HIV-1 late diagnosis: Strategies to overcome the misclassification of individuals acutely infected with HIV-1 as individuals diagnosed late

Objectives Late HIV diagnosis is associated with a higher impact on treatment outcomes and a potential for prolonged transmissibility of HIV-1 infection. The consensus definition for late HIV diagnosis is problematic. It was updated in 2022; however, this definition relies on information that might not be clinically available. This study aimed to assess late HIV diagnosis using alternative parameters, in addition to the definition of clusters of differentiation (CD4) cell count, namely, sequence ambiguity rate and estimated time of infection inferred through phylogenetic analysis. Methods Clinical, socio-demographic, and genotypic information from 3668 antiretroviral therapy–naïve individuals living with HIV was retrieved from the REGA database. Individuals were classified according to three approaches: (i) CD4 cell count, (ii) sequence ambiguity rate, and (iii) phylogenetic reconstruction using TreeTime to estimate the time of most recent common ancestor (MRCA) as a proxy for time of infection. Results Based on CD4 cell count, 53.8% of individuals had a late diagnosis and 46.2% had a non-late diagnosis. Based on sequence ambiguity rate, 57.8% had a chronic and 42.2% had a recent infection, and 86.4% had an estimated time of infection of more than 3 years, whereas 13.6% had less than 3 years. A total of 114 individuals were classified as diagnosed late by CD4 criteria and showed evidence of recent infection based on low ambiguity rates and MRCA estimates under 3 years. These individuals had significantly lower viral loads than those with true late diagnoses (median 61,358 vs 134,730 copies/ml; P <0.001). Overall, 41% of individuals were consistently classified across all three methods. Conclusions The definition of late diagnosis remains a major challenge. Alternative and complementary methods, such as the use of viral loads, combined with some more clinical information, may improve the lack of baseline data.