|Home | About | Journals | Submit | Contact Us | Français|
Intermittent administration of interleukin-2 (IL-2) to human immunodeficiency virus (HlV)-infected patients on antiretroviral therapy (ART) is capable of inducing significant increases in CD4 T cell counts as a result of increased T cell survival and decreased cell turnover. However, its role in the setting of ART interruptions (STI) is less well characterized. We sought to compare the effect of continuous (C) versus intermittent (P) ART on CD4 responses in patients undergoing IL-2 therapy.
CD4 cell responses were compared in 25 patients who underwent IL-2 therapy during periods of continuous ART (n=90 cycles) as well as during STI (n=45 cycles). During STI, patients resumed ART for only 10 days surrounding each IL-2 cycle.
C cycles resulted in a significantly greater CD4 gain than P cycles (Δ156 cells/μL, 95% CI=68–243). In multivariate analyses, baseline CD4/CD25 expression and treatment arm remained strong predictors of CD4 gain while CD8/CD38+, CD8/DR+, and CD4 Ki67+ phenotype were not predictive.
Continuous ART was associated with a statistically significantly greater CD4 cell response to IL-2 therapy than was intermittent ART. These observations may have important implications for the appropriate integration of IL-2 therapy into STI strategies.
Although the human immunodeficiency virus (HIV) epidemic has now passed its 25th anniversary, to date no immune-based therapy has yet been approved for the treatment of what is intrinsically an immunologic disease, albeit one of viral etiology. Of those agents that have been under study, the interleukin (IL) class of immunomodulators, particularly IL-2, have clearly been the best characterized thus far (Allende and Lane 2001). Even for IL-2, however, clinical end-point trials that could potentially lead to licensure are still at least one or more years from completion (Lifson and others 2006). If successful, the two large clinical end-point trials currently underway will establish that IL-2 has a more favorable effect on the rate of clinical disease progression and/or death than that afforded by ART alone. But while this outcome would represent a major milestone in the development and applicability of immune-based therapies for HIV infection, it will not necessarily provide compelling guidance as to how best to incorporate this therapy into modern clinical practice.
In the interim and without this guidance, clinical investigators interested in this approach have been struggling to determine how therapies such as IL-2 might best serve as useful adjuncts to conventional ART. For example, even despite the relative pall cast recently on the field of structured antiretroviral treatment interruptions by the unanticipated results of the international SMART (Strategies for Management of Antiretroviral Therapy) trial, there has been keen interest in attempting to formulate strategies by which an agent such as IL-2 could potentially serve in an effective HAART (Highly Active Antiretroviral Therapy)-sparing capacity (Pett and Emery 2001; Youle and others 2006). That is, it continues to make sense that an agent whose intermittent use has been shown to be capable of reversing the CD4+ cell decline caused by chronic HIV infection could potentially spare the need for chronic persistent reliance on ART. One obvious goal would be to reduce the accompanying toxicities of the latter—including lipodystrophy, hepatic steatosis, hyperlipidemia, glucose intolerance, and neuropathy—as well as to affect favorably such problems as the strict adherence requirements and the emergence of drug-resistant HIV infection associated with their long-term use (Havlir and Currier 2006).
In this regard, some investigators and patients have explored a strategy of combining treatment interruption with intermittent IL-2 therapy in an attempt to boost CD4 counts while minimizing ART exposure (Henry and others 2006). Underlying this approach are the multiple randomized controlled trials that have shown that the administration of intermittent IL-2 therapy can induce selective CD4+ T cell expansions as measured in the peripheral circulation. When compared with ART alone, these increases are significantly higher when IL-2 is co-administered (Levy and others 2003; Farel and others 2004). Initially, this increase primarily reflects increased T cell proliferation but, with chronic intermittent therapy, more likely results from decreased activation coupled with significantly enhanced survival of the expanded population (Sereti and others 2004; Kovacs and others 2005).
Previously, we reported on a cohort of patients treated with long-term intermittent IL-2 therapy who underwent a protocol-defined ART interruption immediately after completing an IL-2 cycle (Keh and others 2006). Among other insights, this analysis demonstrated that the CD4 cell increases induced by IL-2, in this circumstance, were significant but blunted compared with IL-2–induced increases noted during continuous ART. However, the potential effects of IL-2 administered in the setting of an established, ongoing treatment interruption, perhaps the most probable clinical scenario, are yet to be determined.
The goal of the present study was to assess the differential responses to IL-2 therapy when given during conventional prolonged ART interruption periods voluntarily undertaken by patients. Specifically, in a cohort of 25 IL-2 recipients followed longitudinally at our center, who could serve as their own comparators, we sought to compare these responses when IL-2 was administered during continuous suppressive ART versus when it was given during extended treatment interruptions in which only 10 days of pericycle ART was employed. We also sought to determine whether any of the immunologic characteristics of the cohort present at baseline could serve as predictive factors in determining the magnitude of the CD4 changes observed.
Subjects enrolled in an NIAID Institutional Review Board approved protocol evaluating long-term intermittent subcutaneous IL-2 use in HIV-infected patients were considered for this study. Patients enrolled on the parent trial received multiple cycles of IL-2 during the course of their participation, including some who received their initial therapy intravenously. Patients were eligible for participation in this substudy if they had received at least one cycle of subcutaneous IL-2 therapy while on continuous ART as well as at least one cycle of subcutaneous IL-2 therapy during a supervised ART interruption in the era of HAART (between 1996 and 2005). In our institution, patients choosing the latter approach have been required to resume short-course antiretroviral treatment in a 10-day period surrounding IL-2 administration in order to blunt the transient rise in HIV viral load observed in earlier studies involving IL-2 recipients on suboptimal ART (Kovacs and others 1996). A baseline CD4 count of at least 200 cells/μL was required for inclusion in the parent study. All patients were seen and evaluated at the Clinical Center of the National Institutes of Health in Bethesda, MD.
Twenty-five patients met the eligibility criteria. Information about ART status was collected for each patient's cycles: continuous (C) IL-2 cycles were defined as those conducted during a period of uninterrupted ART, whereas pericycle (P) IL-2 cycles were those conducted within a period of ART interruption during which patients were restarted on ART only for a 10-day period surrounding the IL-2 cycle (beginning 3 days prior to and ending 2 days following each 5-day cycle). All subcutaneous cycles administered at least 2 months apart were included in the analysis if the patient's precycle ART status remained unaltered at least until the time of the first postcycle CD4 determination, generally at least 1 to 2 months following cycle completion. While some patients described in the earlier ART interruption trial were also included in the present study, their single “interruption” IL-2 cycles analyzed in that earlier report were specifically excluded from our analysis because they did not meet the definition of P cycles as described above. IL-2 was self-administered subcutaneously by patients in doses ranging from 1.5 MIU to 7.5 MIU twice daily for 5 days. Doses were given at the maximum tolerated dose as individualized for each patient based on their prior observed toxicity profile as well as their IL-2 dose-response relationship.
IL-2 cycling intervals varied significantly between patients depending on the magnitude and duration of their CD4+ response. The algorithm by which IL-2 was administered to study subjects has been described elsewhere (Farel and others 2004). Briefly, serial monitoring of CD4+ cell counts was performed in all subjects every 2–4 months, and IL-2 cycle frequency was designed to maintain the CD4+ cell count at or above a predefined threshold. This threshold was individualized for each subject, on average twice the baseline count present at the time of the IL-2-naïve individual's entry into the study. Patients were offered an IL-2 cycle if they had two consecutive CD4+ determinations at study visits below this threshold. Over time, the individualized thresholds of many of the patients were lowered by a few hundred cells based on patient request—these changes were permitted generally as long as a substantial increment over their baseline counts was still preserved.
Antiretroviral medication history, CD4+ T-cell counts/percentages, plasma HIV RNA levels, and immune activation state variables (phenotypic expression of CD4/CD25, CD8/DR, and CD8/38 as well as intracellular Ki67 staining) were collected at the following time points: On the most proximal clinic visit date preceding the IL-2 cycle prior to the initiation of any pericycle ART (labeled the “baseline” time point), on day 1 of each IL-2 cycle (labeled the “precycle” time point) by which time the P cycle participants had received 3 days of pericycle ART, and on the date of the next follow-up clinic visit (labeled the “postcycle” time point, on average 2 months after the cycle) (Fig. 1). For samples obtained prior to August 2005 on which HIV-1 viral load assays with lower limits of detection of either 500 copies/mL or 10,000 copies/mL had initially been performed, cryopreserved plasma samples obtained on the same dates were re-tested by using the ultrasensitive bDNA assay (lower limit of detection <50 copies/mL) in order to permit comparison with later determinations.
Immunophenotypic analysis (CD4/CD25, CD8/CD38, and CD8/DR) was performed on fresh peripheral blood mononuclear cells from all 25 patients. For Ki67 staining of CD4+ cells, cryopreserved specimens were evaluated at all comparable time points unless stored samples were unavailable. The analyses were performed by four-color immunofluorescence as described previously by using a FACSCanto flow cytometer (BD Biosciences, San Jose, CA). For intracellular Ki67 staining, cells were stained with Ki67 PE (BD Biosciences Pharmingen, San Diego, CA) along with each paired IgG control. Cells were gated on CD3+CD4+ or CD3+CD8+ and then analyzed for Ki67 percent positivity as determined by using an integral region set to include <0.5% positive events on the matched Isotype control.
To assess the impact of various factors on the postcycle-precycle gain in CD4 counts, and to take into account multiple observations from each patient, a linear mixed model was used (Laird and Ware 1982). For each cycle, measured factors included phenotypic expression of CD4/CD25, CD8/DR, and CD8/CD38, Ki67 staining, dose of IL2 administered, and use of continuous ART. All factors were measured at each cycle prior to initiation of IL-2 therapy. Univariate associations between precycle factors and CD4 gain were identified using the mixed model, and factors with a p-value less than 0.05 were included in a multivariate analysis.
To assess postcycle-precycle changes in immune markers, a paired difference t-test was used where the average change for each patient was used as the outcome. To assess the correlation between postcycle-precycle change in immune markers and change in CD4 counts, a linear mixed model as described above was used.
To examine robustness of the conclusions from the mixed model analyses of CD4 gain, an analysis of covariance-like approach was used whereby the linear mixed model was rerun using postcycle CD4 count as the response and precycle CD4 as factor. Because the choice to have a C or P IL-2 cycle was not randomized, mixed models incorporating several baseline factors were fit to adjust for possible selection bias, eg, patients with smaller gains being more likely to choose P cycles. A propensity score approach was used separately to provide a different adjustment for possible selection bias (Lunceford and Davidian 2004). To construct the propensity score, a logistic regression model was estimated using baseline immune markers to predict continuous versus pericycle treatment. All p values are two-sided, and p values <0.05 are reported as significant.
Cycles of each type from all 25 eligible patients were included in the analysis. Each had been enrolled as early as 1993 and followed through the time of data analysis in 2005. The majority (96%) of the patients were male, with a mean age of 30 years when enrolled to the study (median 39 years, range 28–70 years). On average, patients had received a total of nine cycles over the course of their IL-2 study participation. Of these, cycles that met criteria for inclusion in the present analysis averaged 3.6 total cycles per patient while receiving continuous ART and 1.8 cycles per patient while undergoing treatment interruptions.
Interestingly, mean “baseline” CD4+ counts measured approximately 2 months before each IL-2 cycle were significantly higher among the C cycles compared with P cycles (756 cells/μL and 588 cells/μL; p<0.001), likely reflecting the immunologic consequences of prolonged ART interruption in P-cycle participants. In contrast, mean “precycle” CD4+ counts measured at the initiation of IL-2 cycles, at which point P-cycle participants had been restarted on ART for only 3 days, were no longer significantly different for C cycles than for P cycles (807 cells/μL and 738 cells/μL, respectively; p=0.18). Because the choice of which time point to select as the most representative pre-IL-2 value was somewhat arbitrary, both baseline and precycle time points were each analyzed separately. This analysis revealed that conclusions about postcycle effects were similar for each parameter regardless of which of these two time points was selected as the true “pre-IL-2” value. For expediency, therefore, only the baseline time points as defined above are represented in the analysis described below. As expected, baseline viral loads were most often <50 copies/mL for continuous ART cycles and consistently >1000 copies/mL for cycles given during prolonged treatment interruptions.
The changes in immunologic and virologic parameters before and after IL-2 therapy in the two arms were assessed using the paired difference t-test. The estimated mean gain and p values for these comparisons are provided in Table 1. CD4+ cell counts increased significantly following both C cycles and P cycles for participants, although the magnitude of the change was substantially higher when IL-2 was administered in the setting of continuous ART (Fig. 2). Consistent with earlier observations on the correlates of IL-2 therapy, expression of the CD4/CD25+ phenotype also increased significantly in all IL-2 recipients. Activation state as measured both by intracellular Ki67 staining of CD4+ cells and by surface phenotypic expression of CD38 and DR in CD8+ cells significantly decreased postcycle in participants while receiving C cycles, again consistent with prior observations on the effects of IL-2 on these parameters. During P cycles, however, changes in these same markers failed to reach statistical significance, suggesting the likelihood that the dampening effect of chronic intermittent IL-2 therapy, even when administered in the setting of partial viral suppression achieved through resumption of pericycle ART, was unable to overcome sufficiently the opposing increase in lymphocyte activation triggered by uncontrolled viremia.
The association between gain in total CD4 cells (difference between postcycle CD4 and baseline CD4) and other variables was examined by univariate analysis using all cycles. Baseline CD4/CD425, CD8/DR, CD8/CD38, baseline viral load, and treatment arm were all significant predictors of CD4 gain by this analysis.
In contrast, when the correlation of change in immune activation markers and change in CD4+ count was then evaluated by the linear mixed effect model separately for the two treatment arms, a different result was obtained: only the group change in Ki67 expression in the pericycle group and the change in CD8/DR for the continuous group achieved significance (p<0.05) in this model. The corresponding p-values for each marker are shown in Table 2. However, none of these comparisons remained statistically significant after Bonferroni correction.
In a multivariate analysis, only baseline CD4/CD25 expression was predictive of CD4 gain (p=0.013). In this adjusted model, neither the dose of IL-2 administered nor indicators of immune activation state (as assessed by CD8/DR, CD8/38, and CD4/Ki67) remained predictive of this change. Therefore, we selected this combination of baseline CD4/CD25 expression and treatment arm as the best prediction model (Fig. 3). Using this model, patients receiving continuous ART during IL-2 therapy had mean increases of 156 cells/μL (95% CI=68–243) greater than when these same patients underwent IL-2 cycles administered while receiving only pericycle antiretrovirals. Given the CD4-driven rationale for which IL-2 therapy is usually undertaken, this difference in outcomes obviously represents a measurable immunologic toll to be paid for the privilege of avoiding chronic ART.
To validate the accuracy of this model, additional separate sensitivity analyses were undertaken. In one, the post-CD4 count was also analyzed as an outcome measure with baseline CD4, CD4/CD25, and treatment group as factors. Using this model, the estimated benefit of continuous antiretroviral treatment over pericycle ART usage during IL-2 therapy was 223 cells/μL (95% CI=140–306). Under a separate propensity score model, the estimated effect of continuous ART over pericycle usage was 175 cells/μL (95% CI=78–367, based on 1000 bootstrap samples). Therefore, the estimated benefit of continuous treatment over interrupted therapy was consistent over several independent methods of analysis.
In this retrospective study of 25 HAART recipients known to be likely to respond to intermittent IL-2 therapy, we have attempted to characterize further what price may be extracted immunologically for employing this investigational approach in the absence of persistent viral suppression. We did so in a setting where each patient had their immunologic profile characterized both before and after each IL-2 treatment and in which all patients participated at different times in both arms of the trial.
The major finding from our investigation, namely, that the use of IL-2 therapy without complete viral suppression still allows for potential immunologic gain but at a lesser degree than during chronic antiretroviral treatment, extends the observations from two prior studies. In one, a small group of patients naive to treatment elected to receive one of two different doses of subcutaneous IL-2 therapy in the complete absence of antiretroviral medications. In those patients, IL-2 therapy was generally administered safely but the observed CD4 cell responses appeared relatively blunted in comparison to expected increases in patients receiving ART, a historical comparison derived from the extensive phase II IL-2 database available at the time. While an intriguing finding, it remains unclear whether most clinicians would choose to utilize IL-2 therapy as singular therapy rather than as an intermittent adjunct to more conventional ART. In a more recent trial involving known IL-2 responders with CD4 counts above 500 cells/μL, an elective subcutaneous IL-2 cycle was administered to all patients after which ART was stopped immediately in the half of the cohort randomized to undergo antiretroviral treatment interruption. In addition to characterizing the differential immunophenotypic, CD4 subset, and activation state changes accompanying IL-2 therapy, the results of this study again showed that CD4 gains were less in that portion of the cohort who had their ART stopped postcycle in comparison to those randomized to remain on chronic antiretroviral suppression. While the trial had the advantage of being performed in a randomized prospective manner, similar to the present study it also involved patients who had prior exposure to IL-2, had an entry CD4 count ≥500 cells/μL, and thus were known to be IL-2 “responders.” However, unlike the present study it also employed a study design perhaps less likely to be as widely exported to the clinical arena, namely that involving prompt cessation of ART immediately after completion of an IL-2 cycle. Thus, while the immunologic insights afforded by that approach were substantial, it could be argued that the direct clinical applicability may be somewhat more limited.
In contrast, the analytical approach employed in the present study was not randomized but rather was focused on a cohort of patients with a broader CD4 base who were also shown to be capable of responding immunologically to IL-2. In addition to surveying a cohort able to serve as their own comparators on both arms, the chief advantage of this more recent analysis was that it compared IL-2 strategies more likely to be embraced in clinical practice; that is, the use of intermittent adjunctive IL-2 therapy either during continuous ART or during periods of prolonged withdrawal from ART. In direct viral terms, it compared IL-2 responsiveness during periods when ART was able to maintain viral suppression below detectable levels on an extended basis versus when short-term reintroduction of ART was only able to partially offset the resurgent viremia characterizing prolonged ART interruptions (STI). This STI approach was the strategy widely preferred by the participating patients and their referring physicians and, with the exception of requiring resumption of pericycle ART for those considering IL-2, was not one mandated by the parent IL-2 protocol on which they were enrolled.
These distinctions notwithstanding, the major immunologic observations were remarkably similar between these last two interruption trials regardless of the relative timing of ART cessation. In both cases, chronic use of suppressive HAART throughout the study period led to optimal CD4 gain from IL-2 therapy, but some benefits from IL-2 could still be demonstrated during periods of active viremia. CD4/CD25+ expression increased in participants in both trials as a correlate of total CD4 cell gain, likely reflecting the preferential expansion and survival of the cytokine-expanded naive subset of cells, as described previously (Sereti and others 2005). Also consonant with our current understanding of the long-term immunologic effects of intermittent IL-2 treatment, the activation states both of CD4+ and CD8+ cells decreased postcycle in those subcohorts of fully virologically suppressed patients from either trial. In contrast, this dampening effect on activation was apparently counteracted in those patients not receiving chronic HAART in whom persistent viremia was present.
In the present study, we extended those observations from the earlier trial by modeling the behavior of CD4 cells postcycle as a complex function of baseline status. By several lines of analysis, we found that the combination of CD4/CD25+ expression and treatment arm was the best predictor of the likelihood of substantial CD4 cell gain postcycle. Unlike in the earlier trial, however, the influence of baseline viral load in this model could not be fully explored because of the marked dichotomy in plasma viremia between the two treatment arms; ie, almost all C-cycle participants had baseline viral loads <50 copies/mL, whereas, as expected, plasma viremia was detectable in all participants during their P-cycle evaluations.
Even if one or both of the two large clinical end-point trials presently underway reach a definitive conclusion in favor of IL-2 treatment, there will still be a challenging gap to fill in terms of how best to integrate this new therapeutic modality into clinical practice. Many questions remain unanswered in this regard. For example, in the present trial, a 10-day pericycle period of ART resumption was required in order to counteract the known potential for resurgent viremia during IL-2 therapy. This was an arbitrary choice made largely on safety grounds, but neither the precise timing nor optimal duration of resuming ART in this manner can be considered evidence-based. Similarly unknown is whether HIV-infected patients naive to IL-2 will respond as readily to intermittent IL-2 treatment during ART interruption as those previously shown to be IL-2 responders; ie, those who have probably already induced a CD25+ subset of cytokine-expanded cells. Indeed, to help fill part of this gap a multicenter trial has recently been launched that is designed to determine the relative value of using ART in the pericycle period relative to administering IL-2 without any concomitant suppressive therapy, all in patients without prior IL-2 experience (Angus 2006). Nonetheless, even should the efficacy of IL-2 therapy be established within the next few years, these and other such questions will likely still need to be answered more fully before the leap from investigational to proven therapeutic adjunct can be considered complete.
This work was presented in part at the 46th Annual Interscience Conference of Antimicrobial Agents and Chemotherapy (ICAAC), September 2006 (Poster #H-1392). The U.S. Government has been granted a use patent for intermittent IL-2 therapy in HIV infection that includes J.A.K. as an inventor.