Background:  Increasing use of HIV-1 genotyping to manage patients has created interest in assessing the performance of sequencing assays.  Most studies of reproducibility have examined inter-laboratory reproducibility but intra-laboratory reproducibility is also a key issue in the reliability of genotyping results.  We present a multi-center evaluation of the intra-laboratory reproducibility of HIV genotyping. 

Methods:   Three panels, each containing 5 coded specimens were sequenced in 37 laboratories using the VIROSEQ kit (n = 15), the TRUGENE kit (n = 19), or both (n = 3) (median:  3 panels/laboratory; minimum: 1).  One panel included 2 replicates of 1 clinical specimen and 3 of another.  Two panels each included 2 pairs of replicates of clinical specimens.  Rates of disagreement, as the percentage of discordant bases in replicate sequences, were calculated separately for the reverse transcriptase (RT) and protease genes (192 sets of replicates/gene).  Any difference, including missing data for a base in one replicate, was counted as a disagreement.  Four sets of replicate RT sequences, each with extensive missing data that resulted in >100 discordant results, were excluded. 

Results:  Among 380 sets of replicates, containing 204,506 sets of replicate bases, the rate of disagreement was <1% in 283 (74%) sets of replicates, including 94 sets with no disagreements at all.  In 55 sets, the rate was 1 to 2% and 42 sets, it was 2 to 6.5%.  Two types of disagreements were noted:  Most (66%) of the 1729 total disagreements involved reporting of mixed bases in 1 replicate and reporting 1e member of the mixture in the other.  These mixture disagreements accounted for all of the disagreements in 192 sets of replicates and some of the disagreements in 84 sets of replicates.  Disagreements of all other types (e.g., 2 different nucleotides) occurred in 94 sets of replicates only 10 of which did not have mixture disagreements.  Mixture disagreements may reflect the sensitivities of the assay at detecting mixtures and the nature of the samples, while the other disagreements may reflect technical problems in assay performance. 

Conclusions:  The data indicate that 100% reproducibility of genotyping may not be achieved routinely with current technology.  However, rates of disagreement were generally low.  Care must be exercised in interpreting genotyping results because disagreements at codons associated with antiretroviral resistance could affect treatment decisions.

Keywords: genotyping; quality assessment; resistance testing