Laboratory evaluation of four HIV/syphilis rapid diagnostic tests

Background Sexually transmitted infections, such as HIV and syphilis, are one of the major health care problems worldwide, especially in low- and middle income countries. HIV screening programmes have been widely used for many years. The introduction of rapid point-of-care tests (RDTs) that can detect both HIV and syphilis, using one single blood specimen, would be a promising tool to integrate the detection of syphilis into HIV programmes and so improve the accessibility of syphilis testing and treatment. Methods As part of the World Health Organization pre-qualification of in vitro diagnostics assessment, the laboratory performance of four dual HIV-Syphilis rapid diagnostic tests (SD Bioline HIV/Syphilis Duo, DPP HIV-Syphilis Assay, Multiplo Rapid TP/HIV Antibody Test and Insti Multiplex HIV-1/HIV-2/Syphilis Antibody Test) was assessed using a well characterized multiregional panel of stored sera specimens. Results In total 400 specimens were tested with each assay, resulting in excellent sensitivities and specificities for HIV, ranging from 99.5 to 100% and from 93.5 to 99.5%, respectively. Results obtained for the Treponema pallidum antibodies were lower, with the lowest sensitivity of 73.5% for Multiplo and the highest of 87% for SD Bioline. Specificities ranged from 99.0 to 100%. Conclusion Although these results suggest that the tests could further improve in accuracy in detection of treponemal antibodies, their introduction into screening programmes to increase the accessibility of HIV/Syphilis diagnosis and treatment for difficult to reach populations in the world is promising.


Background
Sexually transmitted infections (STIs) are one of the major health care problems worldwide, especially in resource-poor settings. HIV infects approximately 1.8 million people a year with 36.7 million people living with HIV at the end of 2016; and it is estimated that on a yearly basis there are 5.6 million new syphilis infections (http://www.who.int/mediacentre/factsheets/fs360/en/). In 2012, more than 900,000 pregnant women were infected with syphilis, which resulted in approximately 35,000 adverse birth outcomes (http://www.who.int/mediacentre/factsheets/ fs110/en/ and http://www.who.int/reproductivehealth/topics/ rtis/syphilis/pregnancy/en/). Untreated syphilis can result in serious adverse outcomes for pregnancy and can increase the risk of mother-to-child transmission of HIV [1]. As both HIV and syphilis are transmittable sexually and from mother to child, it is not surprising that co-infections are common [2].
Like most STIs, HIV and syphilis are often asymptomatic, which makes sensitive diagnostic testing particularly crucial for early detection and diagnosis, and for guidance of treatment and prevention of onward transmission. Diagnostic tests are available but often not accessible for populations living in the areas where they are needed the most, highlighting the importance of accessibility to simple and affordable tests, such as rapid diagnostic tests (RDTs). These are tests that can be used for testing at point of care, with no requirement for sophisticated laboratory equipment, or specific storage conditions; making them useful in all kinds of healthcare facilities. Since results are mostly available within 30 min, an accelerated linkage to treatment and care can be achieved. For many years, HIV screening programmes that use a validated testing algorithm of anti-HIV RDTs, have been widely used in low-and middle-income countries. Combining detection of anti-HIV and anti-treponemal antibodies in one dual RDT can integrate the detection of syphilis into HIV programmes and so improve the accessibility of syphilis testing and treatment.

Assays
As part of the World Health Organization (WHO) pre-qualification of in vitro diagnostics assessment, four dual HIV/Syphilis RDTs were evaluated between 2014 and 2016 by WHO at the Institute of Tropical Medicine (Antwerp, Belgium). SD Bioline HIV/Syphilis Duo (Standard Diagnostics, Republic of Korea, product number 06FK30; version instructions for use (IFU) 2013/05), DPP HIV-Syphilis Assay (Chembio Diagnostic Systems, United States, product number 659525; version IFU 10-6307-0Rev1) and Multiplo Rapid TP/ HIV Antibody Test (Medmira, Canada, product number 815311005145; version IFU MPSIPYZIS0002EN Rev3/ 1) were evaluated simultaneously. The evaluation of Insti Multiplex HIV-1/HIV-2/Syphilis Antibody Test (bioLytical Laboratories, Canada, product number 90-1032; version IFU 50-1143E) was performed at a later time point. The SD Bioline and the DPP assay are lateral flow (immunochromatographic) RDTs while Multiplo and INSTI are RDTs based on the flow through (immunofiltration) principle. All assays were performed by one operator (blinded to the reference results) according to the manufacturer's instructions for use (IFU). The test characteristics of the assays are described in Table 1.

Evaluation protocol
The evaluations were conducted according to the WHO pre-qualification protocol for performance evaluation. All personnel working on the evaluations were trained in performing and/or interpreting the assays. The results, recorded on standardized data collection sheets, were visually interpreted by the performer of the assay and independently by two other readers. Data entry into the standardized excel files was checked by a second person by visually comparing a print-out of the entered data with the raw data.
All specimens were tested in singular (initial testing). Specimens from the WHO specimen reference panel with indeterminate results (very faint doubtful shadow on the test line or discrepancies between the three readers) or results discrepant from the reference result were repeated in duplicate with the same lot (repeat testing) and (if possible) with the other lot. The result that occurred the most (at least two out of three) was recorded as the final result.
Results obtained with the assays under evaluation were compared to the results of the reference testing algorithm for HIV and Syphilis (described below).

WHO specimen reference panel
All four assays were evaluated using the same characterized evaluation panel of 400 serum/plasma specimens from European, African, Asian, South American and Australian origin collected from patients/study participants attending the Institute of Tropical Medicine (ITM) clinic, collaborating sexually transmitted infection (STI) clinics and blood donation centres. All specimens were assigned a unique identification code.
The panel consisted of 200 anti-HIV antibody positive specimens, of which 100 were anti-TP (Treponema pallidum) antibody positive and 100 were anti-TP antibody negative, and 200 anti-HIV antibody negative specimens, of which 100 were anti-TP antibody positive and 100 anti-TP antibody negative. Separate aliquots (approximate 200 μl) of each specimen were kept stored at − 20°C. Specimens were not thawed more than twice.
The WHO specimen reference panel was tested using the following HIV reference testing algorithms: Vironostika HIV Ag/Ab (bioMérieux, France), Enzygnost Anti-HIV 1/2 Plus (Siemens Healthcare Diagnostics, Germany) or Genscreen HIV-1/2 Version 2 (Bio-Rad, France) and INNO-LIA HIV I/II Score (Fujirebio Europe, Belgium). The Innotest HIV Antigen mAb (Fujirebio Europe, Belgium) was used to further characterize specimens in the event of discrepant results between the assay under evaluation and the reference result. The treponemal (TP) antibody status was determined by testing with the Vitros Syphilis TPA Assay (Ortho Clinical Diagnostics, USA), an automated Treponemal enzyme immunoassay (EIA), and Treponema pallidum passive particle agglutination (TPPA) (SERODIA-TP.PA, Fujirebio, Japan). Specimens with discrepant results to the reference result were further characterized using the BD Macro-Vue™ RPR Card Tests (Becton Dickinson, USA) for the determination of non-treponemal antibodies and/or recomLine Treponema IgM (MIKROGEN, Diagnostik, Germany) for the detection of treponemal IgM antibodies.

Commercial sera panels
The ability of the assays to detect various levels of HIV antibodies was tested by using eight HIV seroconversion panels (PRB914, PRB925, PRB926, PRB930, PRB955, PRB965, PRB968 and PRB 969; SeraCare Life Science Inc., USA), one anti-HIV mixed titre performance panel (PRB205; SeraCare Life Science Inc., USA) and the WHO international biological reference preparation panel (catalogue number 02/210; NIBSC). For Treponema pallidum (TP), one seroconversion panel (PSS901-1.2; SeraCare Life Sciences Inc., USA), one anti-TP mixed titre performance panel (PSS202 (M2); SeraCare, Life Sciences Inc., USA) and the WHO international biological reference preparation panel (catalogue number 05/122; NIBSC) were tested. All panel specimens were characterized using the same set of assays as described for the WHO specimen reference panel. After characterization the specimens were divided into smaller aliquots (approximately 200 μl) and stored at − 20°C or − 80°C. Specimens were not thawed more than twice.

Performance and operational characteristics/statistics
The sensitivity and specificity values with their 95% confidence intervals (95% CI) were determined in comparison with the reference results, using the exact binominal method. Additionally, the invalid rate and inter-reader variability were calculated for each assay (for the WHO specimen reference panel). The invalid rate was expressed as the number of invalid test results over the total number of tests used (percentage). For each test band the inter-reader variability was expressed as the percentage of specimens for which initial results were differently interpreted (i.e., reactive or non-reactive or indeterminate) by the independent readers. The operational characteristics, such as ease of use, number of steps and ease of interpretation, were assessed by the lab technician who performed the tests.
Statistical analyses (two-independent-samples-t-test, logistic regression and McNemar's test for paired proportions) were performed using R version 3.3.2.

WHO specimen reference panel
All 400 specimens of the WHO specimen reference panel were tested by the four RDTs. The control line was present for the 400 panel specimens on SD Bioline HIV/Syphilis Duo, DPP HIV-Syphilis Assay and Insti Multiplex HIV-1/HIV-2/Syphilis Antibody Test, indicating 100% valid results. However, for the Multiplo Rapid TP/HIV Antibody Test, invalid results were initially obtained for eight specimens, mainly due to non or incomplete absorption of the sample, and of these, two remained invalid after repeat testing (final invalid rate 2/400 = 0.5%).
As shown in Table 2, the inter-reader variability was higher on the TP component in comparison to the HIV component with SD Bioline HIV/Syphilis Duo, DPP HIV-Syphilis Assay and Multiplo Rapid TP/HIV Antibody Test: 4.0, 4.5 and 4.8% for TP versus 0, 1.0 and 0.3% for HIV, respectively. In contrast, the inter-reader variability was lower for the TP dot (1.8%) compared to the HIV dot (4.5%) for the Insti Multiplex HIV-1/ HIV-2/Syphilis Antibody Test.
The overall test performance characteristics per assay are summarized in Table 2. Initial sensitivity and specificity were calculated on the initial test results, and the final sensitivity and specificity on the combined results of the initial and repeat testing.  Table 2).

Anti-HIV detection component
In general, 92.0% of the specimens (368/400) were concordant with the reference result for the HIV antibodies in the four RDTs after first testing. All discordances (n = 32, in one or more RDTs) were in TP positive specimens. One HIV infection was missed in the presence of TP antibodies (Australian origin) and 31 specimens (all of African origin) were falsely reactive for HIV antibodies. Out of these, four specimens were misclassified in two or more assays (see Table 3). No specimens were misclassified for HIV in all four assays. Most specimens remained discrepant from the reference result in one or more assays after re-testing (final result versus initial report) Table 3.  Table 2).

Anti-TP detection component
After initial testing, 83.8% (335/400) of the results were in concordance with the Treponema pallidum antibodies reference results in all assays.
Four anti-TP negative specimens (4/200; one HIV negative and three HIV positive; all from South American origin), were false reactive or indeterminate for Treponema pallidum antibodies in at least one of the RDTs. Two remained discrepant from the reference result after repeat testing.   Independent of the HIV status, 30.5% (61/200) of the TP antibody positive samples were not detected by at least one of the assays after first testing, of which 42 were discrepant from the reference result (false non-reactive or indeterminate) in two or more assays (21%). Fourteen of these were false non-reactive on all four RDTs even after repeat testing, eight on three assays and 20 on two. Most remained false non-reactive after repeat testing (see Table 4). Comparing the TPA results of the specimens that were misclassified (false negative or indeterminate) on two or more RDTs with those that were scored correctly, showed a statistical significant difference between the two groups (p < 0.001, mean TPA values of 36 versus 297, respectively, p-value obtained by two independent samples t-test).

Commercial panels
The results obtained for the different panels are summarized in Table 5. Both for the HIV and the TP seroconversion panels, the SD Bioline assay was more sensitive in early detection of antibodies compared with the other three assays and compared to the reference assays.

Discussion
Introduction of RDTs in healthcare is extremely important as these tests may result in an accelerated linkage to care and treatment for many people. Advantages of RDTs over traditional laboratory-based in vitro diagnostics are their ability to be used in rural settings with limited laboratory access, their simplicity of execution and the shorter time to result. Dual HIV/Syphilis RDTs have the additional advantage of using the same specimen (serum/plasma or whole blood) and test device for testing the two infections simultaneously. Consequently, syphilis antibody screening can be easily added to the already existing HIV screening programmes without the need for extra blood sampling and extra waiting time for the result. This may be an important step forward in controlling HIV and syphilis infections in vulnerable risk groups and pregnant women. In this respect the WHO published a note in early 2017 in order to provide advice for countries planning to introduce the HIV/syphilis dual test in antenatal care settings.
This study presents data from four HIV/Syphilis RDTs, obtained after evaluation on the same, well characterized, specimen panel by experienced laboratory personnel in a WHO accredited testing laboratory with ISO 15189 and ISO 17025 accreditation. This makes the sensitivity and specificity calculations very comparable between the four assays. Sensitivities and specificities obtained for anti-HIV antibody detection (Table 2) are in line with results from earlier published laboratory evaluations for all four assays [3-7, 9, 11-14, 16-21]. For Treponema pallidum, however, the sensitivities obtained in our study were lower compared to earlier published data for all assays, while specificities were comparable. One study, by Fakile et al. [21], reported similar lower sensitivities and specificities for treponemal antibodies.
The sensitivities for Treponema pallidum detection of three of the four assays were higher than the minimal clinical sensitivity of 80% as set by the WHO in their product profile, but none of the assays achieved the optimal desired clinical sensitivity of 90%. For all four assays the Treponema pallidum specificities were superior to the optimal desired clinical specificity of 95%. Ref http:// www.who.int/reproductivehealth/POTC-TPPs-2016.pdf.
The different geographical origins of the study populations, as well as the stage of the syphilis infection at time of sampling, might explain these differences in sensitivities. The natural history of syphilis is complex in that the treponemal and non-treponemal antibody profile varies in the different syphilis stages. Treponemal antibodies (IgG and IgM) appear earlier than the non-treponemal antibodies, they remain detectable for life and do not protect against new infection. On the other hand, non-treponemal antibodies decrease and may disappear over time, especially after successful treatment. For SD Bioline and the Multiplo assay the IFU state that they detect IgG as well as IgM antibodies. Additionally, as per the IFU of INSTI, patients in the early primary stage of infection may test negative due to the test's lower affinity to IgM antibodies as compared to IgG. The IFU of the DPP assay does not specify the type of antibody but states that individuals with syphilis who are receiving antibacterial therapy may produce false negative results. For our study we made sure that the panel was composed of specimens from different geographic origins but no selection was made based on the syphilis stage.
Laboratory evaluations executed in an ideal situation give a good reflection of the test with optimal performance characteristics. On the other hand, this strength may also be a limitation because it may not reflect the situation in reality, when these RDTs will be performed in the field by health care workers using whole blood finger stick specimens. So far, data obtained from field evaluations are rather scarce. Bristow et al. [8,11] and Black et al. [13] found sensitivities and specificities for HIV antibodies for SD Bioline HIV/Syphilis Duo ranging from 98.8 to 99.2% and 97.0 to 100%, respectively, and for Treponema pallidum from 66.7 to 96.5% and 90.8 to 98.8%, respectively. In field settings, the performance characteristics are lower than those observed in laboratory settings, especially for the TP component (Table 6).
For the Treponema pallidum positive but HIV negative specimens, 31% were false reactive for HIV antibodies in at least one of the four evaluated assays. The       [5, 7, 12-14, 16, 17, 20, 21, 23, 24]; SD Bioline previous field evaluations: references [8,11,15]; DPP previous laboratory evaluations: references [5,[17][18][19]23]; Multiplo previous laboratory evaluations: references [5,6,17,23]; Multiplo previous field evaluations: reference [10]; Insti previous laboratory evaluation: reference [9] NA Not available a range from lowest obtained sensitivity or specificity to highest intensity of the HIV test line was much weaker as compared to true positive results. These weakly reactive bands may be a result of non-specific serological cross-reactivity as described in a review by Klarkowski et al., where possible causes of false reactivity on RDTs are discussed [23]. They emphasize that stimulation of immune activation (B-lymphocyte activation), which produces broad-spectrum antibodies, might be a significant cause of cross-reactivity and thus false reactive results. In the same review, no studies were described in which syphilis caused HIV false reactivity. The high rate of HIV false reactivity strengthens the need to ensure the quality of testing and underlines the fact that one reactive test result may not be considered as definitive. Any reactive test should always be confirmed by additional testing, as recommended by WHO [24]. Notably, 7% of the 200 anti-TP positive specimens were false non-reactive for TP antibodies in all four assays, all were TPA and TPPA positive and RPR negative, and another 28 were misclassified by two or three assays. This rather high misclassification rate, by two or more assays, may be the result of some characteristics of the setup of our evaluation. Firstly, the consistency of the specimens. Two of the four evaluated assays (Multiplo and INSTI) are flow through principle RDTs. Flow through (or vertical flow) immunoassays rely on the same basic principles as the more common lateral flow immunoassays (such as DPP and SD Bioline), with the flow of the fluid, vertically versus laterally, as the most obvious difference. Because they are based on vertical flow/absorption of the specimen, the assay's performance can be influenced by the composition of the specimens, especially when working with stored serum/ plasma, as was the case in our evaluation. As specified in the Multiplo's Instructions For Use (IFU), all specimens were centrifuged and only the clear supernatant was used on the device. Centrifugation was not described in the other IFU and was therefore not performed for INSTI, SD Bioline and DPP. However, approximately 85% of the specimens that were false non-reactive with INSTI (using un-centrifuged specimens) were also misclassified with Multiplo (using clear supernatant), we therefore suggest that the use of centrifuged and clear supernatant did not have an influence on the final results.
Secondly, the number of freeze/thaw cycles that a specimen had already passed through before being used in this evaluation could differ between the RDTs evaluated, and also between initial testing and eventual repeat testing. Castro et al. [25] found that 10 cycles appear to have a minimal effect on the sensitivity of IgG and IgM for serological testing. As all specimens in our panel were stored at − 20°C in small aliquots, no specimen exceeded more than 2 freeze/thaw cycles, which makes it likely that there was no loss of antibodies.
Another limitation of the current study is the low number of HIV or syphilis non-reactive specimens (200 for each) included in the evaluation. A larger sample size would have given the study results more power. However, we believe that this evaluation reflects the performance of the four assays. A table (Table 7) shows positive and negative predictive values at three different prevalence's for both markers HIV and Treponema. They are also part of each of the individual reports available for the public on the WHO website.
Rapid diagnostic tests are mainly developed to be used in the field, by healthcare workers. For that reason it is important that not only the performance characteristics are evaluated but also the operational characteristics. During this study, the operational aspects, such as ease of use, number of steps and ease of interpretation, were assessed by the lab technician who performed the tests ( Table 1). The Multiplo and INSTI assay are both vertical flow immunoassays and therefore need more steps to be performed before the result can be interpreted: the technician must be very alert. In our experience, in some cases the test bands/ dots were hardly visible, making interpretation difficult and therefore resulting in false reactive, false non-reactive or indeterminate results. Proper training is essential, not only for the vertical flow but also for the lateral flow RDTs, especially when working with non-lab personnel.

Conclusion
In conclusion, the results of this laboratory evaluation suggest that combined HIV/Syphilis rapid diagnostic tests are a useful tool for the detection of both infections using the same device, and can increase the accessibility of HIV/Syphilis diagnosis for difficult to reach populations in the world. However, further evaluations should be conducted to assess the feasibility and acceptability of such assays among healthcare workers in the field. The tests under investigation could further improve in accuracy, especially in the detection of treponemal antibodies, but are promising enough to be introduced into screening programmes.