Introduction
Regulatory T cells (T
regs) are Foxp3
+CD25
highCD4
+ T cells with diverse immunosuppressive functions. Subsets of T
regs include natural T
regs (nT
regs), that are thymus-derived but undergo further expansion in peripheral tissues, and induced T
regs (iT
regs) that are converted from conventional T cells (T
cons) in the periphery [
1‐
3]. Both subsets have been shown to suppress autoreactive lymphocytes and thus to limit the magnitude of innate and adaptive immune responses [
2,
4,
5]. Accordingly, impaired T
reg function aggravates autoimmune diseases while T
reg-mediated immunosuppression may inhibit pathogen clearance and promote chronic infection [
6,
7]. In addition to controlling autoimmunity, T
regs have been ascribed a role as mediators of cancer-related immunosuppression. Studies in murine models show that T
regs accumulate in several forms of experimental cancer and that depletion of T
regs or strategies to target their immunosuppressive features reduce cancer growth [
8,
9]. In many solid human cancers, T
regs accumulate in the tumor microenvironment and their presence typically, albeit not invariably, heralds advanced disease and poor survival [
10‐
13].
Acute myeloid leukemia (AML) is characterized by rapid expansion of immature myeloid cells in bone marrow and other organs [
14]. In AML, the malignant clone is reportedly controlled by cellular immunity, including natural killer (NK) cells and subsets of cytotoxic (CD8
+) T cells [
15]. While relatively little is known about the role of T
regs for the efficiency of anti-leukemic immunity in AML [
16,
17], several other immunosuppressive pathways of relevance to the course of disease have been described [
18‐
20] including immunosuppression exerted by NOX2-derived reactive oxygen species (ROS) released from myeloid cells [
21]. Under conditions of NOX2-related oxidative stress, targeting of ROS formation using the NOX2 inhibitor histamine dihydrochloride (HDC) upholds NK cell and T cell function and improves the efficiency of NK- and T cell-activating agents such as interleukin-2 (IL-2) [
22‐
25]. Monotherapy with IL-2 has yielded disappointment in several clinical trials in AML [
26‐
31]. However, phase III trial results showed that the combination of HDC and low-dose IL-2 improves the leukemia-free survival (LFS) of AML patients in complete remission (CR) after chemotherapy [
32], thus supporting the clinical relevance of NOX2-mediated immunosuppression in AML.
The IL-2 component of the HDC/IL-2 regimen may expand T
regs as these cells express high-affinity IL-2 receptors (CD25) and rely on exogenous IL-2 for proliferation [
33,
34]. Treatment with IL-2 has been shown to increase the population of T
regs and reduce graft-versus-host manifestations in cancer patients receiving allogeneic stem cell transplants (allo-SCT) [
35‐
38]. It is thus conceivable that IL-2-driven T
reg expansion may limit the anti-leukemic efficiency of HDC/IL-2 immunotherapy. For the present study, we monitored T
reg number and function in AML patients in first CR who received HDC/IL-2 for relapse prevention in a phase IV trial. Our results imply that treatment with HDC/IL-2 entails pronounced accumulation of nT
regs in blood and that aspects of T
reg function are relevant to relapse risk in AML.
Patients, materials and methods
Patients, study design and objectives
The Re:Mission trial (NCT01347996, registered at
http://www.clinicaltrials.gov) was a single-armed multicenter phase IV study that enrolled 84 patients (age 18–79) with confirmed AML in first CR who were not eligible for allo-SCT. The patients received ten consecutive 21-day cycles of HDC/IL-2 during 18 months or until relapse or death. Each cycle comprised 0.5 mg histamine dihydrochloride (HDC; Ceplene
®) and 16,400 U/kg human recombinant IL-2 (aldesleukin) that were administered by subcutaneous injection twice daily. The off-treatment periods in cycle 1–3 were 3 weeks, while the off-treatment periods between cycle 4–10 were extended to 6 weeks. All patients were followed for at least 24 months after enrollment. Fourteen patients discontinued prematurely from the study and were censored at the last captured follow-up date. The exclusion criteria for enrollment were identical to those employed in a previous phase III trial [
32]. The primary endpoints comprised assessment of the quantitative and qualitative pharmacodynamic properties of HDC/IL-2, including monitoring of T and NK cell phenotypes before and after treatment cycles while analyses of aspects of immunity versus outcome (LFS and overall survival; OS) were performed post hoc. Patient characteristics, including details regarding previous induction and consolidation therapy and risk group distribution are accounted for in previous publications [
39‐
41] and in Table
1. The trial was approved by the Ethical Committees of each participating institution and all patients gave written informed consent before enrollment.
Table 1
Patient characteristics
Sex |
Female | 44 (52) |
Male | 40 (48) |
Risk group |
Favorable risk | 34 (40) |
Intermediate I | 25 (30) |
Intermediate II | 13 (15) |
High risk | 7 (8) |
Not done | 5 (6) |
Karyotype |
Normal | 44 (52) |
Favorable | 14 (17) |
Unfavorable | 7 (8) |
Other | 15 (18) |
Not done | 4 (5) |
Mutation status |
NPM1 |
n = 69 25 (36) |
FLT3 |
n = 72 6 (8) |
CEBPA |
n = 42 3 (7) |
Induction courses |
1 | 63 (75) |
>1 | 21 (25) |
Consolidation courses |
0–2 | 41 (49) |
>2 | 43 (51) |
Isolation of PBMCs from healthy donors and patient samples
Buffy coats from healthy donors were obtained from the Blood Center at the Sahlgrenska University Hospital, Gothenburg, Sweden. To remove erythrocytes, the blood was mixed at a 1:1 ratio with 2% dextran and left to sediment. The upper phase was transferred to tubes containing Ficoll/Lymphoprep (Alere AB, Lidingö, Sweden) and peripheral blood mononuclear cells (PBMCs) were isolated by density gradient centrifugation. The PBMCs were cryopreserved until further use. Peripheral blood was collected from patients in the Re:Mission trial before and after the first and third treatment cycles, i.e., cycle 1, day 1 (C1D1) and cycle 1, day 21 (C1D21), cycle 3, day 1 (C3D1) and cycle 3, day 21 (C3D21). Patient PBMCs were isolated and cryopreserved at local sites and shipped on dry ice to the central laboratory (the TIMM Laboratory, Sahlgrenska Cancer Center, University of Gothenburg, Sweden) for analysis.
Staining and flow cytometry
Cryopreserved samples were quickly thawed, washed and stained with LIVE/DEAD fixable yellow stain (Life technologies, Grand Island, NY, USA). Thereafter, cells were washed and incubated with an antibody cocktail for surface markers in PBS containing 0.5% BSA and 0.1% EDTA or in Brilliant stain buffer (BD Biosciences, Stockholm, Sweden). The following anti-human monoclonal antibodies were purchased from BD Biosciences: CD3-FITC (HIT3a), CD3-Brilliant Violet 711 (UCHT1), CD4-APC-H7 (RPA-T4), CD8-PerCP-Cy5.5 (SK1), CD14-FITC (MϕP9), CD25-Brilliant Violet 421 (M-A251), CD56-PE-Cy7 (NCAM16.2) and CD127-AF647 (HIL-7R-M21). CTLA-4-PE-Cy7 (L3D10) was obtained from Biolegend (San Diego, CA, USA) and CD14-Qdot655 (TüK4) from Life Technologies. For intracellular staining with Foxp3-PE (3G3; Miltenyi Biotec, Auburn, CA) and Helios-AF647 (22F6; BD Biosciences), cells were fixed and permeabilized using the Foxp3 fixation/permeabilization kit (eBioscience, San Diego, CA, USA) according to the manufacturer’s protocol. A 4-laser BD LSRFortessa SORP flow cytometer (405, 488, 532, and 640 nm; BD Biosciences) was employed to analyze samples. Data analysis was performed using the FlowJo software, version 7.6.5 or later (TreeStar, Ashland, OR, USA), or FACSDiva software, version 6 or later (BD Biosciences). Samples with less than 25% viability were excluded.
Blood samples were available from 81 out of 84 patients. Differential counts of whole blood were performed at local sites and were utilized to calculate absolute counts of T
regs in blood. Notably, the definition of T
regs in this study was restricted to Foxp3
+CD25
highCD4
+ cells. All available samples were analyzed for T
reg content. If an analysis failed according to pre-defined criteria (experimental failure, few cells, poor cellular viability), a second sample was thawed for re-analysis. If the second attempt also failed to generate data, the sample was excluded from analysis. A thorough analysis of expression of T
reg markers (including CTLA-4 and Helios) was performed in 25 randomly selected patients. These patients were largely representative of all participating patients in terms of age (mean age for selected group 57.7 years (23.8–76.5 years) vs. mean age for all patients 58.6 years (19–77 years), risk group classification according to recommendations by the European LeukemiaNet [
42] [among the selected patients 6 (24%) belonged to the favorable group, 14 (56%) to the intermediate group and 3 (12%) to the adverse group, 2 (8%) not done, whereas among all patients 34 (40.5%) belonged to the favorable group, 38 (45.2%) to the intermediate group and 7 (8.3%) to the adverse group, 5 (6%) not done] and French American British (FAB) classification (data not shown). All successfully analyzed samples, according to the pre-defined criteria stated above, were included in this report.
Treg methylation analysis
T
regs (CD4
+CD14
−CD25
hiCD127
low) were sorted from blood samples recovered at the end of treatment cycle 3 (C3D21) and from healthy subjects. Sorted cells (at least 40,000 cells per assay) were washed before being frozen in 200 µl PBS. The DNA methylation status of 15 CpG-motifs within the T
reg-specific demethylated region (TSDR) was analyzed by bisulphite sequencing performed by Epiontis GmbH (Berlin, Germany) as previously described [
43]. Only male subjects were included in analyses of T
reg methylation status cells since the
FOXP3 gene locus is located on the X-chromosome [
44] and X-chromosome inactivation in females would likely influence results.
Treg suppression assay
Patient samples collected on C3D21 with a Treg content of 15–40% of the CD4+ population were used in Treg suppression assays ex vivo. PBMCs collected from healthy donors served as control. Cells were stained with anti-human monoclonal antibodies as described above. Tregs (CD4+CD14−CD25hiCD127low) and conventional CD4+ T cells (Tcons; CD4+CD14−CD25lowCD127hi) were sorted on a 3-laser BD FACSAriaIII flow cytometer (405, 488 and 640 nm; BD Biosciences). The gating strategy is shown in Supplementary Fig. 2. The sorted Tcons were stained with CellTrace™ violet (Life Technologies) and 35,000 cells per well were seeded together with 2 µg/ml soluble anti-CD28, in X-VIVO™ 15 serum-free medium (Lonza Group Ltd, Basel, Switzerland) to a 384-well plate coated with anti-CD3 (OKT3; eBioscience). An equal number of Tregs (35,000/well) was added to half of the wells. After 4–5 days of culture the proliferation of Tcons was determined by measuring the intensity of the CellTrace™ violet staining on an LSRFortessa SORP flow cytometer (BD Biosciences).
Quantitative PCR telomere length assay
T
regs (CD4
+CD25
hiCD127
low) were sorted from patient blood samples recovered at C3D1 and C3D21 or from healthy controls. Cells were sorted into 96-well plates (Life Technologies) for direct cell lysis and kept at −80 °C until analysis. Optimally, four technical replicates of 400 cells/well were obtained from all blood samples. Protease from
Streptomyces griseus (2 μg; Sigma-Aldrich) diluted in PBS (Life Technologies) was added to each well followed by incubation at 37 °C for 10 min and enzyme inactivation at 95 °C for 15 min. The plates were centrifuged at 3000 rpm for 5 min. Quantitative PCR (qPCR) was performed using a CFX384 Touch Real-Time PCR Detection System (Bio-Rad). Primers designed by Cawthon [
45] were used for amplification of a short fixed-length product at a copy number proportional to telomere length, and of the single copy gene albumin, in separate wells. Each 10-µl qPCR reaction contained 1X TATAA SYBR GrandMaster Mix (TATAA Biocenter), 400 nM of each primer, and 2 µl protease-treated DNA. Each technical replicate was assayed in duplicate. The thermal cycling profile was 95 °C for 1 min, 2 cycles of 94 °C for 15 s and 49 °C for 15 s, and 40 cycles of amplification (94 °C for 15 s, 62 °C for 10 s and 74 °C for 15 s). Formation of the correct PCR products was confirmed by melting-curve analysis. Relative telomere lengths were determined by normalizing the telomere qPCR signals against signals observed in the corresponding albumin gene assays.
Statistical analyses
Single comparisons of Treg, Tcon and NK cell phenotypes were performed by paired Student’s t test in accordance with the pre-defined statistical plan. Patients were dichotomized by the median Treg cell number, frequency and telomere length for analyses of LFS (log-rank test). LFS was defined as the time in days from start of immunotherapy with HDC/IL-2 to relapse or death from any cause using data available at the trial closing date (October 13, 2014), i.e., when patients had been followed-up for at least 24 months. Cox multivariable regression analysis that included age and number of induction cycles as potential confounders was utilized to further determine the impact of Treg distribution on LFS. Statistical analyses were performed using Graphpad Prism (Graph Pad Software, La Jolla, CA, USA) and IBM SPSS Statistics (IBM Corp., Armonk, NY, USA) software. All indicated p values are two-sided.
Discussion
Upon diagnosis, AML patients receive induction chemotherapy aiming to achieve CR, which is defined as the microscopic disappearance of leukemic cells and the return of normal hematopoiesis. Despite additional courses of chemotherapy (consolidation), relapse in CR is common and significantly explains why the long-term survival of adult AML patients remains in the range of 30–40% [
14]. A large body of evidence, including the graft-versus-leukemia reaction that mediates relapse prevention after allo-SCT, implicates functions of cytotoxic T cells and NK cells in controlling the malignant clone in AML [
15,
39‐
41]. The purported role of cell-mediated immunity for the surveillance of leukemic cells in AML has inspired the development of immunotherapeutic strategies, in particular for patients in CR who harbor a minimal yet potentially life-threatening burden of leukemia (reviewed in [
15]).
HDC/IL-2 is currently the only documented effective non-transplant immunotherapy for relapse prevention in AML beyond the chemotherapy phase [
15,
32]. As the IL-2 component of this regimen may induce T
regs [
35‐
38] the present study was designed to determine the magnitude of T
reg induction during immunotherapy, the origin and function of accumulating T
regs and the potential impact of T
regs on relapse risk. We therefore analyzed serial blood samples from patients in first CR participating in the phase IV Re:Mission trial (
n = 84) who received ten 3-week cycles of HDC/IL-2 after the completion of consolidation chemotherapy. The frequency of T
regs at the onset of immunotherapy was within or below the range in healthy subjects (3.1 ± 2.2% of CD4
+ T cells; mean ± SD), which is in agreement with a recent study of AML patients in CR [
49]. T
reg counts increased considerably during cycles of HDC/IL-2, in particular during the first treatment cycle. At the end of the first cycle, T
regs typically comprised 15–25% of the CD4
+ cell population in blood. These results concur with previous reports of T
reg induction during treatment of cancer patients with IL-2 [
35,
37,
38] and is likely explained by IL-2 acting via the high-affinity IL-2 receptor CD25 that is constitutively expressed by nT
regs. However, randomized comparisons are required to exclude the possibility that the HDC component contributed to T
reg induction. While we did not have access to bone marrow samples in this study, we reason that a similar increase in T
reg counts is likely to occur also in the bone marrow, since the number of T
regs in blood and bone marrow were previously reported to be highly correlated [
50].
We then asked whether the expanded population of T
regs showed stable or transient expression of Foxp3. In these cells, the TSDR in the
FOXP3 gene locus was highly demethylated implying stable Foxp3 expression and suggesting that the reduction of T
reg counts between cycles was explained by T
reg apoptosis rather than the T
regs being reprogrammed into T
cons. Moreover, there was no increase in the number of T
cons during or between treatment cycles (data not shown). The thymus-derived nT
regs are known to have a demethylated TSDR in the
FOXP3 gene locus while this region generally is more methylated in iT
regs. With the precaution that the TSDR region may become demethylated also in iT
regs in response to antigen stimulation in the presence of IL-2 [
51], we propose that the expanded T
regs were mainly derived from proliferating nT
regs.
We observed that the T
regs accumulating at the end of a HDC/IL-2 treatment cycle expressed elevated levels of CTLA-4, which reportedly contributes to the immunosuppression exerted by these cells [
48]. Also, the expanded T
regs suppressed the proliferation of T
cons in co-culture assays ex vivo. While it is conceivable that T
reg induction may dampen the development of cell-mediated immunity of relevance to elimination of residual leukemia, our initial analysis did not reveal associations between the magnitude of T
reg induction during initial cycles of immunotherapy and clinical outcome. It is conceivable, however, that the lack of association between T
reg induction and clinical outcome may result from effects of HDC—a NOX2 inhibitor—on the immunosuppressive properties of T
regs. This possibility is supported by a previous study showing that immunosuppressive features of CD8
+ T
regs rely on functional NOX2 [
52]. However, monotherapy with IL-2 has been reported to increase T
reg counts and limit the extent of graft-versus-host disease (GvHD) after allo-SCT in cancer patients, apparently without negatively affecting survival [
35]. In accordance, results presented by Martelli et al. implied that allo-transplanted patients with acute leukemia who received donor-derived T
regs in conjunction with T
cons for protection against GvHD did not show increased relapse risk [
53].
A more detailed analysis of T
reg kinetics during treatment with HDC/IL-2 revealed that aspects of T
reg function may indeed impact on clinical outcome. We observed that the magnitude of T
reg induction was frequently blunted in later treatment cycles and that a reduced T
reg accumulation in cycle 3 weakly but significantly prognosticated low relapse risk, thus supporting that sustained presence of T
regs may adversely impact on prognosis. In contrast, the induction of NK cells in blood remained largely stable throughout cycles of immunotherapy. The mechanisms underlying the different kinetics of NK cell and T
reg induction should be further studied. However, in people over the age of 45 the supply of thymic nT
regs is minimal and is sustained mainly by peripheral proliferation [
54]. We thus speculate that the supply of nT
regs may become exhausted during repeated cycles of immunotherapy, in contrast to the bone marrow supply of NK cells. In support of this assumption, we observed a significantly reduced accumulation of T
regs in later treatment cycle only in patients >45-years-old (Supplementary Fig. 1a).
The proliferation of normal somatic cells is limited by the length of telomeres, which typically progressively shorten with increasing age [
55]. Accordingly, we observed a significant correlation between short T
reg telomere length and age among the participating patients (Supplementary Fig. 1b). Despite high age being a dominant predictor of relapse risk in AML [
56], short T
reg telomeres at the end of a treatment cycle was observed mainly in older patients and significantly prognosticated favorable LFS. In agreement with the above-referenced hypothesis of T
reg exhaustion during immunotherapy, we propose that short T
reg telomere length may reflect a reduced capacity of nT
regs to undergo proliferation and, hence, exert immunosuppression in subsequent treatment cycles.
While the preliminary nature of these findings should be emphasized, we speculate that immunosuppressive nT
regs may be targeted for improved anti-leukemic efficacy of HDC/IL-2 immunotherapy. This view gains support from previous studies in which T
regs were targeted during immunostimulation with IL-2 in experimental leukemia using the combination of anti-CD25, aiming to deplete T
regs, and IL-2. This combination significantly improved the survival of leukemia-bearing mice over either treatment alone [
57]. In further support for a role of T
regs in AML immunotherapy, Bachanova et al. reported that patients with relapsed or refractory AML showed encouraging CR rates and disease-free survival following depletion of host T
regs prior to the adoptive transfer of haploidentical NK cells and IL-2 [
58]. Targeting T
regs, for example by use of antibodies blocking CTLA-4, may thus be considered in IL-2-based AML immunotherapy. An alternative approach to minimize a potential negative impact of T
regs may be to replace the IL-2 component with modified IL-2 variants or IL-15 that activate anti-leukemic effector cell populations with reduced or absent expansion of CD25
high expressing T
regs [
59‐
61].