Abstract Body

Assay based staging of HIV for recency of infection has multiple purposes: at the population level, it enables cross-sectional incidence estimation. At the individual level, at diagnosis (as available in some national programmes and being contemplated in other large scale systems) recent/non-recent infection classification helps guide psychosocial support strategies, contact tracing, or inclusion in clinical trials. However, as recency testing is currently a specialist service using custom incidence assays, there is additional expense, and a delay between HIV diagnosis and the delivery of the recency result. Given the high dynamic range inherent in widely used (primarily chemiluminescent) diagnostic platforms, we explore whether this unutilized information (signal/cutoff ratio, S/CO) can immediately stage new diagnoses as recent/non-recent, or at least identify specimens that need not be referred for specialised recency testing, allowing prioritisation of specimens.

2500 specimens with good clinical characterisation were tested diagnostically on the Abbott Architect 4th generation assay and by Sedia Limiting Antigen Assay for recency determination. We compared the recency classifications based on the two platforms through a number of regressions and correlations, and estimated mean duration of recent infection (MDRI) for a number of thresholds on Architect S/CO values.

At Architect S/CO < 150, MDRI was 163 days, and ART naïve False Recent Rate (FRR) was 2.7%, comparable with previously published LAg values. Individual Architect/LAg results were highly correlated (r=0.81). Figure 1 shows the probability of specimens classifying as LAg recent (ODn<1.5) as a function of ARCHITECT S/CO: more than 80% of specimens with S/CO<100, less than 5% with S/CO>400, and less than 1% of specimens with S/CO>500 scored recent by LAg (ODn<1.5).

An unmodified chemiluminent HIV immunoassay assay can provide comparable information to a custom recency staging assay. In the context of population HIV incidence surveillance, this can lead to simple, readily transferable and cheaper testing protocols. At patient diagnosis, this could enirely eliminate the costs and delays associated with additional recency testing, or at least enable selection of a subset of specimens (high- or low- enough reactivity) for which final staging assignments do not require additional testing.