Comparative evaluation of subtyping tools for surveillance of newly emerging HIV-1 strains

HIV-1 non-B subtypes/circulating recombinant forms (CRFs) are increasing worldwide. Since subtype identification can be clinically relevant, we assessed the added value in HIV-1 subtyping using updated molecular phylogeny (Mphy) and the performance of routinely used automated tools. Updated Mphy (2015 updated reference sequences), used as a gold standard, was performed to subtype 13,116 HIV-1 protease/reverse transcriptase sequences and then compared with previous Mphy (reference sequences until 2014) and with COMET, REGA, SCUEAL, and Stanford subtyping tools. Updated Mphy classified subtype B as the most prevalent (73.4%), followed by CRF02-AG (7.9%), C (4.6%), F1 (3.4%), A1 (2.2%), G (1.6%), CRF12-BF (1.2%), and other subtypes (5.7%). A 2.3% proportion of sequences were reassigned as different subtypes or CRFs because of misclassification by previous Mphy. Overall, the tool most concordant with updated Mphy was Stanford-v8.1 (95.4%), followed by COMET (93.8%), REGA-v3 (92.5%), Stanford-old (91.1%), and SCUEAL (85.9%). All the tools had a high sensitivity (≥98.0%) and specificity (≥95.7%) for subtype B. Regarding non-B subtypes, Stanford-v8.1 was the best tool for C, D, and F subtypes and for CRFs 01, 02, 06, 11, and 36 (sensitivity, ≥92.6%; specificity, ≥99.1%). A1 and G subtypes were better classified by COMET (92.3%) and REGA-v3 (98.6%), respectively. Our findings confirm Mphy as the gold standard for accurate HIV-1 subtyping, although Stanford-v8.1, occasionally combined with COMET or REGA-v3, represents an effective subtyping approach in clinical settings. Periodic updating of HIV-1 reference sequences is fundamental to improving subtype characterization in the context of an effective epidemiological surveillance of non-B strains.