This year's resistance Workshop was heavily weighted towards biochemical and structural mechanisms of nucleoside analog resistance (see related article, "Mechanisms of HIV Drug Resistance: An Update from the 5^th International Workshop on HIV Drug Resistance & Treatment Strategies"). There was less information this year on genotypic-phenotypic correlates of nucleoside resistance and clinical outcomes of patients based on the degree of nucleoside resistance. Nevertheless, there was a fair amount of clinically relevant new data on nucleoside resistance, which is reviewed here.

Tenofovir resistance
Data were presented by MD Miller on tenofovir susceptibility (abstract 8) [1]. In treatment-naive patients, he reported, 98% had a <3-fold change in susceptibility using Virco's Antivirogram. In 5,000 clinical isolates from predominantly treatment-experienced patients, 88% of patients had <3-fold reduced susceptibility to tenofovir and only 1% had >10-fold tenofovir resistance. In contrast 18% had >10-fold decreases in susceptibility to AZT. Of the 51 patients with >10-fold reductions in tenofovir susceptibility, 48 had multiple nucleoside analog mutations (NAMS) and 18 had insertions at position 69 (associated with multinucleoside resistance). No data comparing resistance to clinical outcomes were presented to give us the all important "clinically relevant" cutoff, but Virco is reportedly working on generating that data.

ddI resistance
Phenotypic-genotypic correlates of ddI resistance were analyzed using the Virco Virtual Phenotype database (Larder BA, Bloor S, abstract 47) [2]. Once again, we learned that phenotypic resistance to ddI and d4T is difficult to detect, even when using Virco's new cutoffs of 3.5-fold increase in IC₅₀ for ddI and 3-fold increase in IC₅₀ for d4T (only 12% were resistant to both vs. 33% to AZT). These cutoffs were based on the variation of the assay in 98% of treatment-naive patients and not on clinical outcome data. 60% of all phenotypic ddI resistance (>3.5 fold change) was related to multiple NAMS, 25% due to L74V, and 20% due to 69 inserts or Q151M: more evidence to use the term NAMS rather than TAMS (thymidine analog mutations).

T215 reversion
T215 reversion mutants are the most common RT mutants seen in drug-naive persons (3.3%). The drug resistant mutant T215Y reverts to a D, C, or S by changing only one nucleotide. Alternatively, T215Y to T requires two nucleotide changes. A presentation from researchers at the CDC showed that these reversion mutants are more fit than the T215Y mutants (Garcia-Lerma JG, et al, abstracts 21, 22) [3,4]. The presence of other TAMS, M41L and L210W enhances the fitness of the D and C mutants more than the S, whereas T215S is more fit than D and C mutants when this is the only mutation. These mutants are nearly as fit as wild-type strains in in vitro growth competition assays, which may be why they persist in patients for long periods of time in the absence of treatment. These reversion mutants, even when present with other TAMS, are not phenotypically resistant to nucleosides, but if they are cultured in the presence of AZT they rapidly mutate to T215Y, which confers resistance. The take home message here is that these reversion mutations, although they do not confer nucleoside resistance, increase the likelihood of developing nucleoside resistance and should be considered when selecting NRTI drugs.

d4T susceptibility
The susceptibility of clinical samples to d4T was predictive of 8-week virologic responses to d4T monotherapy in patients previously treated with AZT monotherapy. Baseline and post-treatment genotypic and phenotypic data were analyzed for 24 patients who received d4T after AZT in ACTG 302 (Shulman N, et al, abstract 124) [5]. (Genotypic data were previously presented at the 8^th CROI [6].) The 7 responders (>0.3 log₁₀ copies/mL drop in HIV RNA at 8 weeks) had only one NAM, in most cases the K70R mutation, and had d4T fold changes of 0.9-1.1. Not surprisingly, non-responders as a group had more mutations and greater increases in d4T IC₅₀ values; several non-responders had no mutations and low fold changes in d4T susceptibility, however. Using a cutoff of 1.5-fold (which accounts for variability of the assay), 53% of patients with <1.5-fold reduced susceptibility to d4T had positive virologic responses to d4T therapy versus 0% of those with >1.5-fold d4T resistance. Number of NAMS, AZT, and d4T susceptibility were all highly correlated. According to this study, the clinical cutoff for d4T resistance should be 1.5-fold for the ViroLogic assay (PhenoSense).

Clinically relevant cutoffs
The importance of establishing clinically relevant cutoffs for drug susceptibility was emphasized in CCTG 575 (Haubrich R, et al, abstract 80) [7]. This study compared phenotype-guided treatment (using ViroLogic's PhenoSense assay) with standard of care. No significant differences in outcomes were observed. Importantly, though, the phenotype-guided cutoffs used were the old ones that arbitrarily applied a 2.5-fold reduction for every drug as evidence reduced susceptibility. In the current iteration of the PhenoSense report, the clinical cutoffs for d4T and ddI are now approximately 1.5-fold, while the clinical cutoff for abacavir susceptibility is 4.5-fold. Therefore "susceptibility" to d4T and ddI was over-predicted, and "susceptibility" to abacavir was under-predicted. Confounding these results further, ddI and d4T were more commonly used in the phenotype arm and abacavir was more commonly used in the standard of care arm. NNRTI and PI susceptibility, but not NRTI susceptibility, were both predictive of response in the two groups, again evidencing a big problem with d4T and ddI susceptibility cutoffs.

Many of the new and improved genotypic interpretation algorithms can predict response/non-response to abacavir with >75% accuracy, including the Stanford HIV RT and Protease Sequence Database Web site that is frequently referenced and sourced on HIVresistanceWeb. Fine-tuning of these algorithms will need to be done in order to pick up the 30% of patients who were "resistant" according to the algorithm used in CCTG 575, but had a significant clinical response.

References
Abstracts can be accessed by hyperlink after registering for the 5th International Workshop on HIV Drug Resistance & Treatment Strategies Webcast at Mediscover.net

Miller MD, McKenna P, Larder BA, Harrigan PR. Characteristics of tenofovir phenotypic susceptibility. Antiviral Ther. 2001;6(Suppl 1):8. Abstract 8
Larder BA, Bloor S. Analysis of clinical isolates and site-directed mutants reveals the genetic determinants of didanosine resistance. Antiviral Ther. 2001;6(Suppl 1):47. Abstract 47
Garcia-Lerma, et al. Fitness analysis of viruses with unique T215D/C/S mutations from treatment-naïve persons: implications on persistence in vivo and mechanisms of reversion of T215Y. Antiviral Ther. 2001;6(Suppl 1):21. Abstract 21
Garcia-Lerma, et al. Increased ability for selection of zidovudine resistance in a distinct class of wild-type HIV-1 from drug-naïve persons. Antiviral Ther. 2001;6(Suppl 1):22. Abstract 22
Shulman NS, Winters MA, Shafer RW, Zolopa AR, Liou SH, Hughes M, Whitcomb JW, Petropoulos CJ, Hellman NS, Katzenstein DA. Subtle changes in susceptibility to stavudine predict virologic response to stavudine monotherapy after zidovudine treatment. Antiviral Ther. 2001;6(Suppl 1):124. Abstract 124
Shulman N, Shafer R, Winters M, and others. Genotypic predictors of virologic response to stavudine after zidovudine monotherapy (ACTG 302). 8th Conference on Retroviruses and Opportunistic Infections. 2-4 Feb 2001, Chicago, IL. Abstract 437
Kemp, et al. A mutation in HIV-1 reverse transcriptase at codon 318 (Y-F) confers high-level non-nucleoside reverse transcriptase inhibitor in clinical samples. Abstract 44.
Haubrich R, Keiser P, Kemper C, et al. CCTG 575: a randomized, prospective study of phenotype testing versus standard of care for patients failing antiretroviral therapy. Antiviral Ther. 2001;6(Suppl 1):80. Abstract 80