- Email: [email protected]
Specificity of two HIV screening tests detecting simultaneously HIV-1 p24 antigen and antibodies to HIV-1 and −2
Journal of Virological Methods 249 (2017) 143–146
Contents lists available at ScienceDirect
Journal of Virological Methods journal homepage: www.elsevier.com/locate/jviromet
Specificity of two HIV screening tests detecting simultaneously HIV-1 p24 antigen and antibodies to HIV-1 and −2
MARK
⁎
Annette Blaicha, , Andreas Buserb, Marcel Stöcklee, Christian Gehringera,d, Hans H. Hirschc,e, Manuel Battegaye, Thomas Klimkaitc, Reno Freia a
Division of Clinical Microbiology, University Hospital Basel, University of Basel, Petersgraben 4, 4031 Basel, Switzerland Blood Transfusion Centre, Swiss Red Cross, Hebelstrasse 10, 4056 Basel, Switzerland c Division of Infection Diagnostics, Department of Biomedicine (Haus Petersplatz), University of Basel, Petersplatz 10, 4009 Basel, Switzerland d Division of Internal Medicine, University Hospital Basel, University of Basel, Petersgraben 4, 4031 Basel, Switzerland e Division of Infectious Diseases and Hospital Epidemiology, University Hospital Basel, University of Basel, Petersgraben 4, 4031 Basel, Switzerland b
A R T I C L E I N F O
A B S T R A C T
Keywords: Fourth generation immunoassays HIV Specificity False reactive results
This study aimed at assessing the specificity of the Elecsys® HIV combi PT in comparison to the ARCHITECT® HIV Ag/Ab Combo. With both of these assays, 3997 unselected sera from patients of a tertiary health care centre in Basel, Switzerland, were screened for HIV. Reactive sera were reanalysed on the VIDAS® HIV Duo Ultra to identify false-reactive specimens prior to confirmation by quantitative PCR and line immunoassay. The Elecsys® compared to the ARCHITECT® shows a similar specificity (99.7% versus 99.8%) but a slightly lower positive predictive value (71.8% versus 80%). Samples tested with a cut-off index (COI) between 0.91 and 4.85 (cutoff < 0.9) with the Elecsys® and with a signal to cut-off index (S/CO) between 1.09 and 12.49 (cut-off < 1.0) with the ARCHITECT® were false–reactive. There was no false-reactive result with the VIDAS®. Of the falsereactive samples, 66.7% could be related to patient-specific underlying conditions. The HIV two-tiered diagnostic algorithm proposed in this work improved the positive predictive values of the Elecsys® or ARCHITECT® to 100% when the results of the VIDAS® were included. Values just above the cut-off are highly suspicious to be false-reactive and high COI or S/CO ratios are associated with true positivity.
In settings of low HIV prevalence, it is crucial that a HIV screening test is not only highly sensitive, but also specific. HIV screening tests with limited specificity are prone to false-reactive results causing unnecessary concerns in individuals and raise of healthcare costs. The former Elecsys® HIV combi assay had been reported to have a relatively low specificity of 99% and in our low prevalence population a positive predictive value (PPV) of only 55% (Song et al., 2012; Blaich et al., 2011). This assay has been in the meantime replaced by the Elecsys® HIV combi PT assay and differs from its predecessor by a pre-treatment step. Thereby, HIV virions are lysed and antigen is released. Thus, sensitivity for detecting HIV-1 p24 antigen is improved and allows earlier detection of a primary HIV infection (Song et al., 2012). Additionally, the new Elecsys® assay has an improved specificity as well (Song et al., 2012; Mühlbacher et al., 2013; Liu et al., 2016). Therefore, we compared the specificity of the new Elecsys® assay on the cobas® platform with that of the ARCHITECT® HIV Ag/Ab Combo using a large number of unselected samples. In accordance with our routine algorithm, any samples with an initially reactive result, on either platform,
was retested with the VIDAS® HIV Duo Ultra, a semi-automated system that does not allow high − throughput screening (Fig. 1). Furthermore, we aimed at identifying patient-related conditions that might be associated with false-reactive results of the assays. The serum samples were mainly obtained from inpatients and outpatients of the University Hospital Basel, a 900-bed tertiary centre, during a one-year period. The samples had been submitted for HIV screening by the treating physicians. Screening was performed with the Elecsys® HIV combi PT assay on the cobas e601 platform (Roche Diagnostics, Penzberg, Germany). After storage at 4–8 °C for 1-5 days, sera were investigated with the ARCHITECT® HIV Ag/Ab Combo assay on the ARCHITECT i2000 analyser (Abbott AG, Diagnostic Division, Baar, Switzerland) in the Blood Transfusion Centre Basel. The Elecsys® HIV combi PT and the ARCHITECT® HIV Ag/Ab Combo are “4th generation assays” for simultaneous detection of HIV-1 antigen and antibodies against HIV-1 and HIV-2). Sera showing reactive results with any screening assay were subsequently subjected to the VIDAS® HIV Duo Ultra assay on the VIDAS® analyser (bioMérieux, Marcy l'Etoile, France).
⁎
Corresponding author. E-mail addresses: [email protected] (A. Blaich), [email protected] (A. Buser), [email protected] (M. Stöckle), [email protected] (C. Gehringer), [email protected] (H.H. Hirsch), [email protected] (M. Battegay), [email protected] (T. Klimkait), [email protected] (R. Frei). http://dx.doi.org/10.1016/j.jviromet.2017.09.005 Received 27 June 2017; Received in revised form 7 September 2017; Accepted 8 September 2017 Available online 08 September 2017 0166-0934/ © 2017 Elsevier B.V. All rights reserved.
Journal of Virological Methods 249 (2017) 143–146
A. Blaich et al.
Fig. 1. Testing algorithm of the study. Abbreviations: qPCR − quantitative polymerase chain reaction; LIA − line immunoassay.
but a negative result in LIA and qPCR. All assays were performed according to the manufacturers’ instructions. Binary diagnostic test analyses were performed in R 3.3.0 (R Core Team, 2016, Vienna, Austria) using the DTComPair package (Stock and Hielscher, 2014, Heidelberg, Germany) in a paired study design. Specificity was compared using McNemar’s test in R 3.3.0 (Table 2). For every sample giving a false-reactive screening result, we reviewed the corresponding patient files focusing on conditions previously described as related to HIV false-positivity, such as auto-reactive antibodies, various bacterial and viral infections, rheumatoid factor, C-reactive protein > 3.0 mg/L, recent exposure to certain vaccines, pregnancy and history of multiple pregnancies, Stevens-Johnson syndrome, severe or alcoholic hepatic disease, blood transfusion, immunoglobulin infusion, or monoclonal gammopathy (Song et al., 2012; Mühlbacher et al., 2013; Dewar et al., 2006). Approval from the Ethics
Our 6-year experience has shown that the algorithm involving the Vidas® assay as second screening method in our daily routine can distinguish preliminary false-reactive from true-positive screening results. Therefore, we maintained this algorithm in our study. The VIDAS® assay is a 4th generation assay as well. The main features of the screening tests of this study are briefly summarized in Table 1. All of the tests have been described in detail in studies of Kwon et al. (2006), Mühlbacher et al. (2013) and Song et al. (2012). Reactive screening results were confirmed by a line immunoassay (LIA, INNO-LIA HIV I/II, Innogenetics, Ghent, Belgium) and by a quantitative PCR (qPCR, COBAS® AmpliPrep/TaqMan® HIV-1 Test, version 2.0, Roche Diagnostics, Rotkreuz, Switzerland), preferably from a 2nd blood sample (EDTA plasma). Samples were considered HIVpositive when the LIA and/or qPCR were positive. A false-positive result was defined by a reactive result in at least one of the screening assays, 144
Journal of Virological Methods 249 (2017) 143–146 Table 2 Results of 3997 samples submitted for detection of HIV infection.
Abbreviations: mAb = monoclonal antibody; Abs = antibodies; Ag = antigen; ECLIA = electro-chemiluminescence immunoassay; CMIA = chemiluminescent magnetic microparticle immunoassay; ELFA = enzyme-linked fluorescence assay.
bioMérieux/ Vidas HIV Duo Ultra
ELFA
Abbott/ Architect HIV combo Ag/Ab
CMIA
HIV-1/HIV-2 Abs detection: peptides and gp41 proteins of HIV-1 group M and O, HIV polymerase, and gp36 of HIV-2 HIV-1 Ag detection: anti-p24 mAb HIV-1/HIV-2 Abs detection: recombinant antigens and synthetic peptides derived from the HIV transmembrane gp41 proteins of HIV-1 group M and O and gp36 of HIV-2 HIV-1 Ag detection: anti-p24 mAb HIV-1/HIV-2 Abs detection: gp160 (HIV-1) protein, HIV-1 group O and HIV-2 specific synthetic peptides
provides separate results for Ag and Ab detection
HIV combi PT: HIV is lysed and Ag is released by a pretreatment step, for what the PT stands for HIV-1 Ag detection: anti-p24 mAb Roche/Cobas or Elecsys or Modular HIV combi/ HIV combi PT
ECLIA
Manufacturer of the test/device Name of the screening test
Table 1 Tests used in this study.
Principle
Targets
Special features
A. Blaich et al.
Parameter
Elecsys® HIV combi PT
ARCHITECT® HIV Ag/Ab Combo
non-reactive false-reactive true-reactive total % specificity (95% CI) % PPV (95% CI)
3958 11 28 3997 99.7% (99.6–99.9%)
3962 7 28 3997 99.8% (99.7–100%)
71.8% (57.7–85.9%)
80.00% (66.8–93.3%)
p-value
0.346
Abbreviations: CI = confidence interval; PPV = positive predictive value.
Committee had been obtained (Study number: 325/12). A total of 3997 serum samples from 3639 patients were analysed, 2201 (55.1%) samples from 2032 (55.8%) women and 1796 (44.9%) samples from 1607 (44.2%) men. Nearly all (98.3%) samples were obtained from patients admitted to the University Hospital Basel. The median age of the patients was 36 years (IQR: 28–50). Eighteen patient sera from 18 patients showed a false-reactive result, 11 were reactive with the Elecsys® and 7 with the ARCHITECT® only (Fig. 1). There was no false-reactive result with the VIDAS®. Twenty-eight of 46 (60.9%) samples were reactive with all three HIV screening assays and were afterwards confirmed by qPCR and LIA as true HIV positive (Fig. 1). None of the patient sera tested false-reactive by both the Elecsys® and the ARCHITECT®. Among the false-reactive 18 samples, the number of women was approximately equal to the number of men (8 versus 10). In contrast, among the 28 confirmed infected patients, twice as many were male (19 versus 9). Furthermore, we could observe that samples tested with a cut-off index (COI) between 0.91 and 4.85 (cut-off < 0.9) with the Elecsys® were false–reactive and those with a COI of at least 84.25 were all true-positive. Samples tested with a signal to cut-off index (S/CO) between 1.09 and 12.49 (cut-off < 1.0) with the ARCHITECT® were false-reactive and with a S/CO ≥45.65 all true-positive. This is in accordance with the study of Chacón et al. who found that samples with a S/CO ≤10 were false-reactive and all with ≥50 true-positive with the ARCHITECT® (Chacón et al., 2017). Our data confirm that the Elecsys® HIV combi PT assay has an improved PPV (71.8%) and specificity (99.7%) compared to its predecessor. However, the Elecsys® HIV combi (PPV: 55%, specificity: 99%), is still showing a lower PPV and a similar specificity as the ARCHITECT® HIV Ag/Ab Combo assay (80% and 99.8%, respectively) which is in accordance with other reports (Tao et al., 2013; Song et al., 2012; Liu et al., 2015). Like in the present study, Song et al. reported a similar specificity of the ARCHITECT® compared to the Elecsys® on clinical samples. In contrast, others found a better specificity of the Elecsys® compared to the ARCHITECT® on routine samples and on samples with potentially interfering substances (Tao et al., 2013; Mühlbacher et al., 2013). However, the specificity of the ARCHITECT® was assessed on a smaller sample size in these studies (1004 and 2751 samples) compared to the 3997 samples in this work. The HIV twotiered diagnostic algorithm proposed in this work improved PPVs of the Elecsys® or ARCHITECT® to 100% when the results of the VIDAS® were included. Values just above the cut-off are highly suspicious to be falsereactive and high COI or S/CO ratios are associated with true positivity. Furthermore, in 12/18 (66.7%) of the false-reactive HIV results, we found patient-specific underlying conditions (Table 3). Most prominently, irrespective of the system, we identified a concurrent or previous herpes virus infection (in 5 out of 12 cases) or pregnancy (in 4 out of 12 cases). The detailed information about the underlying conditions is summarized in Table 3. In the remaining 6/18 (33.3%) of patients we did not identify any known or new conditions. These results support the algorithm to re-test all Elecsys®- or ARCHITECT®-reactive screening results with the VIDAS® HIV Ultra Duo screening assay. Our findings 145
Journal of Virological Methods 249 (2017) 143–146
A. Blaich et al.
screening results is most important. This is meaningful since in many regions of the world diagnosis and antiretroviral therapy are based solely on screening test positivity.
Table 3 Underlying conditions found for 12 out of 18 patients with HIV false-reactive results. Elecsys®
Underlying condition
Conflict of interest
Patient# 1
alcoholic and HCV induced liver cirrhosis anti-HSV IgG positive EBV reactivation hepatitis A/B and pneumococcal vaccine CRP 9 mg/L
Although Abbott Diagnostics provided free HIV Ag/Ab Combo kits for the study, none of the manufacturers/kit providers had any influence on design, conduct and analyses of this study.
2
pulmonary and lymph-node tuberculosis CRP 22.2 mg/L
Acknowledgements
3
urinary tract infection Escherichia coli anti-HSV IgG positive
4
primary EBV-infection CRP 29.1 mg/L
5
pregnant multiple previous pregnancies anti-CMV IgG positive anti-HSV IgG positive blood transfusion every 2–3 weeks
6
diabetes mellitus type 1 anti-GAD antibodies anti-IA-2 antibodies
7
pregnant
8
glaucoma and cataract surgery hypergammaglobulinemia
ARCHITECT® Patient # 9
We thank Claudia Döpfner and the team of the Blood Transfusion Centre, Tanja Volmut, Anne Köhn and Angelika Geschke from the Division of Clinical Chemistry, University Hospital Basel, for testing the samples. Angelika Aebli and Leona Balaj-Jones from the Division of Infection Diagnostics, Department of Biomedicine, University of Basel, are acknowledged for performing LIA and qPCR. We are grateful to Abbott Diagnostics (Baar, Switzerland, C. Kluge) for the supply of HIV Ag/Ab Combo kits in this study. References Blaich, A., Klimkait, T., Volmut, T., et al., 2011. Comparison of four HIV screening tests of the 4th generation. Annual Assembly of the Swiss Society of Clinical Chemistry & TriNational Congress of Laboratory Medicine, Zurich. (Poster). Chacón, L., Mateos, M.L., Holguín, A., 2017. Relevance of cutoff on a 4th generation ELISA performance in the false positive rate during HIV diagnostic in a low HIV prevalence setting. J. Clin. Virol. 92, 11–13. Dewar, R., Highbarger, H., Davey, R., et al., 2006. Principles and procedures of human immunodeficiency virus: serodiagnosis: cautionary remarks. In: Detrick, B., Hamilton, R.G., Folds, J.D. (Eds.), Manual of Molecular and Clinical Laboratory Immunology, 7th ed. ASM Press, Washington, DC 843–846. Kwon, J.A., Yoon, S.Y., Lee, C.K., et al., 2006. Performance evaluation of three automated human immunodeficiency virus antigen-antibody combination immunoassays. J. Virol. Methods 11, 20–26. Liu, Y., Li, D., Wang, T., et al., 2015. Clinical application evaluation of two fourth-generation human immunodeficiency virus (HIV) screening assays in West China Hospital. J. Clin. Lab. Anal. 29, 146–152. Mühlbacher, A., Schennach, H., van Helden, J., et al., 2013. Performance evaluation of a new fourth generation HIV combination antigen-antibody assay. Med. Microbiol. Immunol. 202 (1), 77–86. R Core Team, 2016. R: A Language and Environment for Statistical Computing. R Foundation for Statistical Computing, Vienna, Austria. https://www.R-project.org/. Song, E.Y., Hur, M., et al., 2012. Performances of four fourth-generation human immunodeficiency virus-1 screening assays. J. Med. Virol. 84, 1884–1888. Stock, C., Hielscher, T., 2014. DTComPair: Comparison of Binary Diagnostic Tests in a Paired Study Design. R Package Version 1.0.3. http://CRAN.R-project.org/ package=DTComPair. Tao, C.M., Cho, Y., Ng, K.P., et al., 2013. Validation of the Elecsys HIV combi PT assay for screening and reliable early detection of HIV-1 infection in Asia. J. Clin. Virol. 58, 221–222.
Underlying condition
multiple previous pregnancies paronychia
10
lues latens hemophilia
11
pregnant multiple previous pregnancies
12
pregnant multiple previous pregnancies anti-HSV IgG positive with recurrent HSV reactivations
Abbreviations: HCV = hepatitis C virus; CRP = C reactive protein; EBV = Epstein Barr virus; HSV = herpes simplex virus; CMV = cytomegalovirus; GAD = glutamic acid decarboxylase; IA = islet antigen.
indicate that newer HIV tests of the 4th generation are still impaired by false reactive results. This points to the fact that screening results must be interpreted in the clinical context and that confirmation of reactive
146
Contents lists available at ScienceDirect
Journal of Virological Methods journal homepage: www.elsevier.com/locate/jviromet
Specificity of two HIV screening tests detecting simultaneously HIV-1 p24 antigen and antibodies to HIV-1 and −2
MARK
⁎
Annette Blaicha, , Andreas Buserb, Marcel Stöcklee, Christian Gehringera,d, Hans H. Hirschc,e, Manuel Battegaye, Thomas Klimkaitc, Reno Freia a
Division of Clinical Microbiology, University Hospital Basel, University of Basel, Petersgraben 4, 4031 Basel, Switzerland Blood Transfusion Centre, Swiss Red Cross, Hebelstrasse 10, 4056 Basel, Switzerland c Division of Infection Diagnostics, Department of Biomedicine (Haus Petersplatz), University of Basel, Petersplatz 10, 4009 Basel, Switzerland d Division of Internal Medicine, University Hospital Basel, University of Basel, Petersgraben 4, 4031 Basel, Switzerland e Division of Infectious Diseases and Hospital Epidemiology, University Hospital Basel, University of Basel, Petersgraben 4, 4031 Basel, Switzerland b
A R T I C L E I N F O
A B S T R A C T
Keywords: Fourth generation immunoassays HIV Specificity False reactive results
This study aimed at assessing the specificity of the Elecsys® HIV combi PT in comparison to the ARCHITECT® HIV Ag/Ab Combo. With both of these assays, 3997 unselected sera from patients of a tertiary health care centre in Basel, Switzerland, were screened for HIV. Reactive sera were reanalysed on the VIDAS® HIV Duo Ultra to identify false-reactive specimens prior to confirmation by quantitative PCR and line immunoassay. The Elecsys® compared to the ARCHITECT® shows a similar specificity (99.7% versus 99.8%) but a slightly lower positive predictive value (71.8% versus 80%). Samples tested with a cut-off index (COI) between 0.91 and 4.85 (cutoff < 0.9) with the Elecsys® and with a signal to cut-off index (S/CO) between 1.09 and 12.49 (cut-off < 1.0) with the ARCHITECT® were false–reactive. There was no false-reactive result with the VIDAS®. Of the falsereactive samples, 66.7% could be related to patient-specific underlying conditions. The HIV two-tiered diagnostic algorithm proposed in this work improved the positive predictive values of the Elecsys® or ARCHITECT® to 100% when the results of the VIDAS® were included. Values just above the cut-off are highly suspicious to be false-reactive and high COI or S/CO ratios are associated with true positivity.
In settings of low HIV prevalence, it is crucial that a HIV screening test is not only highly sensitive, but also specific. HIV screening tests with limited specificity are prone to false-reactive results causing unnecessary concerns in individuals and raise of healthcare costs. The former Elecsys® HIV combi assay had been reported to have a relatively low specificity of 99% and in our low prevalence population a positive predictive value (PPV) of only 55% (Song et al., 2012; Blaich et al., 2011). This assay has been in the meantime replaced by the Elecsys® HIV combi PT assay and differs from its predecessor by a pre-treatment step. Thereby, HIV virions are lysed and antigen is released. Thus, sensitivity for detecting HIV-1 p24 antigen is improved and allows earlier detection of a primary HIV infection (Song et al., 2012). Additionally, the new Elecsys® assay has an improved specificity as well (Song et al., 2012; Mühlbacher et al., 2013; Liu et al., 2016). Therefore, we compared the specificity of the new Elecsys® assay on the cobas® platform with that of the ARCHITECT® HIV Ag/Ab Combo using a large number of unselected samples. In accordance with our routine algorithm, any samples with an initially reactive result, on either platform,
was retested with the VIDAS® HIV Duo Ultra, a semi-automated system that does not allow high − throughput screening (Fig. 1). Furthermore, we aimed at identifying patient-related conditions that might be associated with false-reactive results of the assays. The serum samples were mainly obtained from inpatients and outpatients of the University Hospital Basel, a 900-bed tertiary centre, during a one-year period. The samples had been submitted for HIV screening by the treating physicians. Screening was performed with the Elecsys® HIV combi PT assay on the cobas e601 platform (Roche Diagnostics, Penzberg, Germany). After storage at 4–8 °C for 1-5 days, sera were investigated with the ARCHITECT® HIV Ag/Ab Combo assay on the ARCHITECT i2000 analyser (Abbott AG, Diagnostic Division, Baar, Switzerland) in the Blood Transfusion Centre Basel. The Elecsys® HIV combi PT and the ARCHITECT® HIV Ag/Ab Combo are “4th generation assays” for simultaneous detection of HIV-1 antigen and antibodies against HIV-1 and HIV-2). Sera showing reactive results with any screening assay were subsequently subjected to the VIDAS® HIV Duo Ultra assay on the VIDAS® analyser (bioMérieux, Marcy l'Etoile, France).
⁎
Corresponding author. E-mail addresses: [email protected] (A. Blaich), [email protected] (A. Buser), [email protected] (M. Stöckle), [email protected] (C. Gehringer), [email protected] (H.H. Hirsch), [email protected] (M. Battegay), [email protected] (T. Klimkait), [email protected] (R. Frei). http://dx.doi.org/10.1016/j.jviromet.2017.09.005 Received 27 June 2017; Received in revised form 7 September 2017; Accepted 8 September 2017 Available online 08 September 2017 0166-0934/ © 2017 Elsevier B.V. All rights reserved.
Journal of Virological Methods 249 (2017) 143–146
A. Blaich et al.
Fig. 1. Testing algorithm of the study. Abbreviations: qPCR − quantitative polymerase chain reaction; LIA − line immunoassay.
but a negative result in LIA and qPCR. All assays were performed according to the manufacturers’ instructions. Binary diagnostic test analyses were performed in R 3.3.0 (R Core Team, 2016, Vienna, Austria) using the DTComPair package (Stock and Hielscher, 2014, Heidelberg, Germany) in a paired study design. Specificity was compared using McNemar’s test in R 3.3.0 (Table 2). For every sample giving a false-reactive screening result, we reviewed the corresponding patient files focusing on conditions previously described as related to HIV false-positivity, such as auto-reactive antibodies, various bacterial and viral infections, rheumatoid factor, C-reactive protein > 3.0 mg/L, recent exposure to certain vaccines, pregnancy and history of multiple pregnancies, Stevens-Johnson syndrome, severe or alcoholic hepatic disease, blood transfusion, immunoglobulin infusion, or monoclonal gammopathy (Song et al., 2012; Mühlbacher et al., 2013; Dewar et al., 2006). Approval from the Ethics
Our 6-year experience has shown that the algorithm involving the Vidas® assay as second screening method in our daily routine can distinguish preliminary false-reactive from true-positive screening results. Therefore, we maintained this algorithm in our study. The VIDAS® assay is a 4th generation assay as well. The main features of the screening tests of this study are briefly summarized in Table 1. All of the tests have been described in detail in studies of Kwon et al. (2006), Mühlbacher et al. (2013) and Song et al. (2012). Reactive screening results were confirmed by a line immunoassay (LIA, INNO-LIA HIV I/II, Innogenetics, Ghent, Belgium) and by a quantitative PCR (qPCR, COBAS® AmpliPrep/TaqMan® HIV-1 Test, version 2.0, Roche Diagnostics, Rotkreuz, Switzerland), preferably from a 2nd blood sample (EDTA plasma). Samples were considered HIVpositive when the LIA and/or qPCR were positive. A false-positive result was defined by a reactive result in at least one of the screening assays, 144
Journal of Virological Methods 249 (2017) 143–146 Table 2 Results of 3997 samples submitted for detection of HIV infection.
Abbreviations: mAb = monoclonal antibody; Abs = antibodies; Ag = antigen; ECLIA = electro-chemiluminescence immunoassay; CMIA = chemiluminescent magnetic microparticle immunoassay; ELFA = enzyme-linked fluorescence assay.
bioMérieux/ Vidas HIV Duo Ultra
ELFA
Abbott/ Architect HIV combo Ag/Ab
CMIA
HIV-1/HIV-2 Abs detection: peptides and gp41 proteins of HIV-1 group M and O, HIV polymerase, and gp36 of HIV-2 HIV-1 Ag detection: anti-p24 mAb HIV-1/HIV-2 Abs detection: recombinant antigens and synthetic peptides derived from the HIV transmembrane gp41 proteins of HIV-1 group M and O and gp36 of HIV-2 HIV-1 Ag detection: anti-p24 mAb HIV-1/HIV-2 Abs detection: gp160 (HIV-1) protein, HIV-1 group O and HIV-2 specific synthetic peptides
provides separate results for Ag and Ab detection
HIV combi PT: HIV is lysed and Ag is released by a pretreatment step, for what the PT stands for HIV-1 Ag detection: anti-p24 mAb Roche/Cobas or Elecsys or Modular HIV combi/ HIV combi PT
ECLIA
Manufacturer of the test/device Name of the screening test
Table 1 Tests used in this study.
Principle
Targets
Special features
A. Blaich et al.
Parameter
Elecsys® HIV combi PT
ARCHITECT® HIV Ag/Ab Combo
non-reactive false-reactive true-reactive total % specificity (95% CI) % PPV (95% CI)
3958 11 28 3997 99.7% (99.6–99.9%)
3962 7 28 3997 99.8% (99.7–100%)
71.8% (57.7–85.9%)
80.00% (66.8–93.3%)
p-value
0.346
Abbreviations: CI = confidence interval; PPV = positive predictive value.
Committee had been obtained (Study number: 325/12). A total of 3997 serum samples from 3639 patients were analysed, 2201 (55.1%) samples from 2032 (55.8%) women and 1796 (44.9%) samples from 1607 (44.2%) men. Nearly all (98.3%) samples were obtained from patients admitted to the University Hospital Basel. The median age of the patients was 36 years (IQR: 28–50). Eighteen patient sera from 18 patients showed a false-reactive result, 11 were reactive with the Elecsys® and 7 with the ARCHITECT® only (Fig. 1). There was no false-reactive result with the VIDAS®. Twenty-eight of 46 (60.9%) samples were reactive with all three HIV screening assays and were afterwards confirmed by qPCR and LIA as true HIV positive (Fig. 1). None of the patient sera tested false-reactive by both the Elecsys® and the ARCHITECT®. Among the false-reactive 18 samples, the number of women was approximately equal to the number of men (8 versus 10). In contrast, among the 28 confirmed infected patients, twice as many were male (19 versus 9). Furthermore, we could observe that samples tested with a cut-off index (COI) between 0.91 and 4.85 (cut-off < 0.9) with the Elecsys® were false–reactive and those with a COI of at least 84.25 were all true-positive. Samples tested with a signal to cut-off index (S/CO) between 1.09 and 12.49 (cut-off < 1.0) with the ARCHITECT® were false-reactive and with a S/CO ≥45.65 all true-positive. This is in accordance with the study of Chacón et al. who found that samples with a S/CO ≤10 were false-reactive and all with ≥50 true-positive with the ARCHITECT® (Chacón et al., 2017). Our data confirm that the Elecsys® HIV combi PT assay has an improved PPV (71.8%) and specificity (99.7%) compared to its predecessor. However, the Elecsys® HIV combi (PPV: 55%, specificity: 99%), is still showing a lower PPV and a similar specificity as the ARCHITECT® HIV Ag/Ab Combo assay (80% and 99.8%, respectively) which is in accordance with other reports (Tao et al., 2013; Song et al., 2012; Liu et al., 2015). Like in the present study, Song et al. reported a similar specificity of the ARCHITECT® compared to the Elecsys® on clinical samples. In contrast, others found a better specificity of the Elecsys® compared to the ARCHITECT® on routine samples and on samples with potentially interfering substances (Tao et al., 2013; Mühlbacher et al., 2013). However, the specificity of the ARCHITECT® was assessed on a smaller sample size in these studies (1004 and 2751 samples) compared to the 3997 samples in this work. The HIV twotiered diagnostic algorithm proposed in this work improved PPVs of the Elecsys® or ARCHITECT® to 100% when the results of the VIDAS® were included. Values just above the cut-off are highly suspicious to be falsereactive and high COI or S/CO ratios are associated with true positivity. Furthermore, in 12/18 (66.7%) of the false-reactive HIV results, we found patient-specific underlying conditions (Table 3). Most prominently, irrespective of the system, we identified a concurrent or previous herpes virus infection (in 5 out of 12 cases) or pregnancy (in 4 out of 12 cases). The detailed information about the underlying conditions is summarized in Table 3. In the remaining 6/18 (33.3%) of patients we did not identify any known or new conditions. These results support the algorithm to re-test all Elecsys®- or ARCHITECT®-reactive screening results with the VIDAS® HIV Ultra Duo screening assay. Our findings 145
Journal of Virological Methods 249 (2017) 143–146
A. Blaich et al.
screening results is most important. This is meaningful since in many regions of the world diagnosis and antiretroviral therapy are based solely on screening test positivity.
Table 3 Underlying conditions found for 12 out of 18 patients with HIV false-reactive results. Elecsys®
Underlying condition
Conflict of interest
Patient# 1
alcoholic and HCV induced liver cirrhosis anti-HSV IgG positive EBV reactivation hepatitis A/B and pneumococcal vaccine CRP 9 mg/L
Although Abbott Diagnostics provided free HIV Ag/Ab Combo kits for the study, none of the manufacturers/kit providers had any influence on design, conduct and analyses of this study.
2
pulmonary and lymph-node tuberculosis CRP 22.2 mg/L
Acknowledgements
3
urinary tract infection Escherichia coli anti-HSV IgG positive
4
primary EBV-infection CRP 29.1 mg/L
5
pregnant multiple previous pregnancies anti-CMV IgG positive anti-HSV IgG positive blood transfusion every 2–3 weeks
6
diabetes mellitus type 1 anti-GAD antibodies anti-IA-2 antibodies
7
pregnant
8
glaucoma and cataract surgery hypergammaglobulinemia
ARCHITECT® Patient # 9
We thank Claudia Döpfner and the team of the Blood Transfusion Centre, Tanja Volmut, Anne Köhn and Angelika Geschke from the Division of Clinical Chemistry, University Hospital Basel, for testing the samples. Angelika Aebli and Leona Balaj-Jones from the Division of Infection Diagnostics, Department of Biomedicine, University of Basel, are acknowledged for performing LIA and qPCR. We are grateful to Abbott Diagnostics (Baar, Switzerland, C. Kluge) for the supply of HIV Ag/Ab Combo kits in this study. References Blaich, A., Klimkait, T., Volmut, T., et al., 2011. Comparison of four HIV screening tests of the 4th generation. Annual Assembly of the Swiss Society of Clinical Chemistry & TriNational Congress of Laboratory Medicine, Zurich. (Poster). Chacón, L., Mateos, M.L., Holguín, A., 2017. Relevance of cutoff on a 4th generation ELISA performance in the false positive rate during HIV diagnostic in a low HIV prevalence setting. J. Clin. Virol. 92, 11–13. Dewar, R., Highbarger, H., Davey, R., et al., 2006. Principles and procedures of human immunodeficiency virus: serodiagnosis: cautionary remarks. In: Detrick, B., Hamilton, R.G., Folds, J.D. (Eds.), Manual of Molecular and Clinical Laboratory Immunology, 7th ed. ASM Press, Washington, DC 843–846. Kwon, J.A., Yoon, S.Y., Lee, C.K., et al., 2006. Performance evaluation of three automated human immunodeficiency virus antigen-antibody combination immunoassays. J. Virol. Methods 11, 20–26. Liu, Y., Li, D., Wang, T., et al., 2015. Clinical application evaluation of two fourth-generation human immunodeficiency virus (HIV) screening assays in West China Hospital. J. Clin. Lab. Anal. 29, 146–152. Mühlbacher, A., Schennach, H., van Helden, J., et al., 2013. Performance evaluation of a new fourth generation HIV combination antigen-antibody assay. Med. Microbiol. Immunol. 202 (1), 77–86. R Core Team, 2016. R: A Language and Environment for Statistical Computing. R Foundation for Statistical Computing, Vienna, Austria. https://www.R-project.org/. Song, E.Y., Hur, M., et al., 2012. Performances of four fourth-generation human immunodeficiency virus-1 screening assays. J. Med. Virol. 84, 1884–1888. Stock, C., Hielscher, T., 2014. DTComPair: Comparison of Binary Diagnostic Tests in a Paired Study Design. R Package Version 1.0.3. http://CRAN.R-project.org/ package=DTComPair. Tao, C.M., Cho, Y., Ng, K.P., et al., 2013. Validation of the Elecsys HIV combi PT assay for screening and reliable early detection of HIV-1 infection in Asia. J. Clin. Virol. 58, 221–222.
Underlying condition
multiple previous pregnancies paronychia
10
lues latens hemophilia
11
pregnant multiple previous pregnancies
12
pregnant multiple previous pregnancies anti-HSV IgG positive with recurrent HSV reactivations
Abbreviations: HCV = hepatitis C virus; CRP = C reactive protein; EBV = Epstein Barr virus; HSV = herpes simplex virus; CMV = cytomegalovirus; GAD = glutamic acid decarboxylase; IA = islet antigen.
indicate that newer HIV tests of the 4th generation are still impaired by false reactive results. This points to the fact that screening results must be interpreted in the clinical context and that confirmation of reactive
146