Drug-resistant HIV-1 emerges in many patients experiencing virological rebound during antiretroviral therapy. The transmission of such drug-resistant strains is now well recognized with estimates of 10–20% prevalence in acute [1–4] and established drug-naive patient cohorts [5,6].
This phenomenon has important implications. First, it may prolong time to virological suppression on therapy, and thus compromise the benefit of HAART to the population [1,2,7], even though in some studies [1,2] rates of suppression were equivalent over the period of follow-up. Second, transmitted drug-resistant viruses may lead to an altered progression of HIV disease compared with viruses without baseline resistance. One case of a rapidly progressive, multi-class, drug-resistant infection has recently been reported . By contrast, others have observed that some drug-resistant viruses have reduced ‘fitness’ in comparison to wild-type viruses , and the CD4 cell count is higher and HIV RNA levels lower around time of infection following transmission of drug-resistant virus compared with wild-type virus [2,10]. There remains a dearth of information on the long-term implications of transmitted drug-resistance (TDR) with regard to HIV natural history, characterized through CD4 cell decline.
We now report CD4 cell decline, and response to therapy in patients with transmitted drug-resistant virus compared with patients with wild-type virus within the CASCADE collaboration of seroconverters. This represents the longest follow-up of such individuals reported to date.
We analysed data from ‘Concerted Action on SeroConversion to AIDS and Death in Europe’ (CASCADE), a collaboration of 22 seroconverter cohorts in 10 European countries, Australia and Canada. Details of CASCADE appear elsewhere ; briefly, all cohorts include people infected with HIV-1 for whom the date of seroconversion can be reliably estimated.
We considered a subset of persons in CASCADE, who were genotyped prospectively or retrospectively while naive to therapy within 18 months of a negative HIV antibody test or the date of laboratory evidence of seroconversion, where known. The Stanford algorithm was used to analyse each viral nucleotide sequence and mutation data (Stanford HIV Drug Resistance Database, http://hivdb.stanford.edu, updated October 2004). Drug-specific scores were summed across all mutations in the query sample, and drug susceptibility was classified, based on the total score, as ‘sensitive’, ‘potential low-level resistance’, ‘low-level resistance’, ‘intermediate resistance’ or ‘high-level resistance’. In our primary analysis, when the algorithm inferred at least ‘intermediate resistance’ to one or more antiretroviral drugs, that person was considered to have TDR. Further, for individuals going on to start HAART, if any such drug was included in their initial HAART regimen, that drug was considered ‘inactive’.
Natural history: CD4 cell decline
In order to investigate the effect of TDR on the natural history of HIV infection, we first considered individuals with at least two CD4 cell count measurements following HIV seroconversion available prior to the initiation of HAART. We used random effects models to describe CD4 cell decline before antiretroviral therapy (ART) initiation (modelled continuously on the square-root scale), allowing different rates of decline in the first year and subsequent years . We considered the effect of TDR on both the estimated CD4 cell count at seroconversion and the rate of CD4 cell decline, adjusting for: age at seroconversion, sex, exposure category, and presentation during acute infection. P-values for the difference in CD4 cell count at seroconversion and the rate of decline between those with and without TDR were calculated using F tests.
Response to therapy: time to viral suppression
We then considered the effect of TDR on response to antiretroviral therapy. We included all individuals who started HAART, defined as a combination of three or more drugs from at least two different classes or containing abacavir (and adding at least two new drugs to any previous sub-optimal ART regimen), providing at least one HIV RNA measurement was available after starting HAART. As it was important to maximize the follow-up period, we included patients who were first treated with mono or dual therapy before initiating HAART. This was considered in our statistical analysis.
Using Kaplan–Meier methods we examined the impact of having less than three active drugs in the initial HAART regimen on the time to reaching HIV RNA < 500 copies/ml following HAART initiation, adjusting for: age at seroconversion, sex, exposure category, previous ART experience, the time interval between seroconversion and starting HAART (within 6 months, 6–12 months, longer than 12 months), and calendar time of starting HAART (before 2000, 2000 or later).
We further investigated factors influencing the time to reaching HIV RNA < 500 copies/ml using multivariate Cox proportional hazards models and adjusted for potential prognostic factors as above. We adjusted for the effect of TDR in several ways: having less than three active drugs in the initial HAART regimen, having any form of TDR, and having at least one inactive drug in the initial HAART regimen. P-values are presented from Wald tests.
We assessed the robustness of our conclusions to differing definitions of resistance by applying a looser definition, in which ‘low-level’ resistance, as well as ‘intermediate’ and ‘high-level’ resistance to a drug (as inferred from the Stanford algorithm) was considered to make that drug inactive.
Finally, our methods had assumed an individual to have reached HIV RNA < 500 copies/ml on the date of the next test at which a viral load below this threshold was observed. In reality, the exact point at which the threshold was reached was not usually known. We therefore also used a spline-based proportional hazards model allowing for interval censoring  to test the sensitivity of our conclusions to this assumption.
Resistance testing was undertaken retrospectively on stored plasma samples, using in-house and commercial methodologies and was not used to guide HAART therapy for those patients included within this study. The minimal sequence obtained for all patients included the complete protease gene and positions 39–225 of reverse transcriptase. Each laboratory ran experimental controls to ensure no polymerase chain eaction contamination, and raw sequence data was submitted to a central web-based CASCADE Virology database, which then implemented a further series of quality control measures in order to further exclude potentially spurious sequence data. Such exclusion was based on the proportion of positions demonstrating multiple mixtures, and length of runs of indeterminate bases.
Statistical analyses were performed in STATA (StataCorp. 2003. Stata Statistical Software: Release 8. College Station, Texas, USA: StataCorp LP) and SAS Version 8 (Copyright 1999–2001 SAS Institute Inc., Cary, North Carolina, USA).
Natural history of CD4 cell count levels
Population characteristics and prevalence of resistance
Three hundred individuals (seroconverting between 1987 and 2003) with at least two CD4 cell count measurements available before the initiation of HAART were genotyped from samples taken within 18 months of the last seronegative date (Table 1). The median [interquartile range (IQR)] time from last negative date to the resistance test was 185 days (112–279). The demographics of individuals with a resistance test differed from the parent population: there were a total of 5925 individuals with mean age 30 years in the contributing cohorts, 3334 (56%) of whom were gay men, and 1628 (27%) injecting drug users (IDUs). Thus, as we might expect, those with a resistance test were more likely to be gay men and less likely to be IDUs. Of the 300 included in analysis, 153 initiated therapy and were, therefore, censored at that point. There were few differences in demographic factors between those with and without TDR. A slightly larger proportion were gay men (83 and 73%, respectively). The mean age at seroconversion was 34.0 years. 29 individuals (9.7%) had evidence of TDR, defined as intermediate/high level resistance. Nineteen patients carried virus considered resistant to one or more nucleoside reverse transcriptase inhibitors (NRTIs), nine carried non-nucleoside reverse transcriptase inhibitor (NNRTI) resistance, and six had protease inhibitor (PI) resistance.
Effect of TDR on the CD4 cell profile
In a random effects model for CD4 cell decline, individuals with TDR had a higher estimated CD4 cell count at seroconversion (698 compared with 604 cells/μl, P = 0.14) but a steeper decline in the first year (Fig. 1). Estimated rates of decline in the first year were 5.0 [95% confidence interval (CI), 2.8–7.3] and 1.7 (95% CI, 0.8–2.6) √CD4 cells per year for TDR and no TDR, respectively (P = 0.005). For an individual with a CD4 cell count of 500 cells/μl at seroconversion, these rates correspond to a CD4 cell loss of 199 and 73 cells/μl, respectively, in the first year. Thereafter we found no evidence of a difference in the rate of CD4 cell decline (P = 0.32) between those with and without TDR; for an individual with a CD4 cell count of 500 cells/μl at 1 year following seroconversion, we estimated the loss in the subsequent year to be 67 cells/μl.
Multiple class resistance
Two persons were considered to be resistant to one or more antiretroviral drugs from all three major classes. One of these patients (infected with virus containing the following mutations: reverse transcriptase M41L, E44D, D67N, Y181C, G190S, L210W, T215Y; protease I84V, L90M) had experienced a steady loss of CD4 cells from 900 cells/μl at 5 weeks following seroconversion, to a last recorded count of 350 cells/μl at 142 weeks, an average decline of 289 cells/μl in the first year and 157 cells/μl thereafter. The second individual (infected with virus containing the following mutations: reverse transcriptase M41L, E44D, K103N, L210W, T215Y; protease L33F, V82A, L90M), saw their CD4 cell count fall from 419 cells/μl at 28 weeks to 294 cells/μl at 68 weeks (average decline of 163 cells/μl per year). Neither individual was known to have developed AIDS or to have started antiretroviral therapy.
Response to therapy: time to viral suppression
Data were available on 201 individuals initiating HAART between April 1996 and October 2003 (median = January 2000) (Table 2). The median time from estimated seroconversion to HAART initiation was 244 days (IQR, 61–697 days). For 102 individuals (51%) the HAART regimen was PI-based, whereas 78 and 14 individuals (39 and 7%), respectively, began NNRTI-based and triple-NRTI regimens, and seven (3%) started a HAART combination including all three major classes. Nine percent had received prior ART.
A total of 23 individuals initiating HAART (11%) had evidence of TDR (intermediate/high level resistance), of whom 13 were predicted to have a pattern of resistance that compromised their initial HAART regimen leaving less than three fully active drugs (nine, three and one patients with respectively two, one and none of their initial drugs fully active), whereas 10 patients still had at least three drugs predicted to be fully active. Among those included in this analysis, individuals with TDR were again somewhat more likely to have been exposed to HIV through sex between men than those without (83 versus 74%), and were correspondingly more commonly male (96 versus 84%). There were no major differences in demographic characteristics between those with and without TDR (Table 2). The median CD4 cell count at the time of starting HAART was 366 cells/μl (IQR, 250–504) and was similar in those with and without TDR (376.5 and 366 cells/μl, respectively).
A median of 4 HIV RNA tests (IQR, 2–6) were available in the first year after HAART initiation among individuals with evidence of transmitted drug resistance, and 4 (IQR, 3–6) among those without.
Time from initiating HAART to HIV viral load suppression
Effect of TDR
A total of 181 individuals (90%) achieved viral load < 500 copies/ml after starting HAART. Viral load data were available for a median of 397 days (IQR, 220–822) on the 20 people who did not reach this level. In a multivariate Cox model, we found no evidence of a difference in time to viral load suppression in individuals with a resistance pattern implying less than three fully active drugs in their initial treatment regimen [relative risk (RR), 1.17, 95% CI, 0.65–2.11, P = 0.60, Fig. 2). Two sensitivity analyses were performed. We repeated our initial model using a less conservative definition of resistance based on any drug scoring low-level, intermediate, or high-level resistance using the Stanford algorithm. This included a further 13 patients in the TDR group, infected with viruses containing the following single mutations; 215 variants (n = 6), M41L (n = 2), D67N (n = 1)Y188D (n = 1) in reverse transcriptase, and M46V(n = 1) in protease. Then, to address concerns of bias introduced by interval censoring (i.e. the fact that the exact time of HIV viral load falling below 500 copies/ml is not usually known), we fitted a multivariate semi-parametric model allowing for this. In both models, we found no evidence of a difference in time to viral load < 500 copies/ml in individuals with less than three fully active drugs in their initial HAART regimen compared with those with three or more fully active drugs: the RRs were 0.94 (95% CI, 0.57–1.54) and 1.14 (95% CI, 0.64–2.05), respectively.
We re-fitted the initial Cox model using two different measures of resistance, and in both cases we found no evidence of an effect on time to suppression of HIV viral load. These alternative measures of resistance were: resistance to any drug regardless of the initial regimen (RR, 1.09; 95% CI, 0.69–1.72; P = 0.72); and resistance to one or more drugs in the initial regimen (RR, 1.17; 95% CI, 0.65–2.11; P = 0.60).
The results were similar when we excluded the 18 individuals who had received sub-optimal antiretroviral treatment prior to first starting HAART (RR of viral load suppression for those with less than three drugs active, 1.18; 95% CI, 0.66–2.13).
Effect of other factors
Interestingly, there was strong evidence of shorter time to viral load suppression among individuals starting HAART in 2000 or later (RR, 1.76; 95% CI, 1.26–2.45; P = 0.001). This effect persisted even after adjusting for the type of HAART used in the initial regimen (i.e. NNRTI-based, PI-based, triple-NRTI, triple-class; data not shown). In a univariate analysis, there was a suggestion that persons initiating HAART while treatment naive were more likely to achieve a viral load of < 500 copies/ml than those who had received previous ART, however this effect was not statistically significant after adjusting for other factors (RR, 1.30; 95% CI, 0.71–2.36; P = 0.39). No other demographic factors included in our analyses appeared to predict time to HIV viral load < 500 copies/ml (Table 3).
We describe, for the first time, the impact of transmitted HIV-1 drug resistance on the CD4 cell count decline up to 5 years following HIV seroconversion. Although no difference has been observed between those infected with resistant compared to wild-type virus in the medium to long term, some intriguing differences between these groups have been identified. As has been previously observed, baseline CD4 cell counts are higher in those with TDR, than those infected with wild-type virus . This finding has been ascribed to the reduced fitness of viruses containing drug resistance mutations . However, if this was indeed the case, we would expect the CD4 cell count differences between the two groups at baseline to be mirrored over time, particularly as transmitted resistance mutations appear, on the whole, to persist over the long term with lack of significant reversion of viruses back to wild type (drug sensitive) [9,14]. By contrast, we find that the initial decline is significantly faster in those with TDR, followed by a longer period of equilibration between the two groups, such that, at 5 years post infection, absolute CD4 cell counts appear to be similar. Virtually all in-vivo studies on the fitness cost to resistance have been undertaken in drug-treated individuals, in whom wild-type (non-resistant) virus also remains archived. In this context, fitness reflects the competition between co-existing wild type and resistant viruses. However, in the case of transmitted drug resistance, no such competition exists, since we assume transmission of a narrow, or homogeneous, viral population. Further, the pathogenesis of HIV infection, reflected by peripheral total CD4 cell count changes, is a result of complex host and viral determinants [15,16]. Specifically, the early period of acute infection is characterized by huge infection of memory CD4 cells, possibly outside of the blood compartment , followed by immune activation-induced CD4 cell depletion . In this context, small fitness differences between viruses may impact on initial CD4 cell count, but not play a major role in the subsequent pathogenic process. In the absence of any such laboratory measurements in our cohorts, it is difficult to provide a hypothesis to explain our observation, other than to note that some form of immune homeostasis may be responsible for the CD4 cell count adjustment over the first year following infection.
Of course we are mindful of the relatively small number of patients infected with resistant virus in our study, and that our observations require confirmation. Nevertheless, the strength of our data is based on what is one of the longest follow-up periods for such individuals.
We must also be wary of assuming that all drug-resistant viruses are similar in this regard. Indeed, the recent demonstration of a rapidly progressing infection in a man infected with triple class-resistant virus  illustrates the potential variation in pathogenicity of transmitted viruses. In that particular case, co-receptor tropism (X4) as well as polymerase gene changes may have contributed to disease progression, whereby CD4 cell count fell by at least 800 cells/μl within 20 months of infection. Of interest, the two individuals within the CASCADE cohorts infected with similar triple class-resistant viruses demonstrated an average decline of 289 and 163 cells/μl in the first year following infection, compared with an average of 199 cells/μl for the TDR group as a whole, suggesting that the very fast progressor previously identified by Markowitz et al.  is not necessarily representative of this population.
The key message provided by our analysis on CD4 cell count decline is that rapid declines in the first year following infection with TDR may not necessarily persist, and care should be taken when using these short-term changes to guide the initiation of antiretroviral therapy.
We also addressed the impact of TDR on response to triple therapy. Since resistance is only likely to impact on response where compromised drugs are included in the treatment regimen, we stratified patients by the number of assumed active drugs prescribed (based on the resistance pattern). Surprisingly, when defining resistance by either a conservative or more liberal definition, the number of active drugs used did not impact on the time to suppression (viral load < 500 copies/ml). Does this finding contradict previous published data? Both Little et al.  and Grant et al.  showed that the time to virological suppression was prolonged in patients infected with resistant virus, and in whom treatment was initiated during primary infection. However, viral load sampling was undertaken very frequently in both those studies (at least every 4–6 weeks). By contrast, the median number of viral load determinations in our study was four per year, more akin to routine clinical practice, which may have precluded the detection of small but significant ‘time-to-undetectable’ differences. More importantly, however, both these previous studies demonstrated equivalent suppression of viral load between TDR and non-TDR during the period of follow-up, as indeed shown in our current analysis. It is also intriguing that Little et al.  did not identify an impact of the number of active drugs on virological outcome. Taken together, these observations suggest that TDR impacts on response to first-line therapy in a subtle manner, if at all, and that longer term follow-up, and the study of second-line and salvage therapy will be required to identify clear correlates with the initial presence of resistance. The specific resistance mutations present may also determine the impact on treatment response. For instance, Violin et al.  demonstrated that viruses containing the position 215 variants in reverse transcriptase alone, appear to reduce response to thymidine analogue (stavudine or zidovudine)-containing regimens, although the co-existence of other key resistance mutations appeared to obviate this effect.
This highlights another source of difficulty in addressing issues of treatment response following TDR, namely the precise definition of resistance. All currently available algorithms (relating sets of viral mutations to drug susceptibility/therapy response) derive from treated patients. Since there appears to be selective transmission of resistance-associated mutants, such that the spectrum of mutations in newly infected individuals are different to the spectrum observed in treated patients with resistance , there is a requirement for consensus on the interpretations of the unique resistance profiles observed in such individuals. The data we present in this study illustrates the difficulty in producing clinically derived drug response cut-offs (virological response) for individuals infected de novo with resistant viruses. The small number of viruses containing each individual set of resistance mutations limits our ability to discern mutation-specific effects. Unfortunately, as resistance test-guided, first-line therapy becomes more standard, further studies of this nature become difficult.
In conclusion, TDR is associated with a rapid CD4 cell decline soon after infection, suggesting that these viruses represent a specific group of variants. However, we find no evidence that these viruses confer either a long term advantage or disadvantage in the absence of therapy. In addition, any impact of TDR on response to first line HAART is likely to be very small, possibly because most of these viruses only confer partial resistance to a small number of drugs. We cannot exclude that more significant effects may become apparent during subsequent regimens or whether the effect is larger when patients are infected with specific class-resistant viruses.
Sponsorship: CASCADE is funded through a grant from the European Union [QLK2-2000-01431]. This work is funded through a parallel grant, CASCADE Virology [QLRT-2001-01708].
1. Little SJ, Holte S, Routy JP, Daar ES, Markowitz M, Collier AC, et al
. Antiretroviral-drug resistance among patients recently infected with HIV. N Engl J Med 2002; 347:385–394.
2. Grant RM, Hecht FM, Warmerdam M, Liu L, Liegler T, Petropoulos CJ, et al
. Time trends in primary HIV-1 drug resistance among recently infected persons. JAMA 2002; 288:181–188.
3. UK Collaborative Group on Monitoring the Transmission of HIV Drug Resistance. Analysis of prevalence of HIV-1 drug resistance in primary infections in the UK.BMJ
4. Yerly S, Kaiser L, Race E, Bru JP, Clavel F, Perrin L. Transmission of antiretroviral-drug-resistant- HIV-1 variants. Lancet 1999; 354:728–733.
5. Descamps D, Chaix ML, Andre P, Brodard V, Cottalorda J, Deveau C, et al
. French national sentinel survey of antiretroviral drug resistance in patients with HIV-1 primary infection and in antiretroviral-naive chronically infected patients in 2001–2002. J Acquir Immune Defic Syndr 2005; 38:545–552.
6. Weinstock HS, Zaidi I, Heneine W, Bennett D, Garcia-Lerna JG, Douglas JL Jr, et al
. The epidemiology of antiretroviral drug resistance among drug-naive HIV-1-infected persons in 10 US cities. J Infect Dis 2004; 189:2174–2180.
7. Violin M, Cozzi Lepri A, Velleca R, Vincenti A, D'Elia S, Chiodo F, et al
. Risk of failure in patients with 215 HIV-1 revertants starting their first thymidine analog-containing highly active antiretroviral therapy. AIDS 2004; 18:227–235.
8. Markowitz M, Mohri H, Mehandru S, Shet A, Berry L, Kalyalaramin R, et al
. Infection with multidrug resistant, dual-tropic HIV-1 and rapid progression to AIDS: a case report. Lancet 2005; 365:1031–1038.
9. Brenner BG, Routy JP, Petrella M, Moisi D, Oliveira M, Detorio M, et al
. Persistence and fitness of multidrug-resistant human immunodeficiency virus type 1 acquired in primary infection. J Virol 2002; 76:1753–1761.
10. Bezemer D, Jurriaans S, Prins M, van der Hoek L, Prins JM, de Wolf F, et al
. Declining trend in transmission of drug-resistant HIV-1 in Amsterdam. AIDS 2004; 18:1571–1577.
11. CASCADE Collaboration. Changes in the uptake of antiretroviral therapy and survival in people with known duration of HIV infection in Europe: results from CASCADE.HIV Med
12. Touloumi G, Pantazis N, Babiker AG, Walker SA, Katsarou O, Karafoulidou A, et al
, on behalf of the CASCADE Collaboration. Differences in HIV-RNA levels before the initiation of antiretroviral therapy among 1864 individuals with known HIV-1 seroconversion dates. AIDS 2004; 18:1697–1705.
13. Royston P, Parmar MK. Flexible parametric proportional-hazards and proportional-odds models for censored survival data, with application to prognostic modelling and estimation of treatment effects. Stat Med 2002; 21:2175–2197.
14. Pao D, Andrady U, Clarke J, Dean G, Drake S, Fisher M, et al
. Long-term persistence of primary genotypic resistance after HIV-1 seroconversion. J Acquir Immune Defic Syndr 2004; 37:1570–1573.
15. Douek DC, Picker LJ, Koup RA. T cell dynamics in HIV-1 infection. Annu Rev Immunol 2003; 21:265–304.
16. Ribeiro RM, Mohri H, Ho DD, Perelson AS. In vivo dynamics of T cell activation, proliferation, and death in HIV-1 infection: why are CD4+ but not CD8+ T cells depleted? Proc Natl Acad Sci USA 2002; 99:15572–15577.
17. Mattapallil JJ, Douek DC, Hill B, Nishimura Y, Martin M, Roederer M. Massive infection and loss of memory CD4(+) T cells in multiple tissues during acute SIV infection.Nature
18. Brenchley JM, Schacker TW, Ruff LE, Price DA, Taylor JH, Beilman GJ, et al
. CD4+ T cell depletion during all stages of HIV disease occurs predominantly in the gastrointestinal tract. J Exp Med 2004; 200:749–759.
Deenan Pillay, Krishnan Bhaskaran, Suzanne Jurriaans, Maria Prins, Bernard Masquelier, Francois Dabis, Robert Gifford, Claus Nielsen, Court Pedersen, Claudia Balotta, Giovanni Rezza, Marta Ortiz, Carmen de Mendoza, Claudia Kücherer, Gabriele Poggensee, John Gill, Kholoud Porter.
MRC Clinical Trials Unit (Abdel Babiker, Krishnan Bhaskaran, Janet Darbyshire, Kholoud Porter, A. Sarah Walker) and Royal Free & University College Medical School Windeyer Institute (Rob Gifford and Deenan Pillay).
CASCADE Collaborators contributing data to Virology study
Aquitaine cohort, France (Eric Balestre, Sophie Capdepont, Geneviève Chêne, Francois Dabis, Hervé Fleury, Bernard Masquelier, Rodolphe Thiébaut); German cohort, Germany (Osamah Hamouda, Claudia Kücherer, Gabriele Poggensee); Italian Seroconversion Study, Italy (Claudia Balotta, Benedetta Longo, Giovanni Rezza, Lorenzo Deho); GEMES, Spain (Carmen Rodriguez, Vicente Soriano, Alfredo García-Saiz, Julia del Amo, Jorge del Romero, Josephina Belda, Marta Ortiz, Carmen de Mendoza); Amsterdam Cohort Studies among homosexual men and drug users, The Netherlands (Nicole Back, Roel Coutinho, Maria Prins, Lia van der Hoek); Copenhagen cohort, Denmark (Louise Bruun Jørgensen, Claus Nielsen, Court Pedersen); UK Register of HIV Seroconverters, United Kingdom (Abdel Babiker, Janet H Darbyshire, Noël Gill, Anne M. Johnson, Andrew N. Phillips, Kholoud Porter); South Alberta clinic, Canada: (M. John Gill, Sonia Gingues).
19. Turner D, Brenner B, Routy JP, Moisi D, Rosberger Z, Roger M, et al
. Diminished representation of HIV-1 variants containing select drug resistance-conferring mutations in primary HIV-1 infection. J Acquir Immune Defic Syndr 2004; 37:1627–1631.