Study population
Individuals were recruited at the National Reference Laboratory for Viral Hepatitis (NRLVH) in the Oswaldo Cruz Institute (Rio de Janeiro, Brazil) to give paired serum and oral fluid samples for the optimization of assay conditions. These individuals were recruited in a non-probabilistic method using consecutive sampling and these samples were used only for optimization of the assay conditions.
To evaluate the anti-HBc assay for oral fluid under real life conditions, a total of 1296 individuals were recruited from different serological profiles and different regions.
Serological profiles showed 57 individuals with active infection (HBsAg+/anti-HBc+/anti-HBs− or HBsAg+/anti-HBc−/anti-HBs−), 37 individuls with anti-HBc isolate (HBsAg−/Anti-HBc+/anti-HBs-), 119 individuals with previous HBV exposure (HBsAg−/anti-HBc+/anti-HBs+), 347 individuals vaccinated for HBV (HBsAg−/ anti-HBc−/Anti-HBs+) and 736 susceptible individuals (HBsAg−/Anti-HBc−/Anti-HBs−).
The individuals were recruited from different sample collection events as explained below:
Group I (GI) was composed by 291 individuals recruited from the NRLVH ambulatory. The inclusion criteria for this group were acute, chronic or suspected cases of hepatitis B infection and aged more than 18 years. Samples were collected in a non-probabilistic fashion using consecutive sampling.
Group II (GII) was composed by 1005 individuals living in different regions of Brazil. Of these: 441 individuas from Southeast (95 from Macaé and Petrópolis cities, 277 professional beauticians, and 69 crack-cocaine users, all of them residents of Rio de Janeiro state), 336 individuals from North (Tocantins State) and 228 individuals from Midwest (Mato Grosso do Sul State). None of these individuals were recruited in viral hepatitis ambulatory care settings or had been previously diagnosed as HBV infected.
According to the Brazilian Health Ministry, the HBV prevalence rates per 100,000 inhabitants were 2.8 in Rio de Janeiro, 4.6 in Mato Grosso do Sul and 6.3 cases in Tocantins State [
2]. In previous reports, incidence of HBsAg in these groups varied from 0.2 to 0.7% and the prevalence of anti-HBc/anti-HBs varied from 9.7 to 12.6% [
7,
33,
34]. In Brazil, HBsAg prevalence among crack cocaine users was 6.2% [
35], and among beauty professionals prevalence ranged from 0 to 8% [
36,
37].
Those recruited from Macaé/RJ, Petrópolis/RJ, Tocantis and Mato Grosso do Sul lived in remote areas and/or deprived communities and reported neither parenteral exposure (i.e. did not inject drugs) nor repeated unprotected sexual intercourse. Recruitment of these individuals was previously described [
7,
33,
34].
Beauticians more than 18 years of age were recruited at a fair aiming to promote knowledge, to encourage technical improvement and to stimulate entrepreneurship among beauticians while crack-cocaine users aged 18–24 were recruited when they reported using crack-cocaine on 3 or more days/week in the last 3 months. Further information about recruitment was described previously [
35,
37].
A questionnaire comprising demographic (gender and age) and socioeconomic (education level, family income, and home characteristics) status was applied to these individuals to assess associations in the HBV groups evaluated. Data collection took place directly before sample collection.
Samples were collected in a non-probabilistic manner using consecutive sampling. Data concerning the severity of HBV infection in infected participants were unknown at the time of collection. All study participants gave informed consent obtained from the Ethics Committee of Oswaldo Cruz Institute under CAAE number 34055514.9.0000.5248. Each participant (or legal guardian) gave informed consent before entering the study. Laboratory results were sent to participants and, in the case of carriers, they were referred to health services for orientation and treatment.
Sample collection and laboratory analysis
Blood samples were collected by venipuncture and centrifuged to obtain serum. Oral fluid was obtained using a commercial device (Salivette, Sarstedt, Germany) and processed as previously described [
28]. All samples were stored at − 20 °C until analysis.
All serum samples were submitted to commercial Enzyme immunoassays (EIAs) to detect total anti-HBc antibodies directed against HBV surface antigen (anti-HBs) and HBsAg, (ETI-AB-COREK-PLUS, ETI-MAK-4, and ETI-AB-AUK-3, Diasorin, Italy, respectively) according to the manufacturer’s instructions. Reactive samples were retested to confirm these results.
All oral fluid samples were also tested with the EIA ETI-AB-COREK-PLUS (Diasorin, Italy), designed to detect total anti-HBc in serum. The cut off was calculated according to the manufacturer’s instructions for both serum and oral fluid samples. Samples with optical density / cutoff values (OD/CO) above 1.100 were considered non-reactive and those below 0.900 were considered reactive Samples with values between 0.900 and 1.100 were considered indeterminate and retested in duplicate, those that remained undetermined were excluded from the analysis.
Serum samples from the field study were also tested for the presence of antibody against hepatitis C virus (Murex anti-HCV -version 4.0, Diasroin, Italy) and anti-HCV positivity was evaluated in the performance of anti-HBc detection in oral fluid samples.
Optimization of anti-HBc assay in the panel of oral fluid samples
The first parameter evaluated to optimize EIA using oral fluid to detect anti-HBc was the transport buffer. In this analysis ten paired serum and oral fluid samples were obtained from five anti-HBc reactive individuals and five anti-HBc negative individuals [
38]. Five transport buffers were evaluated: (T1) phosphate buffered saline (PBS) pH 7.2; (T2) PBS/Tween 20 0.05%; (T3) PBS/Tween 20 0.05%/ 0.005% sodium azide; (T4) PBS/Tween20 0.2%/ bovine serum albumin (BSA) 5%, and (T5) PBS/BSA 0.5%. These buffers were chosen as they have previously been used to evaluate HBsAg marker in oral fluid [
28].
The second parameter was sample volume and in this analysis, 15 anti-HBc reactive and 16 anti-HBc negative individuals were tested. Two volumes were tested: (V1) 100 μL of oral fluid sample + 50 μL of neutralization buffer; (V2) 100 μL of oral fluid sample + 25 μL of neutralization buffer (in sera: 50 μL of sample + 50 μL of neutralization buffer + 50 μL of sample dilution).
All assays were done in duplicate and positive results in serum samples were retested to confirm the results. Only reactive samples were included in this analysis.
Data analysis
Using anti-HBc detection in serum samples from commercial standard EIA as the benchmark, we cross-compared standard results with actual findings with respect to sensitivity, specificity, and positive and negative predictive values. In addition, ROC (Receiver Operating Characteristics) curves were fitted seeking optimal cut-offs, as explained in the classic paper by Van der Schouw et al. [
39].
Descriptive statistics comprise the mean ± the standard deviation, with a preliminary assessment using contingency tables and respective statistics. Categorical variables were compared between groups using the chi-square test or Fisher’s exact test, and continuous variables were analyzed using the Mann–Whitney U test. A p-value of < 0.05 was considered significant.
Concordance between the results obtained for the paired oral fluid and sera samples was assessed using the Kappa index (k). According to international standards, findings should be interpreted as follows: < 0.20 corresponds to poor agreement; 0.21–0.40 as fair agreement; 0.41–0.60 as moderate agreement; 0.61–0.80 as good agreement, and 0.81–1.00 corresponds to very good agreement [
40].
Bivariate analysis addressed and cross-compared sociodemographic characteristics, stratifying data for groups I, II and II. The serological profile of seromarkers (i.e. HBsAg, anti-HBc and anti-HBs) of the patients was analyzed by subgroup and serological status.
Analyses were performed using GraphPad InStat 3.01 (GraphPad Software, San Diego, CA), MedCalc 9.2.1.0 (MedCalc Software, Mariakerke, Belgium), as well as the Statistical Package for the Social Sciences (SPSS for Windows, release 10.0; SPSS Inc., Chicago, IL).
Each figure combines one scatter plot (“dotplot”), with the respective correlation index (R2), as well as the relevant ROC curve (plotting standard and specific D.O., yielding a curve cross-comparing false-positive and true-positive samples). Graphs were fitted using the open source software R 3.5.0, specifically using the ggplot 2 and plotROC libraries.