Epidemiological studies and transmission of resistance: evidence for optimism or issues with interpretation?
Simon Collins, HIV i-Base
Overview of epidemology studies
Several epidemiological studies contributed to a generally optimistic atmosphere at the workshop, by suggesting that the potential for widespread multi drug resistance has generally been limited, and that rates of resistance to 2- or 3-classes of drugs may be still remain consistently low.
Trends in resistance, as expected, generally mirror changes in ARV use, with prevalence of PI-associated mutations dropping as NNRTI-based regimens became more widely used – and within class changes generally reduction in TAMS and ddI-related mutations and an increase in K65R reflecting wider use of tenofovir. Similarly, D30N decreased and I50L increased reflecting lower use of nelfinavir and higher use of atazanavir. As usual, regional and national variations were reported among studies. [1, 2, 3]
However, lower prevalence in many of these cross-sectional studies, may miss the cumulative increase of resistance, that would be picked up from longitudinal studies. The importance of this methodology was raised at the 2004 Workshop by researchers from the UK-CHIC cohort. 
The importance of collecting longitudinal data tracked in individual patients, was addressed by Akash Shah from Yale University in a poster at this years meeting, which reported longitudinal retrospective data (every 6 months) from around 400 US patients enrolled in the two clinics in Connecticut. 
Although complete data from month 0, 6, 12, and 18 was only available for 396, 204, 153 and 84 patients respectively, the cumulative presence of resistance increased from around 32% to 46% in patients with 18 months data (p=0.08), and from 31% to 50% in patients with 12 months data (p=0.001). The study reported that genotypic resistance increased at approximately 5% per 6 months, but with current follow-up appeared to plateau at 50%.
Previous presentations from the UK CHIC group showed cumulative resistance to a single drug of approximately 10% occurring every two years, and importantly linked this to a similar increase in risk of virological failure over a similar period. While incidence of 3-drug resistance was ~ 1% the cumulative risk of 3-drug resistance increased to ~ 4% over 6 years. 
In looking at the results from the following studies that reported a generally stable incidence of resistance, including multi-drug resistance, it is important to bear these methological differences in mind.
Lisa Ross and colleagues from GSK presented results from the largest US prevalence study at the meeting, with samples from 1795 treatment naive patients from 33 states, who enrolled into clinical trials from 2000-2004. 
Incidence of IAS mutations for this cohort was around 10% (2000/2004: 5/13% – NRTI 3/4%, NNRTI 2/7%, PI 2/5%, by year). Geographical differences were reported for increases in the South (4/14%) and Northeast (8/16%) compared to more constant incidence in the Midwest (10/10%) and the West (6/6%). While multi-class resistance mutations didnt increase over time in this cohort the incidence of overall resistance doubled from 2000 to 2004 in the South and Northeast.
Viktor von Wyl and colleagues from the Swiss cohort presented data from just over 2,500 patients, followed for over 6,000 patient years, and with 1443 resistance tests linked to treatment failure.  They reported no increasing trend over time of resistance to one or more drug class, and that the prevalence of triple class resistance slightly fell over a five year period from 1999. However, a second Swiss study from the same research group, looking at around 10% of all recent infections from 1996 to 2005, while reporting a generally stable rate of transmission of resistant virus (NRTI being highest at 5.8%, and 2- and 3-class resistance present in 1.8% samples), did report a significant increase in NNRTI transmitted virus in 2005. 
Deenan Pillay and colleagues calculated the viral burden of drug resistance in a largely MSM clinic cohort of 1482 patients in follow-up from 1998 who had median of 9 viral load results, and a total of almost 500 resistance tests. 
Looking at common mutations (at 41, 103, 184 and 215 in RT; and at 90 in protease) they multiplied presence of each mutations by related viral load, accounting for differences over time, and calculated an overall burden of 6.8% resistant virus. This proportion reduced over the study period (2000-2003). The relative frequency of specific mutation burden in the cohort was T215any > M41L > K103N > M184V > L90M. Over the same period, resistance was transmitted to 24/150 acutely infected individuals, with frequency of K103N > 215any > M41L > L90M and no cases of M184V. The viral burden of resistant virus has fallen during this period, with greater viral control, and the researchers used this data to support an optimistic prediction that the transmission of resistance is likely to continue to fall.
Hong-Ha Truong and colleagues from San Francisco reported a reduction of drug resistant in just over 100 recently diagnosed individuals in 2004 (8-12% depending on site) compared with rates in 1996-2001 of between 18 and 27% depending on year.  Data from 2004 came from a primary HIV infection study and from patients seen at STD clinics, but showed no statistically significant differences between the two settings.
In Europe, Annemarie Wensing from Utrecht Medical Centre, and the EU-sponsored SPREAD programme, reported that prevalence of resistance in recent diagnoses was 9.1%, most of whom (71%) showed only single mutations (although acknowledging that viral reversion could underestimate these rates). 
These findings were from almost 1100 patients, recruited prospectively and diagnosed during 2002-3 in 17 European countries. The prevalence of NRTI, NNRTI and PI mutations was 5.4%, 2.6% and 3.0%. Interestingly, people from high prevalence countries had half the risk of transmission of drug resistant infection (OR=0.49; 0.24-0.99, p=0.046). 52% of resistant virus included a mutation or mixed virus at RT 215. 1.3% of patients were infected with three-class resistant virus.
Susan Little and colleagues from University of San Diego, California, reported an increasing in transmission of NNRTI resistance in North America and Australia in a cohort of 1535 patients who were largely male (94%), non-Hispanic white (70%) and who were diagnosed with recent infection following sexual exposure. 
Genotype and phenotype results were available for almost 1200 and 1000 patients respectively, with 650 patients having results from both tests. When comparing the periods 1995 to June 2000 and July 2000 to 2006, they found high level NNRTI resistance increased from 6 to 11% (p=0.014), although overall resistance did not (9 to 15%, p=0.096). The group reported higher NNRTI resistance in the Californian patients (compared to New York or other areas), and that a decrease in NRTI resistance was only significant in New York. MDR remained stable at 2-4%, by both genotype and phenotype, throughout the study.
Methamphetamine use linked to transmission of drug resistance
A second presentation from Susan Littles group, reported results from a preliminary study, that linked use of recreational drug methamphetamine (crystal meth) with acquiring drug-resistant HIV. 
Among 214 MSM diagnosed with recent infection between 2002 and 2006, and who completed self-assessed interviews on substance use, 12.6% had transmitted drug resistance. Among those with drug resistance, 19% reported use of methamphetamine and 9.5% did not, and this was statistically significant in a multivariate analysis (OR=4.29, p=0.01). Use of other substances was not significant, though annual income (>$10k, or $10-30k, compared to >50k or higher) had a similar impact.
Joe Eron from University of North Carolina, looked at prevalence of triple-class resistance (defined as at least one mutation to each of the NRTI, NNRTI and PI classes) in a cohort of just under 1600 treatment-experienced patients, half of whom (n=789) had been exposed to 3 classes. 
The first treatment for 50% of these patients had been 3-drug HAART, with 20% and 30% having started with either dual- or monotherapy respectively. In 609 patients with genotype results, the overall prevalence of triple-class resistance was 20%; but this was only 10% in HAART initiators (over median 4 years follow-up) compared to 26% in patients who first treatment was mono- or dual-therapy. Only the number of prior ARVs and non-HAART exposure were independent predictors of triple-class resistance.
Transmission of resistance to T-20
Bernard Masqulelier from Laboratoire de Virologie, CHU de Bordeaux, reported on 56 recently infected (seroconversion during 2004-2005) treatment-naive French patients in the Aquitaine cohort, and reported an overall prevalence of 16% transmitted resistance. Importantly, they reported the first two cases of transmission of resistance to T-20 (enfuvirtide). 
The first case had N42D mutation in the HR region of gp-41, along with additional mutations in protease (D30N, M36I, N88D) and RT (M41L, L210W, T215D). Phylogenetic analysis from the sexual partner (who was also treatment naive) showed the same genotype, suggesting consecutive MDR transmission. The second case had G36D mutation in the HR1 region, with no additional protease of RT changes.
The authors suggested that broadening the genotype range to include resistance to fusion inhibitors is important for surveillance studies, and that genotyping the gp-41 region should be considered before starting patients on T-20.
Finally, Mark Oette from University of Düsseldorf, prospectively tested 831 patients who started their first HAART regimen from 2001-2005. 
Pre-treatment samples with resistance increased from 4.8% in 2001, to 7.3%, 8.7%, 11.6% and 9.0% in each subsequent year to 2005, although this trend did not reach statistical significance. Importantly, the researchers concluded that further surveillance of primary drug resistance was essential, and that genotypic resistance testing prior to starting HAART should be regarded as standard of care for all patients.
Similarly, Davey Smith and colleagues, after reporting one of the highest prevalence of resistance in newly diagnosed individuals (24.5% in 106 new diagnoses), reported that modeling healthcare costs and showed that resistance testing was at least as cost-effective as other healthcare interventions, when resistance in treatment-naive patients was 8-10% or higher. 
Unless stated otherwise, all references to abstracts relate to the Programme and Abstracts from the XV International Drug Resistance Workshop, 13-17 June 2006, Sitges, Spain. The abstract book is published as a supplement to Antiviral Therapy 2006, Volume 11.
- Balotta C et al. Study of antiretroviral resistance in treated patients with virological failure (START study): an Italian survey over the period 20032005. Abstract 119.
- Kagan R et al. Newer antiretroviral treatment regimens drive HIV-1 RT and PR mutational patterns in a national reference laboratory database. Abstract 120.
- Camacho R et al. Rise and fall of the RT K65R incidence in the Portuguese resistance database. Abstract 121.
- Collins S. Resistance in the UK: new approach to epidemiological studies. HIV Treatment Bulletin, July 2004.
- Shah AD et al. Cumulative prevalence of HIV drug resistance in HIV+ patients in 122 clinical care. Abstract 122.
- Ross L et al. Prevalence of HIV-1 drug resistance-associated mutations in a large cohort of ART-naive HIV-infected individuals in the United States from 2000-2004. Abstract 107.
- von Wyl V, Yerly S, Boni I et al. Prevalence of HIV-1 drug resistance in Switzerland between 1999 and 2004: no trend for an increase. Abstract 104.
- Yerly S, von Wyl V, Boni J et al. Transmission of HIV-1 drug resistance in Switzerland: a 10-year molecular epidemiology approach. Abstract 105.
- Pillay D, Sabin C, Poa D et al. What is the drug resistance mutational infectious burden in an HIV-1 prevalent cohort and the relationship to incidence of transmitted resistance? Abstract 101.
- Truong H-H et al. Reduced levels of primary resistance to nRTIs in San Francisco is discernable using two independent sentinel populations. Abstract 102.
- Wensing AM, Vercauteren J, van de Vijver et al. Transmission of drug-resistance in Europe is characterised by single mutations and revertants. Abstract 98.
- Little S et al. Increases n transmitted NNRTI drug resistance among recently HIV infected patients from North America and Australia. Abstract 97.
- Drumright LN, Gorback PM, Little SJ et al. Transmitted HIV drug resistance is associated with methamphetamine use among recently HIV infected MSM in southern California, USA. Abstract 99.
- Eron J et al. Prevalence and predictors of triple-class antiretroviral viral drug resistance in routine HIV primary care. Abstract 78.
- Masqulelier B, Peuchant O, Capdepont S et al. Primary resistance of enfuvirtide in recently infected, antiretroviral naive patients, ANRS CO3 Aquitaine Cohort. Abstract 95.
- Oette M, Kaiser R, Däumer M et al. Trends of primary drug resistance in chronically HIV-infected patients in Germany, 2001-2005. Abstract 112.
- Smith D, Pesano R, Cachay E et al. Prevalence of HIV drug resistance among antiretroviral naive individuals of unknown infection duration. Abstract 108.