People screened for human immunodeficiency virus (HIV) using rapid diagnostic tests (RDTs) in Africa remain generally unaware of their status for hepatitis B (HBV) and hepatitis C (HCV) infections. We evaluated a two-step screening strategy in Burkina Faso, using both HIV RDTs and Dried Blood Spot (DBS) assays to confirm an HIV-positive test, and to test for HBV and HCV infections. HIV counselling and point-of-care testing were performed at a voluntary counselling and testing centre with HBV, HCV status and HIV confirmation using DBS specimens, being assessed at a central laboratory. Serological testing on plasma was used as the reference standard assay to control for the performance of DBS assays. Nineteen out of 218 participants included in the study were positive for HIV using RDTs. A fourth-generation HIV ELISA and immunoblot assays on DBS confirmed HIV status. Twenty-four out of 25 participants infected with HBV were found positive for hepatitis B surface antigen (HBsAg) using DBS. One sample with a low HBsAg concentration on plasma was not detected on DBS. Five participants tested positive for HCV antibodies were confirmed positive with an immunoblot assay using DBS specimens. Laboratory results were communicated within 7 days to participants with no loss to follow up of participants between the first and second post-test counselling sessions. In conclusion, DBS collection during HIV point-of-care testing enables screening and confirmation of HBV, HCV and HIV infections. Diagnosis using DBS may assist with implementation of national programmes for HBV, HCV and HIV screening and clinical care in middle- to low-income countries.
Diagnosis of human immunodeficiency virus (HIV) infection in resource-limited settings has increased substantially in recent years through implementation of voluntary counselling and testing programmes [1, 2], with particular reference to the use of point-of-care testing. Hepatitis B (HBV) and C (HCV) infections share common routes of transmission with HIV, but because of underdeveloped laboratory facilities and geographic inaccessibility, diagnosis of these infections remains problematic for many sub-Saharan African populations. Consequently, most individuals remain unaware of their HBV and HCV status until advanced stages of the disease, such as liver cirrhosis or hepatocellular carcinoma, become evident. In Burkina Faso, the prevalence rates of HIV, HBV and HCV infections are estimated at 1.0–1.5% , 12–18% and 2–7%, respectively [4-6].
In Africa, rapid diagnostic tests (RDTs) are largely used for HIV testing. These types of tests offer the opportunity to screen for HIV, to provide counselling regarding the implications of infection or its absence and to communicate the screening results during a single visit. RDTs can be performed on capillary blood, which requires less staff training, is less invasive and involves smaller blood volumes than venous blood collection. However, screening strategies based on HIV RDTs have some limitations: (i) poor sensitivity in early stages of HIV infection , (ii) suboptimal positive predictive values in low-incidence countries , (iii) double reactivity of HIV-1 and HIV-2 in settings like West Africa where HIV-2 co-circulates, and (iv) lack of screening for HBV and HCV infections. Dried blood spot (DBS) collection represents a promising alternative to decrease these limitations. Capillary blood collection on a filter card can be performed with relative ease in parallel with HIV rapid testing, and sent with minimal further processing, apart from optimal storage, to a central laboratory for performance of additional screening and confirmatory assays. Several studies have demonstrated the effective use of DBS in HIV diagnosis and monitoring [9-13]. DBS have been evaluated in developed countries for HBV  as well as for HCV screening. We have previously demonstrated that DBS specimens are useful for screening and confirmation of HCV infection .
In this pilot study, we evaluated a screening strategy based on DBS sampling in parallel with HIV point-of-care testing to both confirm HIV RDT-positive results and to establish the HBV and HCV status for persons attending a voluntary counselling and testing in Bobo-Dioulasso, Burkina Faso.
Materials and Methods
Study population and specimen collection
This cross-sectional survey was conducted in Bobo-Dioulasso (the second largest city in Burkina Faso) among people attending the “Centre d'Accueil de Dépistage et d'Information sur le SIDA” (CADI). During pre-test counselling sessions, HBV and HCV testing were offered to all the CADI clients as an addition to HIV screening. Participants providing written informed consent were enrolled and screened for the three infections. Fig. 1 describes the operational procedure for enrolment in the study and testing. For each participant, paired plasma and DBS were simultaneously collected. For DBS preparation, whole blood was spotted (5 spots of 50 μL) on to filter-paper cards (Whatman 903™, Protein Saver Card; Whatman GmbH, Dassel, Germany).
Results of HIV point-of-care testing were given during post-test counselling sessions within 1 h. Participants were invited to return to the CADI 1 week later for results of HBV and HCV testing. Individuals who tested positive for HIV, HCV or HBV infection were referred to a dedicated hospital centre for care and support. Screening and confirmatory assays performed on plasma were used as the reference standard to establish serological status for HIV, HBV and HCV obtained on DBS.
HIV screening at the voluntary counselling and testing centre
HIV screening was performed by social counselling workers at the CADI in accordance with a standard procedure recommended by the National Guidelines on HIV testing that used two RDTs in a sequential algorithm. Samples were tested using a first RDT (Determine® HIV1/2; Inverness Medical, Chiba, Japan). Samples found to be reactive with this first RDT were further assessed by a second immunochromatographic assay as confirmation (SD Bioline® HIV1/2 3.0; Standard Diagnostics Inc, Yongin-Si, South Korea) and to discriminate between HIV type 1, type 2 and type 1 + 2 infections.
HIV screening and confirmation at the central laboratory
Dried blood spot and paired plasma samples from each participant were tested at the Virology Laboratory of Centre Muraz in Bobo-Dioulasso using the Genscreen™ ULTRA HIV Ag-Ab (Bio-Rad, Marnes-La-Coquette, France) which is a fourth-generation commercial enzyme immunoassay (EIA) for simultaneous detection of HIV p24 antigen and HIV antibodies (HIV Ag/Abs). The lower limit of detection for p24 Ag with this assay is 0.65 IU/mL in serum or plasma. For DBS testing, a punched disc of 6-mm diameter was eluted overnight into 300 μL phosphate-buffered saline (PBS) solution for HIV EIA test. The assays were performed and interpreted according to the manufacturer's instructions.
The HIV ELISA-positive specimens were analysed using a line immunoassay (Inno-Lia™ HIV I/II Score; Innogenetics N.V., Gent, Belgium) on an automated system (Auto-Lia™, Innogenetics) to confirm or not the presence of antibodies against HIV and to differentiate between HIV-1, HIV-2 and HIV-1/2 infections. For DBS in this confirmatory assay, a punched disc of 6 mm was eluted overnight in 150 μL PBS.
Dried blood spot and paired plasma samples from each participant were tested at the Central Laboratory using the ETI-MAK-4 HBsAg EIA (DiaSorin S.p.A, Saluggia, Italy) for HBV surface antigen (HBsAg) detection and the Monolisa™ Anti-HBc PLUS (Bio-Rad) for anti-HBV core antibodies (anti-HBc Abs) detection. For DBS testing, a punched disc of 6-mm diameter was eluted overnight into 300 μL PBS for each HBsAg test and into 150 μL PBS for anti-HBc Abs testing.
The assays were performed and interpreted according to the manufacturer's instructions with slight modifications for DBS specimens. For anti-HBc Abs testing on DBS, the microplate was saturated overnight with 200 μL of bovine serum albumin 10% in each well of the microplate before performing the test. The cut-off absorbance values were calculated according to the following formula: (i) DBS HBsAg cut-off value = MAPC/2 + 0.3 and (ii) DBS anti-HBc Abs cut-off value = MANC/3 + 2 SD; where MAPC is the mean absorbance of positive control, MANC is the mean absorbance of negative control, and SD is the standard deviation.
From HBsAg-positive individuals exhibiting low signal to cut-off ratio by ELISA test (OD <3), the concentration of HBsAg in plasma was measured using an Architect HBsAg QT assay (Abbott, Chicago, IL).
HCV testing and confirmation
Dried blood spot and paired plasma samples from each participant were tested at the central Laboratory using the Monolisa™ HCV Ab-Ag ULTRA assay (Bio-Rad), which is a commercial EIA method for detecting HCV antigen and anti-HCV Abs. For DBS testing, a punched disc of 6-mm diameter was eluted overnight into 300 μL of ready-for-use conjugate 1 (provided by the kit manufacturer Bio-Rad). The assays were performed and interpreted according to the manufacturer's instructions.
Plasma and DBS HCV ELISA-positive specimens were further assessed by the Inno-Lia™ HCV Score assay (Innogenetics). For DBS in this HCV confirmatory assay, a 6-mm punched disc was eluted overnight in 150 μL PBS.
To confirm HCV infection, all the plasma-positive specimens by the first ELISA were also tested using another HCV EIA (COBAS® Anti-HCV assay; Roche Diagnostics GmbH, Mannheim, Germany) and referred for PCR (COBAS® AmpliPrep/COBAS® TaqMan®, HCV Test; Roche Diagnostics) when the immunoblot assay was inconclusive.
Analyses were performed on Stata/SE 11.0 (Statcorp, College Station, TX, USA). Descriptive statistics were computed as means or proportions. Distributions of continuous variables were plotted to assess normality and fit to adequate mathematical functions to obtain a normal distribution. We compared mean optical densities (OD) of DBS versus plasma using the non-parametric Wilcoxon rank sum test.
We calculated sensitivity, specificity, positive predictive value and negative predictive value of DBS versus plasma for each of the laboratory assays (HBsAg, anti-HBc, HCV and HIV assays) as simple proportions and obtained their corresponding 95% CIs. Agreement between DBS results and that of plasma was assessed using a kappa coefficient and its 95% CI was obtained.
Characteristics of study participants
From 21 June to 1 July 2011, a total of 218 consecutive individuals counselled and accepting to be tested for HIV in the CADI, consented to be enrolled in the study. The mean age of study participants was 29.8 ± 11.0 years. Women represented 62.8% with a sex ratio (M/F) of 0.6.
Testing for HIV, HBV and HCV on plasma served as the reference standard to establish the serological status of each participant for HIV, HBV and HCV infections. Of 218 samples, 23 (10.6%) tested positive for HIV using the fourth-generation HIV Ag-Ab ELISA. Four EIA-positive samples (1.8%) yielded indeterminate results by immunoblotting assay (with only the presence of a gp41 band) and were considered as negative after assessment of p24 Ag, HIV Abs (using the Roche Cobas® HIV Combi PT assay) and HIV-RNA PCR (data not shown).
Twenty-five (11.5%) plasma samples were positive for HBsAg and 140 (64.2%) were positive for anti-HBc Abs (Table 1).
Table 1. Hepatitis B surface antigen (HBsAg), anti-hepatitis B virus core (HBc) antibody, hepatitis C virus (HCV) and human immunodeficiency virus (HIV) prevalence rates and performance of Dried Blood Spot (DBS) as compared with plasma for HBV, HCV and HIV detection among patients attending voluntary counselling and testing clinics in Bobo-Dioulasso, Burkina Faso
Seven plasma samples were found to be positive for anti-HCV Abs. Immunoblotting confirmed the detection of anti-HCV Abs in plasma in five cases but it was negative for two plasma samples that had low HCV screening signal to cut-off ratio (OD 0.800 and 0.972, respectively, compared with the cut-off value established at 0.439). These two samples were considered as HCV negative because HCV RNA detection and a second HCV Abs EIA were negative. Two participants were co-infected by HIV and HBV, and one by HBV and HCV.
HIV rapid testing
Nineteen of 218 participants (8.7%) tested positive for HIV using the first screening RDT (Determine® HIV1/2) and all of these were confirmed to be positive by the second RDT (SD Bioline® HIV1/2 3.0). One of the HIV-positive participants was positive for both HIV-1 and HIV-2 Abs using the latter RDT.
Use of DBS for HIV screening and immunoblot confirmation
HIV testing on DBS specimens using Genscreen™ ULTRA HIV Ag-Ab assay and immunoblotting confirmed all the HIV infections detected by RDTs. As illustrated in Fig. 2a, a clear difference in ELISA results was observed between HIV-positive and HIV-negative DBS samples (mean plus SD of the OD 3.443 ± 0.430 and 0.131 ± 0.030, respectively; p <0.0001).
Clearly visible bands were obtained by DBS immunoblotting (Fig. 3a) and they were similar to plasma immunoblotting in terms of intensity and the number of visible bands. Immunoblotting on both DBS and plasma for the one result found to be dually positive for HIV-1 and HIV-2 by HIV RDT (SD Bioline®), could not be confirmed because no staining was observed on the HIV-2 bands (gp105 and gp36).
HCV screening and confirmation on DBS
All positive plasma samples for HCV were positive by DBS testing (five samples, 2.3%) and all HCV-negative samples were also accurately classified by the DBS assay (213 samples, 97.3%) (Fig. 2b). Positive HCV status by DBS was confirmed by immunoblotting and was similar to plasma in terms of intensity and number of visible bands (Fig. 3b).
Detection of HBV infection and anti-HBc status on DBS
Compared with plasma, 24/218 (11.0%) DBS samples were appropriately classified as positive for HBsAg and 193/218 (88.5%) as negative. The signal to cut-off ratio was ≥3 in most of the cases, except for three DBS samples (Fig. 2c). HBsAg quantification in plasma indicated that these three participants had low HBsAg concentrations (0.53, 2.58 and 2.86 log10 IU/mL). One sample found to be positive for HBsAg in plasma was negative using DBS (OD 0.268 in DBS versus 1.406 in paired plasma); however, the plasma HBsAg level was especially low in this sample (0.59 log10 IU/mL, data not shown).
For anti-HBc Abs, 139 (63.8%) DBS and paired plasma samples were positive and 77 (35.2%) were negative. Anti-HBc Abs were detected in DBS samples for all HBsAg-positive participants. However, 10th centiles of OD for anti-HBc Ab-positive samples and 90th centiles of OD for anti-HBc Ab-negative samples tended to overlap, allowing for the misclassification of 9.2% (20/218) of samples (Fig. 2d). One (0.5%) false-negative result (DBS OD = 0.072 versus 0.484 for plasma) and one (0.5%) false-positive result (DBS OD = 0.652 versus 0.059 in plasma) were observed (data not shown) when compared with the anti-HBc Abs cut-off value estimated at 0.075 in DBS.
Cost and analytical performances of the DBS assays for HBV, HCV and HIV testing
Sensitivity, specificity, positive predictive value, negative predictive value and kappa coefficient of DBS assays in comparison with plasma assays for HBsAg, anti-HBc, HCV and HIV detection are summarized in Table 1. The cost of the consumables for HIV, HCV and HBV combined testing based on capillary blood collection onto filter-paper card, including the HIV RDTs, HIV Ag-Ab assay, HBsAg assay, anti-HBc Abs assay and anti-HCV Ab assay was estimated to be €12.00 (€1.24 for the lancet and filter-paper card, €1.56 for other consumables, €1 for the HIV RDTs, €8.20 for reagents of the four ELISAs with €4.48 for HCV ELISA reagent). The cost of HIV and HCV confirmation tests by immunoblot was estimated to be €30 each.
Loss to follow-up of participants between their first and second post-test counselling sessions
All 218 participants included in this study returned for follow up at the CADI to receive results on their HCV and HBV infection status, as well as counselling once a week after enrolment.
This field study conducted in West Africa showed that DBS specimens may be successfully used in a two-step strategy based on simple algorithms and commercial assays to confirm HIV RDTs and to complete screening for HBV and HCV infections.
Blood collection on DBS can be performed with relative ease in parallel with HIV rapid testing using a finger-prick in adults and older children or a heel-prick in neonates and infants. In this study, DBS was used to confirm HIV RDT results using fourth-generation ELISAs and immunoblotting. The positive predictive value of RDTs has recently been estimated as 98.0% (96.4–99.1%) in settings with a low prevalence of HIV . Hence, in low-incidence countries such as Burkina Faso, where HIV prevalence is estimated to be between 1.0 and 1.5% , false-positive results constitute a significant proportion of the positive results found with HIV screening tests. Previous studies performed in African pregnant women using HIV RDTs reported that >1% of the results were indeterminate [16, 17]. Spiking venous blood onto DBS filter cards to perform confirmatory tests may be especially interesting in such cases.
In countries where HIV-2 infection is prevalent, positive samples should be tested by confirmatory assays that are able to differentiate HIV-1 from HIV-2 Abs . In our study one participant tested positive for both HIV-1 and HIV-2 Abs by the SD Bioline assay. This result remained unconfirmed by the immunoblotting, suggesting a possible non-specific cross-reaction with anti-HIV-1 Abs on the RDT. Immunoblotting using assays such as the Inno-Lia HIV-1/HIV-2 assay may also provide information on recent infection (up to three bands), or on advanced disease (loss of the p24 Ag). When comparing paired bands obtained from DBS and plasma, results showed a good concordance. This supports the use of DBS in reference laboratories for quality control of HIV results obtained by point-of-care testing performed in voluntary counselling and testing centres, and for the discrimination between HIV-1 and HIV-2 infections before the initiation of antiretroviral therapy. HIV diagnostic strategies including fourth-generation ELISA combining HIV Ag/Abs detection on DBS  may also improve the diagnosis of primary HIV infection. Evidence supports the fact that primary HIV infection is indeed frequently missed during the seroconversion phase by the HIV RDTs  even when using a p24/Abs combined RDT . The prevalence rate of HIV in our study is similar to those observed in 2010 and 2011 in the CADI (8.5% and 6.5%, respectively, personal data). However, it is not representative of the estimated nationwide prevalence in Burkina Faso (around 1%) , suggesting that participants undergoing voluntary counselling and testing tend to be at higher risk for HIV infection.
Using commercial EIA tests, our study shows that assays on DBS were able to successfully detect HBsAg in HBV-infected participants. As expected, the lower limit of detection of antibodies and antigen is higher using DBS than plasma  and one individual with low HBsAg concentration was found to be negative for HBsAg on DBS but positive on plasma in this study. Data on HBsAg concentrations in participants chronically infected with HBV demonstrate that the level of HBsAg below 100 IU/mL is unusual [22-24], suggesting that DBS may be sufficient for HBsAg detection in the vast majority of cases.
Overall, the ODs of DBS samples from participants reactive for HIV, HCV Abs or HBsAg did not overlap with ODs of DBS samples from negative individuals. In contrast, detection of anti-HBc Abs was less obvious with the lowest ratio of the anti-HBc Abs positive/negative OD. Villar et al.  recently reported a relatively low sensitivity (90.5%) for anti-HBc antibody detection using DBS. This may be partially explained by the suboptimal intrinsic performance of competitive immunoassays aimed to detect anti-HBc Abs . Regarding the population size in our study, the total number of positive samples was low for HIV and HBV and least for HCV which may impact on the accuracy of the assays' performance observed in our study. However, the good analytical performance of DBS assays obtained in our study is similar to those previously reported for HCV  and HBV [21, 26] screening using DBS.
Testing for HIV, HBV and HCV using DBS samples may not necessarily be less expensive than classical serology, but in developing countries DBS testing may be much more feasible to perform compared with classical serology requiring venous blood sampling, which requires infrastructural support (electrical power, centrifuge, etc.), and storage and transportation requirements (cooling chain, ice box, dry ice, space consumption, etc.). Besides voluntary counselling and testing centres, our algorithm may be useful in antenatal clinics by adding HBV and HCV screening to the HIV DNA or RNA testing performed in newborns, and particularly useful for HBsAg screening in pregnant women to optimize the prevention of mother-to-child transmission through early vaccination of neonates.
The costs associated with antiviral therapy for chronic HBV and HCV infections as well as the lack of laboratory facilities for diagnosis and monitoring are some of the main limitations for treatment of chronic viral hepatitis in resource-limited settings. As a consequence, hepatocellular carcinoma is one of the three most commonly described tumours in sub-Saharan Africa .
This survey conducted in field conditions of a resource-limited setting demonstrates the feasibility of screening and confirming HBV, HCV and HIV status using a two-step strategy to combine RDTs plus DBS testing. Viral testing on DBS presents a particular benefit for many developing countries in sub-Saharan Africa, like Burkina Faso, where environmental and logistical conditions are not always favourable for the timely transportation and storage of samples such as plasma or serum that have to be obtained by venous puncture. The DBS can be collected in remote rural areas, sent at room temperature to a central reference laboratory, and results communicated using mobile phone texting as recently reported [28, 29]. DBS diagnosis can also assist in evaluating the implementation of a national programme for HBV and HCV screening and care in middle-income and low-income countries.
This work was financially and technically supported by the Inserm Unit 1058 and the Virology Laboratory of Lapeyronie University Hospital, Montpellier France. We thank all the participants for volunteering to participate and the staff of CADI for their efforts in achieving this study. Thanks to Dr Hama A. Diallo (Centre MURAZ, Burkina Faso) for his assistance with the data analysis and Dr Andrea Low (London School of Hygiene and Tropical Medicine), Cara-Chan Manville (Inserm Unit 1058) and Johannes Viljoen (Africa Centre, Durban, South Africa) for English proofreading. We also acknowledge the Service de Coopération et d'Action Culturelle of the French Embassy in Burkina Faso for facilitating accommodation in Montpellier, France. Finally, we thank the Agence Nationale de Recherches sur le Sida et les Hépatites Virales (ANRS) site in Burkina Faso for its constant encouragement and support.