Methods

Study design. Study participants were selected from a pool of 95 individuals who had participated in open-label National Institute of Allergy and Infectious Diseases-sponsored and institutional review board-approved studies using IL-2 (intravenously or subcutaneously) in combination with ART. Only studies with available stored cryopreserved specimens were screened. All participants had CD4⁺ T cell counts >200 cells/µL at baseline and were receiving ART. None had received IL-2 previously.

These studies used an intermittent regimen of 5-day IL-2 (aldesleukin; Proleukin; Chiron) cycles administered every 8 weeks. IL-2 for intravenous administration was diluted in 5% aqueous dextrose solution containing 0.1% albumin and was given by continuous 24-h intravenous infusion with a compact portable pump. The starting dose of IL-2 ranged from 12 to 18 million IUs (MIUs) a day. This dose was decreased in increments of 1.5 to 6.0 MIUs as needed, to manage clinical or laboratory evidence of toxicity.

IL-2 administered subcutaneously was diluted in sterile water and injected subcutaneously at rotating sites. Administration of IL-2 was initiated at 7.5 MIUs twice a day subcutaneously for 5 consecutive days every 8 weeks. To manage toxicity, the IL-2 dose was decreased in increments of 1.5 MIUs/dose. Thereafter, dosage could be adjusted stepwise in either direction but was not to exceed the originally assigned dose or to decrease to <1.5 MIUs/dose.

A matched case-control study was designed. HIV-infected patients from this cohort with a ⩽10% increase in CD4⁺ T cell count (nonresponders [NRs]) were selected and matched for baseline (week 0, before IL-2 immunotherapy) CD4⁺ T cell count with contemporaneous study participants (control subjects) who experienced a ⩾50% increase in CD4⁺ T cell count at the same time point. Response was determined based on CD4⁺ T cell values obtained 2 months after completing the third cycle of IL-2 (week 24).

Baseline CD4⁺ T cell counts represented the mean of the 3 values obtained before the initial administration of IL-2. Control subjects and NRs were additionally matched on the basis of their antiretroviral regimen (nucleoside analogues vs. highly active ART [HAART]) as well as the route of IL-2 administration (intravenous vs. subcutaneous). Patients who had initially received low doses of IL-2 (1.5 MIUs subcutaneously twice a day) due to the study design [1] were excluded from the selection process. These participants had consistently low CD4⁺ T cell responses, which improved after transition to a higherdose regimen. All participating patients had provided written informed consent.

Flow cytometry. Immunophenotypic analysis was performed on cryopreserved peripheral blood mononuclear cells from all study participants by use of 4-color immunofluorescence as described elsewhere [6]. The following monoclonal antibodies were used for staining: CD25 phycoerythrin (PE) (clone 2A3), CD38 PE (clone HB7), CD4 peridinin chlorophyll-a protein (clone SK3), CD3 allophycocyanin (APC), or fluoresceine isothiocyanate (FITC) (clone SK7), CD27 FITC (clone M-T271), HLA-DR FITC (clone L243), programmed death (PD)-1 PE (clone MIH4), and CD45RO PE or APC (clone UCHL-1) from BD Immunocytometry Systems and Pharmingen and forkhead box protein P3 (FoxP3) PE (clone 206D) from BioLegend. To study naive and memory phenotypes, subsets were defined by surface staining with CD3, CD4, CD45RO, and CD27 antibodies as described elsewhere [11]. Naive cells were defined as CD45RO⁻CD27⁺, central memory as CD45RO⁺CD27⁺, effector memory as CD45RO⁺CD27⁻, and effectors as CD45RO⁻CD27⁻. Intracellular staining for the nuclear antigen Ki67 and the cytoplasmic protein Bcl-2 was performed using Ki67 PE (clone B56) and Bcl-2 FITC (clone Bcl- 2/100) from BD Immunocytometry Systems and Pharmingen following the protocols recommended by the manufacturer.

Samples were collected on a FACSCalibur (BD Immunocytometry Systems and Pharmingen) using CellQuest software. Approximately 1.0×10⁵–1.2×10⁵ total events with a minimum of 5000 events in the CD4⁺ or CD8⁺ gate were collected per sample. Flow cytometry data were analyzed using FlowJo software (TreeStar).

T cell receptor excision circle (TREC) measurement. TREC measurement was performed by real-time polymerase chain reaction (PCR) as described elsewhere [12]. Bead-separated CD4⁺ and CD4⁻ T cell subsets were used for determination of TREC levels. A value of TREC per microliter of blood was calculated on the basis of the value of TREC per 1×10⁶ T cells and the absolute CD4⁺ and CD8⁺ T cell counts at the same time point.

Statistical analysis. Median values were used for all analyses, and median values with interquartile ranges (IQRs) are reported. The significance of paired differences (within each group) between weeks 0 and 24 was determined by the sign test. Unpaired analyses (between the 2 groups) were performed by the Wilcoxon rank-sum test. The association between variables was determined by Spearman's rank correlation. Because of the exploratory nature of the present study, there was no adjustment for multiple testing, and calculated P values are reported.

Results

Patient characteristics and CD4⁺ T cell responses to IL-2. There were no statistically significant differences between NRs and control subjects by the 2-sample Wilcoxon rank-sum test (table 1). The viral load of the NR group was higher than that of the control group (4862 vs. 728 copies/mL; P = 0.1) (table 1). When the between-group comparison was performed as a paired test, NRs had a statistically higher HIV load than did the control subjects (P = .02, sign test). The a priori decision had been to perform an independent analysis to achieve more robust statistical significance, particularly in view of the small size of the cohort, the lack of correction for multiple comparisons, and the lack of a complete set for certain parameters tested. As anticipated on the basis of the definition of NRs and control subjects, a significant difference was evident between the 2 groups in the CD4⁺ T cell count change from baseline at week 24 (−1.3% vs. 116.5%; P ⩽ .0001) (table 1). No statistically significant changes in the viral load were noted in either group throughout the duration of the study (data not shown).

Proportions of naive and memory T cell subsets at baseline. The proportion of CD4⁺ T cells with naive (19.1% [IQR, 9.1%–29.2%] vs. 21.9% [IQR, 16.9%–28.9%]), central memory (55.6% [IQR, 43.3%–67.0%] vs. 59.6% [IQR, 52.1%–63.3%]), effector memory (17.2% [IQR, 13.6%–30.8%] vs. 15.4% [IQR, 7.3%–28.4%]), and effector (0.2% [IQR, 0.1%–0.5%] vs. 0.3% [IQR, 0.0%–2.2%]) phenotypes at week 0 did not differ significantly between the NR and control groups (P > .5, for all) (figure 1A). Similarly, the proportion of CD8⁺ T cells with naive (8.6% [IQR, 4.2%–16.9%] vs. 16.5% [IQR, 13.9%–22.4%]; P = .06), central memory (51.3% [IQR, 38.8%–56.3%] vs. 45.1% [IQR, 35.6%–53.7%]; P = .5), effector memory (21.7% [IQR, 15.7%–35.5%] vs. 20.8% [IQR, 18.0%–27.3%]; P = .7) and effector (14.8% [IQR, 8.6%–19.9%] vs. 13.8% [IQR, 9.1%–18.6%]; P = 1.0) phenotype did not differ significantly at baseline between NRs and control subjects (figure 1B), with the exception of a trend toward a lower proportion of naive CD8⁺ T cells in the NR group.

TREC levels and T cell proliferation in CD4⁺ and CD8⁺ T cells at baseline. The TREC content in CD4⁺ and CD8⁺ T cells was measured at weeks 0 and 24 in both groups. At baseline, NRs had significantly lower TREC content than did control subjects in both CD4⁺ (15,700 [IQR, 6800–29,000] vs. 33,400 [IQR, 23,200–50,700] copies/1×10⁶ T cells; P = .03) and CD8⁺ T cells (5200 [IQR, 1900–17,700] vs. 17,900 [IQR, 10,200–37,200] copies/1×10⁶ T cells; P = .03) (figure 2A). Similar findings were observed when TREC levels were measured per microliter of blood (in CD4⁺ T cells, 4.0 [IQR, 3.4–17.0] vs. 15.9 [IQR, 8.5–24.7] copies/µL; P = .05) (in CD8⁺ T cells, 8.1 [IQR, 3.5–16.8] vs. 24.5 [IQR, 8.6–42.3] copies/µL; P = .03).

T cell proliferation was measured by intracellular Ki67. Consistent with the TREC data, NRs had higher levels of Ki67 expression than did control subjects in both CD4⁺ (9.9% [IQR, 6.6%–11.7%] vs. 5.8% [IQR, 4.9%–6.8%]; ) and CD8⁺ P = .03 (9.2% [IQR, 3.3%–12.3%] vs. 3.9% [IQR, 2.8%–4.6%]; P = .03) T cells (figure 2B). Similar findings were observed when activation of T cells was measured by surface CD38 expression (figure 2C). Expression of PD-1, a marker associated with activation and exhaustion of T cells [13], was higher on CD4⁺ T cells from NRs than on those from control subjects at baseline (16.4% [IQR, 8.5%–20.5%] vs. 9.3% [IQR, 7.8%–11.6%]; P = .06). Similarly, coexpression of HLA-DR with CD38 on CD4⁺ T cells, another marker of T cell activation, was higher for NRs than for control subjects at baseline (27.5% [IQR, 16.8%–34.9%] vs. 19.9% [IQR, 16.8%–27.7%]). Study of CD25 (IL-2 receptor α chain), CD127 (IL-7 receptor α chain), HLADR, Bcl-2, and intracellular cytotoxic T lymphocyte-associated protein 4 expression in CD4⁺ and CD8⁺ T cells, as well as PD-1 or HLA-DR/CD38 coexpression on CD8⁺ T cells, showed no statistically significant differences between the 2 groups at baseline (data not shown).

Expression of FoxP3, a marker of regulatory T (T_reg) cells [14], in CD4⁺ T cells did not differ at baseline between NRs and control subjects (8.2% [IQR, 5.3%–10.4%] vs. 7.1% [IQR, 6.0%–9.6%]; P = .9). The percentage of CD4⁺ T cells with a central memory phenotype expressing FoxP3, which is typically the phenotype of conventional functional memory Treg cells [15–17], was significantly higher in NRs than in control subjects (17.7% [IQR, 15.4%–23.6%] vs. 12.6% [IQR, 7.6%–13.5%]; P = .004) (figure 2D).

Expression of CD25 on CD4⁺ T cells after IL-2 administration. Increases in CD25 expression on CD4⁺ T cells is known to occur after IL-2 immunotherapy, and previous studies have shown a correlation between increases in CD25 expression and CD4⁺ T cell expansions [4, 11]. We sought to determine whether changes in CD25 expression on CD4⁺ T cells after IL-2 administration differed between the 2 groups. As shown in table 2, the percentage of CD4⁺ T cells expressing CD25 increased by 11.8% (P < .001) in NRs and by 28.5% in control subjects (P < .001) from week 0 to 24 (table 2). The increases in NRs were significantly smaller than those in control subjects (P = .001).

Expression of FoxP3 in CD4⁺ T cells with a central memory phenotype did not change after IL-2 immunotherapy in either group. There was a significant increase from baseline in the percentage of naive CD4⁺ T cells expressing FoxP3 in both groups, but this increase was significantly smaller in NRs than in control subjects (22.9% [IQR, 8.6%–26.6%] vs. 39.6% [IQR, 25.8%–51.1%]; P = .008) (figure 3).

Immune activation and TREC levels per 1×10⁶ T cells after IL-2 immunotherapy. T cell proliferation (as measured by expression of Ki67) and activation (as measured by expression of CD38) decreased on T cells after IL-2 administration (week 24) in both groups (table 2). The TREC level per 1×10⁶ T cells decreased significantly at week 24 in CD4⁺ T cells of control subjects (−13,100 copies/1×10⁶ T cells; P = .002) and to a much lesser degree in CD8⁺ T cells of both control subjects (−3400 copies/1×10⁶ T cells; P = .06) and NRs (−1700 copies/1×10⁶ T cells; P = .04). No significant change was observed in the CD4⁺ T cells of the NR group (−1600 copies/1×10⁶ T cells; P = .6) (figure 4A and table 2). In contrast, the TREC level per microliter of blood did not change significantly in control subjects in either CD4⁺ or CD8⁺ T cells. In NRs, there was a decrease in TREC level per microliter in CD8⁺ T cells observed at week 24 relative to baseline (P = .007) (figure 4B).

Discussion

Intermittent IL-2 immunotherapy in patients with HIV infection leads to CD4⁺ T cell expansions that are associated with increases in CD4⁺CD25⁺ T cell levels and decreases in T cell proliferation and activation [5, 11]. In the present study, it was shown that increased baseline immune activation impeded CD4⁺ T cell expansions induced by IL-2. It was also shown that, despite lack of increases in CD4⁺ T cell numbers in the NR group, there was clear evidence of a biological effect of IL-2, as shown by the increase in CD25 expression on CD4⁺ T cells.

Multiple studies have shown that IL-2 can lead to a significant, selective, and sustained CD4⁺ T cell expansion. Recent data showed that newly emerging CD4⁺ T cells are the product of peripheral expansion and do not represent recent thymic emigrants, are polyclonal with some intermediate features between naive and memory cells, and have elevated levels of FoxP3 expression with weak to moderate suppressive potential [5]. Ex vivo and in vivo labeling studies have confirmed that, although intense proliferation occurs during IL-2 administration, the CD4⁺ T cell expansions are supported long term by decreased turnover and improved survival of naive and central memory CD4⁺ T cells [7]. Clinical studies have also shown that some patients fail to experience CD4⁺ T cell expansions after IL-2 immunotherapy or do so only after additional cycles beyond the 3–6 that are most typically used during the conventional duration of clinical trials.

In the present study, patients who had poor CD4⁺ T cell responses to IL-2 were studied and matched by baseline CD4⁺ T cell count to control subjects, who were patients participating in the same study who had good CD4⁺ T cell responses to IL-2. A significantly higher level of immune activation and T cell proliferation in both CD4⁺ and CD8⁺ T cell pools were noted in NRs compared with control subjects. In agreement with this observation, significantly lower TREC content was observed in CD4⁺ and CD8⁺ T cells of poor responders compared with control subjects. Despite the above findings, the proportion of T cells with a naive phenotype did not differ at baseline between the 2 groups, suggesting that the activation state of the CD4⁺ T cell pool may not be adequately addressed by its memory/naive subset composition. It is unclear whether these differences, in the presence of similar peripheral CD4⁺ T cell count, represent longer infection times or interindividual variability with respect to immune activation responses or set point.

Regarding the proportion of CD4⁺ T cells bearing a T_reg phenotype, a higher proportion of T_reg cells was noted in the central memory pool at baseline in those patients who had evidence of higher immune activation and did not have CD4⁺ T cell responses after IL-2 administration, suggesting a compensatory mechanism. It is known that functional T_reg cells have a central memory phenotype, whereas CD4⁺ T cells with a naive phenotype that express FoxP3 and/or CD25 may represent precursor of fully functional T_reg cells and may have weaker T_reg cell function. Cell availability precluded functional assays in our study. Future functional studies will be necessary to clearly delineate the role played by these different phenotypes in HIVinduced immune activation and their potential contribution to the expansions induced by IL-2 or other immune-based therapies. The clinical implications of these findings are important for new clinical trial designs addressing the potential role played by IL-2 in patients with HIV viremia either receiving or not receiving antiretrovirals. It is known that immune activation is directly associated with viral load [18, 19]. Thus, one would predict that HIV viremia, after treatment interruption or in patients with inadequately controlled viral load, would likely hinder CD4⁺ T cell responses to IL-2. In fact, in a study of antiretroviral interruption after an IL-2 cycle, it was shown that the CD4⁺ T cell responses during a viremic IL-2 cycle were blunted, compared with those observed in the same individuals receiving ART [20]. Similarly, the CD4⁺ T cell responses of patients receiving IL-2 without antiretrovirals appear blunted compared with those achieved under HAART [21]. For patients already receiving IL-2, optimization of ART should be attempted if adequate IL-2 responses have not been observed.

Interestingly, in the present study IL-2 itself decreased immune activation and T cell proliferation and increased FoxP3 expression in phenotypically naive CD4⁺ T cells even when CD4⁺ T cell counts did not increase. This may explain why some patients eventually respond to IL-2 after additional cycles and even suggests the possibility that a potential CD4⁺ T cell loss during that period was actually averted. That TREC content in CD4+ T cells did not change in people who did not experience CD4⁺ T cell expansions could be explained by decreased T cell proliferation relative to that in control subjects or preferential survival of TREC-enriched T cell subsets. Labeling studies have clearly shown that intense T cell proliferation is seen with IL-2 regardless of the long-term effect on CD4⁺ T cell expansion, making the latter explanation more plausible [7]. In agreement also with in vivo labeling studies [7], a dilution of TRECs in CD8⁺ T cells was observed, suggesting that, despite CD8⁺ T cell proliferation, there is no effect on CD8⁺ T cell survival and thus no CD8⁺ T cell expansion with IL-2.

Finally, the present data are also consistent with the notion that increased immune activation in the T cell pool can in dependently affect disease prognosis and response to therapy in HIV infection, as has been shown previously [22–25]. It is known that immune activation can have deleterious effects on CD4⁺ T cell function even when CD4⁺ T cell numbers are restored with HAART [26]. In the present study, increased activation of T cells was documented by several different markers, including Ki67, CD38, and PD-1. Increased FoxP3 expression on CD4⁺ T cells with a central memory phenotype was also observed in NRs, suggesting the possibility of a compensatory mechanism. Although activated T cells tend to express lower levels of IL-7 receptors—such that their survival may be affected, hampering their peripheral expansion potential [27, 28]—in the present study a relationship between expression of the IL-7 receptor and IL-7 levels was not seen. It is feasible that CD127 down-regulation is a late event, occurring after loss of cells with a naive phenotype; in our cohort, the proportion of naive T cells was preserved.

In summary, we have shown that immune activation can adversely affect IL-2-induced CD4⁺ T cell expansions. Even in the absence of CD4⁺ T cell increases, IL-2 seems to induce biological changes in the T cell pool, including increases in CD4⁺CD25⁺ T cell levels and decreases in immune activation and proliferation. These findings underscore the limitations of the T cell pool to peripherally expand in settings of persistent immune activation and the potential influence of HIV viremia in successes and failures of immune-based therapies.