730    

Background:  To estimate incidence from cross-sectional population surveys, the Serological Testing Algorithm for Recent HIV Seroconversion (STARHS) uses the results of a standard and a Less-Sensitive (LS) EIA to classify HIV infections as recent or long standing. A modified version of the Vironostika HIV-1 MicroElisa System (V-LS) is currently used as the LS-EIA for STARHS. This assay will soon be replaced by the Vironostika HIV-1 Plus O MicroElisa System (VplusO-LS). The VplusO-LS assay has also been adapted for use in STARHS. To investigate the comparability of VplusO-LS and V-LS in STARHS, longitudinal specimens from the HIVNET Infected Participant Cohort (IPC) were tested using both kits.

Methods:  The IPC specimen set (1009 specimens from 105 recent seroconverters with known dates of last negative and first positive HIV test) was tested in triplicate on both assays. Correlation was assessed by fitting a linear regression of the square root standardized optical densities from the Plus O-LS on the square root standardized optical densities from the V-LS. Assay performance on low positive and high positive control materials was also assessed. A subset of 324 specimens from 59 patients with 3 or more specimens collected before starting ART (mean length of follow-up, 504 days) was used to calculate mean window periods (number of days between detection of HIV antibody with a standard EIA and a threshold standardized optical densities of 1.0 with the LS-EIA) and 95% confidence intervals for both assays.

Results:  V-LS classified 459 samples as incident (standardized optical densities < 1), and VplusO classified 472 as incident (90.2% concordance). The 2 tests showed good overall correlation, with an R² of 0.848, and a slope of 0.996. The VplusO assay had a slightly better coefficient of variation on the control materials (18.9% on low positive control and 19% on high positive control vs 18.4% on low positive control and 26.7% on high positive control for V-LS). With both assays, specimen standardized optical densities’s diminished after patients started ART. The mean window period for V-LS was 218 days (95% confidence interval (CI 193 to 250 days). The mean window period for the VplusO-LS was 203 days (95% CI 185 to 243 days).

Conclusion:  The VplusO-LS gave comparable results to the V-LS for the classification of recent HIV infection in terms of overall specimen response (standardized optical densities) and overlap of window periods. Further validation is needed to determine if the two assays provide comparable estimates of HIV incidence when applied to specimens from cross-sectional surveys.

Keywords: STARHS; HIV Incidence; Enzyme Immuno Assay