Virological rebound after suppression on highly active antiretroviral therapy
AIDS 2003; 17(12):1741-1751
Amanda Mocroft a; Lidia Ruiz b; Peter Reiss c; Bruno Ledergerber d; Christine Katlama e; Adriano Lazzarin f; Frank-Detlef Goebel g; Andrew N. Phillips a; Bonaventura Clotet b; Jens D. Lundgren h; for the EuroSIDA study group. Royal Free Centre for HIV Medicine and Department of Primary Care and Population Sciences, Royal Free and University College Medical School, London, UK; bFundacio IrsiCaixa and HIV Unit, Hospital Universitari (UAB) 'Germans Trias i 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.
Objective: To determine the rate of virological rebound and factors associated with rebound among patients on highly active antiretroviral therapy (HAART) with previously 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 the first of two consecutive viral loads above 400 copies/ml. Incidence rates were calculated using person-years of follow-up (PYFU), Cox proportional hazards models were used to determine factors related to rebound.
Results: Of 2444 patients, 1031 experienced virological rebound (42.2%). The incidence of rebound decreased over time; from 33.5 in the first 6 months after initial suppression to 8.6 per 100 PYFU at 2 years after initial suppression (P <
0.0001). The rate of rebound was lower for treatment-naive compared with treatment-experienced patients. In multivariate models, patients who changed treatment were more likely to rebound, as were patients with higher viral
loads on starting HAART. Treatment-naive patients were less likely to rebound. Among pretreated patients, those who were started on new nucleosides were less likely to rebound.
After adjustment, patients on four or five-drug regimens were significantly more likely to experience virological rebound than those on three drugs [relative hazard (RH) 1.29; 95% CI 1.08-1.54, P = 0.0049 and 1.98; 95% CI 1.48-2.64, P <0.0001, respectively], as were patients who made any change to their HAART regimen (RH 1.59, 95% CI 1.39-1.82, P < 0.0001).
Patients starting a single-NNRTI regimen were more likely to have a rebound in viral load (RH 1.36; 95% CI 1.07-1.73, P = 0.011), when compared with patients starting HAART with a single PI-based regimen. Patients taking NNRTI were more likely to experience virological rebound. The majority of patients were taking nevirapine as their first NNRTI-containing HAART regimen. Several observational studies, including EuroSIDA, have reported a superior virological response to HAART in efavirenz-containing regimens compared with nevirapine,
both in naive and treatment-experienced patients.
Compared with treatment-experienced patients, those who were treatment naive were significantly 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 was known to be below 50 copies/ml at the date of initial suppression were significantly less likely to experience virological rebound than patients whose viral load was undetectable at 400 copies/ml (RH 0.54; 95% CI 0.44-0.67, P < 0.0001).
Older patients, those who started HAART more recently, and those with higher current CD4 lymphocyte counts were significantly less likely to experience virological rebound.
The risk of virological rebound was 19% lower for each new NRTI added to the HAART regimen (RH 0.81, 95% CI 0.74-0.88, P < 0.0001).
After adjustment for the factors related to virological rebound, there remained a 27% decreased risk of virological rebound with each additional 6 months since initial suppression (RH 0.73, 95% CI 0.64-0.84, P < 0.0001).
Conclusion: The rate of virological rebound decreased over time, suggesting that the greatest 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 with a good immunological response.
EuroSIDA is one of the largest European observational studies of patients with HIV-1 infection. The study has found a substantial number of patients who had a rebound in viral load after initial virological suppression on HAART regimens,
but the rate of virological rebound decreased over time, suggesting that if the viral load does not rebound in the initial months after virological suppression, there is less risk of virological rebound over time as the time from suppression
increases. Whereas treatment-naive patients had the lowest risk of virological rebound, treatment-experienced patients who could add new antiretroviral drugs to their HAART regimen were also significantly less likely to experience
virological rebound than those who did not add new antiretroviral drugs to their HAART regimen.
Adherence may play a critical role in virological rebound, as the results from both clinical trials and observational studies have shown that adherence plays a role in virological failure. However, we do not have that information
available in the EuroSIDA study.
Patients who swapped any drug in their HAART regimen were also more likely to experience virological rebound. This may raise questions about compliance or side-effects, as physicians may change drugs in order to increase compliance or reduce side-effects, but problems of drug resistance may already have started, or tolerance to side-effects may already have decreased substantially.
The EuroSIDA study is a prospective, European study of patients with HIV-1 infection in 70 centres across Europe (including Israel) and now including Argentina. Details of the study have been published elsewhere. In brief, centres provided data on consecutive patients seen in the outpatient clinic from 2 May 1994 until a predefined number of patients was enrolled from each centre. This cohort of 3116 patients was defined as the EuroSIDA I cohort. The enrolment of a second cohort of 1365 patients began in December 1995. In April 1997, a further 2839 patients were recruited and were defined as the EuroSIDA III cohort. Cohort IV, consisting of 1225 patients, was enrolled from April 1999, and a fifth cohort, cohort V, consisting of 1256 patients, was recruited from September 2001. For cohorts I-III, eligible patients were those with a CD4 lymphocyte count of below 500 cells/mm3 in the previous 4 months, a booked clinic appointment and who were older than 16 years at the time of enrolment. The CD4 lymphocyte count restriction was removed for cohorts IV and V. Information was provided on a standardized data collection form at baseline and every 6 months thereafter. Follow-up was to Spring 2002, with information from up to 16 forms available for cohort I, 13 for cohort II, 10 for cohort III, five for cohort IV and one for cohort V. At each follow-up visit, details on all CD4 lymphocyte counts measured since the last follow-up and viral load measurements were collected. For each patient, the date of starting and stopping each antiretroviral drug was recorded, as was the use of drugs for prophylaxis against opportunistic infections. The dates of diagnosis of all AIDS-defining illnesses have also been recorded, including those diagnoses made subsequent to the initial diagnosis, using the 1993 clinical definition of AIDS from the Centers for Disease Control. Members of the coordinating office visited all centres to ensure correct patient selection and that accurate data were provided by checking the information provided against case-notes for a proportion of patients.
Of 2444 patients, 623 (25.5%) were treatment naive before starting HAART. There were some differences between groups according to previous
treatment. Over 70% of patients from eastern Europe were treatment naive (79 patients, 73.1%) compared with 28.3% of patients from northern Europe, 20.5% from central Europe and 20.5% from southern Europe (P < 0.001, chi-squared test). The more intensive treatment regimens tended to have been in patients who were treatment experienced before HAART, with 100% of those on five antiretroviral drugs being treatment experienced before starting HAART (P
< 0.001, chi-squared test). Treatment-naive patients tended to start HAART later in time, experience a higher percentage increase in CD4 lymphocyte count by the time of viral suppression, and have 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 [interquartile range (IQR) 33.3-45.2), the median CD4 cell count at starting HAART was 244 cells/mm3 (IQR 125-358),
and the median CD4 cell count at virological 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 diagnosed with AIDS (22.9%), and the most
common nucleoside combination used was zidovudine and lamivudine (853 patients, 34.9%).
The most commonly used PI was indinavir (928 patients, 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%) and efavirenz in 166 patients
(6.8%), whereas abacavir was used by 125 patients (5.1%). The first treatment with antiretroviral drugs occurred a median of 37 months before starting HAART (IQR 18-61). The majority of patients had been treated with both
monotherapy and dual combination therapy before HAART (1069 patients, 58.7%), and almost all patients had previously received zidovudine (1704 patients, 93.6%). A total of 701 patients (38.5%) did not start any new NRTI
(i.e. NRTI they had never previously been treated with) at starting HAART, and 479 patients (26.3%) were able to start HAART with two or more 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. Patients with previous antiretroviral treatment experienced viral load rebound at a significantly faster rate than those patients who were treatment naive at starting HAART (P < 0.0001, log-rank test).
Among all patients, regardless of previous treatment experience, there was a clear, statistically significant trend for a decreasing rate of virological rebound as the time since initial suppression increased [rate ratio (RR) 0.63; 95% CI
0.60-0.67, P < 0.0001]. Overall, the rate of virological rebound in the first 6 months after initial suppression was 33.5 per 100 PYFU (95% CI 30.1-36.9), almost four times higher than the rate of virological rebound at or after 24 months after initial suppression (incidence of 8.6 per 100 PYFU, 95% CI 7.4-9.8). The incidence of rebound was significantly higher among treatment-experienced patients, but both groups showed a similar decline in incidence
with increasing time from initial suppression. Overall, the incidence of virological rebound among treatment-naive patients was less than half that of treatment-experienced patients (RR 0.47; 95% CI 0.40-0.46, P < 0.0001). In all time periods the rate of virological rebound among treatment-naive patients remained approximately half that of treatment-experienced patients.
There are several caveats to note. We required that patients respond to their initial regimen within 12 months. It could be argued that patients should respond to HAART within the first 6 months of therapy, although previous work
has shown that patients continue to achieve undetectable levels of viraemia up to 12 months after starting HAART and without changes in HAART, especially for those with high viral loads. In addition, treatment guidelines are generally
based on clinical trials, in which the frequency of viral load measurements may be more frequent than in clinical practice. In sensitivity analyses, when we required that the initial HAART success occurred within the first 6 months, our
conclusions remained unaltered. In addition, we used a global definition of virological rebound using 400 copies/ml, even though in some cases, we had viral loads measured using a sensitivity of 200 or 50 copies/ml. However, we
repeated our analyses using different definitions of virological rebound, including an analysis that selected patients whose viral load was measured using a sensitivity of 50 copies/ml, and by defining virological failure on the basis of a single viral load above the limit of detection, and all showed remarkably similar results.