- Email: [email protected]
Evaluation of the sensitivity and specificity of six HIV combined p24 antigen and antibody assays
Journal of Virological Methods 122 (2004) 185–194
Evaluation of the sensitivity and specificity of six HIV combined p24 antigen and antibody assays Thoai Duong Lya,∗ , Syria Lapercheb , Catherine Brennanc , Ana Vallaric , Anne Ebela , Jeffrey Huntc , Lynn Martinc , David Daghfalc , Gerald Schochetmanc , Sushil Devarec a
Laboratoire LCL, 78, Avenue de Verdun, 94200 Ivry sur Seine, France b National Institute of Blood Transfusion, Paris, France c Abbott Laboratories, Abbott Park, IL, USA
Received 28 June 2004; received in revised form 13 August 2004; accepted 16 August 2004 Available online 13 October 2004
Abstract In this study, we evaluated the performance of six HIV combined p24 antigen and antibody (Ag/Ab) assays versus two third-generation anti-HIV antibody assays. The assays were evaluated using p24 antigen panel of 31 HIV-1 subtypes (n = 124), 25 HIV-1 seroconversion panels (n = 176), HIV-1 antibody positive samples including group M subtypes and group O (n = 559), HIV-2 positive samples (n = 110), and unselected HIV negative samples from four French private laboratories (n = 1005). The results showed that overall HIV combined Ag/Ab assays present better performance, when compared to antibody-only assays. However, some differences were observed in the sensitivity of the six HIV combined Ag/Ab assays evaluated. The AxSYM and Murex Combo assays had the best sensitivity score in this study and reduced the window period by 2.0–2.35 days relative to antibody only assays and 1–2.17 days relative to the other combined Ag/Ab assays. Among combined HIV Ag/Ab assays, Genscreen Plus and AxSYM Combo presented the highest specificity, with 99.9% and 99.8%, respectively. © 2004 Elsevier B.V. All rights reserved. Keywords: Seroconversion; HIV combined antigen and antibody assay; p24 antigen
1. Introduction Since the introduction of HIV combined antigen (Ag) and antibody (Ab) screening assays (referred to as fourthgeneration assays) by the end of 1997 (Weber et al., 1998; G¨urtler et al., 1998; Van Binsbergen et al., 1998), it is estimated that these assays reduce the seroconversion window period by a few days to as much as two weeks in comparison with third-generation antibody assays (Courouc´e et al., 1999; Brust et al., 2000; Ly et al., 2001a). Despite their increased sensitivity, numerous reports indicate that the specificity of HIV Ag/Ab assays vary significantly (Weber et al., 1998; Ly et al., 2000). In addition, combined assays licensed in ∗
Corresponding author. Tel.: +33 149591720; fax: +33 149591798. E-mail address: [email protected] (T.D. Ly).
0166-0934/$ – see front matter © 2004 Elsevier B.V. All rights reserved. doi:10.1016/j.jviromet.2004.08.018
Europe between 1997 and 2000 exhibit a relatively high detection limit for p24 antigen (Courouc´e et al., 1999; Ly et al., 2001a). The detection limit of fourth-generation assays (20 to >100 pg of p24 antigen per ml) is higher than that of antigen assays (3.5–10 pg of p24 antigen per ml). In our previous study, we highlighted that the new fourth-generation assays may be improved by increasing their sensitivity of p24 detection (Ly et al., 2001a). In the present study, six commercially available HIV combined antigen and antibody assays and two third-generation HIV antibody assays were evaluated for (i) sensitivity to detect p24 antigen from diverse virus isolates including HIV-1 group M subtypes and HIV-1 group O, (ii) seroconversion sensitivity using 25 HIV seroconversion panels, (iii) ability to detect antibodies to HIV-1 group M, group O, and HIV-2 and (iv) specificity.
186
T.D. Ly et al. / Journal of Virological Methods 122 (2004) 185–194
2. Materials and methods 2.1. Fourth-generation HIV combined Ag/Ab assays The assays evaluated included: AxSYM HIV Ag/Ab Combo (Abbott Laboratories, IL, USA), Enzygnost HIV Integral (Dade Behring, Penzberg, Germany), Genscreen Plus HIV Ag/Ab (Bio-Rad, Marnes La Coquette, France), Murex HIV Ag/Ab Combo (Abbott, Dartford Kent, England), VIDAS HIV DUO (bioMerieux, Marcy l’Etoile, France) and Vironostika HIV Uniform II Ag/Ab (Organon Teknika, Boxtel, The Netherlands). The procedures used for these six HIV combined p24 antigen and antibody assays have been described in detail previously (Weber et al., 1998; Van Binsbergen et al., 1998; Ly et al., 2001a,b). Five of the fourth-generation assays use a ‘sandwich’ format technique for the detection of anti-HIV antibody, while the VIDAS HIV DUO uses an indirect format. The assays vary slightly in the techniques used for detection of the HIV-1 p24 Ag; the Abbott, Murex, and Vironostika tests use anti-HIV-1 p24 monoclonal Abs for both solid phase capture and probe conjugate for detection. The Enzygnost test uses anti-HIV1 p24 polyclonal Abs for the solid phase and anti-HIV1 p24 monoclonal Abs for conjugate and the Genscreen Plus and the VIDAS DUO tests use anti-HIV-1 p24 monoclonal Abs for the solid-phase and anti-HIV-1 p24 polyclonal Abs for conjugate. All tests were performed and interpreted in accordance with the manufacturer’s recommendations. 2.2. Third-generation anti-HIV Ab assays The two third-generation tests, Genscreen HIV-1/2 version 2 (Bio-Rad, France) and Ortho HIV-1/2 Ab Capture ELISA (Ortho Clinical Diagnostics, USA) are based on a double-Ag sandwich enzyme immunoassay technology used for IgG and IgM Ab detection. All tests were performed and interpreted in accordance with the manufacturer’s recommendations. 2.3. Specimens A blinded panel of well-characterized samples was used and performed in France for evaluation of the various serologic assays. (i) Thirty-one HIV-1 isolates were propagated in tissue culture. The HIV-1 group M subtypes A–G and circulating recombinant forms (CRF01 AE and CRF02AG) virus isolates were kindly provided by Nelson Michael and Merlin Robb (The Walter Reed Army Institute of Research, Rockville, Maryland) (Jagodzinski et al., 2000). The HIV-1 group O virus was obtained from Serologicals Corporation, Clarkston, Ga. SRA Technologies, Rockville, Maryland, propagated all viruses in tissue
culture. The virus stocks were diluted in negative human plasma to provide p24 antigen concentrations of 2, 5, 10, and 25 pg/ml using the Abbott HIV antigen assay (HIV Ag-1 monoclonal) and Abbott HIV-1 p24 antigen (viral) quantitation panel (Abbott Laboratories Diagnostics Division, Abbott Park, IL.). The values for detection limits were calculated using statistical regression (y = ax + b), where the limit is represented by s/co = 1.0. (ii) Twenty-five HIV-1 seroconversion panels with 176 samples were obtained from commercial suppliers: 10 panels were from Boston Biomedica Inc. (BBI), West Bridgewater, Massachusetts (PRB 923, 932, 941, 944, 951, 952, 953, 954, 957, 958), 11 were from North American Biologicals, Inc. (NABI), Boca Raton, Florida (NABI 251, 261, 271, 281, 321, 331, 401, 404, 4888, 37748, 122399), and four were from BioClinical Partners (BCP), Franklin, Massachusetts (BCP 6243, 9013, 9016, 9017). (iii) The antibody positive panel consisted of 669 plasma or sera samples collected from various HIV endemic areas of the world. HIV-1 group M subtype panel (n = 553) included: 113 subtypes A (Cameroon, France, Ghana, Uganda, United Kingdom), 75 subtypes B (Brazil, Cameroon, France, South Africa, Thailand, United Kingdom), 71 subtypes C (South Africa, Uganda, United Kingdom, France), 40 subtypes D (Cameroon, France, South Africa, Uganda, United Kingdom), 11 subtypes F (Argentina, Brazil, Cameroon, France), 9 subtypes G (Cameroon, Ghana), 83 CRF01 AE (Thailand), 92 CRF02 AG (Cameroon, United Kingdom), and 59 mosaic (Argentina, Brazil, Cameroon, Ghana, Uganda, United Kingdom). Six HIV-1 group O (Cameroon, Equatorial Guinea, Spain, France, USA) and 110 HIV-2 (West Africa) completed the antibody positive panel. The subtype determination of HIV-1 is based on sequence analysis of the envelope and gag p24 genes described previously (Brennan et al., 1997) except for the French samples that were characterised by serological subtyping (Barin et al., 1996). The HIV-2 specimens were not subtyped but classified as HIV-2 based on Western blot (HIV-1 and HIV-2) and HIV-2 EIA. (iv) One thousand and five unselected HIV negative samples were collected from four French private laboratories. For the calculation of specificity, specimens were considered HIV negative if all eight screening assays were negative or in the case of the EIA-reactive samples, if they were negative or indeterminate on a Western blot (NEW LAV Blot I, NEW LAV Blot II; Bio-Rad, Marnes La Coquette, France) and negative on HIV p24 Ag assay (Cobas Core Ag HIV EIA, Roche Diagnostics). NEW LAV Blot and HIV p24 Ag assay were used, performed, and interpreted in accordance with the manufacturer’s recommendations.
T.D. Ly et al. / Journal of Virological Methods 122 (2004) 185–194
3. Results 3.1. Sensitivity analysis using 31 HIV-1 p24 antigen panels The distinguishing feature of the fourth-generation assays is their ability to detect the HIV p24 antigen. The results of antigen detection limits using the HIV-1 group M and group O antigen panel are summarized in Table 1. The six tests showed significant differences. Murex Combo (limit detection 2.5–6.5 pg/ml) and AxSYM Combo (limit detection 4.4–10.2 pg/ml) were the most sensitive in detection of p24 Ag for all genotypes. In contrast, VIDAS DUO could detect 6.6 pg/ml CRF02 AG and for the others subtypes, the limit of detection was 7–16.9 pg/ml. However, VIDAS DUO failed to detect three of five subtype C panel members and the group O strain at p24 concentrations from 2 to 25 pg/ml. The limit of p24 Ag detection of Genscreen Plus ranged from 12.1 to 23.9 pg/ml except for three subtype B, two subtype F, one subtype C, one subtype CRF01 AE and group O, which were not detected at the highest concentration tested, 25 pg/ml. Vironostika and Enzygnost did
187
not detect any of the HIV strains at p24 concentrations from 2–25 pg/ml. 3.2. Detection of HIV infection in 25 seroconversion panels All eight HIV assays were tested against 25 HIV-1 seroconversion panels. The results are presented in Table 2 and are expressed as the day of first positive bleed. Analysis of the results of the 25 seroconversion panels indicated several differences between the performance of the various assays. The results have been divided into a number of groups, which highlight essential features of these differences. First, for five of 25 seroconversion panels, HIV-1 infection was detected 2–9 days earlier with HIV-1 p24 Ag assays (data from BBI, BCP, and NABI) than with any of the fourthgeneration HIV Ag/Ab assays (panels BCP 9013, NABI SV-0404 and 122399, and PRB 941 and 957). The fourthgeneration AxSYM Combo assay detected 19 and Murex Combo detected 17 of the 25 panels earlier or on the same bleeds as the HIV-1 p24 Ag assays. In contrast, VIDAS DUO, Genscreen Plus, Vironostika, Enzygnost, Genscreen v. 2 and
188
T.D. Ly et al. / Journal of Virological Methods 122 (2004) 185–194
Table 3 Time delay for detection of primary HIV infection in comparison with the most sensitive assay for each panel Seroconversion panel
Fourth-generation AxSYM HIV Ag/Ab Combo
p24 Aga
Third-generation Murex HIV Ag/Ab Combo
Genscreen Plus HIV Ag/Ab
Vidas HIV DUO
Vironostika HIV Ag/Ab
Enzygnost HIV Integral Ag/Ab
Genscreen HIV-1/2 v. 2 Ab
Ortho HIV-1/2 Ab Capture
0 1 0 5
0 1 0 15
1 3 1 5
6 2 1 15
6 3 1 19
6 3 2 19
1 3 2 0
1 3 2 1
0 0 0 15
NABI 122399 NABI 37748 NABI 4888 NABI SV-0251 NABI SV-0261 NABI SV-0281 NABI SV-0321 NABI SV-0331 NABI SV-0371 NABI SV-0401 NABI SV-0404
1 0 0 0 0 0 0 0 0 0 1
1 0 1 1 0 0 0 0 0 0 1
6 0 1 1 0 1 0 0 0 1 1
1 0 1 5 0 1 0 0 0 1 1
6 0 3 5 0 1 0 1 0 1 2
6 0 1 1 1 1 1 1 0 4 2
6 0 1 1 1 1 1 6 1 1 2
6 1 1 1 1 1 1 6 1 1 2
0 0 0 0 0 0 0 0 0 0 0
PRB 923 (W) PRB 932 (AG) PRB 941 (AQ) PRB 944 (AT) PRB 951 (BA) PRB 952 (BB) PRB 953 (BC) PRB 954 (BD) PRB 957 (BG) PRB 958 (BH)
0 0 1 0 0 0 0 0 1 1
0 0 1 0 0 0 1 0 1 0
1 0 1 1 1 0 1 1 1 7
1 0 5 1 1 0 1 0 1 1
1 0 1 1 1 0 4 1 1 7
1 0 1 8 1 4 4 1 1 9
38 0 1 3 9 1 4 1 1 7
1 0 1 8 9 1 4 1 1 7
0 0 0 0 0 0 0 0 0 0
Total no. of days Mean no. of days
11
23
35
45
65
78
92
62
15
0.44
0.92
Total no. of daysb Mean no. of daysb
6
8
0.26
0.35
a
Ag p24 data from BBI, NABI, and BCP. If panels BCP9017 and PRB 923 were excluded.
29 1.26
1.8 29 1.26
2.6 45 1.96
3.12 58 2.52
3.68 54 2.35
2.48 60 2.61
0.6 0 0
189
b
1.4
T.D. Ly et al. / Journal of Virological Methods 122 (2004) 185–194
BCP 6243 BCP 9013 BCP 9016 BCP 9017
190
T.D. Ly et al. / Journal of Virological Methods 122 (2004) 185–194
Ortho detected the first positive bleed (or earlier) in only 9, 8, 6, 3, 3 and 2 panels, respectively. These results emphasize the importance of p24 Ag detection for early diagnosis. The most sensitive combined Ag/Ab assays for antigen detection detect the first positive bleed by antigen assays in the majority of seroconversion panels. However, some combined assays are less sensitive than antigen assays for the detection of early seroconversion samples in the majority of panels and these also have higher detection limits for p24 antigen. The second significant finding was the improved sensitivity of the fourth-generation assays compared to the thirdgeneration assays. The AxSYM and Murex fourth-generation HIV combined Ag/Ab assays detected infection 2–47 days earlier than the third-generation HIV antibody tests in 20 and 18 of the 25 seroconversion panels, respectively (Table 2). These results confirm the improved efficacy of the fourthgeneration assays to reduce the diagnostic window. However, it is important to note that there is one panel (BCP 9017) in which the third-generation assays were more sensitive than all the fourth-generation and p24 antigen assays (Table 2). The third important observation was the differences in the sensitivity of the six fourth-generation assays. The AxSYM and Murex Combo assays consistently performed better than the other fourth-generation assays. In contrast to the AxSYM and Murex assays that detected infection earlier than thirdgeneration antibody assays in 18 and 20 of the seroconversion panels, Vidas DUO was earlier than antibody assays in 14 panels, Genscreen Plus in 10 panels, Vironostika in eight panels, and Enzygnost in only three panels. The difference in sensitivity between assays is evident based on the observation that for three panels (BCP 9013, 9016, NABI SV-0404), at least three to five assays (Genscreen Plus, Vironostika, Enzygnost, Genscreen v. 2, and Ortho) were not able to detect any bleeds. Based on percentage of positive bleeds for each assay out of a total number of 176 bleeds from 25 panels (Table 2),
the assays can be ranked in the following order: AxSYM HIV Ag/Ab Combo (55.1%), Murex HIV Ag/Ab Combo (52.2%), Genscreen Plus HIV Ag/Ab (47.1%), VIDAS HIV DUO (45.4%), Genscreen HIV-1/2 v. 2 (41.5%), Vironostika HIV Uniform II Ag/Ab (40.9%), Ortho HIV-1/2 Ab Capture (40.3%), and Enzygnost HIV Integral (38.1%). When the time interval between two consecutive bleeds within a panel is relatively short, it is possible to estimate the delay (expressed as the number of days) between the first positive bleed detected by the most sensitive assay on a given seroconversion panel with each of the other assays on this panel. The time delay between blood sampling points in the commercial seroconversion panels used for the present study is on average relatively short (2–7 days) but may last up to 37 days, for example, panel PRB 923. The calculation model for time delays between assays established by the Paul Ehrlich Institute (Courouc´e et al., 1999) was used. This method considers that seroconversion is theoretically possible the day after the last negative follow-up sample. The total and the average number of days of time delay for the 25 panels were calculated in comparison with the most sensitive assay (Table 3). The mean time delay in days for detection of primary HIV infection in comparison with the most sensitive assay for each panel (last negative sample plus one day) ranged from 0.44 (AxSYM Combo) to 3.68 days (Genscreen HIV-1/2 v. 2). The mean time delay between p24 Ag detection assay and AxSYM Combo positive reactivity was 0.16 day (Table 3). If panels BCP 9017 and PRB 923 were excluded from analysis, the mean time delay between p24 Ag assay detection and the first positive of the different tests were AxSYM 0.26, Murex 0.35, Genscreen Plus and VIDAS DUO 1.26, Vironostika 1.96, Enzygnost 2.52, Genscreen v. 2 2.35, and Ortho 2.61 days. The mean time delay (excluding panels BCP9017 and PRB923) between the third-generation antibody Genscreen v. 2 assay and combined Ag/Ab assays was 2.09 days for AxSYM, 2.0 days for Murex, 1.09 days for
T.D. Ly et al. / Journal of Virological Methods 122 (2004) 185–194
191
DUO (13 of 1005), and Genscreen HIV-1/2 v. 2 (19 of 1005) showed a high percentage of initial false-positive results.
Genscreen Plus and Vidas DUO, and 0.39 day for Vironostika. The Enzygnost combined assay detected infections later than the Genscreen v. 2 assay with a mean delay of 0.17 day.
4. Discussion
3.3. Detection of 669 HIV positive samples
This study was designed to compare the sensitivity and specificity of six HIV combined antigen and antibody assays against two third-generation HIV antibody assays. The analysis of the results from this study has allowed us to make several clear conclusions about performance of the different assays. Seroconversion panels from patients during primary infection are informative in order to compare the sensitivity of screening tests. Consistently, the threshold of p24 Ag detection was 3–5 pg/ml with the HIV p24 Ag assays and ranged from 3.3 to more than 25 pg/ml with the fourth-generation assays. When compared with HIV p24 antigen assays, fourthgeneration tests were unable to detect the earliest positive panel member in five of 25 seroconversion panels. Consequently, the fourth-generation assays are not as sensitive as p24 Ag assays for detection of primary infection. However, the difference between the most sensitive combination assays compared to antigen assays is relative short. For the AxSYM
All 553 HIV-1 group M, 6 HIV-1 group O and 110 HIV-2 positive samples were detected by all assays except Genscreen v. 2 missed 1 HIV-1 group O, VIDAS DUO missed 1 HIV-1 subtype F, Enzygnost missed 1 HIV-1 subtype C, and Genscreen v. 2 missed 1 HIV-1 subtype B antibody positive sample (Table 4). 3.4. Specificity analysis using 1005 unselected HIV negative samples False-positive results obtained with the eight assays on 1005 HIV negative samples are shown in Table 5. The three assays with the best specificity based on repeat reactivity included Ortho HIV-1/2 Ab Capture with 100% specificity, Genscreen Plus HIV Ag/Ab with 99.9% specificity, and AxSYM HIV Ag/Ab Combo with 99.8% specificity. Vironostika HIV Uniform II Ag/Ab (28 of 1005), VIDAS HIV Table 5 Specificity of eight HIV screening assaysa Assays
No. of false-positive samples
Specificity (%) Initially reactive
Initially reactive
Repeatedly reactive
Fourth-generation Genscreen Plus HIV Ag/Ab AxSYM HIV Ag/Ab Combo Enzygnost HIV Integral Vironostika HIV Uniform II Ag/Ab Murex HIV Ag/Ab Combo VIDAS HIV DUO
1 2 4 28 5 13
1 2 3 4 4 5
Third-generation Ortho HIV-1/2 Ab Genscreen HIV-1/2 v. 2
0 19
0 6
a
99.9 99.8 99.6 97.21 99.5 98.7 100 98.1
Repeatedly reactive 99.9 99.8 99.7 99.6 99.6 99.5 100 99.4
Total number of HIV negative samples tested n = 1005.
Table 6 Summary of HIV detection for the 23 seroconversion panels Time delay in comparison with p24 Ag Assay n
Mean
Ranking
Performancea
Ranking
Fourth-generation AxSYM HIV Ag/Ab Combo Murex HIV Ag/Ab Combo Genscreen Plus HIV Ag/Ab Combo VIDAS HIV DUO Vironostika HIV Uniform II Ag/Ab Enzygnost HIV Integral
23 23 23 23 23 23
0.26 0.35 1.26 1.26 1.96 2.52
1 2 3 3 5 7
55.1 52.2 47.1 45.4 40.9 38.1
1 2 3 4 6 8
Third-generation Genscreen HIV-1/2 v. 2 Ortho HIV-1/2 Ab Capture
23 23
2.35 2.61
6 8
41.5 40.3
5 7
n = Total number of seroconversion panels (BCP9017 and PRB 923 excluded). a Performance = mean percentage of positive bleeds per panel.
192
T.D. Ly et al. / Journal of Virological Methods 122 (2004) 185–194
and Murex Combo assays, the detection of infection was delayed only 0.26–0.35 days relative to HIV p24 Ag assays. In a blood bank setting, a large majority of HIV infection would be detected by highly sensitive combination assays. Recently, the blood banks have implemented screening of blood products by nucleic acid tests (NAT). It is of interest to note that in the US, during the 3-year period starting in March 1999, NAT detected five antibody negative samples out of more than 19 million donations. Of these five NAT positive samples, one sample was also positive for antigen (Dodd et al., 2002). More recently, studies in low HIV endemic areas, Australia (Cunnigham et al., personal communication), Sweden, and Norway (Andersson et al., manuscript in preparation), and a high HIV endemic region, South Africa (Rensburg et al., personal communication), have demonstrated utility of HIV combined Ag/Ab assays for identifying early seroconversion samples. Antibody assays were used as the primary screening assay in Australia, Sweden, and Norway, while an antigen assay was used in addition to an antibody assay for detection of HIV infection in South Africa. At all sites, antigen positive early seroconversion samples were identified in side-by-side analysis of samples using the AxSYM Ag/Ab Combo assay and an antibody only assay. The incremental pick-up of antigen positive samples further exemplifies utility of combined assays in the clinical setting. Earlier detection of HIV-1 infection by the fourthgeneration assays compared to the third-generation assays is concordant with previous studies (Weber et al., 1998; G¨urtler et al., 1998; Van Binsbergen et al., 1998; Brust et al., 2000; Ly et al., 2001a,b). The earlier detection can be attributed to p24 antigen detection, reinforcing the importance of p24 Ag detection in earlier diagnostics (Courouc´e et al., 1988; Busch et al., 1990). It was of interest to note that there was no statistical difference in sensitivity with Enzygnost and Vironostika fourth-generation tests compared to Genscreen HIV-1/2 v. 2, which is one of the most sensitive commercially available third-generation screening EIAs (Laperche et al., 2000), and that this lack of sensitivity correlates with poor p24 antigen detection by the Enzygnost and Vironostika combined assays. The sensitivity of fourth-generation tests differed because of heterogeneous ability to detect either p24 Ag and/or antibodies (Tables 1–3). The AxSYM and/or the Murex Combo assays detected primary infection prior to all the other fourthgeneration combined assays in 12 out of 25 seroconversion panels (Table 2). This superior performance of the AxSYM and Murex Combo correlates with higher sensitivity for p24 Ag detection (Table 1). The average HIV Ag detection limit was 7 pg of p24 Ag/ml, for the AxSYM and 5 pg/ml for Murex Combo; these assays reduced diagnostic window by 1.6–2.17 days (Table 2) in comparison to the least sensitive combined p24 Ag and antibody assays (Enzygnost, and Vironostika assays; detection limit >25 pg of p24 Ag/ml). In a former study (Ly et al., 2001a), the antigen p24 detection limits for the Vironostika and the Enzygnost assays were 37.09 and 30.17 pg/ml, respectively, by using the HIV1 Ag panel SFTS 96 standard (French Society of Blood
Transfusion). The Genscreen Plus and VIDAS HIV DUO assays had average p24 detection limits of >14 and >11 pg/ml respectively (Table 1) and display similar seroconversion sensitivity (Tables 2 and 3). The AxSYM and Murex combined assays shorten delay for detection of infection 0.91–1.0 days relative to the Genscreen Plus and VIDAS HIV DUO assays. Interestingly, with panel BCP 9017 (Table 2), both thirdgeneration tests (Genscreen HIV v. 2 and Ortho Capture) detected infection earlier than the fourth-generation assays. This reflects a very rare clinical scenario in which the antiHIV IgM antibodies appear prior to detectable p24 Ag. Physicians using fourth-generation tests should be aware of this very rare circumstance as a possible source of false negative. Among six combination assays evaluated, the AxSYM Combo and Genscreen Plus assays identified the earliest seroconversion bleed (bleed 5, day 15) resulting in a day 10 earlier detection of the first positive result compared to the other combined assays. These data indicates that the AxSYM Combo and the Genscreen Plus assays have better antibody detection compared to other combined assays evaluated in this study. With panel PRB 952, VIDAS HIV DUO detected infection at day 10, i.e. earlier than third-generations tests (Genscreen HIV v. 2 and Ortho Capture). Opposite performance is found for later detection (day 14 and 17). This unexpected serological window for the VIDAS HIV DUO test probably reflects the transition from anti-HIV IgM to IgG antibodies. This could impair the accuracy of VIDAS HIV DUO, even if this kind of serological situation is rare. The genetic variability of HIV-1 may represent a major challenge for early antigen detection during primary HIV infection with fourth-generation tests (Apetrei et al., 1996). This study did not compare the ability of these eight HIV assays to detect early infection with different HIV-1 subtypes, because all of the seroconversion samples are HIV-1 group M, subtype B. However, using 31 HIV-1 group M subtypes and group O strains derived from cell culture, an impaired sensitivity for HIV-1 subtype C antigen detection was observed with VIDAS HIV DUO (Table 1). Our previous study with VIDAS HIV DUO Ultra also showed poor detection of subtype C antigen (Ly et al., 2001b) in contrast to the findings of Weber et al. (Weber et al., 2002). The Genscreen Plus failed to detect three of seven HIV-1 subtype B strains (Table 1). These results may reflect the quality of monoclonal antibodies used for the capture of p24 Ag and corroborate observations of Tersmette et al., 1989, who reported failure of monoclonal antibody to detect p24 Ag from certain strains of HIV. Among the fourth-generation assays, the Enzygnost and Vironostika assays showed poor antigen sensitivity using the viral isolate panel and also detected a lower number of seroconversion samples (Table 2). The prevalence of HIV-1 group M non-B subtypes and HIV-2 infections is increasing in Europe (Couturier et al., 2000). The genetic variability represents a challenge for HIV screening assays. In our study, all 669 HIV positive samples,
T.D. Ly et al. / Journal of Virological Methods 122 (2004) 185–194
of different genotypes and geographical origins, were detected by all eight assays with a high signal to cutoff except for the following samples. Genscreen v. 2 missed 1 group O sample perhaps due to absence of a specific reagent for group O detection. VIDAS DUO missed a subtype F that was discovered in France in an African woman; this case was notified to AFSSAPS (France Sanitary Safety of Health Products Agency). Enzygnost missed 1 subtype C and Genscreen v. 2 missed 1 subtype B; these two samples are indeterminate with Western blot HIV-1 and negative with Western blot HIV-2, but strongly positive with all other assays. Since fourth-generation EIAs combine two different test principles in one assay, the potential risk for non-specific reactivity may be higher than for second- and third-generation assays. Previous studies consistently reported a lower specificity for combined tests compared to third-generation assays (G¨urtler et al., 1998; Ly et al., 2000). In this study, we evaluated the specificity of third- and fourth-generation tests using sera sampled from 1005 unselected patients. Among the fourth-generation tests, Genscreen Plus and AxSYM Combo displayed the highest specificity, 99.9% and 99.8% respectively. The other fourth-generation displayed specificity ranging from 99.5% to 99.7%. We report here specificity for VIDAS DUO that is similar to that reported in a previous study with 29,657 unselected samples from five French public and private laboratories (Ly et al., 2000). To our knowledge, this is the first comparative side-by-side study of eight HIV assays to evaluate the specificity on the same panel of unselected routine samples. In summary, the present evaluation indicates that among these six combination assays, the AxSYM and Murex Combo assays have the best overall sensitivity followed by the Genscreen Plus and VIDAS HIV DUO (Table 6). The AxSYM Combo and Genscreen Plus also have good specificity based on the panel of 1005 unselected negative samples.
Acknowledgements We thank A. Bianchi (Laboratoire D´epartemental de Bondy), B. Lecolier, J.M. Hadjez (Centre d’Exploration Fonctionnelle, Paris), J. Lupu (Centre Suchet, Paris), J.Y. Mascart, A.J. Morvan (Laboratoire Morvan Mascart, Paimpol), J.P. Tr´eguier and J.J. Lemoine (Laboratoire Tr´eguier Lemoine, Lannion) for providing French HIV negative and positive samples. The authors also acknowledge Pierre Leroux (Abbott Laboratories, France) for his help in setting up the collaborative studies.
References Apetrei, C., Loussert-Ajaka, I., Descamp, S.D., Damond, F., Saragosti, S., Brun-Vezinet, F., Simon, F., 1996. Lack of screening test sensitivity during HIV-1 non-subtype B seroconversions. AIDS 14, 57–60.
193
Barin, F., Lahbabi, Y., Buzelay, L., Lejeune, B., Baillou-Beaufils, A., Denis, F., Mathiot, C., M’Boup, S., Vithayasai, V., Dietrich, U., Goudeau, A., 1996. Diversity of antibody binding to V3 peptides representing consensus sequences of HIV type 1 genotypes A to E: an approach for HIV type 1 serological subtyping. AIDS Res. Hum. Retroviruses 13, 1279–1289. Brennan, C.A., Lund, J.K., Golden, A., Yamaguchi, J., Vallari, A.S., Phillips, J.F., Kataaha, P.K., Jackson, J.B., Devare, S.G., 1997. Serologic and phylogenetic characterization of HIV-1 subtypes in Uganda. AIDS 11, 1823–1832. Brust, S., Duttmann, H., Feldner, J., Gurtler, L.G., Thorstensson, R., Simon, F., 2000. Shortening of the diagnostic window with a new combined HIV p24 antigen and anti-HIV-1/2/O screening test. J. Virol. Methods 90, 153–165. Busch, M.P., Taylor, P.E., Lenes, B.A., Kleinman, S.H., Stuart, M., Stevens, C.E., Tomasulo, P.A., Allain, J.P., Hollingsworth, C.G., Mosley, J.W., and the transfusion Safety Study Group, 1990. Screening of selected male blood donors for p24 antigen human immunodeficiency virus type 1. N. Engl. J. Med. 8, 1308– 1312. Courouc´e, A.M., Bouchardeau, F., Jullien, A.M., Faucher, V., Lenty, M., 1988. Blood transfusion and human immunodeficiency virus (HIV) antigen. Ann. Intern. Med. 108, 771–772. Courouc´e, A.M., The Groupe de Travail Retrovirus de la S.F.T.S., 1999. Tests de depistage combin´e des anticorps anti-VIH et de l’Age p24. Gazette Transfus 155, 4–17. Couturier, E., Damond, F., Roques, P., Fleury, H., Barin, F., Brunet, J.B., Brun-Vezinet, F., Simon, F., 2000. HIV-1 diversity in France, 1996–1998. The AC 11 laboratory network. AIDS 3, 289– 296. Dodd, R.Y., Notari IV, E.P., Stramer, S.L., 2002. Current prevalence and incidence of infectious disease markers and estimated window-period risk in the American Red Cross blood donor population. Transfusion 42, 975–979. G¨urtler, L.G., M¨uhlbacher, A., Michl, U., Hofmann, H., Paggi, G.G., Bossi, V., Thorstensson, R., G-Villaescusa, R., Eiras, A., Hernandez, J.M., Melchior, W., Doni´e, F., Weber, B., 1998. Reduction of the diagnostic window with a new combined p24 antigen and human immunodeficiency virus antibody screening assay. J. Virol. Methods 75, 27–38. Jagodzinski, L.L., Wiggins, D.L., McManis, J.L., Emery, S., Overbaugh, J., Robb, M., Bodrug, S., Michael, N., 2000. Use of calibrated viral load standards for group M subtypes of human immunodeficiency virus type 1 to assess the performance of viral RNA quantitation tests. J. Clin. Microbiol. 38, 1247– 1249. Laperche, S., Maniez-Montreuil, M., Courouce, A.M., 2000. Les tests de d´epistage combin´e de l’antig`ene p24 et des anticorps anti-VIH dans l’infection pr´ecoce a` VIH-1. Transfus. Clin. Biol. 7 (Suppl. 1), 18–24. Ly, T.D., Edlinger, C., Vabret, A., Morgan, O., Greuet, B., 2000. Contribution of combined detection assays of p24 antigen and antihuman immunodeficiency virus (HIV) antibodies in diagnosis of primary HIV infection by routine testing. J. Clin. Microbiol. 38, 2459– 2461. Ly, T.D., Laperche, S., Courouce, A.M., 2001a. Early detection of human immunodeficiency virus infection using third- and fourth-generation screening assays. Eur. J. Clin. Microbiol. Infect. Dis. 20, 104– 110. Ly, T.D., Martin, L., Daghfal, D., Sandridge, A., West, D., Bristow, R., Chalouas, L., Qiu, X., Lou, S.C., Hunt, J.C., Schochetman, G., Devare, S.G., 2001b. Seven human immunodeficiency virus (HIV) antigen-antibody combination assays: evaluation of HIV serconversion sensitivity and subtype detection. J. Clin. Microbiol. 39, 3122– 3128. Tersmette, M., Winkel, I., Groenick, M., Gruters, R.A., Spence, R.P., Saman, E., Van der Groen, G., Miedema, F., Huisman, J.G., 1989. Detection and subtyping of HIV-1 isolates with a panel of characterized
194
T.D. Ly et al. / Journal of Virological Methods 122 (2004) 185–194
monoclonal antibodies to HIV p24 gag. Virology 171, 149– 155. Van Binsbergen, J., Keur, W., Siebelink, A., Van de Graaf, M., Jacobs, A., De Rijk, D., Nijholt, L., Toonen, J., G¨urtler, L.G., 1998. Strongly enhanced sensitivity of a direct anti-HIV-1/2 assay in seroconversion by incorporation of HIV p24 Ag detection: a new generation Vironostika HIV Uniform II. J. Virol. Methods 76, 59–71.
Weber, B., Mbargane Fall, E.H., Berger, A., Doerr, H.W., 1998. Reduction of diagnostic window by new fourth-generation human immunodeficiency virus screening assay. J. Clin. Microbiol. 36, 2235–2239. Weber, B., Berger, A., Rabenau, H., Doerr, H.W., 2002. Evaluation of a new combined antigen and antibody human immunodeficiency virus screening assay, Vidas HIV Duo Ultra. J. Clin. Microbiol. 40, 1420–1426.
Evaluation of the sensitivity and specificity of six HIV combined p24 antigen and antibody assays Thoai Duong Lya,∗ , Syria Lapercheb , Catherine Brennanc , Ana Vallaric , Anne Ebela , Jeffrey Huntc , Lynn Martinc , David Daghfalc , Gerald Schochetmanc , Sushil Devarec a
Laboratoire LCL, 78, Avenue de Verdun, 94200 Ivry sur Seine, France b National Institute of Blood Transfusion, Paris, France c Abbott Laboratories, Abbott Park, IL, USA
Received 28 June 2004; received in revised form 13 August 2004; accepted 16 August 2004 Available online 13 October 2004
Abstract In this study, we evaluated the performance of six HIV combined p24 antigen and antibody (Ag/Ab) assays versus two third-generation anti-HIV antibody assays. The assays were evaluated using p24 antigen panel of 31 HIV-1 subtypes (n = 124), 25 HIV-1 seroconversion panels (n = 176), HIV-1 antibody positive samples including group M subtypes and group O (n = 559), HIV-2 positive samples (n = 110), and unselected HIV negative samples from four French private laboratories (n = 1005). The results showed that overall HIV combined Ag/Ab assays present better performance, when compared to antibody-only assays. However, some differences were observed in the sensitivity of the six HIV combined Ag/Ab assays evaluated. The AxSYM and Murex Combo assays had the best sensitivity score in this study and reduced the window period by 2.0–2.35 days relative to antibody only assays and 1–2.17 days relative to the other combined Ag/Ab assays. Among combined HIV Ag/Ab assays, Genscreen Plus and AxSYM Combo presented the highest specificity, with 99.9% and 99.8%, respectively. © 2004 Elsevier B.V. All rights reserved. Keywords: Seroconversion; HIV combined antigen and antibody assay; p24 antigen
1. Introduction Since the introduction of HIV combined antigen (Ag) and antibody (Ab) screening assays (referred to as fourthgeneration assays) by the end of 1997 (Weber et al., 1998; G¨urtler et al., 1998; Van Binsbergen et al., 1998), it is estimated that these assays reduce the seroconversion window period by a few days to as much as two weeks in comparison with third-generation antibody assays (Courouc´e et al., 1999; Brust et al., 2000; Ly et al., 2001a). Despite their increased sensitivity, numerous reports indicate that the specificity of HIV Ag/Ab assays vary significantly (Weber et al., 1998; Ly et al., 2000). In addition, combined assays licensed in ∗
Corresponding author. Tel.: +33 149591720; fax: +33 149591798. E-mail address: [email protected] (T.D. Ly).
0166-0934/$ – see front matter © 2004 Elsevier B.V. All rights reserved. doi:10.1016/j.jviromet.2004.08.018
Europe between 1997 and 2000 exhibit a relatively high detection limit for p24 antigen (Courouc´e et al., 1999; Ly et al., 2001a). The detection limit of fourth-generation assays (20 to >100 pg of p24 antigen per ml) is higher than that of antigen assays (3.5–10 pg of p24 antigen per ml). In our previous study, we highlighted that the new fourth-generation assays may be improved by increasing their sensitivity of p24 detection (Ly et al., 2001a). In the present study, six commercially available HIV combined antigen and antibody assays and two third-generation HIV antibody assays were evaluated for (i) sensitivity to detect p24 antigen from diverse virus isolates including HIV-1 group M subtypes and HIV-1 group O, (ii) seroconversion sensitivity using 25 HIV seroconversion panels, (iii) ability to detect antibodies to HIV-1 group M, group O, and HIV-2 and (iv) specificity.
186
T.D. Ly et al. / Journal of Virological Methods 122 (2004) 185–194
2. Materials and methods 2.1. Fourth-generation HIV combined Ag/Ab assays The assays evaluated included: AxSYM HIV Ag/Ab Combo (Abbott Laboratories, IL, USA), Enzygnost HIV Integral (Dade Behring, Penzberg, Germany), Genscreen Plus HIV Ag/Ab (Bio-Rad, Marnes La Coquette, France), Murex HIV Ag/Ab Combo (Abbott, Dartford Kent, England), VIDAS HIV DUO (bioMerieux, Marcy l’Etoile, France) and Vironostika HIV Uniform II Ag/Ab (Organon Teknika, Boxtel, The Netherlands). The procedures used for these six HIV combined p24 antigen and antibody assays have been described in detail previously (Weber et al., 1998; Van Binsbergen et al., 1998; Ly et al., 2001a,b). Five of the fourth-generation assays use a ‘sandwich’ format technique for the detection of anti-HIV antibody, while the VIDAS HIV DUO uses an indirect format. The assays vary slightly in the techniques used for detection of the HIV-1 p24 Ag; the Abbott, Murex, and Vironostika tests use anti-HIV-1 p24 monoclonal Abs for both solid phase capture and probe conjugate for detection. The Enzygnost test uses anti-HIV1 p24 polyclonal Abs for the solid phase and anti-HIV1 p24 monoclonal Abs for conjugate and the Genscreen Plus and the VIDAS DUO tests use anti-HIV-1 p24 monoclonal Abs for the solid-phase and anti-HIV-1 p24 polyclonal Abs for conjugate. All tests were performed and interpreted in accordance with the manufacturer’s recommendations. 2.2. Third-generation anti-HIV Ab assays The two third-generation tests, Genscreen HIV-1/2 version 2 (Bio-Rad, France) and Ortho HIV-1/2 Ab Capture ELISA (Ortho Clinical Diagnostics, USA) are based on a double-Ag sandwich enzyme immunoassay technology used for IgG and IgM Ab detection. All tests were performed and interpreted in accordance with the manufacturer’s recommendations. 2.3. Specimens A blinded panel of well-characterized samples was used and performed in France for evaluation of the various serologic assays. (i) Thirty-one HIV-1 isolates were propagated in tissue culture. The HIV-1 group M subtypes A–G and circulating recombinant forms (CRF01 AE and CRF02AG) virus isolates were kindly provided by Nelson Michael and Merlin Robb (The Walter Reed Army Institute of Research, Rockville, Maryland) (Jagodzinski et al., 2000). The HIV-1 group O virus was obtained from Serologicals Corporation, Clarkston, Ga. SRA Technologies, Rockville, Maryland, propagated all viruses in tissue
culture. The virus stocks were diluted in negative human plasma to provide p24 antigen concentrations of 2, 5, 10, and 25 pg/ml using the Abbott HIV antigen assay (HIV Ag-1 monoclonal) and Abbott HIV-1 p24 antigen (viral) quantitation panel (Abbott Laboratories Diagnostics Division, Abbott Park, IL.). The values for detection limits were calculated using statistical regression (y = ax + b), where the limit is represented by s/co = 1.0. (ii) Twenty-five HIV-1 seroconversion panels with 176 samples were obtained from commercial suppliers: 10 panels were from Boston Biomedica Inc. (BBI), West Bridgewater, Massachusetts (PRB 923, 932, 941, 944, 951, 952, 953, 954, 957, 958), 11 were from North American Biologicals, Inc. (NABI), Boca Raton, Florida (NABI 251, 261, 271, 281, 321, 331, 401, 404, 4888, 37748, 122399), and four were from BioClinical Partners (BCP), Franklin, Massachusetts (BCP 6243, 9013, 9016, 9017). (iii) The antibody positive panel consisted of 669 plasma or sera samples collected from various HIV endemic areas of the world. HIV-1 group M subtype panel (n = 553) included: 113 subtypes A (Cameroon, France, Ghana, Uganda, United Kingdom), 75 subtypes B (Brazil, Cameroon, France, South Africa, Thailand, United Kingdom), 71 subtypes C (South Africa, Uganda, United Kingdom, France), 40 subtypes D (Cameroon, France, South Africa, Uganda, United Kingdom), 11 subtypes F (Argentina, Brazil, Cameroon, France), 9 subtypes G (Cameroon, Ghana), 83 CRF01 AE (Thailand), 92 CRF02 AG (Cameroon, United Kingdom), and 59 mosaic (Argentina, Brazil, Cameroon, Ghana, Uganda, United Kingdom). Six HIV-1 group O (Cameroon, Equatorial Guinea, Spain, France, USA) and 110 HIV-2 (West Africa) completed the antibody positive panel. The subtype determination of HIV-1 is based on sequence analysis of the envelope and gag p24 genes described previously (Brennan et al., 1997) except for the French samples that were characterised by serological subtyping (Barin et al., 1996). The HIV-2 specimens were not subtyped but classified as HIV-2 based on Western blot (HIV-1 and HIV-2) and HIV-2 EIA. (iv) One thousand and five unselected HIV negative samples were collected from four French private laboratories. For the calculation of specificity, specimens were considered HIV negative if all eight screening assays were negative or in the case of the EIA-reactive samples, if they were negative or indeterminate on a Western blot (NEW LAV Blot I, NEW LAV Blot II; Bio-Rad, Marnes La Coquette, France) and negative on HIV p24 Ag assay (Cobas Core Ag HIV EIA, Roche Diagnostics). NEW LAV Blot and HIV p24 Ag assay were used, performed, and interpreted in accordance with the manufacturer’s recommendations.
T.D. Ly et al. / Journal of Virological Methods 122 (2004) 185–194
3. Results 3.1. Sensitivity analysis using 31 HIV-1 p24 antigen panels The distinguishing feature of the fourth-generation assays is their ability to detect the HIV p24 antigen. The results of antigen detection limits using the HIV-1 group M and group O antigen panel are summarized in Table 1. The six tests showed significant differences. Murex Combo (limit detection 2.5–6.5 pg/ml) and AxSYM Combo (limit detection 4.4–10.2 pg/ml) were the most sensitive in detection of p24 Ag for all genotypes. In contrast, VIDAS DUO could detect 6.6 pg/ml CRF02 AG and for the others subtypes, the limit of detection was 7–16.9 pg/ml. However, VIDAS DUO failed to detect three of five subtype C panel members and the group O strain at p24 concentrations from 2 to 25 pg/ml. The limit of p24 Ag detection of Genscreen Plus ranged from 12.1 to 23.9 pg/ml except for three subtype B, two subtype F, one subtype C, one subtype CRF01 AE and group O, which were not detected at the highest concentration tested, 25 pg/ml. Vironostika and Enzygnost did
187
not detect any of the HIV strains at p24 concentrations from 2–25 pg/ml. 3.2. Detection of HIV infection in 25 seroconversion panels All eight HIV assays were tested against 25 HIV-1 seroconversion panels. The results are presented in Table 2 and are expressed as the day of first positive bleed. Analysis of the results of the 25 seroconversion panels indicated several differences between the performance of the various assays. The results have been divided into a number of groups, which highlight essential features of these differences. First, for five of 25 seroconversion panels, HIV-1 infection was detected 2–9 days earlier with HIV-1 p24 Ag assays (data from BBI, BCP, and NABI) than with any of the fourthgeneration HIV Ag/Ab assays (panels BCP 9013, NABI SV-0404 and 122399, and PRB 941 and 957). The fourthgeneration AxSYM Combo assay detected 19 and Murex Combo detected 17 of the 25 panels earlier or on the same bleeds as the HIV-1 p24 Ag assays. In contrast, VIDAS DUO, Genscreen Plus, Vironostika, Enzygnost, Genscreen v. 2 and
188
T.D. Ly et al. / Journal of Virological Methods 122 (2004) 185–194
Table 3 Time delay for detection of primary HIV infection in comparison with the most sensitive assay for each panel Seroconversion panel
Fourth-generation AxSYM HIV Ag/Ab Combo
p24 Aga
Third-generation Murex HIV Ag/Ab Combo
Genscreen Plus HIV Ag/Ab
Vidas HIV DUO
Vironostika HIV Ag/Ab
Enzygnost HIV Integral Ag/Ab
Genscreen HIV-1/2 v. 2 Ab
Ortho HIV-1/2 Ab Capture
0 1 0 5
0 1 0 15
1 3 1 5
6 2 1 15
6 3 1 19
6 3 2 19
1 3 2 0
1 3 2 1
0 0 0 15
NABI 122399 NABI 37748 NABI 4888 NABI SV-0251 NABI SV-0261 NABI SV-0281 NABI SV-0321 NABI SV-0331 NABI SV-0371 NABI SV-0401 NABI SV-0404
1 0 0 0 0 0 0 0 0 0 1
1 0 1 1 0 0 0 0 0 0 1
6 0 1 1 0 1 0 0 0 1 1
1 0 1 5 0 1 0 0 0 1 1
6 0 3 5 0 1 0 1 0 1 2
6 0 1 1 1 1 1 1 0 4 2
6 0 1 1 1 1 1 6 1 1 2
6 1 1 1 1 1 1 6 1 1 2
0 0 0 0 0 0 0 0 0 0 0
PRB 923 (W) PRB 932 (AG) PRB 941 (AQ) PRB 944 (AT) PRB 951 (BA) PRB 952 (BB) PRB 953 (BC) PRB 954 (BD) PRB 957 (BG) PRB 958 (BH)
0 0 1 0 0 0 0 0 1 1
0 0 1 0 0 0 1 0 1 0
1 0 1 1 1 0 1 1 1 7
1 0 5 1 1 0 1 0 1 1
1 0 1 1 1 0 4 1 1 7
1 0 1 8 1 4 4 1 1 9
38 0 1 3 9 1 4 1 1 7
1 0 1 8 9 1 4 1 1 7
0 0 0 0 0 0 0 0 0 0
Total no. of days Mean no. of days
11
23
35
45
65
78
92
62
15
0.44
0.92
Total no. of daysb Mean no. of daysb
6
8
0.26
0.35
a
Ag p24 data from BBI, NABI, and BCP. If panels BCP9017 and PRB 923 were excluded.
29 1.26
1.8 29 1.26
2.6 45 1.96
3.12 58 2.52
3.68 54 2.35
2.48 60 2.61
0.6 0 0
189
b
1.4
T.D. Ly et al. / Journal of Virological Methods 122 (2004) 185–194
BCP 6243 BCP 9013 BCP 9016 BCP 9017
190
T.D. Ly et al. / Journal of Virological Methods 122 (2004) 185–194
Ortho detected the first positive bleed (or earlier) in only 9, 8, 6, 3, 3 and 2 panels, respectively. These results emphasize the importance of p24 Ag detection for early diagnosis. The most sensitive combined Ag/Ab assays for antigen detection detect the first positive bleed by antigen assays in the majority of seroconversion panels. However, some combined assays are less sensitive than antigen assays for the detection of early seroconversion samples in the majority of panels and these also have higher detection limits for p24 antigen. The second significant finding was the improved sensitivity of the fourth-generation assays compared to the thirdgeneration assays. The AxSYM and Murex fourth-generation HIV combined Ag/Ab assays detected infection 2–47 days earlier than the third-generation HIV antibody tests in 20 and 18 of the 25 seroconversion panels, respectively (Table 2). These results confirm the improved efficacy of the fourthgeneration assays to reduce the diagnostic window. However, it is important to note that there is one panel (BCP 9017) in which the third-generation assays were more sensitive than all the fourth-generation and p24 antigen assays (Table 2). The third important observation was the differences in the sensitivity of the six fourth-generation assays. The AxSYM and Murex Combo assays consistently performed better than the other fourth-generation assays. In contrast to the AxSYM and Murex assays that detected infection earlier than thirdgeneration antibody assays in 18 and 20 of the seroconversion panels, Vidas DUO was earlier than antibody assays in 14 panels, Genscreen Plus in 10 panels, Vironostika in eight panels, and Enzygnost in only three panels. The difference in sensitivity between assays is evident based on the observation that for three panels (BCP 9013, 9016, NABI SV-0404), at least three to five assays (Genscreen Plus, Vironostika, Enzygnost, Genscreen v. 2, and Ortho) were not able to detect any bleeds. Based on percentage of positive bleeds for each assay out of a total number of 176 bleeds from 25 panels (Table 2),
the assays can be ranked in the following order: AxSYM HIV Ag/Ab Combo (55.1%), Murex HIV Ag/Ab Combo (52.2%), Genscreen Plus HIV Ag/Ab (47.1%), VIDAS HIV DUO (45.4%), Genscreen HIV-1/2 v. 2 (41.5%), Vironostika HIV Uniform II Ag/Ab (40.9%), Ortho HIV-1/2 Ab Capture (40.3%), and Enzygnost HIV Integral (38.1%). When the time interval between two consecutive bleeds within a panel is relatively short, it is possible to estimate the delay (expressed as the number of days) between the first positive bleed detected by the most sensitive assay on a given seroconversion panel with each of the other assays on this panel. The time delay between blood sampling points in the commercial seroconversion panels used for the present study is on average relatively short (2–7 days) but may last up to 37 days, for example, panel PRB 923. The calculation model for time delays between assays established by the Paul Ehrlich Institute (Courouc´e et al., 1999) was used. This method considers that seroconversion is theoretically possible the day after the last negative follow-up sample. The total and the average number of days of time delay for the 25 panels were calculated in comparison with the most sensitive assay (Table 3). The mean time delay in days for detection of primary HIV infection in comparison with the most sensitive assay for each panel (last negative sample plus one day) ranged from 0.44 (AxSYM Combo) to 3.68 days (Genscreen HIV-1/2 v. 2). The mean time delay between p24 Ag detection assay and AxSYM Combo positive reactivity was 0.16 day (Table 3). If panels BCP 9017 and PRB 923 were excluded from analysis, the mean time delay between p24 Ag assay detection and the first positive of the different tests were AxSYM 0.26, Murex 0.35, Genscreen Plus and VIDAS DUO 1.26, Vironostika 1.96, Enzygnost 2.52, Genscreen v. 2 2.35, and Ortho 2.61 days. The mean time delay (excluding panels BCP9017 and PRB923) between the third-generation antibody Genscreen v. 2 assay and combined Ag/Ab assays was 2.09 days for AxSYM, 2.0 days for Murex, 1.09 days for
T.D. Ly et al. / Journal of Virological Methods 122 (2004) 185–194
191
DUO (13 of 1005), and Genscreen HIV-1/2 v. 2 (19 of 1005) showed a high percentage of initial false-positive results.
Genscreen Plus and Vidas DUO, and 0.39 day for Vironostika. The Enzygnost combined assay detected infections later than the Genscreen v. 2 assay with a mean delay of 0.17 day.
4. Discussion
3.3. Detection of 669 HIV positive samples
This study was designed to compare the sensitivity and specificity of six HIV combined antigen and antibody assays against two third-generation HIV antibody assays. The analysis of the results from this study has allowed us to make several clear conclusions about performance of the different assays. Seroconversion panels from patients during primary infection are informative in order to compare the sensitivity of screening tests. Consistently, the threshold of p24 Ag detection was 3–5 pg/ml with the HIV p24 Ag assays and ranged from 3.3 to more than 25 pg/ml with the fourth-generation assays. When compared with HIV p24 antigen assays, fourthgeneration tests were unable to detect the earliest positive panel member in five of 25 seroconversion panels. Consequently, the fourth-generation assays are not as sensitive as p24 Ag assays for detection of primary infection. However, the difference between the most sensitive combination assays compared to antigen assays is relative short. For the AxSYM
All 553 HIV-1 group M, 6 HIV-1 group O and 110 HIV-2 positive samples were detected by all assays except Genscreen v. 2 missed 1 HIV-1 group O, VIDAS DUO missed 1 HIV-1 subtype F, Enzygnost missed 1 HIV-1 subtype C, and Genscreen v. 2 missed 1 HIV-1 subtype B antibody positive sample (Table 4). 3.4. Specificity analysis using 1005 unselected HIV negative samples False-positive results obtained with the eight assays on 1005 HIV negative samples are shown in Table 5. The three assays with the best specificity based on repeat reactivity included Ortho HIV-1/2 Ab Capture with 100% specificity, Genscreen Plus HIV Ag/Ab with 99.9% specificity, and AxSYM HIV Ag/Ab Combo with 99.8% specificity. Vironostika HIV Uniform II Ag/Ab (28 of 1005), VIDAS HIV Table 5 Specificity of eight HIV screening assaysa Assays
No. of false-positive samples
Specificity (%) Initially reactive
Initially reactive
Repeatedly reactive
Fourth-generation Genscreen Plus HIV Ag/Ab AxSYM HIV Ag/Ab Combo Enzygnost HIV Integral Vironostika HIV Uniform II Ag/Ab Murex HIV Ag/Ab Combo VIDAS HIV DUO
1 2 4 28 5 13
1 2 3 4 4 5
Third-generation Ortho HIV-1/2 Ab Genscreen HIV-1/2 v. 2
0 19
0 6
a
99.9 99.8 99.6 97.21 99.5 98.7 100 98.1
Repeatedly reactive 99.9 99.8 99.7 99.6 99.6 99.5 100 99.4
Total number of HIV negative samples tested n = 1005.
Table 6 Summary of HIV detection for the 23 seroconversion panels Time delay in comparison with p24 Ag Assay n
Mean
Ranking
Performancea
Ranking
Fourth-generation AxSYM HIV Ag/Ab Combo Murex HIV Ag/Ab Combo Genscreen Plus HIV Ag/Ab Combo VIDAS HIV DUO Vironostika HIV Uniform II Ag/Ab Enzygnost HIV Integral
23 23 23 23 23 23
0.26 0.35 1.26 1.26 1.96 2.52
1 2 3 3 5 7
55.1 52.2 47.1 45.4 40.9 38.1
1 2 3 4 6 8
Third-generation Genscreen HIV-1/2 v. 2 Ortho HIV-1/2 Ab Capture
23 23
2.35 2.61
6 8
41.5 40.3
5 7
n = Total number of seroconversion panels (BCP9017 and PRB 923 excluded). a Performance = mean percentage of positive bleeds per panel.
192
T.D. Ly et al. / Journal of Virological Methods 122 (2004) 185–194
and Murex Combo assays, the detection of infection was delayed only 0.26–0.35 days relative to HIV p24 Ag assays. In a blood bank setting, a large majority of HIV infection would be detected by highly sensitive combination assays. Recently, the blood banks have implemented screening of blood products by nucleic acid tests (NAT). It is of interest to note that in the US, during the 3-year period starting in March 1999, NAT detected five antibody negative samples out of more than 19 million donations. Of these five NAT positive samples, one sample was also positive for antigen (Dodd et al., 2002). More recently, studies in low HIV endemic areas, Australia (Cunnigham et al., personal communication), Sweden, and Norway (Andersson et al., manuscript in preparation), and a high HIV endemic region, South Africa (Rensburg et al., personal communication), have demonstrated utility of HIV combined Ag/Ab assays for identifying early seroconversion samples. Antibody assays were used as the primary screening assay in Australia, Sweden, and Norway, while an antigen assay was used in addition to an antibody assay for detection of HIV infection in South Africa. At all sites, antigen positive early seroconversion samples were identified in side-by-side analysis of samples using the AxSYM Ag/Ab Combo assay and an antibody only assay. The incremental pick-up of antigen positive samples further exemplifies utility of combined assays in the clinical setting. Earlier detection of HIV-1 infection by the fourthgeneration assays compared to the third-generation assays is concordant with previous studies (Weber et al., 1998; G¨urtler et al., 1998; Van Binsbergen et al., 1998; Brust et al., 2000; Ly et al., 2001a,b). The earlier detection can be attributed to p24 antigen detection, reinforcing the importance of p24 Ag detection in earlier diagnostics (Courouc´e et al., 1988; Busch et al., 1990). It was of interest to note that there was no statistical difference in sensitivity with Enzygnost and Vironostika fourth-generation tests compared to Genscreen HIV-1/2 v. 2, which is one of the most sensitive commercially available third-generation screening EIAs (Laperche et al., 2000), and that this lack of sensitivity correlates with poor p24 antigen detection by the Enzygnost and Vironostika combined assays. The sensitivity of fourth-generation tests differed because of heterogeneous ability to detect either p24 Ag and/or antibodies (Tables 1–3). The AxSYM and/or the Murex Combo assays detected primary infection prior to all the other fourthgeneration combined assays in 12 out of 25 seroconversion panels (Table 2). This superior performance of the AxSYM and Murex Combo correlates with higher sensitivity for p24 Ag detection (Table 1). The average HIV Ag detection limit was 7 pg of p24 Ag/ml, for the AxSYM and 5 pg/ml for Murex Combo; these assays reduced diagnostic window by 1.6–2.17 days (Table 2) in comparison to the least sensitive combined p24 Ag and antibody assays (Enzygnost, and Vironostika assays; detection limit >25 pg of p24 Ag/ml). In a former study (Ly et al., 2001a), the antigen p24 detection limits for the Vironostika and the Enzygnost assays were 37.09 and 30.17 pg/ml, respectively, by using the HIV1 Ag panel SFTS 96 standard (French Society of Blood
Transfusion). The Genscreen Plus and VIDAS HIV DUO assays had average p24 detection limits of >14 and >11 pg/ml respectively (Table 1) and display similar seroconversion sensitivity (Tables 2 and 3). The AxSYM and Murex combined assays shorten delay for detection of infection 0.91–1.0 days relative to the Genscreen Plus and VIDAS HIV DUO assays. Interestingly, with panel BCP 9017 (Table 2), both thirdgeneration tests (Genscreen HIV v. 2 and Ortho Capture) detected infection earlier than the fourth-generation assays. This reflects a very rare clinical scenario in which the antiHIV IgM antibodies appear prior to detectable p24 Ag. Physicians using fourth-generation tests should be aware of this very rare circumstance as a possible source of false negative. Among six combination assays evaluated, the AxSYM Combo and Genscreen Plus assays identified the earliest seroconversion bleed (bleed 5, day 15) resulting in a day 10 earlier detection of the first positive result compared to the other combined assays. These data indicates that the AxSYM Combo and the Genscreen Plus assays have better antibody detection compared to other combined assays evaluated in this study. With panel PRB 952, VIDAS HIV DUO detected infection at day 10, i.e. earlier than third-generations tests (Genscreen HIV v. 2 and Ortho Capture). Opposite performance is found for later detection (day 14 and 17). This unexpected serological window for the VIDAS HIV DUO test probably reflects the transition from anti-HIV IgM to IgG antibodies. This could impair the accuracy of VIDAS HIV DUO, even if this kind of serological situation is rare. The genetic variability of HIV-1 may represent a major challenge for early antigen detection during primary HIV infection with fourth-generation tests (Apetrei et al., 1996). This study did not compare the ability of these eight HIV assays to detect early infection with different HIV-1 subtypes, because all of the seroconversion samples are HIV-1 group M, subtype B. However, using 31 HIV-1 group M subtypes and group O strains derived from cell culture, an impaired sensitivity for HIV-1 subtype C antigen detection was observed with VIDAS HIV DUO (Table 1). Our previous study with VIDAS HIV DUO Ultra also showed poor detection of subtype C antigen (Ly et al., 2001b) in contrast to the findings of Weber et al. (Weber et al., 2002). The Genscreen Plus failed to detect three of seven HIV-1 subtype B strains (Table 1). These results may reflect the quality of monoclonal antibodies used for the capture of p24 Ag and corroborate observations of Tersmette et al., 1989, who reported failure of monoclonal antibody to detect p24 Ag from certain strains of HIV. Among the fourth-generation assays, the Enzygnost and Vironostika assays showed poor antigen sensitivity using the viral isolate panel and also detected a lower number of seroconversion samples (Table 2). The prevalence of HIV-1 group M non-B subtypes and HIV-2 infections is increasing in Europe (Couturier et al., 2000). The genetic variability represents a challenge for HIV screening assays. In our study, all 669 HIV positive samples,
T.D. Ly et al. / Journal of Virological Methods 122 (2004) 185–194
of different genotypes and geographical origins, were detected by all eight assays with a high signal to cutoff except for the following samples. Genscreen v. 2 missed 1 group O sample perhaps due to absence of a specific reagent for group O detection. VIDAS DUO missed a subtype F that was discovered in France in an African woman; this case was notified to AFSSAPS (France Sanitary Safety of Health Products Agency). Enzygnost missed 1 subtype C and Genscreen v. 2 missed 1 subtype B; these two samples are indeterminate with Western blot HIV-1 and negative with Western blot HIV-2, but strongly positive with all other assays. Since fourth-generation EIAs combine two different test principles in one assay, the potential risk for non-specific reactivity may be higher than for second- and third-generation assays. Previous studies consistently reported a lower specificity for combined tests compared to third-generation assays (G¨urtler et al., 1998; Ly et al., 2000). In this study, we evaluated the specificity of third- and fourth-generation tests using sera sampled from 1005 unselected patients. Among the fourth-generation tests, Genscreen Plus and AxSYM Combo displayed the highest specificity, 99.9% and 99.8% respectively. The other fourth-generation displayed specificity ranging from 99.5% to 99.7%. We report here specificity for VIDAS DUO that is similar to that reported in a previous study with 29,657 unselected samples from five French public and private laboratories (Ly et al., 2000). To our knowledge, this is the first comparative side-by-side study of eight HIV assays to evaluate the specificity on the same panel of unselected routine samples. In summary, the present evaluation indicates that among these six combination assays, the AxSYM and Murex Combo assays have the best overall sensitivity followed by the Genscreen Plus and VIDAS HIV DUO (Table 6). The AxSYM Combo and Genscreen Plus also have good specificity based on the panel of 1005 unselected negative samples.
Acknowledgements We thank A. Bianchi (Laboratoire D´epartemental de Bondy), B. Lecolier, J.M. Hadjez (Centre d’Exploration Fonctionnelle, Paris), J. Lupu (Centre Suchet, Paris), J.Y. Mascart, A.J. Morvan (Laboratoire Morvan Mascart, Paimpol), J.P. Tr´eguier and J.J. Lemoine (Laboratoire Tr´eguier Lemoine, Lannion) for providing French HIV negative and positive samples. The authors also acknowledge Pierre Leroux (Abbott Laboratories, France) for his help in setting up the collaborative studies.
References Apetrei, C., Loussert-Ajaka, I., Descamp, S.D., Damond, F., Saragosti, S., Brun-Vezinet, F., Simon, F., 1996. Lack of screening test sensitivity during HIV-1 non-subtype B seroconversions. AIDS 14, 57–60.
193
Barin, F., Lahbabi, Y., Buzelay, L., Lejeune, B., Baillou-Beaufils, A., Denis, F., Mathiot, C., M’Boup, S., Vithayasai, V., Dietrich, U., Goudeau, A., 1996. Diversity of antibody binding to V3 peptides representing consensus sequences of HIV type 1 genotypes A to E: an approach for HIV type 1 serological subtyping. AIDS Res. Hum. Retroviruses 13, 1279–1289. Brennan, C.A., Lund, J.K., Golden, A., Yamaguchi, J., Vallari, A.S., Phillips, J.F., Kataaha, P.K., Jackson, J.B., Devare, S.G., 1997. Serologic and phylogenetic characterization of HIV-1 subtypes in Uganda. AIDS 11, 1823–1832. Brust, S., Duttmann, H., Feldner, J., Gurtler, L.G., Thorstensson, R., Simon, F., 2000. Shortening of the diagnostic window with a new combined HIV p24 antigen and anti-HIV-1/2/O screening test. J. Virol. Methods 90, 153–165. Busch, M.P., Taylor, P.E., Lenes, B.A., Kleinman, S.H., Stuart, M., Stevens, C.E., Tomasulo, P.A., Allain, J.P., Hollingsworth, C.G., Mosley, J.W., and the transfusion Safety Study Group, 1990. Screening of selected male blood donors for p24 antigen human immunodeficiency virus type 1. N. Engl. J. Med. 8, 1308– 1312. Courouc´e, A.M., Bouchardeau, F., Jullien, A.M., Faucher, V., Lenty, M., 1988. Blood transfusion and human immunodeficiency virus (HIV) antigen. Ann. Intern. Med. 108, 771–772. Courouc´e, A.M., The Groupe de Travail Retrovirus de la S.F.T.S., 1999. Tests de depistage combin´e des anticorps anti-VIH et de l’Age p24. Gazette Transfus 155, 4–17. Couturier, E., Damond, F., Roques, P., Fleury, H., Barin, F., Brunet, J.B., Brun-Vezinet, F., Simon, F., 2000. HIV-1 diversity in France, 1996–1998. The AC 11 laboratory network. AIDS 3, 289– 296. Dodd, R.Y., Notari IV, E.P., Stramer, S.L., 2002. Current prevalence and incidence of infectious disease markers and estimated window-period risk in the American Red Cross blood donor population. Transfusion 42, 975–979. G¨urtler, L.G., M¨uhlbacher, A., Michl, U., Hofmann, H., Paggi, G.G., Bossi, V., Thorstensson, R., G-Villaescusa, R., Eiras, A., Hernandez, J.M., Melchior, W., Doni´e, F., Weber, B., 1998. Reduction of the diagnostic window with a new combined p24 antigen and human immunodeficiency virus antibody screening assay. J. Virol. Methods 75, 27–38. Jagodzinski, L.L., Wiggins, D.L., McManis, J.L., Emery, S., Overbaugh, J., Robb, M., Bodrug, S., Michael, N., 2000. Use of calibrated viral load standards for group M subtypes of human immunodeficiency virus type 1 to assess the performance of viral RNA quantitation tests. J. Clin. Microbiol. 38, 1247– 1249. Laperche, S., Maniez-Montreuil, M., Courouce, A.M., 2000. Les tests de d´epistage combin´e de l’antig`ene p24 et des anticorps anti-VIH dans l’infection pr´ecoce a` VIH-1. Transfus. Clin. Biol. 7 (Suppl. 1), 18–24. Ly, T.D., Edlinger, C., Vabret, A., Morgan, O., Greuet, B., 2000. Contribution of combined detection assays of p24 antigen and antihuman immunodeficiency virus (HIV) antibodies in diagnosis of primary HIV infection by routine testing. J. Clin. Microbiol. 38, 2459– 2461. Ly, T.D., Laperche, S., Courouce, A.M., 2001a. Early detection of human immunodeficiency virus infection using third- and fourth-generation screening assays. Eur. J. Clin. Microbiol. Infect. Dis. 20, 104– 110. Ly, T.D., Martin, L., Daghfal, D., Sandridge, A., West, D., Bristow, R., Chalouas, L., Qiu, X., Lou, S.C., Hunt, J.C., Schochetman, G., Devare, S.G., 2001b. Seven human immunodeficiency virus (HIV) antigen-antibody combination assays: evaluation of HIV serconversion sensitivity and subtype detection. J. Clin. Microbiol. 39, 3122– 3128. Tersmette, M., Winkel, I., Groenick, M., Gruters, R.A., Spence, R.P., Saman, E., Van der Groen, G., Miedema, F., Huisman, J.G., 1989. Detection and subtyping of HIV-1 isolates with a panel of characterized
194
T.D. Ly et al. / Journal of Virological Methods 122 (2004) 185–194
monoclonal antibodies to HIV p24 gag. Virology 171, 149– 155. Van Binsbergen, J., Keur, W., Siebelink, A., Van de Graaf, M., Jacobs, A., De Rijk, D., Nijholt, L., Toonen, J., G¨urtler, L.G., 1998. Strongly enhanced sensitivity of a direct anti-HIV-1/2 assay in seroconversion by incorporation of HIV p24 Ag detection: a new generation Vironostika HIV Uniform II. J. Virol. Methods 76, 59–71.
Weber, B., Mbargane Fall, E.H., Berger, A., Doerr, H.W., 1998. Reduction of diagnostic window by new fourth-generation human immunodeficiency virus screening assay. J. Clin. Microbiol. 36, 2235–2239. Weber, B., Berger, A., Rabenau, H., Doerr, H.W., 2002. Evaluation of a new combined antigen and antibody human immunodeficiency virus screening assay, Vidas HIV Duo Ultra. J. Clin. Microbiol. 40, 1420–1426.