XIV International HIV Drug Resistance Workshop: Epidemiology

An interesting, albeit inconclusive, theme that emerged at the workshop was the detection of minor viral populations and what that means for our understanding of epidemiology, the transmission of drug resistance including multi-drug resistant strains and their potential impact on future treatment outcomes.

Karin Metzner from the University of Erlangen-Nuremberg substantiated a recent conclusion by Scott Hammer from Columbia University following his review of PMTCT data in the Journal of Infectious Diseases (link to aidsmap news report) that “the more you look, the more you find”. Indeed with the proliferation of single genome and ‘ultra-deep sequencing’ techniques we can expect the emergence of a myriad of mutations previously undetected. This German study found that using allele-specific RT-PCR primed for K103N, M184V and L90M as surrogates for non-nucleoside reverse transcriptase inhibitor (NNRTI), nucloside analogue reverse transcriptase inhibitor (NRTI) and protease inhibitor (PI) resistance respectively they were able to detect drug resistant variants in acute seroconverters. In particular, 50% of those with drug resistant virus harboured viral populations that reflected the presence of quasispecies not detected by direct sequencing.

Using similar techniques Jeffrey Johnson from the US Centres for Disease Control and Prevention showed that some NRTI mutations such as D67N and K70R are more readily detected at the low-frequency variation compared to the PI mutation L90M as well as the RT mutation M184V. These mutations were selected because they serve as surrogate markers for class resistance but also because they actively emerge with many background polymorphisms. The data analysis involved 147 drug naïve individuals from the United States and Canada during 1997-2002 who had documented resistance. With more detailed analysis, they found discrepancy of prevalence between mutations in all classes of drugs including an increased detection rate of 20% for M184V, 25% for the L90M, 56% for K70R and 71% for the D67N mutation. Clonal sequencing of samples with newly detected mutants confirmed these findings. In total these data suggest that MDR strains for two class resistance may be underestimated by as much as 25% and for 3 class drug resistance by around 4%.

NNRTI resistance still detectable two years after infection in semen

Compounding these findings, Davey Smith from University of California San Diego (UCSD) confirmed that transmitted drug resistance can persist not only in plasma but also in seminal fluid for longer than three years. Their PHI cohort, which comprised of MSM, reported K103N, Y181 or Y188K mutations. The study amplified paired blood and semen samples including sequencing of the pol region. NNRTI resistance was sustained in both blood and semen in most individuals throughout the follow-up period (mean 622 days, range 95-1179 days). Only one individual displayed full reversion of the K103N mutation in the seminal compartment.

Marie-Laure Chaix from Necker Enfant-Malades in Paris showed compelling evidence that the persistence of transmitted drug resistance in plasma for long periods may indeed result in a suboptimal response to initial therapy. Their study compared individuals from the PRIMO Cohort and the AC11-ANRS Resistance Group who were on treatment during PHI following infection with either resistant or wild-type (WT) virus commenced at a median of 39 days post-infection. Results of the resistance tests were not available to physicians at time of initiation of therapy and outcomes of the treatment were assessed three months after starting treatment.

Of the 35 patients enrolled who had genotypic resistance, a total of 21 had virus resistant to one drug (21 patients), two drugs (10 patients) or three drugs (4 patients) in their regimen. Although at enrolment the individuals with genotypic resistance had lower viral load and higher CD4 counts (5.0 versus 5.3 log copies/ml and 512 versus 475 cells/mm³), after three months on therapy the viral load drop was greater in the WT group (

National guidelines are becoming ever more stringent in prescribing genotyping at infection for public health surveillance, but the French guidelines now recommend prospective testing in patients with PHI with genotyping results made available to prescribers shortly after initiation of treatment. Whilst forceful, what is missing from this study as observed by Doug Richman from UCSD, is that the results were not correlated against the number of active drugs contained in the first regimen. This may be an additional and significant prognostic factor as was documented in earlier studies from UCSD.

Multi-drug resistance and survival

Finally, further analysis from the UK Collaborative Group on HIV Drug Resistance and UK Collaborative HIV Cohort Study (UK CHIC) correlated predictors of survival and treatment response in patients with three drug multi-drug class resistance (3DR), defined as at least one primary mutation from each antiviral class. Regression analysis determined response following 24-28 weeks after diagnosis of 3DR. The cohort included 628 patients with 3DR HIV (85% males, median age 43 years, median CD4 238 copies, median VL 4.15 log copies/ml, median number of previously exposed drugs 8). Of these patients, 321 switched regimens following diagnosis of 3DR; 54 deaths were reported following 3DR with a median follow-up of 2 years. The estimated probability of death was 3% in the first year after 3DR diagnosis, 9% at year 2 and 13% at 3 years. 280 of the MDR patients switched regimens. Genotypic sensitivity scores (GSS) were not significantly correlated with survival although changing treatment for patients with high GSS did reduce the probability of survival. Other factors such as CD4, calendar year at 3DR diagnosis and number of ART drugs in previous regimens were significantly associated with death. The analysis confirms that incremental increases in death were associated with (1) continuing or changing therapy, (2) stopping and restarting therapy and most greatly linked with (3) stopping therapy. Michael Greenberg from Trimeris commented that whilst striking, the impact of newer drugs and treatment strategies on survival should be considered in future analysis.

Much of the discussion centred on three aspects of transmission: the phenomenon of clustering, validation of findings across studies, and their implications for public health strategies. Since HIV is transmitted sexually, vertically or through injecting drug use, clustering is inevitable, Dan Kuritzkes of Harvard Medical School reminded participants. Therefore, although of public health interest, it provides little additional data that is useful.

On the issue of validation, whilst findings from transmission studies remain important and underscore the need for greater public health vigilance, the question of assay variation needs to be urgently addressed. Deenan Pillay from University College London cautioned that assays with a broad range of performance and different sensitivities are utilised, making meta-analysis across studies meaningless. However, as John Mellors from Pittsburgh University noted, despite these variations, results cannot simply be explained away as technical artefacts. In the longer-term it may not simply be the individual mutations that matter but the combined knowledge of viral fitness, transmissibility and durability of resistant viruses that will impact future treatment outcomes, noted Jean-Pierre Routy from McGill University and the Montreal PHI Cohort. These data alert us to greater public health surveillance and towards a clearer understanding of the potential clinical impact of transmission of resistance.