Hepatitis B virus surface antigen and anti-hepatitis C virus rapid tests underestimate hepatitis prevalence among HIV-infected patients
In the case of coinfection with HIV and hepatitis B virus (HBV) and/or hepatitis C virus (HCV), hepatic disease progression is often accelerated, with higher rates of liver cirrhosis and liver-related mortality. We aimed to evaluate the performance of the rapid tests used routinely to detect HBV surface antigen (HBsAg) and anti-HCV among HIV-infected patients in Guinea-Bissau.
Blood samples from HIV-infected patients in Guinea-Bissau were stored after testing for HBsAg and anti-HCV with rapid tests. Samples were subsequently re-tested for HBsAg and anti-HCV in Denmark.
Two rapid tests were used in Guinea-Bissau: HBsAg Strip Ref 2034 (VEDA.LAB, Alençon, France; sensitivity 62.3%; specificity 99.2%) and HEPA-SCAN (Bhat Bio-Tech, Bangalore, India; sensitivity 57.1%; specificity 99.7%). In the two tests the ability to obtain the correct outcome depended on the antigen and antibody concentrations, respectively. Sex, age, CD4 cell count and antiretroviral therapy status did not differ between false negative and true positive samples in either of the tests. The study is limited by a low number of anti-HCV positive samples.
New diagnostic rapid tests should always be evaluated in the setting in which they will be used before implementation.
Globally, more than 240 million people are infected with the hepatitis B virus (HBV) and 150 million people are infected with the hepatitis C virus (HCV) . The estimated numbers of annual deaths attributable to HBV and HCV infection in 2010 were 786 000 and 499 000, respectively . In the case of coinfection with HIV, hepatic disease progression is accelerated, with higher rates of liver cirrhosis and liver-related mortality [3-5].
Testing HIV-infected patients for HBV is essential to ensure optimal treatment against both viruses . Rapid tests detecting the HBV surface antigen (HBsAg) are widely used to provide a fast and low-cost diagnosis. However, the rapid test performance may depend on antigenic variation of HBsAg, and the sensitivity of HBsAg rapid tests has been questioned [6-9]. In the West African country Guinea-Bissau, the rapid test HBsAg Strip Ref 2034 (VEDA.LAB, Alençon, France) is used when testing HIV-infected patients for coinfection with HBV. Although no previous studies have investigated the prevalence of HBsAg among HIV-infected patients in Guinea-Bissau, the neighbouring countries of Senegal and Gambia have reported a high coinfection prevalence of 16.8% and 12.2%, respectively [10, 11].
Several rapid tests detecting HCV circulating antibodies (anti-HCV) have been produced. Independent studies evaluating different anti-HCV rapid tests found a high test sensitivity and specificity [12-16]. However, test sensitivity declined when testing HIV-coinfected patients; this may be attributable to an impaired HCV antibody response [15, 16]. In Guinea-Bissau, HIV-infected patients are tested for HCV using a rapid test (HEPA-SCAN; Bhat Bio-Tech, Bangalore, India). The prevalence of anti-HCV among individuals > 50 years old with HIV infection in Guinea-Bissau was estimated in a recent population-based survey, and a total of 2.0% were coinfected . Patients were tested using Detect-HCV v.3 (Adaltis, Montreal, Canada). The aim of this study was to evaluate the HBsAg and anti-HCV rapid test regimen used at an HIV clinic in Bissau, the capital of Guinea-Bissau.
All patients provided voluntary, signed and dated informed consent, or a fingerprint if illiterate, prior to enrolment in the cohort. The patients consented to give blood and to data resulting from this study being used. The Danish ethics committee gave its consultative approval (case no. 1010050) and the study was finally approved by the UCEPS, the National Ethics Committee of Guinea-Bissau (N. ref. 016/CNES/2011).
Study setting and serum sample collection
The study was conducted at the out-patient antiretroviral therapy (ART) centre at the Hospital National Simão Mendes (HNSM) in Bissau, in collaboration with the Bandim Health Project, the National HIV Programme and the National Public Health Laboratory, between the 28 April 2011 and 30 September 2011. Patients who had a HBsAg or anti-HCV rapid test performed and who provided enough blood to perform subsequent hepatitis analyses (> 0.5 ml of plasma) were enrolled in this study. No patients were excluded from the study. The HIV clinic has previously been described elsewhere .
Patients were tested in the HNSM laboratory for HBsAg (HBsAg Strip Ref 2034) and anti-HCV (HEPA-SCAN) using serum according to the manufacturer's recommendations. The HBsAg Strip Ref 2034 is a rapid immunoassay based on an immunochromatographic sandwich principle. The test uses a combination of conjugated monoclonal staining antibodies (Colloidal Gold) and a solid phase of polyclonal antibodies to selectively identify HBsAg, and the antibody−antigen complex forms a strip line which indicates the presence of HBV in a sample. The HEPA-SCAN is based on a noncompetetive ‘sandwich’ enzyme-linked immunosorbent assay (ELISA). If HCV antibodies are present in the sample, an antibody−antigen complex is formed, and the test displays a blue indicator line. In the case of an inconclusive test result, the package inserts recommended re-testing. All results were continuously entered in a database.
Plasma was separated from blood samples collected in vaccutainers containing EDTA on the same day as the rapid tests were performed, and plasma was frozen at −20°C. The plasma samples were re-tested for HBsAg and anti-HCV at the Department of Clinical Immunology, Aarhus University Hospital, Aarhus, Denmark using commercially available chemiluminescence assays [HBsAg, antibodies to HCV core antigen (anti-HBc), total (IgM and IgG) and anti-HCV; Architect; Abbott Laboratories, Abbott Park, IL]. The chemiluminescence assay provided signal-to-cut-off ratios (S/CO; arbitrary units), which were interpreted as ‘reactive’ or ‘nonreactive’ using predefined cut-offs according to the manufacturer's recommendations. HBsAg reactive samples were furthermore assessed for HBV envelope antigen (HBeAg) and anti-HBe (Architect; Abbott Laboratories). HCV RNA and HBsAg concentration measurements were performed at the Department of Clinical Biochemistry, Section of Molecular Diagnostics, Aalborg University Hospital. The samples were analysed using in-house real-time polymerase chain reaction (PCR) methods as previously described [19, 20]. The limit of detection of HCV RNA was 20 IU/ml. HIV type discrimination was performed with Genie III HIV-1/HIV-2 (Bio-Rad, Steenvorde, France) . Samples testing anti-HCV reactive in the chemiluminescence test but HCV RNA negative were subsequently confirmed with INNO-LIA HCV Score (Innogenetics, Gent, Belgium), which was considered conclusive in the case of a mismatch with the chemiluminescence test.
We analysed the data using a χ2 test for categorical variables. Continuous variables were compared using a two-sample t-test (normal distribution) or Wilcoxon's rank-sum test (nonnormal distribution). The kappa statistic was used to assess the performance of rapid assays in comparison with results obtained using Architect (Abbott Laboratories) confirmed with HCV RNA measurements and INNO-LIA HCV Score. A P-value < 0.05 was considered significant. Data were analysed using Stata IC 11.0 (Stata Corporation, Texas, USA).
A total of 438 patients were tested for HBsAg in both Guinea-Bissau and Denmark using blood samples collected on the same date (Table 1). Among the 77 samples positive for HBsAg in the chemiluminescence assay, 73 (94.8%) were also anti-HBc positive. Three of the remaining four samples were HBeAg positive, and one sample was HBsAg positive only. This HBsAg-positive-only sample had detectable HBV DNA (10 × 103 IU/mL) and could be identified as genotype E. The sensitivity of HBsAg Strip Ref 2034 was 62.3% and the specificity was 99.2% when compared with the chemiluminescence assay (Table 2). The median HBsAg concentration measured as the signal-to-cut-off ratio was significantly lower in the samples that were nonreactive in the rapid test but reactive in the chemiluminescence assay compared with samples that were reactive in the rapid test (5.4 S/CO vs. 3436.6 S/CO, respectively; P < 0.01). The median HBsAg concentration was also higher in samples that were reactive in the rapid assay compared with those that were nonreactive (2.0 × 102 IU/mL vs. 9.596 × 106 IU/mL, respectively; P < 0.01). Detectable HBeAg was more prevalent among patients who tested HBsAg positive in both tests (25.0%) than among those who were only reactive in the chemiluminescence assay (6.9%) (P = 0.05). Age and gender distributions were similar and there was no difference in the CD4 cell count or ART status between patients testing false negative and true positive. All HBsAg-positive patients on ART (45.5%) were treated with a combination containing lamivudine and none were treated with tenofovir. In this setting, the positive predictive value (PPV) of HBsAg Strip Ref 2034 was 0.94 and the negative predictive value (NPV) was 0.93.
Table 1. Patient characteristics at the time of testing for hepatitis B virus surface antigen (HBsAg) and hepatitis C virus antibodies (anti-HCV)
|Number of patients included||438||393|
|Gender [n/N (%)]|| || |
|Male||145/436 (33.3)||130/391 (33.3)|
|Female||291/436 (66.7)||261/391 (66.7)|
|Age [n/N (%)]|| || |
| < 30 years||100/420 (23.8)||95/378 (25.1)|
|30–50 years||237/420 (56.4)||213/378 (56.4)|
| > 50 years||83/420 (19.8)||70/378 (18.5)|
|HIV type [n/N (%)]|| || |
|HIV-1||319/438 (72.8)||287/393 (73.0)|
|HIV-2||80/438 (18.3)||70/393 (9.2)|
|HIV-1/-2||39/438 (8.9)||36/393 (17.8)|
|CD4 cell count [n/N (%)]|| || |
| < 200 cells/μL||147/428 (34.5)||133/383 (34.7)|
|200–350 cells/μL||121/428 (28.3)||110/383 (28.7)|
| > 350 cells/μL||160/428 (37.4)||140/383 (36.6)|
|Patients receiving ART [n/N (%)]|| || |
|Yes||208/438 (47.5)||177/393 (45.0)|
|No||230/438 (52.5)||216/393 (55.0)|
Table 2. Performance of the hepatitis B virus surface antigen (HBsAg) and hepatitis C virus antibody (anti-HCV) rapid tests
|Number of patients tested with rapid test||438||393|
|True positive samples [n/N (%)]||48/77 (62.3)||4/7 (57.1)|
|False positive samples [n/N (%)]||3/361 (0.8)||1/386 (0.3)|
|True negative samples [n/N (%)]||358/361 (99.2)||385/386 (99.7)|
|False negative samples [n/N (%)]||29/77 (37.7)||3/7 (42.9)|
|Measure agreement (%)||92.7||99.0|
In total, 393 patients were included in the evaluation of HEPA-SCAN and seven samples (1.8%) were anti-HCV positive by INNO-LIA HCV Score (Table 2). Twelve samples (3.1%) were anti-HCV reactive in the chemiluminescence test but had undetectable HCV RNA and tested negative by INNO-LIA HCV Score, and these samples were considered false positive in the chemiluminescence test. False positive samples in the chemiluminescence test had a significantly lower median signal-to-cut-off ratio than true positive samples (1.5 S/CO vs. 12.6 S/CO, respectively; P < 0.01). The sensitivity of the HEPA-SCAN rapid test was 57.1% and its specificity was 99.7% compared with INNO-LIA HCV Score. HEPA-SCAN false negative samples (by the INNO-LIA HCV Score) had a lower median chemiluminescence S/CO measurement than true positive samples (P = 0.03). The median HCV viral load was not lower in the false negative samples than in the true positive samples (P = 0.48). There was no difference in sex, age, CD4 count or ART status between patients testing false negative and true positive in the HEPA-SCAN test. The PPV and NPV of HEPA-SCAN were 0.80 and 0.99, respectively.
In this study, we evaluated the performance of rapid tests detecting HBsAg and anti-HCV used in Guinea-Bissau: HBsAg Strip Ref 2034 (sensitivity 62.3%; specificity 99.2%) and HEPA-SCAN (sensitivity 57.1%; specificity 99.7%). The false negative results for both tests were associated with lower S/CO levels. Gender, age, CD4 cell count and ART status did not differ between false negative and positive samples in either of the tests.
The study participants were tested for HBsAg and anti-HCV with rapid tests as a routine part of the national HIV programme policy in Guinea-Bissau. This is a typical setting for the use of rapid tests in many resource-limited countries. In the rapid test pack inserts, both producers state a test sensitivity of > 99%. Evaluating rapid tests in an artificial set-up with selected patient samples may give an unrealistic impression of the performance of the test in other settings. In a study from Malawi, different causes of low HBsAg and anti-HCV test sensitivity were suggested; operator errors, effect of tropical climate on kit stability, or unforeseen technical interference with the test in areas with high HIV and/or malaria prevalence .
The current evaluation took place in a clinical setting, which is a strength of this study. The evaluation of the sensitivity of the HEPA-SCAN anti-HCV test was hampered by the low prevalence of HCV in the study population. Only seven patients had serological signs of current or past infection, and thus the estimation of the sensitivity of the HEPA-SCAN rapid test may not be accurate. Furthermore, HCV antibody levels were not measured, and the analysis was performed by comparing S/CO levels with those determined using the Architect assay and should be interpreted with caution.
We were able to find a study from Ghana which evaluated two HBsAg tests previously produced by VEDA.LAB: a HBsAg dipstick and a latex agglutination test. That study also found suboptimal sensitivities of 71% and 53.8%, respectively . We could not find any published independent evaluations of the anti-HCV test used in this study. Evaluations of other rapid tests detecting HBsAg and anti-HCV found higher sensitivity than in our study [6, 12-16]. For the HBsAg rapid test this may be attributable to a higher affinity of certain HBV genotypes . Selection of HBsAg vaccine escape mutants could potentially reduce the assay performance , but unfortunately we did not perform sequencing of the HBsAg coding gene.
The majority of our study participants were severely immunosuppressed as a result of their HIV infection. Patients with an impaired immune system have been shown to present a lower anti-HCV concentration which decreases the test sensitivity [15, 16]. However, we found no association between immunosuppression as assessed by CD4 cell count and low sensitivity in the rapid test. Information on coinfection with hepatitis delta virus (HDV) was not available, but a previous study found no difference in HBsAg levels when comparing HBV monoinfected patients with HBV/HDV-coinfected patients , and thus we do not expect that HDV coinfection influenced the performance of HBsAg Strip Ref 2034. Ideally, implementation of a test should be evaluated in the laboratory that plans to introduce the test .
HBV infection is prevalent among HIV-infected patients in West Africa, with prevalences of 16.8% and 12.2% in the neighbouring countries of Senegal and Gambia, respectively [10,11], and, in the case of undetected HBV infection, HIV-infected patients may be put on ART containing lamivudine but not tenofovir, which is not efficient for HBV suppression and may lead to liver disease progression, risk of viral transmission and development of HBV resistance . Currently, measurements of HCV RNA and HCV treatment are not available in Guinea-Bissau. Insensitive HCV tests may underestimate the morbidity and mortality caused by this disease and hereby delay adequate roll-out and eventual scale-up of HCV treatment.
In conclusion, screening for HBV and HCV among HIV-infected patients in resource-limited settings such as Guinea-Bissau is important, and screening should be implemented even though optimal sensitivity is not achievable. However, efforts should be made to identify the best option for a certain area because rapid tests may have suboptimal performance. As the World Health Organization recommends, new diagnostic rapid tests should always be evaluated before implementation in a new setting .
The authors are grateful to the patients for their willingness to participate in the study. Special thanks to the clinic assistants Aminata, Quintino, Marques, João Paulo, Imbemba, Zé and the rest of the staff working in the HIV clinic at HNSM for their dedication to their work and their kindness to patients. We are also grateful to the laboratory staff working in the HIV section at the National Public Health Laboratory and to the office staff at the Bandim Health Project for making this study possible. We thank Lone Jakobsen Schmidt, Helle Bøgelund Selmann, Hanne Kjeldsen and Anne-Marie Ovesen for their excellent technical assistance at the Department of Clinical Immunology, Aarhus University Hospital.
Contributions to authorship: BLH, SJ, HK and CE conceived the study; BLH, ALL, HK and CE designed the study protocol; CM, DdaS and ZJdaS carried out the clinical assessments; BLH, SJ, ALL, CW, LØ, HK and CE carried out analysis and interpretation of data. BLH, SJ and CE drafted the manuscript; SJ, CW, HK, CE and LØ critically revised the manuscript for intellectual content. All authors read and approved the final version of the manuscript. BLH, CW and CE are guarantors of the paper.
Conflicts of interest: The authors have no conflicts of interest to declare.
Funding: The study was supported by the Bandim Health Project, Medical Doctor Johan Boserup and Lise Boserup´s grant, Aage and Johanne Louis-Hansen´s foundation, Jacob and Olga Madsen´s foundation and Aase and Ejnar Danielsen´s foundation. The Global Fund to Fight AIDS, TB and Malaria (Global Fund) supported the data collection during 2009–2013 through the National Secretary for the fight against AIDS (SNLS) in Guinea-Bissau. The Ministry of Foreign Affairs of Denmark (DANIDA) provided a travel grant for BLH, and Aarhus University, Denmark supported a 1-year scholarship for BLH. Abbott provided serology reagents. West African Platform for HIV Intervention Research (WAPHIR) and International Epidemiologic Databases to Evaluate AIDS (IeDEA) supported data collection in Bissau.
Appendix: Appendix: the Bissau HIV cohort study group
The Bissau HIV cohort study group comprises: Amabelia Rodrigues, David da Silva, Zacarias da Silva, Candida Medina, Ines Oliviera-Souto, Lars Østergaard, Alex Laursen, Morten Sodemann, Peter Aaby, Anders Fomsgaard, Christian Erikstrup, Jesper Eugen-Olsen and Christian Wejse (chair).