Background:  Since November 2002, North Carolina’s Screening and Tracing Active Transmission (STAT) program has tested all public HIV testing clients for acute or recent HIV infection, and has used sequencing to exclude sample duplication, to explore epidemic structure and clustering and to survey for transmitted drug resistance.

Methods:  All HIV testing form data (demographics, risk factors, and ZIP code of residence) are “anonymized” and maintained in the STAT database. New HIV cases classified as acute (by RNA pooling) or recent (by STARHS) undergo pol sequencing by Genosure; additional sequencing used TruGene. HIV subtype was determined by GASP (www.datamonkey.org/GASP); drug resistance interpretations were generated using the Stanford HIV Drug Resistance Database algorithm, version 4.2.2 (hivdb.stanford.org); and calculation of distance matrices using the TN93 nucleotide distance, and phylogenetic reconstruction, was performed using HyPhy (www.hyphy.org). Sequence clustering was analyzed using multidimensional scaling, and exponential random graph models, controlling for the number of edges and triangles in the network. Spatial clusters were based on inclusion of each client’s residence in core areas defined by standardized HIV prevalence data.

Results:  Of the 231 sequences isolated from individuals with acute or recent HIV infection between November 2002 and December 2005, 227 (98%) were subtype B, and 4 (2%) were subtype A/G recombinants. Of the 227 subtype B sequences, 25 (11%) were associated with primary drug resistance or exhibited “revertant” mutations at position 215 in reverse transcriptase. Overall, the sequences were distantly related (mean divergence, 5.9%), with sequences from different spatial clusters intermingling. However, sequences from 27 individuals (12%) occurred in 12 closely related (≤1% divergence) clusters, comprising 9 pairs (2 of which involved drug resistant virus) and 3 triplets. Individuals from the same spatial cluster were significantly more likely to exhibit phylogenetic clustering than those from different spatial clusters (odds ratio 2.3, p <0.05).

Conclusions:  Using a combination of testing for recent infection, and spatial and phylogenetic sequence analysis, we have identified a number of potential sub-epidemics within North Carolina, including geographic clusters and a possibly novel A/G recombinant subtype. Overall prevalence of transmitted resistance in North Carolina may be low.