The challenge of discriminating between HIV‐1, HIV‐2 and HIV‐1/2 dual infections

Objectives Discrimination between HIV‐1 and HIV‐2 is important to ensure appropriate antiretroviral treatment (ART) and epidemiological surveillance. However, serological tests have shown frequent mistyping when applied in the field. We evaluated two confirmatory tests, INNO‐LIA HIV I/II Score and ImmunoComb HIV 1/2 BiSpot, for HIV type discriminatory capacity. Methods Samples from 239 ART‐naïve HIV‐infected patients from the Bissau HIV Cohort in Guinea‐Bissau were selected retrospectively based on the initial HIV typing performed in Bissau, ensuring a broad representation of HIV types. INNO‐LIA results were interpreted by the newest software algorithm, and three independent observers read the ImmunoComb results. HIV‐1/HIV‐2 RNA and DNA were measured for confirmation. Results INNO‐LIA results showed 123 HIV‐1 positive samples, 69 HIV‐2 positive and 47 HIV‐1/2 dually reactive. There was agreement between INNO‐LIA and HIV‐1/HIV‐2 RNA and DNA detection, although not all HIV‐1/2 dually reactive samples could be confirmed by the nucleic acid results. Overall, the observers found that the ImmunoComb results differed from the INNO‐LIA results, with agreements of 90.4, 91.2 and 92.5%, respectively, for HIV‐1, HIV‐2 and HIV‐1/2. The combined kappa‐score for agreement between the three observers was 0.955 (z‐score 35.1; P < 0.01). Of the HIV‐2 mono‐reactive samples (INNO‐LIA), the three observers interpreted 24.6–31.9% as HIV‐1/2 dually infected by ImmunoComb. None of these samples had detectable HIV‐1 RNA or DNA. Conclusions There was accordance between INNO‐LIA calls and nucleic acid results, whereas ImmunoComb overestimated the number of HIV‐1/2 dually infected patients. Confirmatory typing is needed for patients diagnosed with HIV‐1/2 dual infection by ImmunoComb.