The Virologic, Immunologic, and Clinical Effects of Interleukin 2 With Potent Antiretroviral Therapy in Patients With Moderately Advanced Human Immunodeficiency Virus Infection:  A Randomized Controlled Clinical Trial—AIDS Clinical Trials Group 328 FREE

Human immunodeficiency virus (HIV) infection is characterized by impaired lymphocyte function and decreasing CD4 T-lymphocyte counts, leading to immunodeficiency and reduced survival. Potent antiretroviral drugs can produce long-term suppression of plasma HIV RNA with increases in CD4 T-cell counts and delayed disease progression.^1- 2 In more advanced infection, the durability of virologic suppression may be limited, CD4 cell counts frequently may not be restored to reference levels, and immunologic function remains impaired.^3- 6 Approaches to preserve immune function and immune reconstitution would have value.

Interleukin 2 (IL-2) is a cytokine with potent immunomodulating effects. Interleukin 2 induces proliferation and differentiation of CD4 and CD8⁺ lymphocytes, protects T lymphocytes from cell death, and enhances the cytotoxic activity.^7- 9 Interleukin 2 can induce substantial expansion of CD4 cells, associated with a dramatic prolongation of their half-life.^10- 21 Maintenance therapy in which IL-2 is administered when CD4 cell counts fall to predetermined levels is now commonly used.¹² Patients with relatively high baseline CD4 cell counts have shown the best responses, with less favorable responses reported in more immunosuppressed patients receiving less potent antiretroviral therapy.^14,22- 23 A randomized study in subjects receiving HAART foundonly modest enhancement of CD4 cell counts in advanced (<200/μL) infection with subcutaneous (SC) IL-2 treatment compared with HAART alone (median gains of 65/μL vs 18/μL after 24 weeks [P<.001]).²⁴

This AIDS Clinical Trials Group (ACTG) study evaluated the immunologic, virologic, and clinical effect of intravenous (IV) or SC IL-2 administration with a specified HAART regimen in moderately advanced HIV infection.

Adults infected with HIV type 1 (HIV-1) without active AIDS-defining illnesses and with CD4 cell counts of 50/μL to 350/μL were eligible. Prior protease inhibitor therapy or IL-2 therapy, autoimmune disease, or cardiac disease were exclusionary. Subjects could not have prespecified laboratory abnormalities. The study was approved by institutional review boards, and the patients provided informed consent.

Patients were enrolled at 26 ACTG sites. During step 1, subjects received 12 weeks of indinavir sulfate and specified combinations of 2 nucleosides (minimum of 1 new), including zidovudine, lamivudine, didanosine, or stavudine. Substitution for intolerance or for viral breakthrough was permitted. Patients with no more than 5000 plasma HIV RNA copies/mL at week 12, suggesting responses to HAART, were randomized 1:1:1 to HAART alone for 72 weeks, HAART plus continuous-infusion IL-2 (HAART + IV IL-2) at 9 mIU daily for 5 days every 8 weeks, or HAART + SC IL-2 at 7.5 mIU twice daily for 5 days every 8 weeks. Subjects could receive up to 9 cycles of IL-2 (Figure 1). Recombinant human IL-2 was used (Proleukin; Chiron, Emeryville, Calif). There were up to 2 dose reductions for clinical adverse effects or laboratory toxic effects (6 or 3 mIU daily for IV IL-2 and 4.5 or 2.5 mIU twice daily for SC IL-2). Patients in the IV IL-2 arm could switch to the SC IL-2 arm after 3 or 6 cycles if they achieved greater than a 25% increase and a 100/μL increase in CD4 cell count above the randomization baseline. Postrandomization AIDS-related events were assessed using protocol-specified criteria. All patients underwent chest radiography and physical examination and were seen every 4 weeks until week 20 and every 8 weeks thereafter. A substudy (ACTG 5046) examined immunization responses in 38 patients.^25- 26 Another substudy (ACTG 928) in 15 patients added a tumor necrosis factor inhibitor, etanercept, to minimize the toxic effects of IL-2 and, while not affecting toxicity, it enhanced IL-6 and C-reactive protein expression.²⁷

Safety laboratory procedures and CD4/CD8⁺ T-cell counts were performed every 8 weeks at ACTG-certified laboratories. Batched HIV RNA levels were measured at entry, randomization, and weeks 60 and 84 by means of the branched DNA assay (Chiron) in an ACTG-certified laboratory. The RNA levels below 50 copies/mL were analyzed as if they were 50 copies/mL. Skin testing for delayed-type hypersensitivity was assessed at baseline and weeks 36, 60, and 84. Eighty-seven patients participated in the immunology substudy, wherein enumeration of lymphocyte subpopulations, lymphocyte proliferation, and cytokine and soluble activation marker levels were performed at entry, randomization, and weeks 36, 60, and 84.

Lymphocyte subpopulations were measured using ACTG consensus methods that used fluorochrome-labeled monoclonal antibodies (Becton-Dickinson Pharmingen, San Diego, Calif) to the following antigens: CD3, CD4, CD8, CD16, CD19, CD25, CD28, CD38, CD45RA, CD56, CD62L, CD95, and HLADR. Naïve cells were defined as CD45RA⁺CD62L⁺, memory/effector cells as CD45RA⁻CD62L⁻ plus CD45RA⁻CD62L⁺ plus CD45RA⁺CD62L⁻, IL-2 receptor α chain–positive cells as CD25⁺, and activated cells as HLADR⁺CD38⁺. Functional subsets were defined according to the expression of CD95⁺ or CD28⁺. Lymphocyte proliferation assays used ACTG consensus methods, and cytokine and soluble activation marker levels were measured in cryopreserved specimens.

Randomization was stratified by participation in the ACTG 928 and by the nucleoside analogue combinations. Each IL-2 treatment arm was to be compared (separately) with the HAART alone arm using 2-sided tests with significance levels of .025. (No other correction for multiple comparisons was performed.) The primary efficacy end point was the proportion of patients with CD4 T cells (a week 60 count of ≥50% larger than the randomization count). The accrual goal was 120 randomized patients (40 per arm) with week 12 and week 60 CD4 cell counts who were not enrolled in the ACTG 928. This sample size provided at least 80% power to detect a difference of 35% in the true proportions of patients with a CD4 T-cell response, using the Fisher exact test. The week 12 CD4 and CD8⁺ T-cell counts were those closest to randomization. The week 60 counts were those between weeks 52 and 68 that were closest to week 60.

The protocol specified that the secondary CD4 T-cell analysis should apply random effects models to the slope of log CD4 cell counts between weeks 12 and 60. For other secondary analyses, proportions of patients with an end point were compared using Fisher exact tests. Analyses of variables with many possible values used the Wilcoxon rank sum test.

Step-up logistic regression models using the likelihood ratio test were performed to model the probability of CD4 cell response by week 60 in all randomized patients; treatment variables were always in the model, and candidate exploratory variables included the factors in Table 1 and white blood cell count, lymphocyte count, lymphocyte percentage, and real-time plasma HIV RNA levels (at entry and randomization). Continuous covariates were divided into quartiles for these models. Sets of step-up logistic regression models were performed in the immunology substudy, trying all the variables of entry and randomization percentages and counts for the T-cell subsets measured. All of these measurements were dichotomized at the median observed value.

Two hundred four patients were enrolled and 174 were randomized to step 2: 15 for the ACTG 928 and 159 for the ACTG 328 (Figure 1). Data for the ACTG 928 patients appear only in the tables regarding toxic effects and deaths (Table 2 and Table 3). Reasons why patients were not randomized to step 2 included plasma HIV RNA levels of greater than 5000 copies/mL (n = 10), inability to comply (n = 9), or 1 of multiple other reasons (n = 11). The baseline characteristics are presented in Table 1. The groups were similar, although the HAART alone group was likely to be older than 49 years and to have higher CD4 and CD8⁺ T-cell counts.

Sixteen patients changed their HAART drugs. Forty patients discontinued study drug therapy before week 60 (10 receiving HAART alone, 13 receiving HAART + IV IL-2, and 17 receiving HAART + SC IL-2); the reasons for most of these changes were mild or moderate toxic effects, complexity of medications, clinical events, and noncompliance. Fifty-one patients discontinued study drug therapy before week 84. Four patients died before week 60.²⁷

The median number of cycles in the HAART + IV IL-2 arm was 9; in the HAART + SC IL-2 arm, 8. Only 14 patients (26%) in the IV IL-2 arm received 4 or more cycles and only 4 (8%) received 7 or more cycles. By week 84, most IV IL-2 patients had been receiving SC IL-2 for 1 year (Figure 2).

The median dose per cycle of IV IL-2 was 45 mIU (protocol-specified initial dose); the median dose per cycle of SC IL-2 was 75 mIU (protocol-specified initial dose) for only the first 2 cycles (third cycle, median dose, 60 mIU; from the fourth cycle, median dose, 45 mIU). The median dose per cycle of SC IL-2 received after switching was similar to that in the SC IL-2 group. The median total dose during the first 9 cycles was 427 mIU in the IV IL-2 patients and 413 mIU in the SC IL-2 patients.

Every attempt was made to obtain week 60 CD4 cell counts (primary end point) on all patients, even for subjects who had stopped protocol treatment. Patients who died by week 60 were counted as nonresponders. Of 159 step 2 patients, 141 (89%) were included in this analysis; 16 had stopped protocol therapy more than 8 weeks before their week 60. Shown in Figure 3A, each IL-2 arm had significantly more week 60 responders than did the HAART arm: 13 (29%) of 45 in the HAART arm, 39 (81%) of 48 in the HAART + IV IL-2 arm, and 32 (67%) of 48 in the HAART + SC IL-2 arm (P<.001 for the IV and SC arms). The advantage of IL-2 treatment was irrespective of the nucleoside analogue combination. Similarly, both IL-2 arms were superior to HAART alone in the proportions of responders at week 36 (142 patients [P<.001 for both]) and week 84 (130 patients [P<.001 for both]).

Based on the random-effects models (a protocol-designated secondary end point), the CD4 T-cell slopes were greater in the IV and SC IL-2 arms than in the HAART alone arm from week 12 to weeks 36, 60, and 84 (P<.001 at each time point). Figure 3B shows the average CD4 and CD8⁺ T-cell counts by arm for weeks 12 to 84. From weeks 12 to 60, there were significantly higher increases in CD4 cell counts in both IL-2 arms compared with the HAART arm (P≤.001 for both comparisons). No significant differences were observed in CD8⁺ T-cells counts (data not shown).

No significant differences were observed in the percentages of patients with HIV RNA of less than 50 or less than 500 copies/mL (Figure 4). At randomization, there were fewer patients in the HAART + SC IL-2 arm with less than 50 copies/mL than in the HAART alone arm. No significant differences were seen in median HIV RNA levels or the percentage of patients with virologic failure (>0.7-log increase in HIV RNA levels) from randomization to week 60 or to week 84.

Table 2 shows the timing and CD4 levels at the diagnosis of new AIDS-defining clinical events. More AIDS-related events were observed in the HAART alone arm than in either of the IL-2 arms (7 in the HAART alone arm vs 0 in the IV IL-2 + HAART arm [P = .006] and 1 in the SC IL-2 + HAART arm [P = .03]). The malignancies, which included Kaposi sarcoma, lymphomas, and Castleman disease, occurred in patients with the last CD4 cell counts before diagnosis ranging from 171/μL to 582/μL. There were no significant differences in HIV-related non–AIDS-defining events in the 3 arms (1, 3, and 1) or in the number of deaths from any cause (3, 2, and 2). Inclusion of events for the 15 ACTG 928 participants did not affect the results.

There were no significant differences in skin test reactivity or in lymphocyte proliferation, and the differences in soluble marker levels had large P values (.02-.05; data not shown).

Table 3 shows the median week 12 measurements and median changes from weeks 12 to 60 in the T-cell phenotypic panel for each of the 3 treatment arms. For each phenotype, 51 patients (split among the 3 arms) had results from weeks 12 and 60. The HAART + IV IL-2 arm had significantly (P<.05) greater increases than did the HAART alone arm in the numbers of CD4 memory/effector, CD4 naïve, CD4/CD95⁺, CD4/CD28⁺, and CD8⁺/CD28⁺ cell counts. The HAART + SC IL-2 arm had significantly greater increases than did the HAART alone arm in CD4 memory/effector, CD4/CD95⁺, CD4/CD28⁺, CD4/CD25⁺, and CD8⁺ memory cell counts. These differences persisted through week 84 (data not shown). There were no significant changes in the number or the percentage of CD4 or CD8⁺ T cells expressing activation markers.

In the final step-up logistic model of week 60 CD4 T-cell response for the overall study, both IL-2 arms were highly significantly different from the HAART alone arm (P<.001 for IV IL-2 + HAART and P = .003 for SC IL-2 + HAART). The only other variables significant for larger probability of week 60 CD4 T-cell response were entry (week 0) plasma HIV RNA level of less than 10 000 copies/mL (P = .01) and randomization CD8⁺ T-cell percentage of less than 47% (P = .05); both cutoff values represent the first quartile of these variables. No other variables, notably including a pretreatment IL-2 CD4 T-cell count of less than or greater than 200/μL, were significant in the final model.

Table 4 shows the percentage of patients in each arm with grade 3 or grade 4 laboratory or clinical toxic effects. Only categories of adverse events in which at least 5% of the 174 patients had grade 3 or worse adverse events are reported. Overall, both IL-2 arms were associated with significantly more grade 3 or 4 clinical toxic effects usually associated with IL-2 treatment. The HAART + IV IL-2 arm had more gastrointestinal tract toxic effects (especially nausea and vomiting), skin/mucocutaneous toxic effects, and general toxic effects (specifically fever and chills), and SC IL-2 treatment was associated with more fatigue.

Intermittent IV or SC IL-2 treatment combined with antiretroviral therapy has been shown to produce substantial and sustained increases in CD4 cell counts in patients with early-stage or midstage HIV infection, ie, CD4 cell counts of greater than 250/μL.^13- 21 There are fewer studies in patients with more advanced HIV infection. In this randomized controlled study, the immunologic and virologic effects of beginning HAART and then IL-2 by the intravenous or subcutaneous routes was compared with HAART alone in patients with CD4 cell counts between 50/μL and 350/μL before HAART. In patients with moderately advanced disease who responded virologically to HAART after 12 weeks, significant increases in CD4 cell counts were seen with IL-2 therapy and greater expansion of multiple phenotypic subsets of CD4 T cells was observed. At least a 50% increase in CD4 cells was achieved by week 60 in 81% of patients in the IV IL-2 + HAART arm, 67% of those in the SC IL-2 + HAART arm, and 29% of those in the HAART alone arm. This was achieved without an increase in plasma HIV RNA levels.

At the initiation of this study, it was unclear whether the CD4 cell response to IL-2 treatment would be affected by the administration route, especially in patients with low baseline CD4 cell counts, because the study suggesting equivalent results was conducted in patients with CD4 cell counts greater than 250/μL. The SC route seemed more practical for long-term administration. We sought to address these issues by randomizing patients to receive SC IL-2 or IV IL-2 at the onset of the study but allowing subjects to switch to SC IL-2 later if they reached certain increases in CD4 cell counts; therefore we did not directly compare the routes of IL-2 administration. The slope of CD4 cells was larger in the IV IL-2 arm than in the SC IL-2 arm for 72 weeks, although most patients in the IV arm had been receiving SC IL-2 for a year. Differences in the type and frequency of adverse and toxic effects were observed, with patients tolerating IL-2 treatment reasonably well. This occurred with similar IL-2 dosing per cycle. A study of quality of life in the 3 groups suggested a short-term detrimental effect during courses of IL-2 treatment and some evidence of a long-term beneficial effect in the SC IL-2 arm.²⁸

The sustained 3- to 4-fold greater increases in CD4 cell counts with IL-2 could have clinical relevance in patients with advanced HIV infection or in those with insufficient immunologic responses to suppressive antiretroviral therapy.²⁹ Participants had increases in CD4 cell counts regardless of whether the counts were above or below 200/μL at the time of randomization.

As noted in patients with earlier-stage disease, the increase in CD4 cell count as a result of IL-2 therapy appears to be associated with a general peripheral expansion of CD4 cell populations that includes naïve and memory/effector cells as well as increases in the number of cells expressing CD28, CD25, and CD95.¹⁷

Despite increases in the number of CD4 cells and their subsets, we were unable to demonstrate in vitro evidence of increased functionality. In a substudy of this trial,²⁵ responses to immunization were not enhanced among IL-2 recipients. However, a modest enhancement in vaccine responsiveness in IL-2 recipients has been reported.¹⁷

In this study, the number of CD4 cells expressing CD4/CD28 increased in both of the IL-2 arms. Expression of this coreceptor for T-cell activation is critical for T-cell responsiveness and it predicted responses to immunizations in HIV infection.^30- 31 Although exposure to IL-2 induces expression of the α receptor chain (CD25) that is needed for high-affinity IL-2 binding and promotes cellular expansion, high-level expression of CD25 also is seen among CD4 regulatory T cells that attenuate T-cell immunity.³² While IL-2 administration enhances the expansion of CD25⁺ cells, these cells do not appear to have the same immunosuppressive activities that are characteristic of CD4 regulatory T cells.¹¹

The CD4 cell count is the most reliable predictor of short-term morbidity, with counts below 200/μL associated with a heightened risk of opportunistic infections and death.^3,27,33 While IL-2 administration can increase the CD4 cell count in persons with HIV infection, there has been no convincing evidence that the CD4 T-cell count increases seen after IL-2 administration confer any clinical benefit. Although we were unable to demonstrate laboratory evidence of enhanced immune function, the apparent lower frequency of AIDS-defining events in the IL-2 arms and an earlier meta-analysis³⁴ of several clinical studies suggesting clinical benefits are encouraging but require confirmation. Two large randomized studies powered to discern differences in the occurrence of the clinical end point, the Evaluation of Subcutaneous Proleukin in a Randomized International Trial (ESPRIT)³⁵ and the Phase III Multicenter Randomized Study of the Biological and Clinical Efficacy of Subcutaneous Recombinant, Human Interleukin-2 in HIV-Infected Patients With Low CD4 Counts Under Active Antiretroviral Therapy (SILCAAT),³⁶ are ongoing.

In conclusion, this study provides evidence that IL-2 combined with HAART has the potential to significantly increase CD4 cell counts and partially restore naïve and memory CD4 cells and other phenotypic subpopulations of T cells in patients with moderately advanced HIV infection, without increasing HIV replication. The short-term occurrence of AIDS-defining diseases might have been reduced in patients receiving IL-2. Whether IL-2 produces long-term clinical benefit will be answered by large, ongoing clinical trials.

Correspondence: Richard B. Pollard, MD, Division of Infectious Diseases, University of California–Davis Health System, 4150 V St, Suite G500, Sacramento, CA 95817 (rbpollard@ucdavis.edu).

Accepted for Publication: November 4, 2006.

Author Contributions:Study concept and design: Mitsuyasu, Gelman, Cherng, Fahey, Reichman, Bucy, Kilby, Lederman, and Pollard. Acquisition of data: Mitsuyasu, Landay, Reichman, Erice, Bucy, Kilby, Lederman, Hamilton, Lertora, White, and Pollard. Analysis and interpretation of data: Mitsuyasu, Gelman, Cherng, Bucy, Kilby, Lederman, Hamilton, Tebas, Duliege, and Pollard. Drafting of the manuscript: Mitsuyasu, Gelman, Cherng, Landay, Erice, Kilby, Lederman, and Pollard. Critical revision of the manuscript for important intellectual content: Mitsuyasu, Gelman, Cherng, Fahey, Reichman, Bucy, Kilby, Lederman, Hamilton, Lertora, White, Tebas, Duliege, and Pollard. Statistical analysis: Gelman and Cherng. Obtained funding: Mitsuyasu and Pollard. Administrative, technical, and material support: Landay, Fahey, Reichman, Erice, Kilby, Hamilton, Lertora, and White. Study supervision: Bucy, Kilby, Hamilton, Lertora, Duliege, and Pollard.

Group Members: Participating ACTG investigators and sites in this study include Virginia Waite, BSN, and Liz Connick, MD (University of Colorado, Denver); University of North Carolina, Chapel Hill; Judith Carden, RN, and Susan McCarthy, RN (University of California–Los Angeles); New York University, New York; Sheryl S. Storey, PA-C, and Ann C. Collier, MD (University of Washington, Seattle); David Asmuth, MD, and Suzanne Lanier, RN (University of Texas Medical Branch, Galveston); Kathy Shina, RN, and Michael Chance, RN (Case Western Reserve University, Cleveland, Ohio); Greg Sfakianos, MD, and Michael S. Saag, MD (University of Alabama at Birmingham); Cornell University, New York, NY; University of California–San Francisco; Susan L. Koletar, MD, and Charlotte Mills, RN (Ohio State University, Columbus); Carol Hamilton, MD, and Shelia Tedder, RN (Duke University, Durham, NC); Jane Reid, RN, and Lisa Demeter, MD (University of Rochester, Rochester, NY); Scott Souza, PharmD, and Nancy Hanks, RN (University of Hawaii, Honolulu); Mitchell Goldman, MD, and Kristen Todd, RN, MSN (Indiana University, Indianapolis); Sandra Valle, PA-C, and Debbie Slamowitz, RN, BSN, ACRN (Stanford University, Palo Alto, Calif); David Mushatt, MD, and Erin Plaia, RN (Tulane University, New Orleans, La); Gildon Beall, MD, and Mario Guerro, MD (Harbor General Hospital, Los Angeles); David Ives, MD, and Carol DeLaney, BSN, ACRN (Beth Israel Deaconess-West Campus, Boston, Mass); Michael Dube, MD, and Hannah Edmondson-Melancon, RN (University of Southern California, Los Angeles); Donna Mildvan, MD, and Ron D’Amico, DO (Beth Israel Medical Center, New York); Mt Sinai Hospital, New York; University of Iowa, Iowa City; and M. Conklin, RNP, and M. Royal, RPH (Washington University, St Louis, Mo).

Financial Disclosure: Dr Tebas has received grant support from the National Institutes of Health (NIH), Pfizer, Roche, VIRxSYS, Tibotec, Primagen, Gilead, VGC, and Wyeth; has received honoraria from Bristol-Myers Squibb, GlaxoSmithKline, and Tibotec; and has been a consultant for Bristol-Myers Squibb and Primagen. Dr Pollard was an ad hoc consultant and a member of a speaker's bureau and received research grant support from Bristol-Myers Squibb.

Funding/Support: This work was supported in part by grants AI27660, AI38858, AI25879, AI27658, AI38855, AI27661, AI32782, AI39156, AI25868, AI50410, AI25903, AI25903, AI27661, AI27673, AI25924, AI46370, and AI25859 from the US Public Health Service, NIH, National Institute of Allergy and Infectious Diseases for the Adult AIDS Clinical Trials Group; and by grants RR-00865, RR-00080, RR-00032, RR-05096, RR-00046, and RR-00044 from the US Public Health Service, NIH, National Center for Research Resources for the General Clinical Research Centers.

Acknowledgment: We thank our consultants and collaborators, Richard Davey, MD, Joseph Kovacs, MD, Charles Flexner, MD, Larry Fox, MD, Albert Wu, MD, and Susan Koletar, MD, for their help during the development and conduct of this study; the following pharmaceutical companies for providing drugs for this study and their helpful insights: Chiron Corporation (David Sahner, MD), Bristol-Myers Squibb (Steven Schnittman, MD), and Glaxo Wellcome Inc (Diane Goodwin, PharmD, and Alex Rinehart, PhD); the many investigators and working staff of the AACTG, both at the sites and in the Operations and Data Management offices, who participated in the conduct of this study; and the many patients and their supporters for participating and assisting in making this study possible.