Andrew N. Phillipsa, Bonaventura Clotetb, Jens D. Lundgrenh, for the
EuroSIDA study group�
Objective: To determine the rate of virological rebound and factors associated withrebound among patients on highly active antiretroviral therapy (HAART) with pre-viously undetectable levels of viraemia.
Design: An observational cohort study of 2444 patients from the EuroSIDA study.
Methods: Patients were followed from their first viral load under 400 copies/ml to thefirst of two consecutive viral loads above 400 copies/ml. Incidence rates werecalculated using person-years of follow-up (PYFU), Cox proportional hazards modelswere used to determine factors related to rebound.
Results: Of 2444 patients, 1031 experienced virological rebound (42.2%). Theincidence of rebound decreased over time; from 33.5 in the first 6 months after initialsuppression to 8.6 per 100 PYFU at 2 years after initial suppression (P , 0.0001). Therate of rebound was lower for treatment-naive compared with treatment-experiencedpatients. In multivariate models, patients who changed treatment were more likely torebound, as were patients with higher viral loads on starting HAART. Treatment-naivepatients were less likely to rebound. Among pretreated patients, those who werestarted on new nucleosides were less likely to rebound.
Conclusion: The rate of virological rebound decreased over time, suggesting that thegreatest risk of treatment failure is in the months after initial suppression. Treatment-naive patients were at a lower risk of rebound, but among drug-experienced patients,those who added new nucleosides had a lower risk of rebound, as were patients witha good immunological response. & 2003 Lippincott Williams & Wilkins
AIDS 2003, 17:1741–1751
Keywords: highly active antiretroviral therapy, rebound, response, viral load
From the aRoyal Free Centre for HIV Medicine and Department of Primary Care and Population Sciences, Royal Free andUniversity College Medical School, London, UK; bFundacio IrsiCaixa and HIV Unit, Hospital Universitari (UAB) ‘Germans Triasi Pujol’, Badalona, Spain; cAcademisch Medisch Centrum bij de Universiteit van Amsterdam, Amsterdam, the Netherlands;dUniversity Hospital, Zurich, Switzerland; eHopital Pitie-Salpetriere, Paris, France; fOspedale San Raffaele, Milan, Italy;gMedizinische Poliklinik, Munich, Germany; and hCHIP, Hvidovre Hospital, Hvidovre, Denmark.�Members of the study group are listed in the appendix.
Correspondence to: Dr A Mocroft, Royal Free Centre for HIV Medicine and Department of Primary Care and PopulationSciences, Royal Free and University College London Medical Schools, Royal Free Campus, Rowland Hill Street, London NW32PF, UK.
Received: 7 November 2002; revised: 14 February 2003; accepted: 11 March 2003.
DOI: 10.1097/01.aids.0000076310.76477.27
ISSN 0269-9370 & 2003 Lippincott Williams & Wilkins 1741
Introduction
The virological response to highly active antiretroviraltherapy (HAART) is now well documented in clinicpopulations of patients with HIV infection [1–6].Patients respond with a rapid reduction in plasma viralload with a median time to undetectable levels ofapproximately 4 months. It has been suggested that theachievement and maintenance of prolonged viral sup-pression is directly related to the long-term efficacy ofHAART [7]. However, the viral load may start to risein a substantial proportion of patients [1,5,8–10],which may be related to potentially serious adverseevents, the emergence of drug-resistant viruses, thedifficulties of maintaining long-term adherence, andthe limited number of available drugs [11,12]. Fewstudies evaluating the long-term response to HAART[13,14] and more specifically an investigation of thereasons for viral rebound have been published. Someclinical trials have considered virological rebound[15,16], but these were generally based on small num-bers, and patient populations tend to differ somewhatfrom routine clinic populations [17,18]. In observa-tional studies, the rates of virological rebound havebeen reported to be between 20 and 40%, with higherrates of virological rebound among patients with higherviral loads on starting HAART, in those with pooradherence and those who have undergone previousantiretroviral treatment [19,20].
The aims of this study were to describe the rate ofvirological rebound among 2444 HIV-infected patientsin the EuroSIDA study, and to investigate the factorsrelated to virological rebound.
Methods
PatientsThe EuroSIDA study is a prospective, European studyof patients with HIV-1 infection in 70 centres acrossEurope (including Israel, see Appendix) and nowincluding Argentina. Details of the study have beenpublished elsewhere [21]. In brief, centres provideddata on consecutive patients seen in the outpatientclinic from 2 May 1994 until a predefined number ofpatients was enrolled from each centre. This cohort of3116 patients was defined as the EuroSIDA I cohort.The enrolment of a second cohort of 1365 patientsbegan in December 1995. In April 1997, a further2839 patients were recruited and were defined as theEuroSIDA III cohort. Cohort IV, consisting of 1225patients, was enrolled from April 1999, and a fifthcohort, cohort V, consisting of 1256 patients, wasrecruited from September 2001. For cohorts I–III,eligible patients were those with a CD4 lymphocytecount of below 500 cells/mm3 in the previous 4
months, a booked clinic appointment and who wereolder than 16 years at the time of enrolment. The CD4lymphocyte count restriction was removed for cohortsIV and V. Information was provided on a standardizeddata collection form at baseline and every 6 monthsthereafter. Follow-up was to Spring 2002, with infor-mation from up to 16 forms available for cohort I, 13for cohort II, 10 for cohort III, five for cohort IV andone for cohort V. At each follow-up visit, details on allCD4 lymphocyte counts measured since the lastfollow-up and viral load measurements were collected.For each patient, the date of starting and stopping eachantiretroviral drug was recorded, as was the use ofdrugs for prophylaxis against opportunistic infections.The dates of diagnosis of all AIDS-defining illnesseshave also been recorded, including those diagnosesmade subsequent to the initial diagnosis, using the 1993clinical definition of AIDS from the Centers forDisease Control [22]. Members of the coordinatingoffice visited all centres to ensure correct patientselection and that accurate data were provided bychecking the information provided against case-notesfor a proportion of patients.
Statistical methodsHAART was defined as starting a protease inhibitor(PI) or a non-nucleoside reverse transcriptase inhibitor(NNRTI) for the first time, combined with at leasttwo nucleoside reverse transcriptase inhibitors (NRTI),or starting triple NRTI therapy with abacavir. For viralload measurements, the Roche was the predominantmethod used in 64% of centres, nucleic acid sequence-based amplification was used in 16%, and the Chironbranched DNA approach was used in 20%. For thepurposes of this analysis, a more general definition ofan undetectable viral load was defined as below 400copies/ml, regardless of whether the viral load wasmeasured to a greater degree of sensitivity. Patientswere followed from the first date of viral load below400 copies/ml after the initiation of HAART, provid-ing this occurred within 12 months of startingHAART, until the date of the first of two consecutiveviral loads above 400 copies/ml, or until the date ofthe last viral load measure, for patients whose viral loaddid not rebound. Patients were excluded from theseanalyses if their viral load became undetectable onHAART before prospective follow-up in the EuroSI-DA study began. Patients were grouped according towhether they were treatment naive or experienced atstarting HAART. Chi-squared tests and non-para-metric tests such as the Wilcoxon test were used tocompare the characteristics of patients included inanalyses.
The categorical variables considered were sex, exposuregroup, race, region of Europe, clinical status (i.e.whether patients had ever received an AIDS diagnosis),the number of antiretroviral drugs included in the
HAART regimen, specific drugs included in the regi-men, date of starting HAART, NRTI combinations,HAART regimen (i.e. single PI regimen, dual PIregimen, NNRTI regimen or mixed regimen), limit ofdetection of viral load used and a change (i.e. starting anew antiretroviral drug or stopping any antiretroviraldrug) in any of the antiretroviral drugs included in theinitial HAART regimen. Continuous variables in-cluded the date of starting HAART, CD4 lymphocytecount and viral load at or before starting HAART (andwithin 12 months of starting HAART), CD4 lympho-cyte count at the date of viral suppression, time toachieve viral suppression from starting HAART, age,time with AIDS, and the percentage increase in theCD4 lymphocyte count between starting HAART andachieving suppression. Additional variables were exam-ined for treatment-experienced patients, namely, thenumber of new NRTI started at the date of startingHAART (i.e. the number of NRTI to which patientshad never previously been exposed), the number ofNRTI ever taken, specific NRTI taken, the time sincestarting antiretroviral therapy, and what previous anti-retroviral therapy they had been exposed to (i.e.monotherapy only, dual therapy only or both monoand dual therapy).
Kaplan–Meier estimates were used to examine thecumulative probability of virological rebound overtime, both overall and stratified by previous treatmentexperience. The incidence rates of viral rebound (num-ber of virological rebounds/number of person-yearsfollow-up; PYFU) in different periods of follow-upand according to previous NRTI experience werecalculated and differences between groups were assessedusing Poisson regression. Cox proportional hazardsmodels, stratified by centre, were used to determinewhich factors were independently related to virologicalrebound, both overall among all patients, and withintreatment-naive and experienced patients.
Poisson regression was performed using STATA (ver-sion 7), all other analyses were performed using SAS(Statistical Analysis Software, version 6.12; Cary, NC,USA).
Results
Table 1 describes the patients included in the study,stratified by previous treatment. Of 2444 patients, 623(25.5%) were treatment naive before starting HAART.There were some differences between groups accordingto previous treatment. Over 70% of patients fromeastern Europe were treatment naive (79 patients,73.1%) compared with 28.3% of patients from northernEurope, 20.5% from central Europe and 20.5% fromsouthern Europe (P , 0.001, chi-squared test). The
more intensive treatment regimens tended to have beenin patients who were treatment experienced beforeHAART, with 100% of those on five antiretroviraldrugs being treatment experienced before startingHAART (P , 0.001, chi-squared test). Treatment-naive patients tended to start HAART later in time,experience a higher percentage increase in CD4lymphocyte count by the time of viral suppression, andhave higher levels of viraemia at starting HAART(P, 0.0001, P , 0.0001, P , 0.0001, respectively,Wilcoxon tests). The median age was 38 years [inter-quartile range (IQR) 33.3–45.2), the median CD4 cellcount at starting HAART was 244 cells/mm3 (IQR125–358), and the median CD4 cell count at virologi-cal suppression was 300 cells/mm3 (IQR 180–430).The median time to viral suppression was 3 months(IQR 2–7). A total of 560 patients had been diagnosedwith AIDS (22.9%), and the most common nucleosidecombination used was zidovudine and lamivudine (853patients, 34.9%).
The most commonly used PI was indinavir (928patients, 38.0%), followed by ritonavir (540 patients,22.1%), and saquinavir hard gel (392 patients, 16.0%).Nevirapine was used in 303 patients (12.4%) andefavirenz in 166 patients (6.8%), whereas abacavir wasused by 125 patients (5.1%). Table 2 describes theprevious treatment of the 1821 patients who weretreatment experienced at starting HAART. The firsttreatment with antiretroviral drugs occurred a medianof 37 months before starting HAART (IQR 18–61).The majority of patients had been treated with bothmonotherapy and dual combination therapy beforeHAART (1069 patients, 58.7%), and almost all patientshad previously received zidovudine (1704 patients,93.6%). A total of 701 patients (38.5%) did not startany new NRTI (i.e. NRTI they had never previouslybeen treated with) at starting HAART, and 479 pa-tients (26.3%) were able to start HAART with two ormore new NRTI in the HAART regimen.
The viral load rebounded in 1031 patients (42.2%)during a median follow-up of 23 months (IQR 7–47).Of those who experienced virological rebound, 151(14.6%) were treatment naive. Fig. 1 describes theprogression to virological rebound, using Kaplan–Meier estimates. The progression is shown overall andstratified by treatment experience. There appears to bean initially high rate of virological rebound, whichslows down with increasing time from HAART; forexample, among all patients, at 12 months 27.9% ofpatients were estimated to have experienced virologicalrebound [95% confidence interval (CI) 26.1–29.7)] andat 24 months this increased to 37.5% (95% CI 35.5–39.5). Patients with previous antiretroviral treatmentexperienced viral load rebound at a significantly fasterrate than those patients who were treatment naive atstarting HAART (P, 0.0001, log-rank test).
Virological rebound after suppression on HAART Mocroft et al. 1743
Among all patients, regardless of previous treatmentexperience, there was a clear, statistically significanttrend for a decreasing rate of virological rebound as thetime since initial suppression increased [rate ratio (RR)0.63; 95% CI 0.60–0.67, P , 0.0001]. Overall, therate of virological rebound in the first 6 months afterinitial suppression was 33.5 per 100 PYFU (95% CI30.1–36.9), almost four times higher than the rate ofvirological rebound at or after 24 months after initialsuppression (incidence of 8.6 per 100 PYFU, 95% CI7.4–9.8). Fig. 2 describes the incidence of virologicalrebound according to both time from initial suppres-sion and previous treatment experience. The incidenceof rebound was significantly higher among treatment-experienced patients, but both groups showed a similardecline in incidence with increasing time from initialsuppression. Overall, the incidence of virological re-bound among treatment-naive patients was less thanhalf that of treatment-experienced patients (RR 0.47;95% CI 0.40–0.46, P , 0.0001). In all time periodsthe rate of virological rebound among treatment-naive
patients remained approximately half that of treatment-experienced patients.
In Cox models, the association with outcome wastested for all variables in univariate models and thosethat were significant (P , 0.1) were then included inmultivariate models. All models were stratified bycentre. Both changing treatment and CD4 lymphocytecount were included as time-updated covariates. Theresults are shown in Table 3. After adjustment, patientson four or five-drug regimens were significantly morelikely to experience virological rebound than those onthree drugs [relative hazard (RH) 1.29; 95% CI 1.08–1.54, P ¼ 0.0049 and 1.98; 95% CI 1.48–2.64, P,0.0001, respectively], as were patients who made anychange to their HAART regimen (RH 1.59, 95% CI1.39–1.82, P , 0.0001). Compared with patientswhose viral load was below 400 copies/ml whenstarting HAART, all other patients were more likely toexperience virological rebound (excluding patients forwhom we did not know the viral load at starting
HAART, Highly active antiretroviral therapy; IDU, injection drug user; IQR, interquartile range; NNRTI, non-nucleosidereverse transcriptase inhibitor; NRTI, nucleoside reverse transcriptase inhibitor; PI, protease inhibitor.aViral load in the 12 months before starting HAARTwas available for 2099 patients (85.6%) at a median of 2 months beforestarting HAART (IQR 0–3 months).
AIDS 2003, Vol 17 No 121744
HAART). Patients starting a single-NNRTI regimenwere more likely to have a rebound in viral load (RH1.36; 95% CI 1.07–1.73, P ¼ 0.011), when comparedwith patients starting HAART with a single PI-basedregimen. Compared with treatment-experienced pa-tients, those who were treatment naive were signifi-cantly less likely to experience virological rebound(RH 0.56; 95% CI 0.46–0.68, P , 0.0001). It was also
interesting to note that patients whose viral load wasknown to be below 50 copies/ml at the date of initialsuppression were significantly less likely to experiencevirological rebound than patients whose viral load wasundetectable at 400 copies/ml (RH 0.54; 95% CI0.44–0.67, P , 0.0001). Older patients, those whostarted HAART more recently, and those with highercurrent CD4 lymphocyte counts were significantly lesslikely to experience virological rebound (Table 3).
The analysis was repeated for treatment-experiencedpatients, and the results are also shown in Table 3.Additional variables included in this analysis were thenumber of NRTI ever taken, what regimens thepatients had tried (i.e. mono, dual or both), and alsothe time since starting antiretroviral therapy. None ofthese additional variables were of significance in uni-variate analyses. The only variable related to the risk ofvirological rebound was the number of new NRTIincluded in the HAART regimen. In multivariateanalyses, the results of the analyses were very similar towhen all patients were included in the analysis. Inaddition, the risk of virological rebound was 19% lowerfor each new NRTI added to the HAART regimen(RH 0.81, 95% CI 0.74–0.88, P , 0.0001).
A Cox model including all patients was constructedthat redefined the baseline date to be 1 January 1995,and patients were then left-censored until the date ofinitial virological suppression. This analysis allowed thedecreasing rate of virological rebound over time to be
CI, Confidence interval; HAART, highly active antiretroviral therapy; NNRTI, non-nucleoside reverse transcriptase inhibitor; NRTI, nucleosidereverse transcriptase inhibitor; PI, protease inhibitor; RH, relative hazard.aUnivariate estimates are from Cox models including all patients with the exception of bvariable fitted as a time-dependent covariate, andcvariable fitted among treatment-experienced patients only.
AIDS 2003, Vol 17 No 121746
formally tested after adjustment for the other factorsrelated to virological rebound. In univariate analyses,there was a 15% decreased risk of virological reboundwith each additional 6 months since initial virologicalsuppression (RH 0.85, 95% CI 0.81–0.89, P ,0.0001). This confirmed the results of Fig. 2, in whichthere was strong evidence of a decrease in the rate ofvirological rebound over time. After adjustment for thefactors related to virological rebound shown in Table3, there remained a 27% decreased risk of virol-ogical rebound with each additional 6 months sinceinitial suppression (RH 0.73, 95% CI 0.64–0.84, P,0.0001).
Several sensitivity analyses were also performed toinvestigate how sensitive the results were to minorchanges in the definitions used. In the first sensitivityanalysis, patients were censored at the date of changingany aspect of their initial HAART regimen. As somepatients changed therapy before their initial virologicalsuppression, this reduced the number of patients in-cluded in the analysis to 1944, of whom 404 (20.8%)experienced virological rebound. Highly consistentresults to those shown in Table 3 were found. In afurther analysis, patients were eligible for inclusion inanalyses if they achieved their initial virological sup-pression within 6 months of starting HAART. Thisreduced the number of patients included in the analysisto 1726, of whom 678 experienced virological rebound(39.3%). The results were once again highly consistent.Further analysis excluded patients whose viral load wasunknown or was below 400 copies/ml at startingHAART, excluding all patients who started saquinavirhard gel (because of its poor bioavailability), anddefining failure as a single viral load above 400 copies/ml, all with similar results As a different limit ofdetection was used in different centres, the analyseswere repeated using a definition of virological successand rebound of 200 or 50 copies/ml for all centres.Each analysis produced very similar results, and allresults were consistent with those shown in Table 3(further details available on request).
Discussion
EuroSIDA is one of the largest European observationalstudies of patients with HIV-1 infection. The study hasfound a substantial number of patients who had arebound in viral load after initial virological suppressionon HAART regimens, but the rate of virologicalrebound decreased over time, suggesting that if theviral load does not rebound in the initial months aftervirological suppression, there is less risk of virologicalrebound over time as the time from suppression in-creases. Whereas treatment-naive patients had the low-est risk of virological rebound, treatment-experienced
patients who could add new antiretroviral drugs totheir HAART regimen were also significantly lesslikely to experience virological rebound than thosewho did not add new antiretroviral drugs to theirHAART regimen.
It has been well established that previously treatedpatients who start HAART tend to experience a poorervirological response to HAART [1–5,23]. It is likelythat treatment-experienced patients were at a higherrisk of harbouring drug-resistant mutations [24]. Severalstudies have demonstrated a significant correlation be-tween drug resistance and virological response to a newtreatment regimen in patients who have failed previoustherapy [25,26], and the presence of drug resistance hasbeen shown to be independently related to virologicalresponse [16]. Likewise, adherence may play a criticalrole in virological rebound [27], as the results fromboth clinical trials and observational studies have shownthat adherence plays a role in virological failure[10,15,19,27]. However, we do not have that informa-tion available in the EuroSIDA study. It is not possibleto estimate the extent to which patients have beenadherent to therapy; however, as one of the inclusioncriteria was an initial response to HAART, all patientsappear to have been adherent to therapy to someextent for the first few months of HAART.
Patients on more intensive antiviral treatment regimenswere significantly more likely to have a rebound inviral load, which could be partly explained by com-pliance or side-effects. The treatment schedule andside-effect profile of a five-drug HAART regimen islikely to be significantly more complicated [11,12], andsuch regimens are generally used among patients withprevious treatment failure. Patients who swapped anydrug in their HAART regimen were also more likelyto experience virological rebound. As the lower limitof detection in some patients was less than 200 or 50copies/ml, this may indicate that some patients changedtheir regimen because of low-level increases in viralload. However, even after adjustment for the lowerlimit of detection used, patients who swapped any drugin their HAART regimen were more likely to experi-ence virological rebound, suggesting that this cannotexplain all of the increased risk of virological rebound.Again, this may raise questions about compliance orside-effects, as physicians may change drugs in order toincrease compliance or reduce side-effects, but pro-blems of drug resistance may already have started, ortolerance to side-effects may already have decreasedsubstantially.
Patients with the lowest levels of viral load (i.e. meas-ured to below 50 copies/ml), were significantly lesslikely to experience virological rebound than thosewhose viral load was below 400 copies/ml. Patientswith viral loads measured to a lower level (such as 200
Virological rebound after suppression on HAART Mocroft et al. 1747
or 50 copies/ml) would be included in the analyses andnot defined as a virological failure until their viral loadrebounded above 400 copies/ml, even if individualcentres may classify them as failing treatment with viralrebounds to above 50 copies/ml. There is conflictingevidence on the importance of low levels of viralrebound, or ‘blips’ on long-term virological suppression[28,29], although these results may suggest that it isimportant, in terms of virological rebound, to achievethe lowest possible levels of viraemia. Several studiesthat used ultrasensitive methods of determining viralload have also suggested that the long-term durabilityof virological response depends on achieving low levelsof viraemia [30,31]; however, the long-term clinicalbenefits of achieving very low levels of viraemia haveyet to be determined. It was also interesting to notethat, among treatment-experienced patients, those whocould add new drugs to their HAART regimen weresignificantly less likely to have a rebound in viral load.Adding new drugs when forming a HAART regimenhas previously been shown to affect the virologicalresponse to HAART favourably [1,32]. With longerfollow-up, however, the differences in virological re-bound between treatment-naive and experienced pa-tients who start new drugs may increase, simplybecause experienced patients have a lower number offuture options available to them.
A further important finding was that the rate ofvirological rebound decreased over time, both in naiveand treatment-experienced patients. The difference invirological rebound over time between treatment-experienced and treatment-naive patients persisted be-yond 2 years. This has previously been shown for otherobservational cohorts [20,33]. One possible explanationfor this finding is a selection effect, in which patientswho were more likely to experience virological re-bound have been selected out as the time from initialresponse has increased. Those that experience more orpotentially serious toxicities, are less adherent, or havedeveloped a greater degree of resistance may thus havea rebound in viral load more quickly. Another morespeculative explanation is that the declining rate ofvirological rebound reflects the decline in newlyactivated infected cells, as the pool of latently infectedcells becomes reduced, but this would seem to beinconsistent with the observed low rate of decline inthe numbers of latently infected cells [34].
Patients taking NNRTI were more likely to experiencevirological rebound. The majority of patients weretaking nevirapine as their first NNRTI-containingHAART regimen. Several observational studies, in-cluding EuroSIDA, have reported a superior virologicalresponse to HAART in efavirenz-containing regimenscompared with nevirapine [35–37], both in naive andtreatment-experienced patients. The results of specifictreatment comparisons from observational studies
should always be interpreted with caution, and weawait the results from randomized clinical trials(START and 2NN).
There are several caveats to note. We required thatpatients respond to their initial regimen within 12months. It could be argued that patients should respondto HAART within the first 6 months of therapy[11,12], although previous work has shown that pa-tients continue to achieve undetectable levels of vir-aemia up to 12 months after starting HAART andwithout changes in HAART, especially for those withhigh viral loads [4]. In addition, treatment guidelinesare generally based on clinical trials, in which thefrequency of viral load measurements may be morefrequent than in clinical practice. In sensitivity analyses,when we required that the initial HAART successoccurred within the first 6 months, our conclusionsremained unaltered. In addition, we used a globaldefinition of virological rebound using 400 copies/ml,even though in some cases, we had viral loads meas-ured using a sensitivity of 200 or 50 copies/ml. How-ever, we repeated our analyses using differentdefinitions of virological rebound, including an analysisthat selected patients whose viral load was measuredusing a sensitivity of 50 copies/ml, and by definingvirological failure on the basis of a single viral loadabove the limit of detection, and all showed remarkablysimilar results.
In summary, we found a higher rate of virologicalrebound in treatment-experienced patients comparedwith treatment-naive patients, and a decrease in therate of virological rebound over time. The long termfollow-up of large cohorts such as EuroSIDA isessential to monitor whether virological rebound even-tually results in immunological failure and subsequentclinical failure.
References
1. Paredes R, Mocroft A, Kirk O, Lazzarin A, Barton SE, van LunzenJ, et al. Predictors of virological success and ensuing failure inHIV-positive patients starting highly active antiretroviral therapy.Arch Intern Med 2000, 160:1123–1132.
2. Grabar S, Pradier C, le Corfec E, Lancar R, Allavena C, BentataN, et al. Factors associated with clinical and virological failurein patients receiving a triple therapy including a proteaseinhibitor. AIDS 2000, 14:141–149.
3. Wit FW, van Leeuwen R, Weverling GJ, Jurriaans S, Nauta K,Steingrover R, et al. Outcome and predictors of failure of highlyactive antiretroviral therapy: one-year follow-up of a cohort ofhuman immunodeficiency virus type 1-infected persons. J InfectDis 1999, 179:790–798.
4. Staszewski S, Miller V, Sabin CA, Carlebach A, Berger AM,Weidmann E, et al. Virological response to protease inhibitortherapy in an HIV clinic cohort. AIDS 1999, 13:367–373.
5. Ledergerber B, Egger M, Opravil M, Telenti A, Hirschel B,Battegay M, et al. Clinical progression and virological failure onhighly active antiretroviral therapy in HIV-1 patients: a prospec-tive cohort study. Lancet 1999, 353:863–868.
6. Fatkenheuer G, Theisen A, Rockstroh J, Grabow T, Wicke C,Becker K, et al. Virological failure of protease inhibitor therapyin an unselected cohort of HIV-infected patients. AIDS 1997,11:F113–F116.
7. Staszewski S, Miller V, Sabin C, Schlecht C, Gute P, Stamm S,et al. Determinants of sustainable CD4 lymphocyte count in-creases in response to antiretroviral therapy. AIDS 1999,13:951–956.
8. Mocroft A, Devereaux H, Kinloch-de-Loes S, Wilson D, Madge S,Youle M, et al. Immunological, virological and clinical responseto highly active antiretroviral therapy treatment regimens in acomplete clinic population. AIDS 2000, 14:1545–1552.
9. Deeks SG, Hecht FM, Swanson M, Elbeik T, Loftu R, Cohen PT,et al. HIV RNA and CD4 cell count response to proteaseinhibitor therapy in an urban AIDS clinic: response to bothinitial and salvage therapy. AIDS 1999, 13:F35–F43.
10. Lucas GM, Chaisson RE, Moore RD. Highly active antiretroviraltherapy in a large urban clinic: Risk factors for virological failureand adverse drug reactions. Ann Intern Med 1999, 131:81–87.
11. British HIV Association Writing Committee. British HIV Associa-tion (BHIVA) guidelines for the treatment of HIV-infected adultswith antiretroviral therapy. HIV Med 2001, 2:276–313.
12. Yeni PG, Hammer SM, Carpenter CCJ, Cooper DA, Fischl MA,Gatell JM, et al. Antiretroviral treatment for adult HIV infectionin 2002. Updated recommendations of the International AIDSSociety – USA panel. JAMA 2002, 288:222–235.
13. Gulick RM, Mellors JW, Havlir D, Eron JE, Meibohm A, CondraJH, et al. 3-year suppression of HIV viremia with indinavir,zidovudine, and lamivudine. Ann Intern Med 2000, 133:35–39.
14. Kaufmann GR, Bloch M, Zaunders JJ, Smith D, Cooper DA. Longterm immunological response in HIV-1 infected subjects receiv-ing potent antiretroviral therapy. AIDS 2000, 14:959–969.
15. Descamps D, Flandre P, Calvez V, Peytavin G, Meiffredy V, CollinG, et al. Mechanisms of virologic failure in previously untreatedHIV-infected patients from a trial of induction-maintenance.JAMA 2000, 283:205–211.
16. Havlir DV, Hellmann NS, Petropoulos CJ, Whitcomb JM, CollierAC, Hirsch MS, et al. Drug susceptibility in HIV infection afterviral rebound in patients receiving indinavir-containing regi-mens. JAMA 2000, 283:229–234.
17. Gifford AL, Cunningham WE, Heslin KC, Andersen RM, Nakazo-no T, Lieu DK, et al. Participation in research and access toexperimental treatments by HIV-infected patients. N Engl J Med2002, 346:1373–1382.
18. Madge S, Mocroft A, Wilson D, Youle M, Lipman MCI, Tyrer M,et al. Participation in clinical trials amongst patients infectedwith HIV 1 in a single treatment centre over 12 years. HIV Med2000, 1:212–218.
19. Le Moing V, Chene G, Carrierri MP, Alioum A, Brun-Vezinet F,Proth L, et al. Predictors of virological rebound in HIV-1 infectedpatients initiating a protease-inhibitor containing regimen. AIDS2002, 16:21–29.
20. Phillips AN, Staszewski S, Lampe F, Youle M, Klaike S, Bickel M,et al. Human immunodeficiency virus rebound after suppressionto < 400 copies/ml during initial highly active antiretroviralregimens, according to prior nucleoside experience and dura-tion of suppression. J Infect Dis 2002, 186:1086–1091.
21. Lundgren JD, Phillips AN, Vella S, Katlama C, Ledergeber B,Johnson AM, et al. Regional differences in use of antiretroviralagents and primary prophylaxis in 3122 European HIV-infectedpatients. J Acquir Immune Defic Syndr 1997, 16:153–160.
22. Centers for Disease Control. 1993 revised classification systemfor HIV infection and expanded surveillance case definition forAIDS among adolescents and adults. MMWR 1992, 41 (No. RR-17):1–17.
24. Hirsch MS, Brun-Vezinet F, D’Aquila RT, Hammer SM, JohnsonVA, Kuritzkes DR, et al. Antiretroviral drug resistance testing inadult HIV-1 infection. Recommendations of an InternationalAIDS Society – USA panel. JAMA 2000, 283:2417–2426.
25. Deeks SG, Barbour JD, Martin JN, Swanson MS, Grant RM.Sustained CD4+ T cell response after virological failure ofprotease inhibitor based regimens in patients with human
26. 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.
27. Nieuwkerk PT, Sprangers MAG, Burger DM, Hoetelmans RMW,Hugen PWH, Danner SA, et al. Limited patients adherence tohighly active antiretroviral therapy for HIV-1 infection in anobservational cohort study. Arch Intern Med 2001, 161:1962–1968.
28. Greub G, Cozzi Lepri A, Ledergerber B, Staszewski S, Perrin L,Miller V, et al. Intermittent and sustained low-level HIV viralrebound in patients receiving potent antiretroviral therapy[Letter]. AIDS 2002, 16:1967–1969.
29. Havlir DV, Bassett R, Levian D, Gilbert P, Tebas P, Collier AC,et al. Prevalence and predictive value of intermittent viremiawith combination HIV therapy. JAMA 2001, 286:171–179.
30. Kempf DJ, Rode RA, Xu Y, Sun E, Heath-Chiozzi ME, Valdes J,et al. The duration of viral supression during protease inhibitortherapy for HIV-1 infection is predicted by plasma HIV-1 RNAat the nadir. AIDS 1998, 12:F9–F14.
31. Raboud J, Montaner JSG, Conway B, Rae S, Reiss P, Vella S, et al.Suppression of plasma viral load below 20 copies/ml is requiredto achieve a long-term response to therapy. AIDS 1998,12:1619–1624.
32. Mocroft A, Gill MJ, Davidson W, Phillips AN. Predictors of aviral response and subsequent virological treatment failure inpatients with HIV starting a protease inhibitor. AIDS 1998,12:2161–2167.
33. Phillips AN, Miller V, Sabin C, Cozzi Lepri A, Klauke S, Bickel M,et al. Durability of HIV-1 viral suppression over 3.3 years withmultidrug antiretroviral therapy in previously drug-naive indivi-duals. AIDS 2001, 15:2379–2384.
34. Finzi D, Blankson J, Siliciano JD, Margolick JB, Chadwick K,Pierson T, et al. Latent infection of CD4+ T cells provides amechanism for lifelong persistence of HIV-1, even in patients oneffective combination therapy. Nat Med 1999, 5:512–517.
35. Phillips AN, Pradier C, Lazzarin A, Clotet B, Pradier C, Goebel F-D, et al. Viral load outcome of non-nucleoside reverse transcrip-tase inhibitor regimens for 2203 mainly antiretroviral experi-enced patients. AIDS 2001, 15:2385–2395.
36. Lepri AC, Phillips AN, d’Arminio Monforte A, Piersantelli N,Orani A, Petrosillo N, et al. Virologic and immunologic responseto regimens containing nevirapine or efavirenz in combinationwith 2 nucleoside analogues in the Italian Cohort NaiveAntiretrovirals (ICoNA) Study. J Infect Dis 2002, 185:1062–1069.
37. Matthews GV, Sabin CA, Mandalia S, Lampe F, Phillips AN,Nelson MR, et al. Virological suppression at 6 months is relatedto choice of initial regimen in antiretroviral-naıve patients: acohort study. AIDS 2002, 16:53–61.
Appendix
The multicentre study group on EuroSIDA(national coordinators in parenthesis)Argentina: (M. Losso), A. Duran, Hospital J.M.Ramos Mejia, Buenos Aires, Argentina.
Austria: (N. Vetter) Pulmologisches Zentrum derStadt Wien, Vienna, Austria.
Belgium: (N. Clumeck) P. Hermans, B. Sommereijns,Saint-Pierre Hospital, Brussels, Belgium; R. Colebun-ders, Institute of Tropical Medicine, Antwerp, Bel-gium.
Virological rebound after suppression on HAART Mocroft et al. 1749
Denmark: (J. Nielsen) J. Lundgren, T. Benfield, O.Kirk, Hvidovre Hospital, Copenhagen, Denmark; J.Gerstoft, T. Katzenstein, B. Røge, P. Skinhøj, Rig-shospitalet, Copenhagen, Denmark; C. Pedersen,Odense University Hospital, Odense, Denmark.
Estonia: (K. Zilmer) Tallinn Merimetsa Hospital,Tallinn, Estonia.
France: (C. Katlama) M. De Sa, Hopital de la Pitie-Salpetiere, Paris, France; J.-P. Viard, Hopital Necker-Enfants Malades, Paris, France; T. Saint-Marc, HopitalEdouard Herriot, Lyon, France; P. Vanhems, Univer-sity Claude Bernard, Lyon, France; C. Pradier, Hopitalde l’Archet, Nice, France.
Germany: (M. Dietrich) C. Manegold, Bernhard-Nocht-Institut for Tropical Medicine, Hamburg, Ger-many; J. van Lunzen, H.-J. Stellbrink, EppendorfMedizinische Kernklinik, Hamburg, Germany; V. Mill-er, S. Staszewski, J.W. Goethe University Hospital,Frankfurt, Germany; F.-D. Goebel, Medizinische Po-liklinik, Munich, Germany; B. Salzberger, UniversitatKoln, Cologne, Germany; J. Rockstroh, UniversitatsKlinik, Bonn, Germany.
Greece: (J. Kosmidis) P. Gargalianos, H. Sambatakou,J. Perdios, Athens General Hospital, Athens, Greece;G. Panos, I. Karydis, A. Filandras, 1st IKA Hospital,Athens, Greece.
Hungary: (D. Banhegyi) Szent Laslo Hospital, Buda-pest, Hungary.
Ireland: (F. Mulcahy) St James’s Hospital, Dublin,Ireland.
Israel: (I. Yust) M. Burke, Ichilov Hospital, Tel Aviv,Israel; S. Pollack, Z. Ben-Ishai, Rambam MedicalCenter, Haifa, Israel; Z. Bentwich, Kaplan Hospital,Rehovot, Israel; S. Maayan, Hadassah University Hos-pital, Jerusalem, Israel.
Italy: (S. Vella, A. Chiesi) Istituto Superiore di Sanita,Rome, Italy; C. Arici, Ospedale Riuniti, Bergamo,Italy; R. Pristera, Ospedale Generale Regionale,Bolzano, Italy; F. Mazzotta, A. Gabbuti, Ospedale S.Maria Annunziata, Florence, Italy; R. Esposito, A.Bedini, Universita di Modena, Modena, Italy; A.Chirianni, E. Montesarchio, Presidio Ospedaliero A.D.Cotugno, Naples, Italy; V. Vullo, P. Santopadre,Universita di Roma ‘La Sapienza’, Rome, Italy; P.Narciso, A. Antinori, P. Franci, M. Zaccarelli,Ospedale Spallanzani, Rome, Italy; A. Lazzarin, R.Finazzi, Ospedale San Raffaele, Milan, Italy; A,D’Arminio Monforte, Ospedale L. Sacco, Milan,Italy.
Latvia: (L. Viksna) Infectology Centre of Latvia, Riga,Latvia.
Lithuania: (S. Chaplinskas) Lithuanian AIDS Centre,Vilnius, Lithuania.
Luxembourg: (R. Hemmer), T. Staub, Centre Hos-pitalier, Luxembourg.
Netherlands: (P. Reiss) Academisch Medisch Cen-trum bij de Universiteit van Amsterdam, Amsterdam,the Netherlands.
Norway: (J. Bruun) A. Maeland, V. Ormaasen, UllevalHospital, Oslo, Norway.
Poland: (B. Knysz) J. Gasiorowski, Medical Uni-versity, Wroclaw, Poland; A. Horban, CentrumDiagnostyki i Terapii AIDS, Warsaw, Poland; D.Prokopowicz, A. Wiercinska-Drapalo, Medical Uni-versity, Bialystok, Poland; A. Boron-Kaczmarska, M.Pynka, Medical Univesity, Szczecin, Poland; M. Be-niowski, Osrodek Diagnostyki i Terapii AIDS, Chor-zow, Poland; H. Trocha, Medical University, Gdansk,Poland.
Portugal: (F. Antunes) Hospital Santa Maria, Lisbon,Portugal; K. Mansinho, Hospital de Egas Moniz,Lisbon, Portugal; R. Proenca, Hospital Curry Cabral,Lisbon, Portugal.
Romania: (D. Duiculescu) Spitalul de Boli Infectioasesi Tropicale Dr Victor Babes, Bucarest, Romania; A.Streinu-Cercel, Institute of Infectious Diseases, Buc-arest, Romania.
Spain: (J. Gonzalez-Lahoz) B. Diaz, T. Garcıa-Benayas, L. Martin-Carbonero, V. Soriano, HospitalCarlos III, Madrid, Spain; B. Clotet, A. Jou, J.Conejero, C. Tural, Hospital Germans Trias i Pujol,Badalona, Spain; J.M. Gatell, J.M. Miro, HospitalClinic i Provincial, Barcelona, Spain.
Sweden: (A. Blaxhult) Karolinska Hospital, Stock-holm, Sweden; A. Karlsson, Sodersjukhuset, Stock-holm, Sweden; P. Pehrson, Huddinge Sjukhus,Stockholm, Sweden.
Switzerland: (B. Ledergerber) R. Weber, UniversityHospital, Zurich, Switzerland; P. Francioli, A. Telenti,Centre Hospitalier Universitaire Vaudois, Lausanne,Switzerland; B. Hirschel, V. Soravia-Dunand, HospitalCantonal Universitaire de Geneve, Geneve, Switzer-land; H. Furrer, Inselspital Bern, Bern, Switzerland.
Ukraine: (N. Chentsova) Kyiv Centre for AIDS, Kyiv,Ukraine.
United Kingdom: (S. Barton) St Stephen’s Clinic,Chelsea and Westminster Hospital, London, UK; A.M.Johnson, D. Mercey, Royal Free and UniversityCollege London Medical School, London (UniversityCollege Campus), UK; A. Phillips, C. Loveday, M.A.Johnson, A. Mocroft, Royal Free and UniversityCollege Medical School, London (Royal Free Cam-pus), UK; A. Pinching, J. Parkin, Medical College ofSaint Bartholomew’s Hospital, London, UK; J. Weber,G. Scullard, Imperial College School of Medicine at StMary’s, London, UK; M. Fisher, Royal Sussex CountyHospital, Brighton, UK; R. Brettle, Western GeneralHospital, Edinburgh, Scotland, UK.
Virology groupC. Loveday, B. Clotet (Central coordinators) plus adhoc virologists from participating sites in the EuroSIDAStudy.
Steering committeeF. Antunes, A. Blaxhult, N. Clumeck, J. Gatell, A.Horban, A. Johnson, C. Katlama, B. Ledergerber(Chair), C. Loveday, A. Phillips, P. Reiss, S. Vella.
Coordinating centre staffJ. Lundgren (Project leader), I. Gjørup, O. Kirk, N.Friis-Moeller, A. Mocroft, A. Cozzi-Lepri, D. Moller-up, M. Nielsen, A. Hansen, D. Kristensen, L. Kolte, L.Hansen, J. Kjær.
