU/mL and 0.08–0.10 log10 IU/mL measured by the assays, respectively, compared with the baseline levels

(Fig. 5B and D). One male had an acute HBV infection and spontaneously “cleared” the virus; the Architect and Elecsys assays yielded negative results approximately seven months after his first visit, but the Lumipulse assay detected low-level HBsAg for two additional visits until the HBsAg was undetectable at his latest visit (Fig. 5E). 3.8. Comparison of HBsAg quantification results in samples with concomitant HBsAg and anti-HBs Twelve out of 15 samples had positive HBsAg results tested by the three assays. Three samples had a negative Elecsys testing result but positive Architect and Lumipulse results. The Lumipulse assay did not yield higher HBsAg results than the other assays (Table 4).

Fig. 3. Correlations of the HBsAg quantified by the Lumipulse and the Architect assays (A) and by the Lumipulse and the Elecsys assays (B) in samples with and without HBV S-gene mutants.

44

R. Yang et al. / Journal of Virological Methods 228 (2016) 39–47

Fig. 4. Correlations between serum HBsAg concentration quantified using the three assays and HBV DNA as well as HBeAg levels (A–C), and correlation of serum HBV DNA and HBeAg level (D) in treatment-naïve HBeAg positive CHB patients (n = 112).

4. Discussion In this study, the Lumipulse HBsAg ultrasensitive assay was shown to be useful in screening for HBV infection. It showed a sensitivity of 0.005 IU/mL, which is 10 times more sensitive than the Abbott Architect and Roche Elecsys assays. We validated the precision and accuracy at the limit of detection of HBsAg. The Lumipulse assay showed comparable within-run and total precisions at the level of 0.01 IU/mL as with those of the Architect and Elecsys assays at the level of 0.10 IU/mL. On the other hand, by using the serially diluted HBsAg reference materials, the Lumipulse assay could quantify the extremely low level of HBsAg (0.004 IU/mL) with a CV < 4%. Therefore, it is qualified to detect the trace amount of HBsAg. Choi et al. evaluated the performance of the Lumipulse assay as a screening tool, and found that the consistency of the

qualitative HBsAg results between the Lumipulse and Architect assays was very high (99.80%). But the distribution of the HBsAg levels and the number of weakly reactive results were not provided (Choi et al., 2013). To further validate if the specificity might be compromised by the ultra-high sensitivity of the Lumipulse assay, over two thousand routine clinical samples were tested by the Lumipulse and the Architect assays in our study. As expected, the Lumipulse assay detected more HBsAg positive samples than the Architect assay. There were 24 out of 2043 samples (1.17%) with negative Architect results but with positive Lumipulse results. The HBsAg concentrations of these samples were all low, ranging from 0.005 to 0.32 IU/mL measured by the Lumipulse assay. Twenty samples out of the 24 samples (83.33%) were confirmed as positive. In the 20 samples, 6 samples (30.00%) also yielded positive HBV DNA results, although their viral loads were low. The specificity of the Lumipulse assay was equal to that of the Architect assay (both were

Table 4 HBsAg quantification in samples with concurrent HBsAg and anti-HBs (n = 15). Sample#

HBV DNA (IU/mL)

Anti-HBs (IU/L)

Elecsys HBsAg (IU/mL)

Architect HBsAg (IU/mL)

Lumipulse HBsAg (IU/mL)

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

<15 <15 <15 ND <15 <15 <15 18,500 ND <15 ND 19,200 <15 697,000 <15

45.46 83.74 28.08 12.44 185.23 15.24 10.22 36.55 14.57 11.91 13.62 32.89 24.44 29.9 827.36

48.44 31.86 0.18 15.15 (−) 0.26 600.00 15.38 2.86 (−) 0.09 0.16 0.09 97.91 (−)

71.59 42.31 1.51 31.76 0.14 0.57 238.52 79.48 4.62 0.17 0.26 0.48 0.19 153.42 0.08

24.11 24.42 2.62 20.65 0.43 0.34 580.71 85.10 2.77 0.15 0.06 0.16 0.11 436.55 0.05

ND, not done.

R. Yang et al. / Journal of Virological Methods 228 (2016) 39–47

45

Fig. 5. Dynamics of HBsAg measured by different assays in two patients receiving PEG-IFN alpha-2b treatment (A and B), 2 patients undergoing ETV treatment (C and D) and a 47-year-old male with acute HBV infection and spontaneous viral clearance (E). HBeAg, HBV DNA and anti-HBs status are indicated in boxes. LOD, limit of detection.

99.84%). Interestingly, in our study, all the six samples with false positive HBsAg results (three for the Architect assay and three for the Lumipulse assay) were from adult female individuals, among whom four samples were from pregnant women. Similar results have been reported (Gotstein et al., 2002; Veselsky et al., 2014; Weber et al., 2006). There are many factors contributing to the falsepositive results, such as the cross-reactive endogenous proteins in the specimens (Gotstein et al., 2002). Fortunately these samples accounted for a very small proportion. Further testing for anti-HBc and HBV DNA may help to address such discrepancies. The results of the routine clinical samples indicated that in China, where the prevalence rate of HBsAg is above 7% (Liang et al., 2009), carriers of extremely low level of HBsAg are not uncommon. The proportion of the ultra-low HBsAg carriers was 0.98% (20/2043) based on our data. The significance deserves our further investigation. First, with the implementation of NAT in blood donation screening in developed countries where the HBV infection is usually uncommon, HBsAg might become a less important biomarker for the blood donation screening (Stramer et al., 2012). However, HBsAg testing remains as the first-line choice in most countries, especially in those resource-limited countries where HBV infection is endemic (Allain and Cox, 2011). The implementation of

the ultra-sensitive HBsAg assay might help further strengthen the blood donation security, especially for the blood recipients under immunosuppressive treatment. Second, it can shed light on the formation of the occult HBV infection (OBI), challenging the concept the OBI (Allain and Cox, 2011). According to our findings, out of 2034 subjects, six patients with “occult” infections had lowlevel serum HBsAg with the help of the ultra-sensitive serological assay. As a result, they should not have been diagnosed as OBI any longer. Third, in clinical settings, the association between the extremely low level of HBsAg and the outcome remains largely unknown. There are CHB patients who “achieved” HBsAg seroclearance carrying detectable serum HBsAg, particularly those with negative anti-HBs (Seto et al., 2015; Shinkai et al., 2013). In our study, a patient with a typical acute HBV infection was followed up. Using the ultra-sensitive HBsAg assay, his HBsAg seroclearance was “delayed” for 70 days (Fig. 5E). Togashi et al. reported that extremely low level of HBsAg was seen more frequently in patients with liver diseases than in the general population (Togashi et al., 2008). Our results indicate that approximately 1% more samples will be reported as positive with the Lumipulse G HBsAg-Quant assay. For these individuals, the outcome should be followed up closely, and the potential incidence of liver cirrhosis and HCC

46

R. Yang et al. / Journal of Virological Methods 228 (2016) 39–47

should be monitored. Subsequent serological and NAT testing including anti-HBc, anti-HBc IgM, anti-HBs, and HBV DNA assays should be performed, and the aminotransferases, alpha-fetoprotein and liver histology should also be assessed as well. According to our results, 50% (10/20) of the subjects carrying trace amount of HBsAg had negative anti-HBc antibodies (Table 2). Among them, three individuals (two infants and a young woman) received a recent HBV vaccination. They should be diagnosed as non-HBV carriers because the transient antigenemia caused by the hepatitis B vaccination may be detected by the ultra-sensitive HBsAg assay (Anjum, 2014; Otag, 2003). Consequently, we should inquire the medical history of the subjects before the weakly positive HBsAg results are reported to avoid unnecessary medical visits and psychological harm. The remaining seven subjects may be in the window period of the acute infection. They should be closely observed to see whether the infection could be resolved after six months (EASL, 2012; Lok and McMahon, 2009). On the other hand, half of the 20 subjects with trace amount of HBsAg had positive anti-HBc antibodies, indicating the past or chronic HBV infection. Among them, three immunocompetent patients had detectable HBV DNA. They should be treated as the HBsAg/anti-HBc positive patients with detectable DNA (EASL, 2012). A stem cell transplantation recipient receiving immunosuppression therapy had positive anti-HBc and 361 IU/mL of serum HBV DNA. Using the ultra-sensitive assay, more patients receiving chemotherapy and/or immunosuppression with potential risk of HBV reactivation can be found. They should be followed carefully by means of ALT and HBV DNA monitoring and treated with NA to prevent the HBV reactivation (Hwang and Lok, 2014). On the whole, the Lumipulse assay shed light on the discovery and management of the low-level HBsAg carriers, but prospective studies are still needed in the future. Meanwhile, it can be speculated that the interpretation of the extremely low-positive HBsAg results to the patients will be more challenging for the physicians and laboratory technicians than before. The three quantitative HBsAg assays compared in this study showed excellent correlations. Meanwhile, as expected, the assays reflected similar HBsAg dynamics when they were used for monitoring the treatment or the natural history of acute infection. HBsAg quantification results were slightly different among the assays; the Architect assay yielded 0.19 and 0.12 log10 IU/mL higher HBsAg level than the Elecsys and Lumipulse assays when measuring the serum samples. One of the possible reasons for the difference might be that the Architect assay is calibrated using the 1st generation of international standards while the other two assays are calibrated against the 2nd generation. In clinical settings, such discrepancy might be considered because the change of HBsAg level was often minimal, especially for patients receiving NA therapy or those in the natural history of HBV infection (Martinot-Peignoux et al., 2013; Seto et al., 2013). When the quantitative results were near the cutoff levels predicting the outcome (Ma et al., 2010; MartinotPeignoux et al., 2013; Seto et al., 2013), such discrepancy should also be considered. Quantitative HBsAg results by different assays can be interpreted interchangeably after conversion. Samples with HBV genotypes B, C and C/D were included for correlation evaluation. The correlations were genotype independent. Previous studies also show a high correlation between the Architect and Elecsys assays when testing samples with different genotypes (Sonneveld et al., 2011; Tuaillon et al., 2012). The Sgene mutations within the “a” determinant have been reported to possibly impair the sensitivity of the HBsAg assays (Coleman, 2006; Mizuochi et al., 2010). In this study, 7.14% (8/112) of samples from treatment-naïve CHB patients harbored such mutants. However, these mutations seemed non-harmful to HBsAg quantification. The three assays are based on sandwich principle, where two or more anti-HBs antibodies are applied as the capture and detection antibodies. Theoretically only the co-existence of two or

more mutations in these epitopes can lead to underestimation or even detection failure, whereas such concurrent mutations were not found in the samples. Mizuochi et al. reported that the mutation of T140S impaired the sensitivity of the Lumipulse II HBsAg qualitative assay, which was the prototype of the Lumipulse quantitative assay (Mizuochi et al., 2010). Our results showed that this mutation did not affect the quantification, although only one sample was measured. Serum HBsAg is correlated with the expression of transcriptionally active covalently closed circular DNA (cccDNA), whereas HBV DNA and HBeAg reflect the viral replication during different phases of HBV infection (Brunetto, 2010). HBsAg levels were all highly correlated with the HBV DNA levels and moderately correlated with the HBeAg levels in treatment-naïve CHB patients, whatever the HBsAg assay used. There was a much stronger correlation between the HBsAg and HBV DNA than that in previous studies (Ganji et al., 2011; Tuaillon et al., 2012). This might be due to the higher homogeneity of our patients; only HBeAg positive patients at the immune clearance phase were enrolled (Song et al., 2014). The Lumipulse assay includes a pre-treatment process. Epitopes potentially hidden in the antigen-antibody complex or in the particles (Dane particles or subviral particles) may be exposed for detection (Matsubara et al., 2009; Shinkai et al., 2013). Samples with concurrent HBsAg and anti-HBs were tested, but the Lumipulse assay did not yield significantly higher HBsAg results. Zhang et al. found that co-existing anti-HBs antibodies were reactive to HBsAg but had a lower specific activity than those from immunized persons (Zhang et al., 2007), indicating that they might not compete with the capture or detection antibodies used in the HBsAg assays to bind the epitopes. Therefore, the concomitant anti-HBs antibodies in the serum may not affect the detection of the HBsAg using various assays, and the pre-treatment process included in the Lumipulse assay may not result in a higher HBsAg result. In conclusion, as an ultra-sensitive assay, the Lumipulse G HBsAg-Quant assay is suitable to screen for HBV infection in clinical or blood donation settings, facilitating the identification of HBV carriers with trace amount of HBsAg. Clinical application of such a new assay will lead to a two side consequence. It provides a tool to reveal and investigate the significance of extremely low level of circulating HBsAg, whereas cautious interpretation of the results to the patients would be needed to prevent unnecessary alarm. As a quantitative assay, it yields strongly correlated results with the other commonly used ones and is suitable to monitor the HBsAg kinetics during the natural history of infection or antiviral therapy. However, there are minor biases in quantifying HBsAg among different assays. However, there are minor biases in quantifying HBsAg among different assays. Therefore, careful interpretation of the quantitative results would be needed, especially when the HBsAg quantification is near the cutoff levels predicting the prognosis, or when the change of the HBsAg level is marginal in patients receiving NA therapy or those during the natural history of infection. Acknowledgments This work was supported by grants from the China National Science and Technology Major Project for Infectious Diseases Control during the 12th Five-Year Plan Period (2012ZX10002003 and 2012ZX10002005), Major Project of National Science and Technology “Creation of major new drugs” (2012ZX09303019), Natural Science Foundation of China (81201339). The Lumipulse HBsAg kits were kindly provided by the Fujirebio Inc. References Allain, J.P., Cox, L., 2011. Challenges in hepatitis B detection among blood donors. Curr. Opin. Hematol. 18, 461–466.

R. Yang et al. / Journal of Virological Methods 228 (2016) 39–47 Anjum, Q., 2014. False positive hepatitis B surface antigen due to recent vaccination. Int. J. Health Sci. 8, 189–193. Brunetto, M.R., 2010. A new role for an old marker, HBsAg. J. Hepatol. 52, 475–477. Burdino, E., Ruggiero, T., Proietti, A., Milia, M.G., Olivero, A., Caviglia, G.P., Marietti, M., Rizzetto, M., Smedile, A., Ghisetti, V., 2014. Quantification of hepatitis B surface antigen with the novel DiaSorin LIAISON XL Murex HBsAg Quant: correlation with the ARCHITECT quantitative assays. J. Clin. Virol.: Off. Pub. Pan Am. Soc. Clin. Virol. 60, 341–346. Chen, C.J., Wang, L.Y., Yu, M.W., 2000. Epidemiology of hepatitis B virus infection in the Asia-Pacific region. J. Gastroenterol. Hepatol. 15 (Suppl.), E3–E6. Choi, S.J., Park, Y., Lee, E.Y., Kim, S., Kim, H.S., 2013. Performance evaluation of LUMIPULSE G1200 autoimmunoanalyzer for the detection of serum hepatitis B virus markers. J. Clin. Lab. Anal. 27, 204–206. Coleman, P.F., 2006. Detecting hepatitis B surface antigen mutants. Emerg. Infect. Dis. 12, 198–203. EASL, 2012. EASL clinical practice guidelines: management of chronic hepatitis B virus infection. J. Hepatol. 57, 167–185. Fattovich, G., Bortolotti, F., Donato, F., 2008. Natural history of chronic hepatitis B: special emphasis on disease progression and prognostic factors. J. Hepatol. 48, 335–352. Ganji, A., Esmaeilzadeh, A., Ghafarzadegan, K., Helalat, H., Rafatpanah, H., Mokhtarifar, A., 2011. Correlation between HBsAg quantitative assay results and HBV DNA levels in chronic HBV. Hepat. Monthly 11, 342–345. Gotstein, M.G., Aide, P.M., Coleman, P.F., Sanborn, M.R., 2002. Rate of hepatitis B virus infection in pregnant women determined by a monoclonal hepatitis B surface antigen immunoassay. J. Clin. Microbiol. 40, 3493–3496. Hwang, J.P., Lok, A.S., 2014. Management of patients with hepatitis B who require immunosuppressive therapy. Nat. Rev. Gastroenterol. Hepatol. 11, 209–219. Lee, H.J., Kim, S.Y., Lee, S.M., Heo, J., Kim, H.H., Chang, C.L., Lee, E.Y., Son, H.C., 2012. Elecsys hepatitis B surface antigen quantitative assay: performance evaluation and correlation with hepatitis B virus DNA during 96 weeks of follow-up in chronic hepatitis B patients. Ann. Lab. Med. 32, 420–425. Lee, M.H., Yang, H.I., Liu, J., Batrla-Utermann, R., Jen, C.L., Iloeje, U.H., Lu, S.N., You, S.L., Wang, L.Y., Chen, C.J., Group, R.E.V.E.A.L.H.S., 2013. Prediction models of long-term cirrhosis and hepatocellular carcinoma risk in chronic hepatitis B patients: risk scores integrating host and virus profiles. Hepatology 58, 546–554. Liang, X., Bi, S., Yang, W., Wang, L., Cui, G., Cui, F., Zhang, Y., Liu, J., Gong, X., Chen, Y., Wang, F., Zheng, H., Wang, F., Guo, J., Jia, Z., Ma, J., Wang, H., Luo, H., Li, L., Jin, S., Hadler, S.C., Wang, Y., 2009. Epidemiological serosurvey of hepatitis B in China – declining HBV prevalence due to hepatitis B vaccination. Vaccine 27, 6550–6557. Lok, A.S., McMahon, B.J., 2009. Chronic hepatitis B: update 2009. Hepatology 50, 661–662. Lu, F.M., Zhuang, H., 2009. Management of hepatitis B in China. Chin. Med. J. 122, 3–4. Ma, H., Yang, R.F., Wei, L., 2010. Quantitative serum HBsAg and HBeAg are strong predictors of sustained HBeAg seroconversion to pegylated interferon alfa-2b in HBeAg-positive patients. J. Gastroenterol. Hepatol. 25, 1498–1506. Martinot-Peignoux, M., Lapalus, M., Laouenan, C., Lada, O., Netto-Cardoso, A.C., Boyer, N., Ripault, M.P., Carvalho-Filho, R., Asselah, T., Marcellin, P., 2013. Prediction of disease reactivation in asymptomatic hepatitis B e antigen-negative chronic hepatitis B patients using baseline serum measurements of HBsAg and HBV-DNA. J. Clin. Virol.: Off. Pub. Pan Am. Soc. Clin. Virol. 58, 401–407. Matsubara, N., Kusano, O., Sugamata, Y., Itoh, T., Mizuii, M., Tanaka, J., Yoshizawa, H., 2009. A novel hepatitis B virus surface antigen immunoassay as sensitive as hepatitis B virus nucleic acid testing in detecting early infection. Transfusion 49, 585–595. Mizuochi, T., Mizusawa, S., Nojima, K., Okada, Y., Yamaguchi, K., 2010. Single amino acid substitution in the hepatitis B virus surface antigen (HBsAg) “a”

47

determinant affects the detection sensitivity of an HBsAg diagnostic kit. Clin. Chim. Acta: Int. J. Clin. Chem. 411, 605–606. Otag, F., 2003. False positive HBsAg result in blood donors due to administration of three different recombinant DNA Hepatitis B vaccines. Vaccine 21, 3734–3737. Rijckborst, V., Hansen, B.E., Ferenci, P., Brunetto, M.R., Tabak, F., Cakaloglu, Y., Lanza, A.G., Messina, V., Iannacone, C., Massetto, B., Regep, L., Colombo, M., Janssen, H.L., Lampertico, P., 2012. Validation of a stopping rule at week 12 using HBsAg and HBV DNA for HBeAg-negative patients treated with peginterferon alfa-2a. J. Hepatol. 56, 1006–1011. Seto, W.K., Tanaka, Y., Wong, D.K., Shinkai, N., Cheung, K.S., Liu, K.S., Fung, J., Lai, C.L., Yuen, M.F., 2015. Longitudinal profiles of highly-sensitive hepatitis B surface antigen levels: re-evaluation of HBsAg seroclearance. Liver Int.: Off. J. Int. Assoc. Study Liver, http://dx.doi.org/10.1111/liv.12980. Seto, W.K., Wong, D.K., Fung, J., Huang, F.Y., Lai, C.L., Yuen, M.F., 2013. Reduction of hepatitis B surface antigen levels and hepatitis B surface antigen seroclearance in chronic hepatitis B patients receiving 10 years of nucleoside analogue therapy. Hepatology 58, 923–931. Shinkai, N., Matsuura, K., Sugauchi, F., Watanabe, T., Murakami, S., Iio, E., Ogawa, S., Nojiri, S., Joh, T., Tanaka, Y., 2013. Application of a newly developed high-sensitivity HBsAg chemiluminescent enzyme immunoassay for hepatitis B patients with HBsAg seroclearance. J. Clin. Microbiol. 51, 3484–3491. Song, G., Rao, H., Feng, B., Wei, L., 2014. Prediction of spontaneous HBeAg seroconversion in HBeAg-positive chronic hepatitis B patients during the immune clearance phase. J. Med. Virol. 86, 1838–1844. Sonneveld, M.J., Rijckborst, V., Boucher, C.A., Zwang, L., Beersma, M.F., Hansen, B.E., Janssen, H.L., 2011. A comparison of two assays for quantification of hepatitis B surface antigen in patients with chronic hepatitis B. J. Clin. Virol.: Off. Pub. Pan Am. Soc. Clin. Virol. 51, 175–178. Stramer, S.L., Zou, S., Notari, E.P., Foster, G.A., Krysztof, D.E., Musavi, F., Dodd, R.Y., 2012. Blood donation screening for hepatitis B virus markers in the era of nucleic acid testing: are all tests of value? Transfusion 52, 440–446. Togashi, H., Hashimoto, C., Yokozawa, J., Suzuki, A., Sugahara, K., Saito, T., Yamaguchi, I., Badawi, H., Kainuma, N., Aoyama, M., Ohya, H., Akatsuka, T., Tanaka, Y., Mizokami, M., Kawata, S., 2008. What can be revealed by extending the sensitivity of HBsAg detection to below the present limit? J. Hepatol. 49, 17–24. Tseng, T.C., Kao, J.H., 2013. Clinical utility of quantitative HBsAg in natural history and nucleos(t)ide analogue treatment of chronic hepatitis B: new trick of old dog. J. Gastroenterol. 48, 13–21. Tuaillon, E., Mondain, A.M., Nagot, N., Ottomani, L., Kania, D., Nogue, E., Rubbo, P.A., Pageaux, G.P., Van de Perre, P., Ducos, J., 2012. Comparison of serum HBsAg quantitation by four immunoassays, and relationships of HBsAg level with HBV replication and HBV genotypes. PLoS ONE 7, e32143. Veselsky, S.L., Walker, T.Y., Fenlon, N., Teo, C.G., Murphy, T.V., 2014. Discrepant hepatitis B surface antigen results in pregnant women screened to identify hepatitis B virus infection. J. Pediatr. 165, 773–778. Weber, B., Van der Taelem-Brule, N., Berger, A., Simon, F., Geudin, M., Ritter, J., 2006. Evaluation of a new automated assay for hepatitis B surface antigen (HBsAg) detection VIDAS HBsAg Ultra. J. Virol. Methods 135, 109–117. Yang, R., Cong, X., Xu, Z., Xu, D., Huang, W., Maertens, R., Wei, L., 2010. INNO-LiPA HBV genotyping is highly consistent with direct sequencing and sensitive in detecting B/C mixed genotype infection in Chinese chronic hepatitis B patients and asymptomatic HBV carriers. Clin. Chim. Acta: Int. J. Clin. Chem. 411, 1951–1956. Zhang, J.M., Xu, Y., Wang, X.Y., Yin, Y.K., Wu, X.H., Weng, X.H., Lu, M., 2007. Coexistence of hepatitis B surface antigen (HBsAg) and heterologous subtype-specific antibodies to HBsAg among patients with chronic hepatitis B virus infection. Clin. Infect. Dis.: Off. Pub. Infect. Dis. Soc. Am. 44, 1161–1169. Zoulim, F., Carosi, G., Greenbloom, S., Mazur, W., Nguyen, T., Jeffers, L., Brunetto, M., Yu, S., Llamoso, C., 2015. Quantification of HBsAg in nucleos(t)ide-naive patients treated for chronic hepatitis B with entecavir with or without tenofovir in the BE-LOW study. J. Hepatol. 62, 56–63.