The authors have declared that no competing interests exist.
Conceived and designed the experiments: FCF RN LH CH MT GM. Performed the experiments: FCF RN BM MH OA. Analyzed the data: FCF. Wrote the paper: FCF EL MB.
Co-infection with hepatitis B virus (HBV) is highly prevalent in people living with HIV in Sub-Saharan Africa. Screening for HBV surface antigen (HBsAg) before initiation of combination antiretroviral therapy (cART) is recommended. However, it is not part of diagnostic routines in HIV programs in many resource-limited countries although patients could benefit from optimized antiretroviral therapy covering both infections. Screening could be facilitated by rapid diagnostic tests for HBsAg. Operating experience with these point of care devices in HIV-positive patients in Sub-Saharan Africa is largely lacking. We determined the prevalence of HBV and Hepatitis C virus (HCV) infection as well as the diagnostic accuracy of the rapid test device Determine HBsAg in an HIV cohort in rural Tanzania.
Prospectively collected blood samples from adult, HIV-1 positive and antiretroviral treatment-naïve patients in the Kilombero and Ulanga antiretroviral cohort (KIULARCO) in rural Tanzania were analyzed at the point of care with Determine HBsAg, a reference HBsAg EIA and an anti-HCV EIA.
Samples of 272 patients were included. Median age was 38 years (interquartile range [IQR] 32–47), 169/272 (63%) subjects were females and median CD4+ count was 250 cells/µL (IQR 97–439). HBsAg was detected in 25/272 (9.2%, 95% confidence interval [CI] 6.2–13.0%) subjects. Of these, 7/25 (28%) were positive for HBeAg. Sensitivity of Determine HBsAg was rated at 96% (95% CI 82.8–99.6%) and specificity at 100% (95% CI, 98.9–100%). Antibodies to HCV (anti-HCV) were found in 10/272 (3.7%, 95% CI 2.0–6.4%) of patients.
This study reports a high prevalence of HBV in HIV-positive patients in a rural Tanzanian setting. The rapid diagnostic test Determine HBsAg is an accurate assay for screening for HBsAg in HIV-1 infected patients at the point of care and may further help to guide cART in Sub-Saharan Africa.
Combination antiretroviral therapy (cART) has drastically reduced morbidity and mortality of HIV/AIDS in industrialized countries and recently also in resource-limited settings, including Sub-Saharan Africa (SSA)
Epidemiological data collected in rural and urban areas in various parts of Africa strongly indicate that rates of hepatitis B co-infection are higher in SSA than in Western Europe or the USA
Lamivudine, emtricitabine and tenofovir are nucleoside/−tide reverse transcriptase inhibitors included in standard cART regimens in Tanzania and other countries in SSA. In addition to their suppressive effects against HIV, these drugs feature excellent antiviral activity against HBV
Rapid diagnostic testing (RDT) has successfully facilitated widespread screening for HIV even in rural Africa. Several point of care (POC) products for the detection of HBsAg are currently available but their take-up has been limited so far. The practical evaluation of these RDT devices in HIV-positive patients in an African peripheral setting is key in order to study the utility of including them in diagnostic routines. In this study, we prospectively assessed the diagnostic accuracy and feasibility of a scalable product locally in a large HIV outpatient clinic in rural Tanzania.
Between November 2011 and June 2012, female and male, adult (≥18 years), antiretroviral-naïve, HIV-1 infected patients consecutively attending the Chronic Disease Clinic of the St. Francis Referral Hospital in Ifakara, Kilombero district, Tanzania, were prospectively offered enrollment into this study. Care at the facility was provided according to the guidelines of the national AIDS control program (NACP)
Patient data was retrieved from the electronic patient database of the Kilombero and Ulanga antiretroviral cohort (KIULARCO)
Following inclusion into the study, 8 mL of blood were withdrawn into vacutainers containing ethylenediaminetetraacetic acid (EDTA). After completion of CD4+ cell counts and the index test (Determine HBsAg) on whole blood, samples were centrifuged for 10 min at 2000×g. Alanine aminotransferase (ALT) and creatinine levels were determined, the plasma transferred into freezer vials and stored at −80°C for later serological analysis.
All samples were tested both with the index lateral flow assay Determine HBsAg (Alere Inc., Massachusetts, USA) and the reference enzyme immunoassay Murex HBsAg 3.0 (Abbott Diagnostics, Wiesbaden, Germany). Initial reactivity in the reference assay was confirmed using the Murex HBsAg confirmatory 3.0 neutralization assay. HCV antibodies were screened using Murex anti-HCV 4.0 and HBeAg/anti-HBeAg using DiaSorin ETI-EBK plus and ETI-AB-EBK plus, respectively.
All analyses have been performed on samples obtained from one blood draw per subject. For the evaluation of Determine HBsAg, one index assay was conducted using 50 µL of whole, fresh EDTA blood, followed by a repeat index test using 50 µL of stored plasma of the identical aliquot utilized for the reference assay. Both Determine assays were performed by technicians blinded to the results of the reference test and read by two independent observers (the first being FF or RN, the second being a laboratory technician). In case of discordant readings, the result recorded by the first, i.e. higher qualified observer was used for analysis. Invalid rapid assay results (i.e. non-appearance of control band) underwent repeated testing once with the same sample. The distributors indicate sensitivity and specificity of >99.5% and >99.5% for the reference Murex HBsAg and >95.5% and >99.95% for Determine HBsAg, respectively. All EIAs were performed by FF or RN in the laboratory of IHI in Ifakara according to manufacturer’s instructions. EIA results were analyzed using automated routines in Excel (Microsoft Inc., Redmond, USA). All test kits used in this study were purchased on the open market.
CD4+ cell counting was performed using a CyFlow Counter (Partec GmbH, Münster, Germany) and analysis of ALT and creatinine with Vitrous DTSC II (Ortho-Clinical Diagnostics, Inc., Raritan, NJ).
SAMPLES positive for HBsAg were defined by positive reactivity in the Murex reference test followed by confirmation with the Murex neutralization assay. Prevalence of HBsAg and anti-HCV was defined as count of samples positive divided by total count of samples tested. We calculated sensitivity, specificity, negative and positive predictive values out of a 2×2 cross table using standard definitions
For the study of diagnostic accuracy, a sample size of 270 was calculated for a minimal acceptable sensitivity for Determine of 0.9, an estimated prevalence rate of 0.15, a statistical power of 0.8, and an alpha level of 0.05
During 8 months of enrollment, 279 adult eligible patients were offered to participate in the study as shown in the study profile in
Study profile of the evaluation of Determine HBsAg against a reference HBsAg enzyme immuno assay.
HBsAg negative | HBsAg positive | Total | |||
( |
p-value | ||||
Gender (male) | 88 (35.6%) | 15 (60%) | 103 (37%) | 0.017 | |
Age (y) | 38 (32–48) | 42 (38–56) | 38 (32–47) | 0.18 | |
WHO Stage | I | 153 (61.9%) | 12 (48%) | 165 (60.6%) | |
II | 51 (20.6%) | 6 (24%) | 57 (20.9%) | ||
III | 32 (12.9%) | 6 (24%) | 38 (13.9%) | ||
IV | 11 (4.4%) | 1 (4%) | 12 (4.4%) | 0.417 | |
CD4+ Count (/µL) | 260 (99–441) | 130 (76–349) | 250 (97–439) | 0.25 | |
ALT (U/l) | 32 (23–45) | 42 (30–75) | 32 (23–46) | 0.006 | |
Creatinine (µmol/L) | 73 (58–86) | 73 (50–93) | 73 (58–86) | 0.81 | |
Anti-HCV (positive) | 9 (3.6%) | 1 (4%) | 10 (3.7%) | 0.92 | |
HBeAg (positive) | not tested | 7 (28%) | not tested | – | |
Anti-HBeAg (positive) | not tested | 18 (72%) | not tested | – |
Abbreviations: ALT, alanine aminotransferase; ARV, antiretroviral; HBe-Ag, Hepatitis B envelope antigen; HBsAg, Hepatitis B surface antigen; HCV, Hepatitis C virus; y, years.
Data are presented as median (interquartile range) for continuous variables and as count (column percentage) for categorical variables. P values hypothesis testing between HBsAg positive and HBsAg negative are derived from Mann-Whitney U test for continuous and χ2-test for categorical variables.
HBsAg was detected in 25 of 272 (9.2%, 95% confidence interval [CI] 6.2–13.0%) subjects. Of these, 7/25 (28%, 95% CI 13.5–47.2%) were positive for HBeAg and 18/25 (72%, 95% CI 52.7–86.5%) positive for anti-HBeAg. There was no significant difference in ALT levels between HBsAg+/HBeAg+ (41 U/l, IQR 25–80, n = 7) and HBsAg+/HBeAg- (44 U/l, IQR 32–75, n = 28; p = 0.71) patients. Antibodies to HCV (anti-HCV) were found in 10/272 (3.7%, 95% CI 2.0–6.4%) of patients and 1/272 (0.36%, 95% CI 0.04–1.7%) had serological evidence of both HBsAg and anti-HCV.
All 272 samples were tested both with the index and the reference test using samples of the same venipuncture. Median time to reference and index analysis of the frozen plasma samples was 25 days (IQR 8–35). All samples showed a conclusive result in Determine HBsAg upon first time execution of the assay. The performance of Determine HBsAg is shown in
TP | FP | TN | FN | Sensitivity | Specificity | NPV | PPV | Inter-rater Agreement |
|
|
|
|
|
|
|
|
|
24 | 0 | 247 | 1 | 96 (82.8–99.6) | 100 (98.9–100) | 99.5 (98.1–99.9) | 100 (90.1–100) | 0.97 (0.93–1) |
Abbreviations: CI, confidence interval; FN, false negative; FP, false positive; NPV, negative predictive value; PPV, positive predictive value; TN, true negative; TP, true positive. Agreement is measured between two independent readings of Determine HBsAg.
The main findings of this study on 272 HIV-infected patients in a rural Tanzanian setting were a high prevalence of co-infection with viral hepatitis B and that a rapid diagnostic test for HBsAg is accurate and precise for screening for HBsAg in our setting. According to patients’ characteristics, the study population was representative for HIV-infected patients presenting for care in SSA
The reported prevalence of detectable HBsAg of 9.2% in our study in HIV-positive patients in Ifakara lies within the range reported from previous studies in the same area
A positive HBsAg status was associated with moderately increased ALT levels, which is to be expected due to the hepatic inflammatory state in patients with active replication of HBV. Of the patients with detectable HBsAg, 28% showed HBeAg positivity. A positive HBeAg status is associated with increased HBV DNA levels
This is one of the first studies prospectively evaluating a rapid test for HBsAg in HIV-infected patients entirely at the point of care in an African peripheral health institution. We chose Determine HBsAg for its low waste production, undemanding storage requirements and most importantly for its preexisting supply chains, as Determine HIV rapid tests are already part of the routine HIV testing algorithm in Tanzania
The few previous studies evaluating this assay in HIV-infected patients did not report encouraging findings but had limitations, notably limited sample sizes: In a study in HIV-positive patients in Kumasi, Ghana, the sensitivity was found to be only 69.3% with a specificity of 100%. The majority of samples giving a false negative result using Determine HBsAg had a concentration of HBsAg below the determined rapid test’s detection limit and a substantially lower HBV DNA viral load than samples giving a true positive result. Unfortunately, that study was retrospective in nature and results could not be linked to individual characteristics of the patients of whom the samples were originally obtained. Importantly, no information on treatment status or cART regimen was available. It can be hypothesized that present treatment with the dually-active drug lamivudine partly suppressed replication of HBV in a relevant proportion of patients to the extent that the concentration of HBsAg was below Determine HBsAg’s detection limit. In order to limit the effect of this possible interaction in our study, we only included patients with no prior cART experience. This also realistically depicts the situation where screening for HBsAg would take place mainly, namely at the inclusion into care. Two other studies on Determine HBsAg in HIV patients were consecutively conducted by a research group in Malawi. In the first study published in 2008
A study by Hoffmann et al.
Our study provides evidence that Determine HBsAg can provide testing of good quality in HIV patients in rural Africa. It improved several limitations of previous studies by prospectively including a well characterized population at a relevant point of time, at which the assay would likely be used if adopted in routine screening. Additionally, the study was planned and reported according to accepted standards for studies of diagnostic accuracy, which some previous reports failed to achieve. However, there is a relevant degree of uncertainty about the estimates due to the limited sample size, of which previous studies also suffered. It is not clear whether its findings can be interpolated to other settings as previous evaluation of Determine HBsAg in HIV patients has shown considerable heterogeneity of sensitivity between studies.
The opportunities for HBsAg rapid testing in SSA are many
There are some limitations to be mentioned. Due to the cross-sectional nature of the study, the prevalence of chronic HBV infection, as defined by persistence of HBsAg for >6 months could not be established. Nevertheless, a single measurement reflects the actual field conditions prior to the initiation of cART. Also, we could not fully characterize HBV infection in our population due to the lack of other virological markers. Importantly, the number of HBsAg positive subjects included was lower than anticipated in the sample size calculation, which lead to a higher than desired level of imprecision about the estimates, particularly the sensitivity of Determine HBsAg. However, we quantified and indicated this. It must be considered that we used anticoagulated venous blood for the evaluation of Determine HBsAg and thus no direct conclusion on the performance with capillary blood can be drawn. The prevalence estimate of HCV was entirely based on detection of anti-HCV and was not confirmed by HCV RNA-PCR. This approach is likely to overestimate the true prevalence of co-infection with HCV as false positive results due to cross-reactivity in anti-HCV EIA are common in African samples
This study includes also strengths. Importantly, the study was conducted prospectively and entirely at the point of care at a large out-patient clinic in rural Tanzania. The latter aspect is an essential proof of concept regarding the feasibility of Determine HBsAg at a rural site in SSA.
In conclusion, the data presented in this study demonstrate that Determine HBsAg can offer a readily available and affordable opportunity for accurate screening of viral hepatitis B in HIV patients before the initiation of antiretroviral treatment in a resource-limited setting. This provides the basis for a better quality of care and treatment of cART.
We thank all the staff involved of the Chronic Disease Clinic of the St. Francis Referral hospital in Ifakara and of the Ifakara Health Institute for helping to conduct the study. Finally, we appreciate the participation of all patients without whom this study would not have been possible.