Field validation and application of the luminex triplex HIV assay to estimate HIV prevalence and HIV-1 incidence in Nigeria

Abstract

HIV cross-sectional surveys require multi-layered testing with several tests to estimate HIV prevalence and HIV-1 incidence. We evaluated the performance and accuracy of the newly developed HIV Triplex assay to diagnose HIV-1 and HIV-2 and detect HIV-1 recent infections using plasma samples from the 2018 Nigeria AIDS Indicator and Impact Survey (NAIIS). Plasma samples from consenting HIV-positive (n=2,773) and a subset of HIV-negative samples (n=7,196), as determined by the national rapid testing algorithm, followed by Bio-Rad Geenius HIV-1/2 Supplemental Assay and Western Blot, aged 18 months - 64 years, were tested using the Luminex-based HIV Triplex assay. The assay classified specimens as HIV-1 positive, HIV-2 positive, dual (HIV-1 & 2) infections, or HIV-seronegative. All HIV-1 and dual infections were further classified as either HIV-1 recent (<6 months) or long-term (LT) based on mean fluorescent intensities and compared with the LAg-Avidity EIA as the reference. Multiplex results were analyzed and compared with the final NAIIS survey data for unweighted HIV prevalence and HIV-1 incidence. The diagnostic sensitivity and specificity of the HIV Triplex assay was 99.71% and 99.37%, respectively, with a kappa of 0.987 when compared to NAIIS survey results. Percent agreement between the HIV Triplex assay and the LAg-Avidity EIA for recent and LT classification was 98.86% with a kappa of 0.80 [CI: 0.71-0.89] and a Spearman-ranked correlation (ρ) of 0.689. A small number (n=45; 0.63%) of the subset of negatives tested were classified by the multiplex assay as either HIV-1 positive (n=35) or HIV-2 positive (n=10). Nevertheless, the HIV Triplex assay agreed with NAIIS HIV-negative survey results (99.37%). Using these results as they were, unweighted estimates of HIV prevalence for both HIV Triplex assay and NAIIS test results were similar (1.62% [95% CI: 1.56-1.68] and 1.60% [95% CI: 1.54-1.66], respectively) with overlapping confidence. After adjusting for viral load and anti-retroviral therapy, HIV-1 unweighted incidence for ages ≥15 years, using HIV Triplex assay data, was 0.70 per 1,000 [95% CI: 0.40-0.90]. This is similar to the unweighted incidence using the LAg-based RITA (recent infection testing algorithm) of 0.80 per 1,000 [95% CI: 0.60-1.10]. The HIV Triplex assay combines several assays in one, providing highly accurate results for estimating HIV prevalence and HIV-1 incidence in surveys. This assay has the potential to simplify cross-sectional surveys making them less expensive, easier, and quicker.