The ultimate goal of HAART is to fully suppress HIV replication as this minimizes the risk of developing drug resistance and provides the greatest potential for immune recovery . Viral rebound is often associated with the emergence of resistance [2,3], which may lead to subsequent immunological decline . With accumulating regimen failures the goal of fully suppressed viral replication becomes more difficult to achieve . Factors associated with an increased rate of viral rebound include pre-HAART use of nucleoside analogue reverse transcriptase inhibitors (NRTI) as single or dual therapy [6–11], use of particular antiretroviral drugs [11,12] and poor adherence to medication . Lower rates of viral rebound have also been reported with increased duration of virological suppression [7,10,14]. Analysis of the EuroSIDA cohort in 2444 patients demonstrated that the overall rate of viral rebound during the first 6 months after initial full suppression was 33.5/100 person-years (PY) follow-up, compared with 8.6/100 PY in those who were suppressed for greater than 24 months .
At present, relatively few data exist on the association between the number of previous antiretroviral regimens failed and the rate of viral rebound following suppression to ≤ 50 HIV RNA copies/ml. The aim of this study was to determine whether the rate of viral rebound decreases with increasing duration of viral suppression regardless of prior virological failure and, if so, whether rebound rates in patients who have previously failed one or more regimens ultimately decline to levels as low as those seen in patients receiving first-line therapy. This has important implications for the ongoing treatment support of individuals with prior treatment failure.
The source of data for our analyses was the United Kingdom Collaborative HIV Cohort (UK CHIC) study. This is an observational cohort of some of the largest HIV clinical centres in the UK; the dataset used for the current analysis includes information from seven centres (Chelsea and Westminster, St Mary's NHS Trust, King's College Hospital, the Mortimer Market Centre, the Royal Free, St Bartholomew's and The Royal London Hospital and Brighton and Sussex University Hospital). The creation and design of the UK CHIC Study has been described previously . Data collected include information on patient demographics, antiretroviral history, laboratory findings, AIDS defining events and deaths. All data provided by the participating centres are thoroughly checked and any inconsistencies followed up.
Selection of patients for inclusion in the analysis
In selecting patients who would contribute person-time to the analysis, all individuals who attained an HIV RNA viral load of ≤ 50 copies/ml while receiving HAART (defined as a regimen containing three or more antiretrovirals) were considered. Patients were followed from the date when their HIV RNA first fell below 50 copies/ml while on antiretroviral therapy (ART) until the time of viral rebound (defined as two consecutive values ≥ 400 copies/ml or one value ≥ 400 copies/ml followed by starting two or more new drugs). Patient follow-up was censored (i.e., patients were removed from the risk set for the analysis) prior to this point if the patient discontinued or reduced HAART to fewer than three drugs, or on the date of the patient's last viral load measurement. Patients who experienced viral rebound or whose follow-up was censored on discontinuation/reduction of HAART re-entered the analysis if they subsequently re-suppressed their viral load to ≤ 50 copies/ml. The rate of viral rebound was calculated by dividing the total number of events (viral rebounds) by the person-time spent in a particular category.
Defining failure history at the start of each period of suppression
At the start of each period of viral suppression, the patient's failure history was assessed and the patient was categorized as having failed 0, 1, 2, 3, or 4 or more prior regimens. The analysis included patients who were antiretroviral naive when they started HAART as well as those with pre-HAART NRTI exposure. Any treatment taken before January 1996 counted as having failed one regimen, regardless of viral load. From January 1996, having a viral load of ≥ 400 copies/ml, whilst receiving any treatment and not having been off treatment in the previous 4 months counted as a regimen failure. Each time a viral load of ≥ 400 copies/ml was obtained whilst being on a regimen containing at least one drug (and having been on that drug for over 4 months) that the patient had not previously failed was counted an additional regimen failure. Patients who had temporarily been removed from the risk set after discontinuing HAART were not classified as having failed that regimen as long as they had not met the definition of failure at the time of discontinuation; thus, if these patients subsequently re-entered the analysis their number of failed regimens remained unchanged. Patient follow-up in the analysis was stratified according to the length of time a patient's viral load had been suppressed in the current episode of viral suppression; for individuals re-entering the risk set after treatment discontinuation/reduction, the clock was reset such that this length started once again from zero. The allocation of person years (PY) is further described in Fig. 1.
We used Poisson regression to identify independent predictors of viral rebound after adjusting for other potential confounding variables. The analyses were also adjusted for the following potential confounders: current antiretroviral regimen, calendar year, age, ethnicity, sex, risk group, viral load at initiation of ART, time since initiation of ART and CD4 cell count at time VL ≤ 50 copies/ml.
As each person could contribute more than one period of follow-up (and viral rebound) to the analysis, rate ratios were also estimated using generalized estimating equations to fit univariable and multivariable Poisson regression models (using the GENMOD procedure in SAS), allowing for multiple events in the same individuals. This present analysis is based on patients who received HIV care from January 1996 to December 2003.
Given the focus of these analyses, we also tested for an interaction between number of past regimens failed and time under suppression.
At the time of analysis a total of 21,256 patients were included in the UK CHIC database; 12 648 (59.5%) of these patients received HAART at some time over follow-up. A total of 10 237 (80.9%) patients attained a VL ≤ 50 copies/ml while on HAART. Of those patients, follow-up from the time of first attaining a VL ≤ 50 copies/ml to viral rebound, discontinuation of HAART or date of last viral load totalled 26 494 PY. Most person-time was in males (22 205 PY, 83.8%), 17 900 PY (67.6%) was in those of white ethnicity and 18 760 PY (70.8%) in those with a homosexual risk for HIV infection. At the start of therapy, the median [interquartile range (IQR)] CD4 cell count and viral load over episodes were 190 (81–302) cells/mm3 and 4.6 (3.0–5.3) log10 copies/ml, respectively. At time of first achieving a viral load of ≤ 50 copies/ml, the median (IQR) CD4 cell count was 314 (197–460) cells/mm3. Characteristics of the patients included in the analysis are shown in Table 1.
During follow-up, 1853 (18.1%) patients experienced at least one episode of virological failure. A total of 2460 events occurred in total giving a rebound rate of 9.3 [95% confidence interval (CI), 8.9–9.7]/100 PY. Table 2 shows the rates of viral rebound according to both the number of regimens previously failed and duration of viral suppression. Within the first year of viral suppression, the rate of viral rebound was 8.3/100 PY (95% CI, 7.5–9.1) in patients who had not failed treatment previously, rising to 32.7/100 PY (95% CI, 27.6–37.8) in patients who had failed four or more regimens. Viral rebound rates decreased progressively with increased duration of viral suppression despite one or more previous episodes of failure. In patients who had failed one or more antiretroviral combinations and remain suppressed at 4 years, rebound rates fell to levels nearly as low as those seen in patients who had never experienced virologic failure. After 2–3 years of suppression the rate of rebound was 4.1/100 PY (95% CI, 3.3–4.9) in naive and 6.3/100 PY (95% CI, 3.1–11.2) in patients who had failed four or more regimens; after ≥ 4 years, the corresponding rebound rates were 3.0/100 PY (95% CI, 2.0–4.0) and 1.4/100 PY (95% CI, 0.0–8.1).
Factors associated with viral rebound are shown in Table 3. In a Poisson regression model, patients of black ethnicity had a 48% increase in the rate of viral rebound compared to those of white ethnicity [1.48 (range, 1.26–1.73)]. Patients commencing therapy between 1996 and 1998 were more than twice as likely to experience viral rebound compared to those patients who started therapy after 2002 [2.17 (1.82–2.58)]. Per 10-year increase in age, the rate of rebound decreased by 22% [0.78 (0.74–0.82)]. Patients with a CD4 cell count > 350 cells/mm3 at the time of entering the analyses were less likely to experience viral rebound. The risk of viral rebound increased by 38% [1.38 (1.34–1.42)] for every regimen previously failed and decreased by 32% [0.68 (0.68–0.70)] for every 1-year increase in duration of viral suppression to ≤ 50 copies/ml plasma HIV RNA.
The test for interaction between number of past regimens failed and time under suppression was statistically significant with a P-value of 0.0003. Thus, although rebound rates were higher in those with a greater number of previously failed regimens, this effect waned with increased duration of suppression (as illustrated in Table 2). To illustrate this, Fig. 2 shows the relative rate of rebound, estimated separately after stratifying by the number of regimens failed according to time with viral suppression. Patients who had remained virologically suppressed for less then 1 year had a 43% [1.43 rate ratio (RR) (1.36–1.5)] increased chance of viral rebound per extra regimen failed. This decreased to 23% [1.23 RR (1.02–1.48)] in those who remain suppressed for 1–2 years; there was no association between the number of regimens previously failed and viral rebound among those suppressed for > 4 years [0.99 RR (0.76–1.28)].
This is the first study to specifically focus on the effect of the duration of viral suppression on the rate of viral rebound in relation to the number of antiretroviral regimens previously failed. Although viral rebound rates increased with the number of regimens previously failed they declined progressively with increased duration of viral suppression, regardless of the number of previous failures. Within the first year of viral suppression, the rate of rebound ranged from 8.3/100 PY in patients on first-line therapy, to 32.7/100 PY in patients who had failed four or more regimens. However, this large difference in viral rebound rate observed between those on first line and heavily treatment experienced patients was substantially reduced with increasing time with suppression and there appeared to be little difference in the rebound rate observed in patients who remained suppressed for > 4 years.
These results showing that viral rebound rates decrease progressively with increased duration of viral suppression are consistent with data from other observational cohorts [7,10,15]. This is likely to be due to a selection effect, as patients who are able to suppress their virus for the longest periods tend to be those who are most adherent to therapy, those who experience least drug toxicities, those who achieve the highest plasma drug concentrations or those who have the fewest pre-existing drug resistance mutations. Our findings suggest that there is a period of a few years of higher risk of viral rebound during which close patient monitoring, adherence support, and close monitoring for viral rebound are most important. It seems likely that pre-existing resistance makes a contribution to the raised rate of rebound in those with multiple regimen failures  and so it may be appropriate to conduct randomized trials to assess the value of adding new drugs, if available, in this high-risk group within the first year or two.
In patients starting HAART it is well established that those with prior exposure to mono or dual therapy experience higher rates of viral rebound compared to treatment naive patients [1,7,8,10,12,15,16]. Treatment experienced patients are more likely to harbour mutations in HIV pol that confer decreased sensitivity to therapy and a number of studies have demonstrated a significant correlation between these mutations and virological response to a new treatment regimen in patients who have already failed a different regimen [17,18].
In accordance with the results of a previous study  our results indicate that a lower baseline CD4 cell count was a significant predictor of viral rebound. One potential explanation that requires investigation is that some drugs may not be maintained at optimal levels in those with low CD4 cell counts, due to poorer drug absorption, drug–drug interactions, adherence, or other factors.
Baseline viral load at the start of ART was not associated with an increased risk of rebound. This contradicts the findings of a previous study whereby lower HIV-1 RNA levels at the time of starting ART independently predicted a higher chance of therapeutic success in terms of achieving and maintaining an undetectable viral load .
We  and others  have already shown an association between black ethnicity and an increased risk of viral rebound, which also appears in this study. This may in part be related to socio-economic factors. Interestingly, older age was found to be a significant predictor of a better response. Previous studies have shown that younger age is associated with poorer responses to therapy , and it is already known that in younger patients with chronic conditions such as cystic fibrosis, adherence to medication poses a significant problem . The reasons for these associations are unclear, however social, or behavioural factors may play an important role.
Calendar year was associated with an increased risk of viral rebound. Individuals commencing therapy between the years of 1996 to 1998 were at increased risk of viral rebound. These patients are more likely to have had prior exposure to NRTI mono or dual therapy and have accumulated mutations associated with drug resistance. It is likely that in recent years greater emphasis has been placed on maintaining good adherence and contemporary regimens may now be more intrinsically effective.
The focus of this analysis was to investigate the effect of previous treatment failures on viral rebound rates and duration of viral suppression and we did not focus on rebound rates associated with particular NRTI combinations and ‘third’ agents. Such findings from the UK CHIC Study have recently been described elsewhere .
In conclusion, in those patients with suppression, previous failure is associated with a higher rate of viral rebound, but the likelihood of rebound declines with the duration of suppression. Therefore prolonged viral suppression should always be the goal when designing each new regimen. Achievement of viral suppression in a patient with previous multiple failures is followed by a period of high rebound-risk that requires careful ongoing management and support. Both clinicians and patients can be encouraged that regardless of previous treatment failure, rebound rates even after multiple previous therapy failure fall to levels approaching those of patients on first line therapy after 4 years of suppressive therapy. This has important implications for providing treatment support and advice during those 4 years of follow up.
We thank all the clinicians, data managers and research nurses in participating clinical centres who have assisted with the provision of data for this project.
Sponsorship: This work was funded by the Medical Research Council, UK (Grants G0000199 and G0600337).
1. Staszewski S, Miller V, Sabin CA, Schlecht C, Gute P, Stamm S, et al
. Determinants of sustainable CD4 lymphocyte count increases in response to antiretroviral therapy. AIDS 1999; 13:951–956.
2. Kempf DJ, King MS, Bernstein B, Cernohous P, Bauer E, Moseley, et al
. Incidence of resistance in a double-blind study comparing lopinavir/ritonavir plus stavudine and lamivudine to nelfinavir plus stavudine and lamivudine
. J Infect Dis
3. Johnson V, Shafer R, Smeaton L, DeGruttola V, Winters M, Hazelwood J, et al
. Relative superiority of an initial antiretroviral strategy is not determined only by frequency of emergence of HIV drug resistance and cross resistance
. Eleventh Conference on Retroviruses and Opportunistic Infections.
San Francisco, February 2004 [abstract 662].
4. Parkin NT, Deeks SG, Wrin MT, Yap J, Grant RM, Lee KH, et al
. Loss of antiretroviral drug susceptibility at low viral load during early virological failure in treatment experienced patients. AIDS 2000; 14:2877–2887.
5. The PLATO Collaboration. Predictors of CD4 count trend and mortality among HIV-1-infected persons with virologic failure to all three antiretroviral drug classes
6. Mocroft A, Devereaux H, Kinloch-de-Loes S, Wilson D, Madge S, Youle M, et al
. Immunological, virological and clinical response to highly active antiretroviral therapy treatment regimens in a complete clinic population. AIDS 2000; 14:1545–1552.
7. Mocroft A, Ruiz L, Reiss P, Ledergerber B, Katlama C, Lazzarin A, et al
. Virological rebound after suppression on highly active antiretroviral therapy. AIDS 2003; 17:1741–1751.
8. Ledergerber B, Egger M, Opravil M, Telenti A, Hirschel B, Battegay M, et al
. Clinical progression and virologic failure on highly active antiretroviral therapy in HIV-1 patients: a prospective cohort study. Lancet 1999; 353:863–868.
9. Paredes R, Mocroft A, Kirk O, Lazzarin A, Barton S, Van Lunzen J, et al
. Predictors of virologic success and ensuing failure in HIV-positive patients starting highly active antiretroviral therapy in Europe. Arch Intern Med 2000; 160:1123–1132.
10. Phillips AN, Staszewski S, Lampe F, Youle M, Klauke S, Bickel M, et al
. Human immunodefiency virus rebound after suppression to <400 copies/mL during initial highly active antiretroviral therapy regimens, according to prior nucleoside experience and duration of suppression. J Infect Dis 2002; 186:1086–1091.
11. Phillips AN, Ledergerber B, Horban A, Reiss P, Chiesi A, Kirk O, et al
. Rate of viral rebound according to specific drugs in the regimen in 2120 patients with HIV suppression. AIDS 2004; 18:1795–1804.
12. Smith CJ, Phillips A, Hill T, Fisher M, Gazzard B, Porter K, et al
. The rate of viral rebound after attainment of an HIV load <50 copies/mL according to specific antiretroviral drugs in use: Results from a multicenter cohort study. J Infect Dis 2005; 192:1387–1397.
13. Le Moing V, Chene G, Carrieri MP, Alioum A, Brun-Vezinet F, Piroth L, et al
. Predictors of virological rebound in HIV-1 infected patients initiating a protease inhibitor containing-regimen. AIDS 2002; 16:21–29.
14. The UK Collaborative HIV Cohort Steering Committee. The creation of a large UK-based multicentre cohort of HIV-infected individuals: the UK Collaborative HIV Cohort (UK CHIC) study
. HIV Med
15. Phillips AN, Miller V, Sabin C, Cozzi Lepri A, Klauke S, Bickel A, et al
. Durability of HIV-1 viral suppression over 3.3 years with multi-drug antiretroviral therapy in previously drug-naïve individuals. AIDS 2001; 15:2379–2384.
16. Collaboration of HIV cohorts. Nucleoside analogue use before and during highly active antiretroviral therapy and viral load rebound
. J Infect Dis
17. Durant J, Clevenbergh P, Halfon P, Delgiudice P, Porsin S, Simonet P, et al
. Drug resistance genotyping in HIV-1 therapy: the VIRADAPT randomised controlled trial. Lancet 1999; 353:2195–2199.
18. Deeks SG, Barbour JD, Martin JN, Swanson MS, Grant RM. Sustained CD4+T cell response after virological failure of protease inhibitor based regimens in patients with human immunodefiency virus infection. J Infect Dis 2000; 181:946–953.
19. Miller V, Staszewski S, Sabin CA, Carlebach A, Rottmann C, Weidmann et al
. CD4 lymphocyte count as a predictor of the duration of highly active antiretroviral therapy-induced suppression of human immunodeficiency virus load
. J Infect Dis
20. Parienti JJ, Massari V, Deschamps D, Vabret A, Bouvet E, Larouze B, et al
. Predictors of virologic failure and resistance in HIV-infected patients treated with nevirapine or efavirenz based antiretroviral therapy. Clin Infect Dis 2004; 38:1311–1316.
United Kingdom Collaborative HIV Cohort Study Group
UK CHIC steering committee
J. Anderson, A. Babiker, V. Delpech, D. Dunn, P. Easterbrook, M. Fisher, B. Gazzard (Chair), R. Gilson, M. Gompels, T. Hill, M. Johnson, C. Leen, C. Orkin, A. Phillips, D. Pillay, K. Porter, C. Sabin, A. Schwenk, J. Walsh.
Royal Free and University College, London (L. Bansi, T. Hill, A. Phillips, C. Sabin); Medical Research Council Clinical Trials Unit (MRC CTU), London (A. Babiker, D. Dunn, S. Patel, K. Porter, S. Sheehan).
King's College Hospital, London (P. Easterbrook, S. Duffell, E. Bazuaye, E. Macfarlane); Brighton and Sussex University Hospitals NHS Trust (M. Fisher, N. Perry, A. Pullin, D. Churchill, W. Harris); Chelsea and Westminster NHS Trust, London (B. Gazzard, S. Bulbeck, S. Mandalia, J. Clarke); Mortimer Market Centre, Royal Free and University College Medical School (RFUCMS), London (R. Gilson, J. Dodds, A. Rider, I. Williams); Health Protection Agency —Communicable Disease Surveillance Centre, London (V. Delpech); Royal Free NHS Trust and RFUCMS, London (M. Johnson, M. Youle, F. Lampe, C. Smith, H. Gumley, C. Chaloner, D. Ismajani Puradiredja); St. Mary's Hospital, London (J. Walsh, J. Weber, C. Kemble); Barts and The London NHS Trust, London (C. Orkin, R. Thomas, K. Jones, J. Hand); North Middlesex, London (A. Schwenk, J. Ainsworth); Homerton, London (J. Anderson, S. Gann, K. Jones); Edinburgh (C. Leen, A. Wilson)
21. Zindani GN, Streetman DD, Streetman DS, Nasr SZ. Adherence to treatment in children and adolescent patients with cystic fibrosis. J Adolesc Health 2006; 38:13–17.