Since the use of a serologic assay for estimating HIV-1 incidence from cross-sectional samples was first described in 1998 [1], considerable emphasis has been placed on the development and optimization of laboratory techniques for distinguishing recent from long-term HIV infection. Accurate methods of estimating HIV incidence are essential for monitoring trends in transmission, identifying high-risk populations, and evaluating the efficacy of prevention measures [2]. Typically, tests for recent infection (TRIs) have involved the measurement of HIV-specific antibody responses or avidity, based on the increasing pattern of antibody reactivity observed in HIV-1 infected individuals post-seroconversion [3,4]. To enhance the predictive value between samples from recent and long-term infected individuals, serologic assay approaches have included the modification of commercial kits, to reduce the sensitivity of detection [1,5,6], or a dissociation step [7,8], as a measure of antibody avidity. In the absence of manufacturer support, off-label use of commercial diagnostic tests may be problematic, especially in disseminating a common protocol to all potential users and relying on continued production. Several in-house serologic assays have also been developed, specifically for the purpose of determining recent HIV-1 infection [9-12]. For widespread use, in-house techniques must be validated to assess intra/inter-laboratory assay performance prior to large-scale production.

The BED capture immunoassay (BED-CEIA), the most widely used assay designed solely for the purpose of estimating HIV incidence, measures the proportion of antibody directed against an immunodominant branched gp41 peptide [9,13]. The intra- and inter-laboratory variability of the BED-CEIA assay has been described in detail by Dobbs et al. [14]. The BED-CEIA assay was the first TRI to become commercialized and has been used worldwide for HIV-1 incidence surveillance purposes [15-18]. Despite its widespread availability, the assay has undergone scrutiny based on reports describing high false-recent rate (FRRs) in some populations, which can lead to the overestimation of HIV incidence [19-22]. Recently, the HIV-1 Limiting Antigen (LAg)-Avidity EIA, which measures binding of high-avidity antibodies to a subtype-conserved, recombinant gp41 protein, has been commercialized for HIV-1 surveillance use [10,23]. Given the public health implications of inaccurate incidence estimates, new and developing TRIs must be comprehensively validated prior to their implementation. We have described the development of an in-house, HIV-1-specific Bio-Plex assay for determining recent HIV-1 infection [11]. The multiplex assay measures both HIV-specific antibody levels and avidity to multiple analytes in a single 96-well assay plate. Recent studies have shown the FRR can be eliminated or reduced when using a multi-test algorithm approach for determining recent infection [24,25]. Furthermore, improved HIV-1 incidence estimates have been demonstrated using algorithms of multiple analyte measures obtained from the HIV Bio-Plex assay [26].

Sine initial proof-of-concept studies with the in-house multiplex assay have yielded promising results; we investigated the feasibility of developing a kit for widespread dissemination and use. We evaluated a prototype multiplex kit to measure HIV-1-specific antibody levels and avidity. The performance of the kits and intra-assay variation were compared to our in-house methods [11]. Inter-laboratory performance was assessed amongst three participating laboratories. These results will aid in determining the practicality of developing a bead-based multiplex kit for estimating HIV incidence and will enable further development of the assay, with the intent of producing a consistent product for widespread use.

The HIV-1-specific Bio-Plex assay, described previously [11,26], allows for the measurement of HIV-specific antibody levels and avidity to multiple analytes for improved cross-sectional HIV-1 incidence estimates. Thus far, all of the assay components have been acquired or produced in-house and validation studies have been performed in a single laboratory. Although preliminary studies indicate that the in-house multiplex assay yields promise, the current format may not be practical for widespread dissemination and use. We have demonstrated the feasibility of producing a commercial kit with intra-assay and inter-operator reproducibility, amongst three separate laboratories.

Since the performance of the multiplex assay relies primarily on binding of antibodies to antigen-coupled microspheres, it is essential to demonstrate that microsphere sets and assay reagents can be produced outside the reference laboratory and yield reproducible results. Overall, consistent results were observed between the in-house assay and commercial kit, with longitudinal seroconversion panels. The calibrator is an important component of the assay as it controls for slight variations in MFI readings from run-to-run and may serve to normalize lot-to-lot differences with future commercial kits, as observed with the BED-CEIA assay [14]. Reactivity with the calibrator, however, appeared to affect the range in normalized MFI values observed for each analyte. The kit gp160-n and gp120-n exhibited a greater range of reactivity and approximately half the MFI value for the calibrator, as compared to the in-house assay, which is desirable for increased distinction in reactivity between samples. Since the HIV-1 recombinant proteins coupled to the microspheres were obtained from the same commercial source for both the in-house assay and kit, variations in reactivity may be due to lot-to-lot differences of the recombinant proteins or lab-specific coupling procedures. The calibrator and microsphere sets in the pilot kit will be further validated to ensure maximum separation in reactivity between samples from recent and long-term infections. Furthermore, the current calibrator was created in the reference laboratory by pooling plasma from HIV positive individuals and, therefore, has limited availability. For kit production, we will evaluate the possibility of commercial calibrator production.

In the absence of validated assay cutoffs and mean durations of recency (MDRs), the kit performance is best determined by comparing the reactivity of well-characterized specimens from individuals with known recent and long-term infections. Despite slight differences in the normalized MFI value or avidity index scale between the in-house assay and the kit, there was minimal to no overlap in reactivity between known recent and long-term samples, with the exception of gp41-n, as described previously [11]. Although the current recombinant gp41 protein included in the assay does not provide discrimination based on antibody levels, due to relatively high reactivity with both recent and long-term samples, the avidity measures display a wide dynamic range and may serve as a more robust indicator of recent infection. We are currently evaluating several subtype-conserved recombinant gp41 proteins and peptides to identify an analyte that provides a measurable distinction in both antibody levels and avidity with specimens from individuals infected with diverse HIV-1 subtypes. One of the major benefits of the Bio-Plex format is the relative ease in which analytes can be added or removed from the assay, without affecting the remaining components or assay protocol.

An inter-laboratory assessment of the prototype multiplex kits was performed to evaluate operator performance of the assay protocol and potential variation in assay results, which indicated acceptable concordance amongst the three participating laboratories and similar separation in reactivity between recent and long-term specimens. These results are encouraging, given that the reference laboratory distributed the assay protocol to all operators, but did not provide training on the assay to the test operators. One laboratory exhibited poor correlation with the other laboratories for gp160-n, although the median inter-laboratory CV remained within acceptable limits (Table 2). The divergent results could not be investigated further due to limited kit availability; however, this underscores the importance of operator training prior to initiation of future reproducibility studies. Inter-laboratory correlation for gp41-n was deceptively low due to a limited range of reactivity for the analyte, but the median CV indicated reproducible results. Overall, the avidity measures were robust and highly reproducible, which is not surprising given that avidity index is calculated relative to an untreated control on the same assay plate.

For initial development and validation studies, all samples were tested in duplicate and the average MFI value of the replicates was used to calculate the normalized MFI value or avidity index. However, the low intra-assay CVs reported by all three laboratories indicate that reliable results may be provided with a single sample reading, which can reduce reagent cost. In the future assay format, a cutoff value will be selected for the IgG control bead set to identify sample output values that are invalid due to plating or machine reading error and would require repeat testing. Further assay refinements may include incorporation of subtype-conserved analytes, given that all recombinant proteins used in the current assay format are subtype-B derived. It is not known how the current format of the assay will perform with specimens from non-B subtype infections. Long-term stability studies will be needed to assess the shelf-life and storage temperature for the kit, as well as large scale inter-laboratory evaluations to assess operator acceptability and kit performance.

In summary, we have demonstrated the feasibility of producing a multiplex bead-based assay for measuring HIV antibody levels and avidity in kit form. Given the potential for improved incidence estimates based on intra-assay algorithms of multiple analyte measures, a commercial kit will allow for the transfer of technology to multiple and diverse laboratories for widespread evaluation and use of the assay.