| |
Prolonged Treatment Interruption after Immunologic Response to Highly Active Antiretroviral Therapy
|
| |
| |
Clinical Infectious Diseases 2003;37:1541-1548
Patrick M. Tarwater,1 Michelle Parish,2 and Joel E. Gallant2
1University of Texas Health Science Center, School of Public Health, Houston, Texas; and 2Johns Hopkins University School of Medicine, Baltimore, Maryland
ABSTRACT. Duration of treatment interruption (TI) was investigated in 105 human immunodeficiency virus infected patients whose antiretroviral therapy was interrupted with the intention to resume therapy on the basis of clinical or laboratory indicators. In a mixed cohort study, 57% of patients had not resumed therapy at the time of writing (median TI duration, 114 weeks); the most recent analysis of this group revealed a mean CD4 cell count of 500 cells/mm3. Patients with lower CD4 cell counts at therapy initiation were more likely to resume therapy than were those with counts of >500 cells/mm3 (<200 cells/mm3 [relative hazard, 4.4]; and 200350 cells/mm3 [relative hazard, 2.9]). Patients who met current United States Department of Health and Human Services criteria for starting therapy at the time of therapy initiation were 3 times more likely to resume therapy than were those who did not. Our results have implications for this TI strategy: there may be a subset of patients who can safely discontinue therapy for prolonged periods of time.
BACKGROUND
During the past few years, researchers, clinicians, and patients have begun to question the assumption that antiretroviral therapy, once initiated, must be taken for the duration of the patient's life. Treatment interruption (TI) and intermittent therapy, once euphemisms for nonadherence, have been legitimized on the basis of data from observational studies and randomized clinical trials.
One of the earliest strategies to be studied involved periodic interruption of therapy to improve HIV-specific cellular immunity through repeated exposure of the immune system to low-level viremia. Although structured TIs may still have a role for patients treated during primary infection, there is little enthusiasm for this approach in chronically infected patients, because of discouraging data from clinical trials.
Another approach involves TI cycles that are based on predetermined time intervals. The rationale is to decrease the cumulative time that treatment is received, with the goal of reducing cost and, possibly, toxicity. Although patients in one small pilot trial using a 7-days-on/7-days-off schedule maintained virus suppression and did not develop drug resistance, other trials using the same schedule and, in some instances, longer interruption cycles found that this approach led to the accumulation of drug resistance mutations.
Another rationale for using TI is to improve response to salvage therapy in patients with substantial treatment experience and extensive drug resistance, possibly by allowing reemergence of less-resistant virus. Data on this strategy are conflicting, with some studies showing a short-term benefit associated with TI and others showing no benefit and potential harm. This strategy remains controversial because of the potential dangers associated with TI in patients with advanced HIV disease and the inconsistent data supporting its use.
What is perhaps the most promising approach to intermittent therapy is sometimes referred to as "pulse therapy," in which patients who have successfully responded to HAART discontinue treatment and restart only when they have an indication for treatment, which, in most cases, is based on CD4 cell count and/or HIV RNA load. This strategy has shown promise in several small observational studies and clinical trials. Potential candidates include those who started treatment that was based on earlier, more aggressive guidelines, as well as those who had strong indications for therapy at the time of initiation but subsequently experienced substantial immune reconstitution due to HAART. Clinicians and patients are faced with the question of whether HAART must be continued indefinitely once initiated, regardless of the strength of the original indications or the immunologic response to therapy. In an effort to answer this question, we conducted a mixed (i.e., historical and concurrent) observational cohort study involving patients who discontinued therapy with the intention of restarting on the basis of clinical and/or laboratory criteria.
Study population and data collection
Patients were included in this cohort if they received HIV care at the Johns Hopkins AIDS Service (Baltimore) and interrupted therapy, either at the suggestion of their clinician or with the clinician's approval, with the intention to resume therapy on the basis of clinical criteria (progression of disease) or laboratory criteria (decrease in CD4 cell count and/or increase in HIV RNA load). Patients were not eligible for inclusion if the intention of the clinician was to resume therapy after resolution of drug toxicity, collection of resistance data, or some other end point not associated with the patient's clinical status after stopping therapy. When the medical record was not clear regarding the rationale for TI or the plans for treatment reinitiation (TR), the clinician was asked directly. We did not define a minimum duration of TI because this would have excluded patients who developed immediate symptoms (e.g., symptomatic viremia) following TI.
Study follow-up began at the initiation of TI, and chart review was then performed prospectively. This study was strictly observational: all decisions regarding interruption and resumption of therapy were made by the patient's clinician. Reasons for interruption and resumption of therapy were obtained from patient medical records, as were data on metabolic and morphological complications associated with therapy and the patients' responses to TI.
RESULTS
A total of 105 patients were included in the cohort. The mean age of the patients was 42 years, 59% were white, 36% were African American, 75% were male, 57% were infected with HIV through homosexual or bisexual contact, 15% were injection drug users, and 21% had received a diagnosis of AIDS before receiving HAART.
At HAART initiation, the mean CD4 cell count was 442 cells/mm3 (median, 412 cells/mm3), and the mean HIV RNA load was 30,800 copies/mL (median, 36,000 copies/mL).
At TI, the mean CD4 cell count was 748 cells/mm3 (median, 731 cells/mm3), and the mean HIV RNA load was 300 copies/mL (median, 150 copies/mL).
Before TI, 72 (69%) and 35 (33%) had an HIV RNA load of <400 copies/mL and <50 copies/mL, respectively. At initiation of HAART, 79 (77%) met criteria for therapy that were based only on 1997 DHHS guidelines (a CD4 cell count of <500 cells/mm3 or an HIV RNA load of >20,000 copies/mL), whereas 44 (43%) also met criteria for therapy that were based on 2001 guidelines (a CD4 cell count of <350 cells/mm3 or an HIV RNA load of >55,000 copies/mL).
Among the 105 patients, reported reasons for discontinuation of therapy (not mutually exclusive) included perceived lack of an indication for therapy on the part of the clinician (44%), drug toxicity (15%), nonadherence (14%), performance of resistance testing (15%), and virological failure (8%).
The mean HIV RNA load after TI, defined as the mean of all measurements observed 6 weeks after TI but before TR, was 25,500 copies/mL (median, 32,000 copies/mL). The mean viral rebound after TI was correlated with the mean HIV RNA load before initiation of HAART (r = 0.3; P = .047). No AIDS-defining opportunistic infections or malignancies were observed during TI. Two patients developed asymptomatic thrombocytopenia, and 3, each of whom had started therapy before virus load testing was available, developed symptomatic viremia (an acute retroviral syndromelike illness characterized by fever, sweats, and other constitutional symptoms) in association with high-level viremia (peak HIV RNA load, >750,000 copies/mL). Two of the 3 patients resumed antiretroviral therapy, and their symptoms resolved. The third patient chose to not resume therapy, and symptoms resolved within 12 weeks.
Of the 105 patients in the cohort, 45 (43%) had resumed HAART at the time of writing (i.e., the "resumer" group). Among these 45 patients, the mean interval between TI and TR was 4 months (median, 2.7 months), and, before TR, the mean CD4 cell count was 305 cells/mm3 (median, 264 cells/mm3), and the mean HIV RNA load was 164,000 copies/mL (median, 103,000 copies/mL). In contrast, 60 patients (57%) had not resumed therapy at the time of writing (i.e., the "nonresumer" group), with a mean TI duration of 10 months (median, 9.6 months) and, at the most recent analysis, a mean CD4 cell count of 500 cells/mm3 (median, 472 cells/mm3) and a mean HIV RNA load of 46,000 copies/mL (median, 28,000 copies/mL). Among the 45 resumers, reported reasons for TR (not mutually exclusive) included increasing HIV RNA load (8 patients), decreasing CD4 cell count (11), or both (15); clinical factors (5); patient preference (2); symptomatic rebound viremia (3); and treatment of hepatitis B virus coinfection (1).
Eleven (24%) of the 45 resumers had received a diagnosis of AIDS before the initiation of HAART, compared with 11 (18%) of the 60 nonresumers. Twenty-seven (60%) of the resumers had met 2001 DHHS guidelines at baseline for the initiation of HAART, compared with 17 (28%) of the nonresumers.
Most of the patients who resumed HAART had achieved virus resuppression at the time of writing. Of the 15 who had an HIV RNA load of <50 copies/mL before resuming HAART, 13 had achieved resuppression to that level, 1 had a virus load of <400 copies/mL, and 1 had a virus load of >400 copies/mL but had been receiving HAART for only 10 weeks. Thirteen of the 14 patients who had an HIV RNA load of 50400 copies/mL at TI achieved resuppression to <400 copies/mL, and 8 had an HIV RNA load of <50 copies/mL. Of the 16 who had an HIV RNA load of >400 copies/mL at TI, 8 had an HIV RNA load of <400 copies/mL, with 5 having achieved a virus load of <50 copies/mL.
The hazard estimates for each group are relative to the group of patients with a CD4 cell count of 500 cells/mm3 and an incidence of 1.4 resumptions per 10 person-years. Patients with CD4 cell counts of 200349 cells/mm3 were almost 3 times more likely to resume therapy (P = .018), and those with CD4 cell counts of <200 cells/mm3 at HAART initiation were >4 times more likely to resume therapy (P = .002).
Similar estimates for time to resumption are grouped according to HIV RNA load strata at the initiation of HAART, including a category for the 50 patients (48%) with no pretherapy measurements of HIV RNA load. Hazard estimates for TR in each category were measured relative to those with an HIV RNA load of 4.04.49 log10 copies/mL and an incidence of 0.8 resumptions per 10 person-years. Patients with an HIV RNA load of >5.0 log10 copies/mL were almost 6 times more likely to resume treatment (P = .009).
To investigate the change in CD4 cell count during receipt of HAART as a prognostic factor for resumption, we used the upper quartile as the referent group, which had an incidence of 1.9 resumptions per 10 person-years. No group was significantly different from the referent group, although smaller increases in the CD4 cell count during receipt of HAART were associated with increased likelihood of therapy resumption. To investigate the CD4 cell count at TI as a prognostic factor for resumption, we used the group with CD4 cell counts of >1000 cells/mm3 at TI as the referent group. The incidence of resumption in this group was 1.2 per 10 person-years. These data conservatively suggest (P < .10) that patients with a CD4 cell count of <750 cells/mm3 at the time of TI were >3 times more likely to resume HAART than were those in the referent group. The HIV RNA load at TI was not predictive of resumption of HAART (data not shown).
Those who met the 2001 criteria at the time of initiation were 3 times more likely to resume therapy than were those who did not (P < .001).
As a second outcome measure, we estimated the yearly rate of decrease (slope) in the CD4 cell count following TI. We applied these estimated slopes to the group-wise median CD4 cell counts at the time of TI to estimate the time in years for achieving a CD4 cell count of 200 cells/mm3, assuming a stable slope and no resumption of therapy. Significant differences in rates of decrease in the CD4 cell count were found, on the basis of the CD4 cell count at TI and increase in CD4 cell count while receiving HAART. Specifically, those who had the greatest mean increase in CD4 cell count during receipt of HAART (i.e., 100 cells/mm3 per year) also had the most rapid decrease after TI (mean, 175 cells/mm3 per year), and those with no increase during receipt of HAART had a significantly slower rate of decrease after TI (mean, 35 cells/mm3 per year; P < .001). Similarly, those with the lowest CD4 cell counts at the time of TI (i.e., <500 cells/mm3) had significantly slower rates of decrease (43 cells/mm3 per year), compared with those with higher levels at TI (i.e., 1000 cells/mm3), whose mean rate of decrease was 190 cells/mm3 per year (P = .003).
We reviewed the medical records of each patient to determine the effect of TI on metabolic and morphological complications associated with HAART. Of the 32 patients noted to have some form of lipodystrophy (i.e., lipoatrophy, fat accumulation, or both) during receipt of HAART, 17 (53%) were noted to have some improvement, either by patient report, clinician assessment, or both. Four (57%) of 7 patients with hyperglycemia and 24 (69%) of 35 with hyperlipidemia were noted to have experienced improvement following TI.
AUTHOR DISCUSSION
In the enthusiasm that followed the introduction of protease inhibitors and ushered in the HAART era, treatment guidelines were aggressive, recommending therapy for all HIV-infected patients with CD4 cell counts of <500 cells/mm3 or modest increases in virus load. In 2001, these guidelines were modified, reflecting a growing reluctance on the part of clinicians and patients to prematurely commit patients to life-long therapy and the side effects of long-term toxicity and risk of resistance. As a result of these changes, many patients now receiving therapy would not have been treated, on the basis of the guidelines currently in use. Others, despite having had unequivocal indications for therapy at the time of initiation, have experienced profound immunologic responses to therapy. Many clinicians have chosen to discontinue the administration of HAART to such patients, but data supporting this approach have been lacking.
In our study, 57% of the patients in the cohort had not resumed therapy after a median TI duration of 114 weeks, with a mean CD4 cell count of 500 cells/mm3. The remaining 43% resumed therapy after a median TI duration of 31 weeks. No opportunistic infections or malignancies had occurred during TI at this writing, although there were 3 cases of symptomatic viral rebound. The best predictor of TI duration was the pretreatment CD4 cell count: patients with baseline CD4 cell counts of <200 and 200350 cells/mm3 were 4.4 and 3 times more likely, respectively, to resume therapy than were those who initiated therapy with counts of 500 cells/mm3. Patients who met the current DHHS guidelines criteria for starting therapy at the time of treatment initiation were 3 times more likely to resume treatment than were those who did not, a difference that was affected more strongly by the CD4 cell count component of the guidelines than the virus load component.
Most patients who had maintained virus suppression during receipt of HAARTand some who had notexperienced virus suppression following TR. TI was associated with reported improvements in some of the metabolic and morphological complications associated with HAART, although these findingsespecially those pertaining to morphological complications (e.g., lipodystrophy)must be interpreted with caution, given the observational nature of this study and the fact that data on complications were based on clinician report in the medical record.
The absence of pretreatment measurements of HIV RNA loads for 50% of the patients in the cohort was because HAART was available before HIV RNA loads could be measured. Those patients with missing HIV RNA values had a risk similar to that observed in those with HIV RNA loads of 4.55.0 log10 copies/mL, suggesting that patients who initiated therapy in the earliest years of the HAART erausually through clinical trialshad more advanced disease and higher virus loads.
Our results provide support for an approach that many find intuitive: patients who did not need therapy at the time of therapy initiation might not need it now. However, it should be noted that some patients have done well while not receiving therapy, despite having had strong indications for treatment, even by today's more conservative standards. Our results have implications for the pulse therapy strategy, in which the goal of therapy is to maintain the CD4 cell count above a predetermined threshold using cycles of therapy followed by prolonged TIs. There may be a subset of patients presumably, those with relatively low virus load set points and good CD4 cell count responses to HAARTwho will be able to discontinue therapy safely for prolonged periods of time. Large, randomized, controlled clinical trials are in progress to evaluate this approach. If the CD4 cell count is truly the most important predictor of TI duration, this raises the question of whether a pulse therapy strategy could be combined with immune-based therapies, such as IL-2, to increase the CD4 cell count and prolong the TI duration.
It is clear that this treatment strategy is inappropriate for a large proportion of HIV-infected individuals, namely (1) patients with CD4 cell counts that fail to increase substantially during receipt of therapy, and (2) those with high pretreatment virus loads or low CD4 cell counts who might experience symptomatic viral rebound and/or rapid progression during TI. However, for a subgroup of infected patients, pulse therapy offers the possibility of prolonged reprieves from antiretroviral therapy and its associated toxicities and cost. This strategy may be less likely to promote drug resistance than strategies involving shorter and more frequent TIs. In fact, given the high rate of resistance seen in patients who have received antiretroviral therapy for a long duration, it is possible that pulse therapy approaches may actually decrease the long-term risk of drug resistance.
Longer-term follow-up periods will be needed to determine the duration of TI and whether treatment can be interrupted repeatedly in a pulse therapybased fashion. Most importantly, randomized, prospective clinical trials comparing pulse therapy with continued therapy will be necessary to definitively assess the risks and benefits of this approach before it can be routinely recommended.
|
|
| |
| |
|
|
|
