| |
HAART Interruption: Predictors and Virological and Immunologic Consequences (20% Interrupt HAART in this study)
|
| |
| |
JAIDS Journal of Acquired Immune Deficiency Syndromes. 42(5):554-561, August 15, 2006.
Touloumi, Giota PhD*; Pantazis, Nikos Mr*; Antoniou, Anna*; Stirnadel, Heide A.; Walker, Sarah A. PhD; Porter, Kholoud PhD; on behalf of the CASCADE Collaboration
From the *Athens University Medical School, Athens, Greece; GlaxoSmithKline, Brentford, UK; and MRC Clinical Trials Unit, London, UK.
Abstract
Objective: To characterize the magnitude and the predictors of highly active antiretroviral therapy (HAART) interruption (TI) and to investigate its immunologic and virological consequences.
Methods: Using Concerted Action on Seroconversion to AIDS and Death in Europe data from 8300 persons with well-documented seroconversion dates, we identified subjects with stable first HAART (for at least 90 days) not initiated during primary infection. A TI was defined as an interruption of all antiretroviral therapy drugs for at least 14 days.
Results:
Of 1551 subjects starting HAART, 299 (19.3%) interrupted treatment.
Median (interquartile range) duration of the TI was 189 (101-382) days.
The cumulative probability (95% confidence interval) of TI at 2 years was 15.9% (14.0%-18.1%).
Women were more likely to have a TI than men in the same exposure group (35.8% vs 24.2% among drug users, 22.1% vs 13.3% among heterosexuals; P < 0.05).
Higher baseline viremia and poor immunologic response to HAART were associated with higher probabilities of TI.
Median (interquartile range) individual CD4 cell loss during TI was 94 (1-220) cells/μL.
Older age at HAART (>40 yr), lower pre-HAART nadir (<200 cells/μL), and lower CD4 at start of TI (<350 cells/μL) were significantly associated with greater relative CD4 loss during TI.
Conclusions: We estimate that almost 1 in 6 subjects on HAART interrupts treatment by 2 years. Further research is needed to investigate the reasons why TI is higher in women. We have identified characteristics of subjects with the greatest risk for CD4 loss in whom TI may have greater risks.
Background
Highly active antiretroviral therapy (HAART) has dramatically decreased HIV-1-associated mortality and morbidity in the Western world.1,2 The initial enthusiasm though has been dampened because it was realized that prolonged suppression of viral load does not eradicate HIV-1 infection3; in clinical cohorts, approximately 25% of patients initiating HAART either do not achieve viral suppression or lose it within 2 to 3 years.4-6 Besides, long-term administration of HAART is associated with poor adherence,7 an increased risk for developing drug resistance,8 and undesirable side effects.9 Because of these complications associated with the long-term use of HAART, an increasing number of subjects interrupt treatment.10,11
The initial recommendations suggesting treating HIV-1-infected subjects aggressively and at early stages12 have been updated toward a more conservative approach of delaying HAART initiation.13 As a result, a substantial number of subjects who initiated treatment according to the early recommendations, including subjects with CD4 cell count above 500 cells/μL, would not have started therapy under current recommendations. It is unclear whether these subjects should stop therapy until it is indicated by current guidelines.
In an effort to increase long-term adherence and maintain treatment success, several therapeutic strategies have been proposed, including structured treatment interruptions, in which subjects are on and off treatment for fixed, usually short, periods14,15 and CD4-guided treatment interruption (TI).16,17 Initial results from such studies on safety and benefits remain mixed. Nevertheless, unsupervised TI for the above reasons has become common in clinical practice.18-20 It is unclear though what effect TI has on the immunologic and virological profile during it and after reinitiation of antiretroviral therapy (ART). Results from several studies suggest that prolonged TI in patients with suppressed virus and marked immune reconstitution is generally safe.10,16,17,21 Results are less clear in other groups of subjects.22,23 Identification of factors affecting immunologic and virological changes during TI is of great clinical importance because it helps to classify subjects at high risk.
In this study, we take advantage of the Concerted Action on Seroconversion to AIDS and Death in Europe (CASCADE)24 large cohort collaboration to describe the incidence of TI in clinical practice, to investigate the predictors of a TI, and to assess immunologic and virological changes during TI as well as the determinants of these changes. The well-estimated seroconversion dates in CASCADE enabled us to investigate the effect of the pre-ART history on TI consequences, whereas the wide range of ages, mode of infections, and the representation of both sexes allowed us to study the effect of these demographic characteristics on the probability of TI and its consequences. Furthermore, for the analysis of the data, we applied methods that allow for competing risks and informative censoring due to treatment resumption, which are known to lead to biased results if ignored.
Patients
The study population was derived from the CASCADE study described in detail elsewhere.24 Briefly, data were pooled in August 2003 from 22 cohorts comprising 8300 HIV seroconverters. The date of seroconversion was estimated by the study investigators for each participating study, with most being the midpoint between the last HIV-negative and first HIV-positive test dates. Subjects were included in the current analyses if their date of seroconversion was estimated using the midpoint method or on the basis of laboratory evidence of seroconversion, their HIV test interval was less than 3 years, they had initiated a stable (for at least 90 days) first HAART regimen at least 1 year after seroconversion, and if they had HIV-RNA and CD4 cell count measurements at HAART initiation and at least one additional measurement of each marker during HAART.
First HAART was considered to be any regimen containing 3 or more ART drugs from at least 2 classes, or containing abacavir or tenofovir. Individuals who had previously been on suboptimal therapy also had to have added or changed at least 2 drugs simultaneously. Treatment interruption was defined as a discontinuation of all drugs for at least 14 days. In case of phased drug interruption, the date at which the last drug was terminated was considered as the date of TI initiation. Patients who interrupted HAART but resumed therapy within 14 days were not defined as having TI. Subjects initiating HAART during the first year after seroconversion were not included in this study because they consist of a selective group treated close to primary infection and will be subject to a separate analysis. Switch to a new HAART regimen was defined as a change of at least 2 drugs simultaneously.
DISCUSSION
In this study, we have found that by 2 years after HAART initiation, 15.9% of the study population had stopped all ART drugs, whereas another 16.3% initiated a new HAART regimen. Similar rates were reported by Monforte et al.18 The relative higher rates of TI found in other studies19,20,29 can be explained by different design such as the inclusion only of women19 or ignoring competing risks because of switching HAART regimen, which is known to lead to overestimated rates for TI as the first event after initiating HAART.26
We found that women had higher probabilities of TI than men among both IDUs and MSW, with the gender difference being similar within both risk groups. A similar finding was reported by Monforte et al.18,30 The reason for this gender difference is unclear. Providing that medical care is adequate, women do respond equally well or even better than men to ART.31,32 However, recent studies reported lower ART adherence rates in women compared with men,33 with this difference being explained mainly by social and behavioral factors. Alternatively, sex differences in drug-related adverse events (reported to be higher in women34) can explain this observed difference in rates of TI. Unfortunately, reasons for TI or adherence rates were not available in our study. The higher rates of TI for IDUs (evident in both men and women) can be explained by poor adherence and/or higher rates of treatment failure.35 Our finding of higher viremia at HAART initiation and poor immunologic response to HAART being associated with increased probabilities of TI are in line with results from other studies.19,20,29 Interestingly, history before HAART initiation (ie, time elapsed since seroconversion or rate of CD4 decline) does not seem to be associated with TI.
Age at HAART initiation, pre-HAART nadir, and CD4 cell count at TI initiation were associated with the relative rate of CD4 loss during TI in this study. Subjects aged above 40 years at HAART initiation, with nadir CD4 cell counts below 200 cells/μL and with CD4 cell count at TI initiation below 350 cells/μL had the steepest CD4 declines. Age has also been identified as a significant determinant in other studies.16,36 It has been reported that CD4 cell count decline is more pronounced among subjects who interrupt HAART because of toxicity or side effects,19 but this finding could not be tested in our study.
Nadir CD4 cell count has been recognized as one of the most important predictors of both rate of CD4 loss and time on TI in several studies.17,21,36,37 An inverse correlation between CD4 gain during therapy and loss during TI has been reported by many researchers.17,22,34,38 This finding has been interpreted as suggesting that CD4 cells gained during HAART are lost much more rapidly than the pretreatment CD4 cells. This is in accordance with pre-HAART nadir CD4 cell count being a significant predictor of CD4 loss during TI. We found that subjects with CD4 cell counts higher than 350 cells/μL at TI initiation had lower relative rates of CD4 decline. Although this finding seems to contradict previous findings,17,22,36,38 it should be emphasized that we report proportional rates of drop, whereas the other studies reported absolute rates of drop. Similar proportional rates of drop are translated to larger absolute drops in subjects with higher initial values.
Although many of our results have been reported in previous studies, the strengths of this analysis are its rigorous adjustment for potential confounding factors, notably, a switch in HAART regimen as a potential "competing risk" for patients currently on their first HAART regimen and patients resuming treatment during a TI. Furthermore, by virtue of its size, we were also able to have reasonable power to look jointly at CD4 at TI, HAART initiation, and nadir, and assess the combined effects of these different parameters, whereas the known infection dates enabled us to asses the pre-HAART history effect.
Treatment interruption was not associated with significantly increased risk for clinical progression in our study. Results from other studies are mixed.10,30,38 Treatment discontinuation though may have other potential negative consequences such as development of drug-resistant mutations. However, it has been found that this problem seems to be rare and mainly in subjects who were on suboptimal therapy before the initiation of HAART39,40 and those on regimens that include drugs with long half-lives such as nevirapine and efavirenz (nonnucleoside reverse transcriptase inhibitors).41
In conclusion, in this large observational study, a substantial and increasing number of subjects interrupt HAART. Discontinuing treatment seems to be safe for subjects with a well-retained immune system (note from Jules Levin: this conclusion I think has been discounted by the SMART study results and other studies reported in early 2006 at CROI. Many concluded that these study results find too many risks associated with interruptions. Risks discovered were drug resistance and complications including heart disease, other than just cd4 count decline). However, subjects above 40 years old, with a pre-HAART CD4 below 200 cells/μL or with limited immune reconstitution during HAART (CD4 at TI below 350 cells/μL), had the greatest proportionate decrease in CD4 cell counts during TI. For such subjects, if TI is considered, caution and close monitoring are essential to ensure that risks are minimal.
RESULTS
Of the 8300 subjects in the CASCADE study, 1551 met the inclusion criteria. Most of them were men (83%), with median [interquartile range (IQR)] age at seroconversion and HAART initiation of 28.5 (24.2-34.7) and 35.9 (31.3-41.8) years, respectively. The predominant mode of infection was sex between men in 56% of the subjects, injecting drug use (IDU) in 22.5%, and sex between men and women (MSW) in 15.7% (Table 1). The median (IQR) duration of first HAART was 19.9 (10.7 to 36.0) months. About half (44.9%) of the patients had sustained virological response for a median time (IQR) of 17.6 (8.0 to 33.9) months.
Incidence of TI
Of the 1551 subjects, 299 (19.3%) had a TI, 283 (18.2%) switched to a new HAART regimen, and 969 (62.5%) neither interrupted nor switched to a new HAART regimen throughout their follow-up. The cumulative probabilities [95% confidence interval (CI)] of a TI and of switching to a new HAART regimen by 2 years after HAART initiation were 15.9% (14.0%-18.1%) and 16.3% (14.3%-18.5%), respectively, with the corresponding figures by 5 years after HAART initiation being 29.7% (26.4%-33.4%) and 29.1% (25.7%-33.0%), respectively.
Predictors of TI and of Initiating a New HAART Regimen
The characteristics of the study population by subsequent ART status are shown in Table 1. The median (IQR) individual CD4 increase was 76 (9-160) cells/μL at 3 months since first HAART initiation and 135 (27-288) cells/μL at the end of the first HAART regimen or last follow-up. The corresponding figures were 86 (6-174) and 104 (8-270) cells/μL for subjects having a TI, 68 (10 to 163) and 86 (-1 to 216) cells/μL for those switching to a new HAART regimen, and 77 (8-159) and 177 (51-326) for those who retained their initial HAART regimen, respectively.
Results from multivariate analysis (Table 2) showed that women were more likely to have a TI, and IDUs were more likely to have a TI and less likely to switch to a new HAART regimen. Further analysis revealed that women were more likely to have a TI than men among both IDUs and MSW. Comparing women to men, the adjusted hazard ratio of having a TI was 1.63 (P = 0.020) among IDUs and 1.78 (P = 0.089) among MSW. However, the gender-risk group interaction was not statistically significant (P = 0.411), indicating that the gender effect on TI was similar in both groups. As expected, most subjects without a virological response during first HAART switched to a new HAART regimen. Higher viremia at HAART initiation was associated with higher probabilities of TI, but no differences were found in their risk for switching to a new HAART regimen. Treatment interruption was more likely in persons with higher CD4 nadir pre-HAART and lower last CD4 cell counts during HAART (Table 2). Equivalently, those with larger CD4 increases during HAART had lower probabilities of TI. Time from seroconversion until HAART initiation or rate of CD4 decrease before HAART initiation was not significantly associated with the probability of TI after adjusting for nadir and baseline CD4 cell counts (data not shown).
Immunological Consequences During TI
Of the 299 subjects who interrupted all antiretroviral drugs, 236 (78.9%) had CD4 cell count and HIV-RNA measurements within 6 months before the TI and at least one additional measurement for both markers during the TI. The median (IQR) duration of the TI was 199 (103-416) days. The median (IQR) CD4 cell count and HIV-RNA at the start of TI were 441 (278-632) cells/μL and less than 500 (<50-21,063) copies/mL, respectively. The individual median (IQR) drop in CD4 cell count during TI was 94 (1-220) cells/μL, and the median (IQR) increase in HIV-RNA was 1.34 (0.22-2.74) log10 copies/mL.
Changes in CD4 cell count and HIV-RNA during TI are displayed in Figure 1. A slow decrease (more pronounced during the first 3 months after TI) in CD4 cell count throughout TI is seen. On the contrary, a steep increase in HIV-RNA levels during the first 3 months of TI is also observed, but the levels seem to stabilize thereafter.
CD4 rate of change and time to treatment resumption were highly positively correlated (r = 0.64), indicating that subjects with steeper rates of CD4 loss did in fact tend to reinitiate ART sooner. Therefore, results were based on random effect models, allowing also for informative censoring. The mean (95% CI) annual CD4 decline during TI was estimated as 0.85 (0.73-0.97) on the log scale, corresponding to a mean (95% CI) relative loss in CD4 cells per year of 57.3% (51.8%-62.1%).
Factors associated with the relative rate of CD4 loss during TI are presented in Table 3. There was a trend for steeper relative CD4 drops with increasing age at HAART initiation. Although higher viral loads at the time of initiating a TI were significantly associated with steeper relative declines in CD4 during TI in univariate analysis, this association did not persist after allowing for nadir and CD4 cell counts at TI initiation. Subjects with pre-HAART nadir CD4 counts below 200 cells/μL and with CD4 below 350 cells/μL at the time of starting a TI had steeper relative CD4 declines. The relative rate of CD4 decrease in each group is presented in Figure 2. After accounting for pre-HAART nadir, CD4 at start of TI, and age at HAART initiation, neither time from seroconversion until HAART initiation nor rate of CD4 decline before HAART initiation was significantly associated with relative CD4 decline rates during TI.
Among the 181 subjects whose CD4 cell counts were higher when HAART was discontinued than when it was initiated, 73 experienced a drop in CD4 cells to pre-HAART levels. The cumulative incidence (95% CI) of CD4 drop to pre-HAART level was 37.0% (30.1%-45.6%) at 12 months after a TI.
Virological Consequences During TI
The mean (95% CI) rate of HIV-RNA change during the first 3 months of TI was 0.49 (0.42-0.56) log10 copies/(mL mo), resulting in a mean (95% CI) increase of 1.47 (1.26, 1.68) log10 copies/mL in the first 3 months. The subsequent mean (95% CI) rate of HIV-RNA levels change after the first 3 months was -0.0027 (-0.0115, 0.0062) log10 copies/(mL mo), confirming descriptive results for relatively stable levels after the initial fast increase during the first 3 months of TI.
From the 125 subjects with undetectable viral load at start of TI, 102 (81.6%) experienced a virological rebound during TI, whereas 19 (15.2%) reinitiated treatment without a virological rebound. Over half (57.6%) of the 236 subjects had an increase of at least 1 log10 in their HIV-RNA levels during the TI, whereas 80 (33.9%) reinitiated ART before the occurrence of such an increase in their viral load levels.
Higher viremia at HAART initiation is associated with increased hazard of having a log10 increase in viral load during TI. However, subjects with relatively high HIV-RNA levels at TI initiation (thus with relative small reductions during HAART) had reduced probabilities to further increase their HIV-RNA levels compared with the rest of the subjects (data not shown). Of the 190 subjects with HIV-RNA levels at TI initiation below those at HAART initiation, 98 (51.6%) increased their viral load to pre-HAART levels in a median time of 14.9 months.
Clinical Progression
During first stable HAART, patients had been followed-up for 3115 person/y. During this time, 49 patients developed clinical AIDS, and 9 died before developing AIDS, whereas 15 of the 230 with AIDS at HAART initiation experienced a new AIDS event. The 299 persons who interrupted first HAART had been followed-up for 472.8 days during HAART and 242.8 person/y during TI. Eight patients developed clinical AIDS during HAART and 7 during TI, whereas 1 of the 48 patients with AIDS at HAART initiation experienced a new AIDS event during HAART and 2 during TI. The overall incidence rate (95% CI) of any AIDS event was 0.019 (0.010-0.037) during HAART and 0.037 (0.019-0.071) cases per person per year during TI, the difference being nonsignificant (P = 0.158).
Statistical Analysis
Initial virological response was defined as a decrease in plasma HIV-1 RNA to below 500 copies/mL during first HAART and sustained virological response as retention of the initial virological response throughout subsequent follow-up. Subjects whose first value less than 500 copies/mL was their last measurement during HAART were classified as initial responders. Single HIV-1 RNA values between 500 and 1000 copies/mL followed by counts less than 500 copies/mL were ignored. Clinical AIDS cases were defined in accordance with category C of the 1993 revised classification system of the Centers for Disease Control and Prevention.25
Treatment interruption and switch to a new HAART regimen act as competing risks for the first major change in ART for a patient starting HAART naive, and ignoring this fact may lead to overestimated event incidence and to biased effect estimates.26,27 Therefore, cumulative incidence of TI or switching to a new HAART regimen as well as predictive value of factors affecting the probability of a TI or of switching to a new HAART were estimated using survival curves and Cox-proportional hazards models, allowing also for competing risks.26,27 Specific STATA programs were developed to fit these models. In an analysis of longitudinal changes in CD4 and viral load during TI, patients' follow-up was censored at the time of ART reinitiation or last evaluation if still on TI. It is expected that subjects with steeper CD4 declines during TI reinitiate ART sooner and therefore have fewer CD4 measurements during TI. This phenomenon is known as informative censoring. Given that ignoring potential informative censoring can lead to biased estimates, rates of CD4 decline during TI were estimated through random effect models adjusted for selective drop-outs due to treatment resumption.28 MLn macros (available on request) were developed for these models. Time to CD4 cell count decline to pre-HAART levels or to HIV-RNA increase by 1 log10 copies/mL during TI was analyzed using survival analysis techniques, allowing also for competing risk due to treatment resumption.
|
|
| |
| |
|
|
|
