Discrimination between recent and non-recent HIV infections using routine diagnostic serological assays

The suitability of routine diagnostic HIV assays to accurately discriminate between recent and non-recent HIV infections has not been fully investigated. The aim of this study was to compare an established HIV recency assay, the Sedia limiting antigen HIV avidity assay (LAg), with the diagnostic assays; Abbott ARCHITECT HIV Ag/Ab Combo and INNO-LIA HIV line assays. Samples from all new HIV diagnoses in Ireland from January to December 2016 (n = 455) were tested. An extended logistic regression model, the Spiegelhalter-Knill-Jones method, was utilised to establish a scoring system to predict recency of HIV infection. As proof of concept, 50 well-characterised samples were obtained from the CEPHIA repository whose stage of infection was blinded to the authors, which were tested and analysed. The proportion of samples that were determined as recent was 18.1% for LAg, 6.4% with the ARCHITECT, and 14.5% in the INNO-LIA assay. There was a significant correlation between the ARCHITECT S/CO values and the LAg results, r = 0.717, p < 0.001. ROC analysis revealed that an ARCHITECT S/CO < 250 had a sensitivity and specificity of 90.32% and 89.83%, respectively. Combining the Abbott ARCHITECT HIV Ag/Ab Combo assay and INNO-LIA HIV assays resulted in an observed risk of being recent of 100%. Analysis of the CEPHIA samples revealed a strong agreement between the LAg assay and the combination of routine assays (κ = 0.908, p < 0.001). Our findings provide evidence that assays routinely employed to diagnose and confirm HIV infection may be utilised to determine the recency of HIV infection.