Answers are grounded in this paper and cite their sources.
HIV genotyping facilitates the choice of drug at the point at which highly active antiretroviral therapy (HAART) fails, and confers clinical benefit [1,2]. Sequencing is the current `gold standard' for resistance testing, but there has been recent concern over the accuracy of the technique in detecting pure and mixed populations of virus. An evaluation using the ENVA-2 panel of viruses [3] demonstrated that 66 and 71% of laboratories successfully identified a homogeneous population of mutant, reverse transcriptase, and protease genes, respectively. Fewer than half of the laboratories detected mutations in a 50 : 50 ratio of mutant : wild type (WT), with worsening success as the proportion of mutant : WT decreased. A sequencing reaction might incorrectly identify mixtures of WT and mutant virus as a result of amplification bias during the polymerase chain reaction (PCR). If a primer anneals to a disproportionately greater amount of a minor viral population in the early rounds of PCR, then that population may predominate by the end of the reaction. We studied a patient failing on HAART, whose sequence data suggested the presence of such bias. We identified a 46-year-old HIV-positive Afro-Caribbean man who was taking stavudine, didanosine and nelfinavir and was experiencing virological failure, with a plasma RNA level of 36 345 copies/ml. A sequence of protease revealed the nelfinavir-associated mutations D30N, L63P, V77I and L90M. Codons 36, 46, 71, 84 and 88 were WT. Electropherograms indicated background viral populations at codons 30, 63 and 90. In order to evaluate these mixed codons further, RNA was re-extracted (Qiagen Ltd., UK) in duplicate from the same plasma sample. Ten aliquots of complementary DNA were taken from each RNA eluant to give a total of 21 theoretically identical samples, including the original. After nested PCR to amplify protease, the product was purified (Qiagen) and sequenced using the ABI dRhodamine Ready Reaction Mix (Perkin Elmer, Warrington, UK), both primers from the nested PCR reaction and two ABI 310 Genetic Analysers to give 42 sequences. Nine codons associated with decreased sensitivity to nelfinavir were analysed in each of the 42 sequences. One would expect all sequences to be identical, yet only four codons [36 (WT), 71 (WT), 77 (mutant) and 84 (WT)] produced the same result. Codon 30 was GAT (WT) in 29 out of 42 sequences, but AAT (mutant) in three out of 42 and `N'AT (no predominating base) in 10 out of 42 samples. Codon 46 was ATG (WT) in 37 out of 42 samples and ATN in five out of 42. In these cases, the `N' comprised a WT `G' and a mutant `A', indicating a mixed population not reported in 37 replicates. For codon 63, 33 out of 42 samples were CCA (mutant); although the classical mutation is CCC, this is a common polymorphism. The other nine samples included one CTA (WT) and eight CNA. Had the single WT result been the only sequence produced, the clinician would incorrectly have considered that the position did not contain a dominant mutation. For codon 90, four out of 42 sequences were TTG (WT), 29 out of 42 were ATG (mutant) and nine out of 42 were NTG. This codon showed the greatest intra-sample variability, with a 9.52% probability that a single sequence would miss a dominant mutant population. Fig. 1 shows six of the 42 electropherograms for codon 30 and demonstrates how peaks at the same position can vary in their relative sizes. The first two electropherograms for replicates 1 and 2 are reported as GAT (WT), even though a smaller `A' peak is visible under the `G'. For replicates 3, 4 and 5 the relative sizes of the `G' and `A' peaks are too close for one to dominate and the codon is reported as an `N'. In replicate 6, the `A' peak is larger than the `G', and is interpreted as a mutant codon.Fig. 1.: Six of the 42 electropherogram images for codon 30 of protease taken from the same plasma sample. GAT indicates a wild-type (WT) codon and AAT a mutant (MT) codon. NAT indicates no predominant base. There is clear variation in the relative A : G peak sizes, as indicated by the arrow, with the black `G' dominating in the first two electropherograms, no dominant base in the next three (but with increasing proportions of `A') and the green `A' dominating in the final image.This is the first report demonstrating that a single result for a mixed codon is unreliable. Amplification bias may result in a minority species of HIV being depicted as the majority. As PCR amplification is theoretically exponential, only a small degree of bias in the early cycles would change template proportions. The implications of this are clinically relevant. Firstly, the relative peak sizes in a sequence cannot be used to determine the proportions of a mixed population. Secondly, a dominant mutant population may be missed as a result of the disproportionate amplification of a minor WT population. This could result in a drug to which there is resistance being continued, or an ineffective regime being started. If this bias is occurring regularly it may be necessary to carry out resistance tests in duplicate or by clonal analysis. The cost and labour implications of this are significant. Alexander J. Frater Cecile C. Chaput Jonathan N. Weber Myra O. McClure
No takes yet. Share an insight, caveat, or question.
Chaput et al. (2000) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: