Performance of the BioPlex 2200 Multiplexing Immunoassay Platform for the Detection of Herpes Simplex Virus Type 2 Specific Antibodies in African Settings

Abstract

The BioPlex platform was evaluated for the detection of herpes simplex virus 2 (HSV-2) antibodies in sub-Saharan Africa individuals in comparison to clinicovirological standards and compared to HerpeSelect. The sensitivities and specificities were, respectively, 88.9% and 93.5% for BioPlex and 89.9% and 92.7% for HerpeSelect. The agreement between both assays was 95.7%.

TEXT

The identification of individuals infected with herpes simplex virus (HSV) is critical for the management of herpes-related conditions, such as genital herpes, as well as for epidemiological studies and clinical or intervention trials. The performance of commercially available enzyme-linked immunosorbent assays (ELISAs) in detecting HSV-2-specific glycoprotein G2 (gG2) antibodies varies significantly between different study populations (1, 8, 17). Several gG2-specific ELISAs have been shown to lack specificity in the detection of HSV-2 antibodies among African (4, 17) and Brazilian (15) populations, requiring that the ELISA cutoff value be increased for better specificity in the general population. However, a critical element in evaluating the performance of these assays is the clinical stage of HSV-2 infection (2, 11, 14). For example, the FDA-approved HerpeSelect gG2-specific ELISA (Focus Technologies, Cypress Hill, CA) had high sensitivity in predicting genital HSV-2 infection, particularly first episodes of HSV-2 ulcers, in patients with genital ulcer disease (GUD) from the Central African Republic and Ghana (11).

In the present study, we evaluated the performance of the new BioPlex 2200 immunoassay platform (3) (Bio-Rad Laboratories, Hercules, CA) in detecting HSV-1 and HSV-2 antibodies in populations living in sub-Saharan Africa, including patients with proven genital HSV-2 infection. We used stored sera obtained during cross-sectional studies from two distinct clinicovirological populations. First, sera were obtained between May and July 2009 from 200 HIV-seronegative children (age 0 to 17) seen at the Complexe Pédiatrique of Bangui, Central African Republic, and clinically asymptomatic for genital herpes. Informed consent was obtained from the parents or guardians of these children or from the older children themselves. Second, sera were collected from women presenting with GUD at sexually transmitted infection (STI) clinics in Bangui, Central African Republic, and in Accra and Kumasi, Ghana, who were enrolled in a randomized placebo-controlled trial of acyclovir between May 2003 and October 2005 (ClinicalTrials.gov registry no. NCT00158483) (12, 13). Consenting women with clinically verified GUD were interviewed and examined and submitted blood and genital samples at enrollment for (i) HSV-2 serology using HerpeSelect ELISA and (ii) GUD etiology and the presence of HSV-2 DNA in lesional and cervicovaginal lavages using molecular tests, as described previously (10, 12). From the 226 women enrolled in the trial who had detectable genital HSV-2 DNA and who were either HSV-2 seropositive or seronegative, 208 serum samples were available for this study (12).

Sera were aliquoted, frozen at −20°C, and further tested for HSV-1- and HSV-2-specific antibodies using the BioPlex 2200 HSV-1 and HSV-2 IgG kit. The BioPlex 2200 platform is a fully automated instrument that combines flow cytometric technology with antigen-coated fluoromagnetic bead chemistry. The BioPlex 2200 HSV-1 and HSV-2 IgG kit detects and differentiates IgG antibodies to HSV-1 and HSV-2 by using beads coated with recombinant peptides encompassing the gG1 N-terminal region (amino acids 1 to 173) and the region between amino acids 205 to 240 of the gG2, respectively. For every sample processed, three internal quality control beads are employed that can check for detector fluctuations, sample integrity, and nonspecific binding. The results are reported according to their antibody index (AI), with values of <0.9 considered negative, 0.9 to 1.0 equivocal, and >1.0 positive.

Serum samples were tested in parallel using the HerpeSelect gG2 ELISA, and the results were expressed using AIs of 1.1, as recommended by the manufacturer, and 3.5, as recommended by many authors to improve the assay's specificity in African individuals (4, 6, 7, 9, 15). The kappa statistic was used to assess the concordance between the two assays. The sensitivity and specificity of both assays were determined in comparison with clinicovirological reference standards. Samples positive for HSV-2 DNA were taken as a group with high posterior probability to be HSV-2 seropositive and were used as the clinicovirological standard to determine sensitivity. Samples from children with high posterior probability to be HSV-2 seronegative were used as the clinicovirological standard to determine the specificity. It is customary in this instance to use samples from children over the age of 1 year (to avoid the presence of passive maternal antibodies) and under the age of sexual debut (in practice, before the teenage years). We therefore selected samples from 139 children aged 1 to 10 years from the 200 asymptomatic children as a reference standard in this study.

Using the Bio-Rad BioPlex 2200 immunoassay kit, 158 (79%) and 12 (6.0%) of the 200 asymptomatic children were found to be seropositive for HSV-1 and HSV-2, respectively. Figure 1 shows clear differences in the patterns of HSV-1 and HSV-2 seroprevalence by age. The HSV-1 seroprevalence was already 50% among infants aged <1 year and steadily increased to 100% in young people aged 16 to 17 years. With regard to HSV-2, 25% of the infants aged <1 year had detectable antibodies, likely of maternal origin. The prevalence at older ages was low (below 10%). These observations are consistent with the natural history of HSV-1 and HSV-2 infections as reported in sub-Saharan Africa or elsewhere, with near universal infection by HSV-1 in early childhood and rapid rise of HSV-2 infection around the time of puberty and beginning of sexual activity (5, 16, 18).

Fig. 1.

Distribution of seropositivity to HSV-1 and HSV-2 IgG-specific antibodies in 200 healthy children living in the Central African Republic, by age, using the Bio-Rad BioPlex 2200 system at an antibody index of 1.0 for cutoff positivity.

Among the 208 samples from patients with proven genital HSV-2 infection (true positive cases) and available results for both assays, the sensitivities of serological testing were 88.9% (95% confidence interval [CI], 83.9% to 92.9%) and 89.9% (95% CI, 85.0% to 93.6%) for HSV-2 BioPlex (AI > 1.0) and HerpeSelect (AI > 1.1), respectively (Table 1). The concordance between the two assays was high (200/208; agreement = 96.2%; 95% CI, 92.3% to 98.3%; κ = 0.80; P < 0.0001). The HIV serostatus was available for 207 individuals. For both assays, sensitivity was higher in the 125 HIV-1-seropositive individuals than in the 82 HIV-seronegative individuals. For BioPlex, the sensitivities were 98.4% (95% CI, 94.3% to 99.8%) in HIV-1 seropositive and 74.4% (95% CI, 63.6% to 83.4%) in HIV-seronegative individuals, respectively (P < 0.001); for HerpeSelect, the sensitivities were 96.8% (95% CI, 92.0% to 99.1%) in HIV-1-seropositive individuals and 79.3% (95% CI, 68.9% to 87.4%) in HIV-seronegative individuals (P < 0.001). In this cohort, a high proportion (84%) of first-episode HSV-2 infections (n = 25) occurred in HIV-seronegative women, of whom roughly 75% seroconverted at day 28 (11). This explains why some true HSV-2 infections were HSV-2 seronegative at inclusion and, hence, the apparently reduced sensitivity among HIV-uninfected women. The concordance between assays was higher in the HIV-seronegative individuals than in the HIV-1-seropositive individuals, despite lower agreement (HIV seronegative, agreement = 92.7% and κ = 0.80; HIV-1 seropositive, agreement = 98.4% and κ = 0.66). The slightly better concordance observed in HIV-seronegative individuals is due to a higher expected agreement in HIV positives because of the very high HSV-2 seroprevalence among HIV-positive individuals. Using the 3.5 cutoff for HerpeSelect, the sensitivities were 82.4% (95% CI, 74.6% to 88.6%) in HIV-1-seropositive and 63.4% (95% CI, 52.4% to 73.8%) in HIV-seronegative individuals, respectively (P = 0.001).

Table 1.

Performance of HSV-2 Bio-Rad BioPlex 2200 system and HSV-2-specific HerpesSelect ELISA and overall agreement between both assays

Parametera	Result [% (95% CI)] at indicated cutoff valueb
	Manufacturer's value	>3.5
BioPlex sensitivity	88.9 (83.9–92.9)	82.2 (76.3–87.2)
HerpeSelect sensitivity	89.9 (85.0–93.6)	75.0 (68.5–80.7)
Bioplex specificity	93.5 (88.1–97.0)	97.1 (92.8–99.2)
HerpesSelect specificity	92.7 (87.0–96.4)	97.8 (93.8–99.6)
Agreement between both assays	95.7 (92.9–97.5)	91.9 (88.5–94.5)
κ coefficientc	0.91	0.84

^aSensitivity and specificity were determined in comparison with clinicovirological reference standards. Samples from molecularly documented HSV-2 genital infection were used as the reference to determine sensitivity. Samples from asymptomatic children aged 1 to 10 years were used as the reference to determine specificity.

^bThe manufacturer's recommended positivity cutoff for BioPlex is 1.0 and for HerpeSelect ELISA is 1.1; the use of a higher index value of 3.5 for HerpeSelect has been recommended to increase the specificity in African populations(4).

^cThe concordance between the Bio-Rad BioPlex 2200 system and the HerpeSelect ELISA is evaluated by the kappa coefficient.

Among the 139 samples from children aged 1 to 10 years (assumed to be true negative cases), the specificities of both assays were high and comparable: 93.5% (95% CI, 88.1% to 97.0%) for BioPlex (AI > 1.0) and 92.7% (95% CI, 87.0% to 96.4%) for HerpeSelect (AI > 1.1) (Table 1). The specificity increased to 97.8% (95% CI, 95.4% to 100%) for HerpeSelect when the AI was raised to 3.5. Interestingly, using a higher cutoff for BioPlex (3.5) as well gave slightly better specificity (97.1%; 95% CI, 92.8% to 99.2%). Concordance between the two assays at the lower cutoff was high (130/137; agreement = 94.9%; 95% CI, 89.8% to 97.9%; κ = 0.60; P < 0.0001).

The overall concordance between the two assays was excellent at both the manufacturer's and the 3.5 cutoff (κ = 0.91 and κ = 0.84, respectively; P < 0.0001) (Table 1).

Taken together, our observations demonstrate that the Bio-Rad BioPlex 2200 immunoassay has strikingly similar performance to the HerpeSelect gG2 ELISA, with the advantage of simultaneous detection of HSV-1 and HSV-2 antibodies, thus providing an additional useful tool for HSV type-specific serology which can be used in epidemiological, clinical, or intervention studies in African populations.