Relevance of cutoff on a 4th generation ELISA performance in the false positive rate during HIV diagnostic in a low HIV prevalence setting

Background: Despite the high specificity of fourth-generation enzyme immunoassays (4th-gen-EIA) for screening during HIV diagnosis, their positive predictive value is low in populations with low HIV prevalence. Thus, screening should be optimized to reduce false positive results. Objectives: The influence of sample cutoff (S/CO) values by a 4th-gen-EIA with the false positive rate during the routine HIV diagnosis in a low HIV prevalence population was evaluated. Study design: A total of 30,201 sera were tested for HIV diagnosis using Abbott Architect (R) HIV-Ag/Ab-Combo 4th-gen-EIA at a hospital in Spain during 17 months. Architect S/CO values were recorded, comparing the HIV-1 positive results following Architect interpretation (S/CO >= 1) with the final HIV-1 diagnosis by confirmatory tests (line immunoassay, LIA and/or nucleic acid test, NAT). ROC curve was also performed. Results: Among the 30,201 HIV performed tests, 256 (0.85%) were positive according to Architect interpretation (S/CO >= 1) but only 229 (0.76%) were definitively HIV-1 positive after LIA and/or NAT. Thus, 27 (10.5%) of 256 samples with S/CO >= 1 by Architect were false positive diagnose. The false positive rate decreased when the S/CO ratio increased. All 19 samples with S/CO <= 10 were false positives and all 220 with S/CO > 50 true HIV-positives. The optimal S/CO cutoff value provided by ROC curves was 32.7. No false negative results were found. Conclusions: We show that very low S/CO values during HIV-1 screening using Architect can result HIV negative after confirmation by LIA and NAT. The false positive rate is reduced when S/CO increases.