Limitations in online tools to identify HIV-1 subtypes
Simon Collins, HIV i-Base
Clive Loveday from ICVC collaborative research group, who lead much of the early interest in research into the relevance of sub-type on treatment response in the UK, reported a significant discordance between different online resources used to define HIV sub-type. 
This study submitted 1002 consecutive clinical pol samples to five popular online interpretation tools: Stanford, NCIB, REGA, STAR and Los Alamos RIP 2.0.*
100% of samples were assigned a sub-type in the Stanford, NCIB and Los Alamos databases, with approximately 23% and 13% of the REGA and STAR resources unable to identify sub-type. However, concordant results from all five systems were only produced for 58% of samples with the remaining 42% (417 samples) resulting in 40 combinations of patterns of concordance/discordance: ~ 16% were concordant across four resources and 13% concordant across three resources. 5% samples produced discordant results between all five resources.
The authors concluded that use of any one tool alone, compared with any of the other tools, will result in misclassification of 20% or more patients sub-type, and that unassigned samples from REGA and STAR probably reflected the higher stringency of those tools. Also, that discordance reflected the difficulty of keeping resources updated, particularly with the evolution of new and recombinant viruses.
Implications for epidemiological studies are important: several posters at the workshop highlighted a broadening in European populations from traditionally defined geographically defined populations and HIV sub-type. For example, Ana Garcia-Diaz reported that 46% of new infections at the Royal Free Hospital (105/239) from 2004-2006 were non-B sub-types and that only 80% of these were reported in heterosexuals , and Marie-Laure Chaix reported that non-B sub-types were reported in ~ 11% of newly diagnosed gay and bisexual men in France .
LosAlamos RIP 2.0
This is a high rate of non-concordance and it is unlikely that clinics in the UK testing for HIV sub-type, receive results from more than one database.
While sub-type appeared to have relatively small implications for initial response to treatment, as the studies above indicate, the implications for resistance may be more significant.
Unless stated otherwise, all references to abstracts relate to the Programme and Abstracts from the XV International Drug Resistance Workshop, 13-17 June 2006, Sitges, Spain. The abstract book is published as a supplement to Antiviral Therapy 2006, Volume 11.
- Loveday C, MacRae E. Limitations in using online tools to determine HIV-1 subtype in 116 clinical patients: a comparison of 5 tools. Abstract 116.
- Garcia-Diaz A, Booth C, Nebbia G et al. Transmitted drug resistance clusters with the infecting HIV-1 subtype: a single centre analysis of all new HIV-1 diagnoses in London. Abstract 113.
- ML Chaix ML, Deveau C, Calvez V et al. Increase of non-B HIV-1 resistant virus in primary infected patients: 9 years of French Experience (19962004). Abstract 110.