Background
Type 1 diabetes (T1DM) is an emerging medical problem, since there is no causal treatment and patients inevitably develop full onset of the disease, e.g., in Poland the consequent morbidity doubles every 10 years [
1]. The majority of patients are children and the initial manifestation can often be severe, including deep ketoacidosis or coma. It is, therefore, important to investigate novel treatments, aiming at early intervention, while a significant mass of β-cells is still present and can be preserved.
A common consensus exists that the disease develops as a result of the attack of autoaggressive T-cells that infiltrate pancreatic islets and destroy insulin-producing β-cells [
2]. This autoaggression is usually unleashed when suppressive subsets, such as CD3
+CD4
+FoxP3
+ T regulatory cells (Tregs), are somehow impaired [
3]. Indeed, the adoptive transfer of Tregs was confirmed in animal models as an effective way to stop or delay the progression of the disease [
4]. Translational studies in humans seem to confirm this observation, however, the disease still progresses in patients treated with Tregs preparation. It is, therefore, necessary to identify the factors that influence the efficacy of this therapy in the clinical setting.
Starting from 2009, therapy using T regulatory cells moved to the clinical stage and its efficacy is currently being assessed in various conditions, including T1DM [
5]. Our group performed several such studies, including that for the treatment of T1DM [
5‐
8]. In this paper, apart from the clinical background, we will present some immunity studies in order to identify the factors that possibly influence the efficacy of the adoptive transfer of Tregs in T1DM.
Methods
Protocol and treatment
This was an open-labeled study conducted according to the Declaration of Helsinki principles and was approved by the Ethics Committee of the Medical University of Gdańsk, Poland (NKEBN/8/2010 with amendments). The trial was registered at the Current Controlled Trials database:
http://www.controlled-trials.com/ISRCTN06128462 (Additional file
1). Written informed consent was received from parents of all the participants and from the patients themselves, if above 16-years of age.
As described in earlier reports [
7,
8], 12 Caucasian children from the Polish population with recently diagnosed T1DM were treated with ex vivo expanded autologous Tregs. The general health and metabolic status of the treated individuals were followed for 24 months after inclusion to the study along with those of ten untreated, control patients matched for age, sex and disease duration. The main inclusion criteria were: having autoimmune T1DM diagnosed within 2 months; the presence of at least one type of anti-islet autoantibody anti-GAD, anti-IA2, IAA, or ICA; age–5 to 18 years; fasting plasma C-peptide levels >0.4 ng/mL and proper management of diabetes. The control group was recruited among patients who fulfilled the same criteria, but did not qualify for admission to the treated group due to inadequate venous access (Table
1).
Table 1
Clinical characteristics of the patients
Age (years) [median; min–max] | 12.2; 8–16 | 11.5; 7–16 |
BMI [median; min–max] | 17.1; 12.5–23.5 | 16.7; 14.2–20.8 |
Polydipsia at diagnosis (number of patients) | 5 | 8 |
Polyuria at diagnosis (number of patients) | 5 | 3 |
Loss of weight at diagnosis (number of patients) | 4 | 3 |
pH at diagnosis (capillary blood) [median; min–max] | 7.40; 7.36–7.42 | 7.39; 7.35–7.53 |
pO2 at diagnosis (capillary blood—mmHg) [median; min–max] | 69.3; 24.1–88.0 | 69.5; 56.0–86.6 |
pCO2 at diagnosis (capillary blood—mmHg) [median; min–max] | 39.6; 28.0–46.9 | 38.0; 24.0–40.7 |
HCO3 at diagnosis (capillary blood—mmHg) [median; min–max] | 24.15; 18.8–27.4 | 23.6; 21.3–25.2 |
Acid/base balance at diagnosis (BE—mEq/l) [median; min–max] | 0.05; −7.8–3.2 | −0.5; −3.8–0.9 |
Sat02 at diagnosis (capillary blood—%) [median; min–max] | 94.1; 90.2–97.3 | 95.4; 92.4–97.2 |
Anti-GAD65 antibody (number of patients) | 9 | 9 |
ICA (number of patients) | 7 | 5 |
IAA (number of patients) | 7 | 4 |
Tregs were isolated from the patients’ peripheral blood with a GMP-compliant FACS sorter (Influx; BDBiosciences, USA). The purity of Tregs after sorting was ≈98% (range 97–100%). The expansion was performed under GMP conditions and according to our previously described protocol using anti-CD3/anti-CD28 beads, interleukin 2 (IL-2) and autologous serum. The final product on release kept the FoxP3 expression above 90% [median (min.–max) = 91% (90–97)] [
9].
The dose-escalation scheme of Tregs administration was: 10 × 106 of Tregs/kg b.w. in a single infusion (three patients), 20 × 106 of Tregs/kg b.w. in a single infusion (three patients), and a total of 30 × 106 of Tregs/kg b.w. in two infusions (six patients), with the second dose being administered 6–9 months after the first one. Two patients were lost to follow-up at +6 and +9 months, while ten patients completed the trial.
The primary endpoints of the trial were safety and remission defined as daily dose of insulin (DDI) ≤0.5 UI/kg b.w. and fasting C-peptide levels >0.5 ng/mL 1 year after recruitment. Secondary endpoints included the immune background of the patients treated with the preparation of Tregs.
Fasting C-peptide levels, fasting glucose, HbA1c and T1DM autoantibody [glutamic acid decarboxylase autoantibody (anti-GAD65), insulin autoantibody (IAA), insulin antigen 2 antibody (IA2) and zinc transporter 8 autoantibody (anti-ZnT8)] levels were measured during a 24-month-long follow-up at different time points, as previously described [
7,
8]. The mixed meal tolerance test (MMTT) was performed according to standard criteria on the day of the 24th month of follow-up [
10].
Immune phenotyping was performed using a seven-color panel: CD3/CD4/CD25/CD127/CD45RA/CD62L/FoxP3. Two phenotypes of Tregs were analyzed: CD3
+CD4
+FoxP3
+ and CD3
+CD4
+CD25
highCD127
− T-cells. The gate CD25
highCD127
− from the later population was also used to assess the content of FoxP3
+ T-cells in order to estimate an overlap between the two phenotypes of Tregs. Finally, CD3
+CD4
+FoxP3
+ T-cells were subdivided into naive CD3
+CD4
+FoxP3
+CD62L
+CD45RA
+ (Tn) Tregs, central memory, CD3
+CD4
+FoxP3
+CD62L
+CD45RA
− (Tcm) Tregs, and effector memory, CD3
+CD4
+FoxP3
+CD62L
−CD45RA
− (Tem) Tregs [
9].
The following anti-human monoclonal antibodies were used in this procedure (fluorochrome/class/clone): anti-CD3 (PacificBlue/IgG1/UCHT1), anti-CD4 (PerCP/IgG1/RPA-T4), anti-CD25 (PE/IgG1/M-A251), anti-CD127 (FITC/IgG1/hIL-7R-M21) and, anti-CD45RA (PE-Cy7/IgG1/L48). All of the antibodies were purchased from BDBiosciences, Poland. Anti-CD62L (APC-Cy7/IgG1/3B5) was supplied by Invitrogen, USA, and the FoxP3 staining kit by eBioscience, USA.
Serum levels of IFNγ, VEGF, TNFα, IL1, IL2, IL4, IL6, IL8, IL10 and, IL12 were measured with the Luminex Bead based Multiplex Assay (ebioscience, USA), while BAFF, TGFβ, and, IL17 were measured with Quantikine High Sensitivity ELISA kit (R&D Systems, USA). All assays were performed according to the manufacturers’ instructions.
IL2 dependency tests
Samples of cells from Tregs expansions administered to the patients and autologous CD3+CD4+CD127+ conventional/effector T-cells (Teffs) expanded along with Tregs were cultured for additional 8 days after the release of Treg products to the clinic. Tregs and Teffs were cultured separately or co-cultured in a 1:1 ratio, 1 × 106 cells/well in 24-well plates, in a 5% CO2 atmosphere, at 37.0 °C, in the following concentrations of IL2: 0.0, 10.0, 100.0 and 1000 UI/mL. Survival of the cells was measured every day by flow cytometry using 7-aminoactinomycin D staining (7-AAD, Via-probe BDBiosciences, Poland).
Statistical analysis
Data were computed with the software Statistica 10.0 (Statsoft, Poland). As indicated by the distribution of the variables non-parametric tests were used. The analysis was performed using Kruskal–Wallis ANOVA (KW), the U-Mann–Whitney test (MW), Wilcoxon test, and Spearman’s rank correlation. The p value was considered statistically significant when <0.05 (Additional file
2).
Discussion
Cellular therapy with autologous expanded Tregs seemed to delay the progression of T1DM but its efficacy could be limited by several factors. As the treatment was introduced after the initial onset of T1DM, probably only a small fraction of β-cells was still preserved in the body and could be protected from autoimmune attack, which was a major limitation of the success of the treatment. The advanced stage of the disease also affected the Tregs. The analysis of this population suggested that it was substantially evolving with the progression of the disease, and it was, therefore, beneficial to use autologous Tregs as early as possible in order to obtain an optimal preparation of Tregs for clinical applications. An additional dose of Tregs later in the study improved the results, but the effects were limited. The disease not only modified Tregs, but also the cytokine milieu towards stronger proinflammatory activity, by increasing the levels of IL6 and, to a lesser extent, TNFα and, IL1. Finally, as confirmed in vitro, survival of the expanded Tregs was dependent on IL2 and the interaction with other lymphocytes.
The time of recruitment to the treatment was an obvious medical limitation of the trial. Current diagnostics identify T1DM patients relatively late, when only 10–30% of the islets are still functional, so only this small fraction can be spared from the disease [
11]. Routine markers of autoimmunity in T1DM, such as autoantibodies, may identify patients in danger of disease development. However, their accuracy in prediction of the onset—notably when titers are low or some autoantibodies are not detectable yet—is not sufficient to identify early stage T1DM [
12]. It is therefore imperative to develop credible criteria of imminent T1DM, ideally before clinical manifestation, while there is still a substantial mass of β-cells. An enrichment of the routine algorithm of T1DM diagnosis with HLA haplotyping, non-HLA polymorphisms, and an assessment of islet-autoreactive T-cells or circulating DNA of the islets may allow the identification of patients in a very early stage of the disease [
13‐
15]. The administration of Tregs to such patients would definitely improve the success of the therapy.
The need for early intervention is also justified by changes in the Treg population during T1DM progression. We have found that two phenotypes of Tregs, that is, CD3
+CD4
+FoxP3
+ and CD3
+CD4
+CD25
highCD127
−, overlap, but diverge with time. It implies problems with acquiring Tregs for clinical expansion in the more advanced stage of the disease. It can be performed with either worse purity of putative Tregs or with pure Tregs at the expense of the number of sorted cells. In both cases, this can affect the quality of the preparation that is administered to the patient. Additional proof of the change in the population of Tregs in T1DM was a shifting proportion of naïve and memory Tregs as the disease progressed, notably in the untreated controls. In these subjects, there was a continuous shrinkage of the Tn compartment at the expense of more differentiated memory subsets. Our previous studies suggested that inflammation associated with long-lasting T1DM was responsible for such an effect on Tregs. These cells were characterized by decreased expression of FoxP3 and CD62L under proinflammatory conditions, and this effect could be reverted using anti-inflammatory agents [
16‐
19]. It was intriguing that, in the current study, gradual increase in the level of proinflammatory cytokines was seen early, starting from the clinical onset of T1DM. In addition, it was the only examined immune phenomenon that could not be stopped completely by the adoptive transfer of Tregs.
Interestingly, a swap between naïve and memory Tregs could be also seen in interventional groups immediately after infusion of expanded Tregs. However, it would always revert to the baseline: 2 years after Tregs infusion naïve and memory proportions were close to those observed at the beginning of the study. The patients administered with two doses of Tregs preserved proportions closest to the baseline ratio. This temporary change was probably mostly caused by the content of the administered expanded Tregs, which are usually Tcm FoxP3
high Tregs [
9]. The fact that they reverted to the baseline ratio in treated subjects suggests a possibility that the level and proportions of Tregs, like other lymphocytes, are homeostatically regulated to some ‘baseline’ levels. Indeed, in our study, the Tcm/Tn proportion of Tregs reverted along with the numbers of total Tregs. Interestingly, the increase in the percentage of Tregs after infusions did not clearly correlate with the dose of administered Tregs, but with C-peptide levels, as shown in our previous report [
8]. In addition, the increased level of Tregs, followed by higher C-peptide levels, could be seen after each infusion of the cells. Hence, the number of doses might be an independent factor affecting the efficacy of the treatment, equally important as the cumulative number of Tregs administered. If repetitive doses of Tregs can maintain an increased level of Tregs for a longer period, they can also maintain the C-peptide levels, which was observed in this study in patients treated with two doses of Tregs. The hypothesis on homeostatic regulation is therefore important for the treatment regimen, as splitting the therapy to repetitive smaller doses of Tregs could keep an increased level of Tregs in peripheral blood longer than a single high dose. The assumption on the Tregs homeostat should be completed with the effect of proinflammatory cytokines in T1DM. This proinflammatory activity probably exerts significant influence, as seen in untreated controls, whose sera maintained high levels of proinflammatory cytokines, as well as their Tregs populations were shifted towards the memory subsets in a sustained way.
Finally, we assessed the effect of IL2 on expanded Tregs. This is the cytokine known to influence the survival and function of Tregs [
20] and it was also used in high concentrations to expand Tregs for clinical application in this study. The production of the clinical preparation of Tregs required high doses of IL2 and its sudden deprivation on release of the product resulted in a fast decline in cell survival. Fortunately, there were two factors that probably protected Tregs viability in vivo. Even small doses of IL2 added to cell culture—equal to the concentrations of IL2 measured in sera of the patients—significantly improved Tregs viability. Furthermore, the addition of other lymphocytes, which probably secreted IL2 or supported Tregs through cell-to-cell interactions, improved Tregs viability. Most importantly, we found a synergic effect of IL2 and co-cultured Teffs on Tregs viability, since the survival of Tregs in co-cultures with Teffs and exogenous IL2 was comparable to the survival of other lymphocytes. In vivo, the concentration of IL2 decreased shortly after the infusion of Tregs in the group treated with two doses of Tregs, which might simply reflect an increased consumption of IL2 by the infused Tregs. The dependence of Tregs on IL2 was tested in several clinical trials, in which IL2 was either given alone or administered together with expanded Tregs. The former has already been tested in T1DM, but appeared to be ineffective [
21]. The latter was tested in graft-versus-host disease with some effect; however, although IL2 as a medicinal product is indicated against tumors, the co-administration of IL2 and Tregs was associated with manifestation of malignancy [
22]. Paradoxically, this might be the major threat of systemic injections of IL2 with Tregs, since the exogenous cytokine might disturb trafficking of administered Tregs attracted to local inflammation, as in
insulitis. It is an important argument as we confirmed that such an accumulation of Tregs in human inflamed tissues exists [
23]. Instead of chemotaxis towards inflamed tissues, Tregs injected in combination with exogenous IL2 may impose generalized systemic immunosuppression, facilitating progression of tumors. Hence, such combined interventions should be performed with caution.
Conclusions
Currently, there are two centers that completed their first studies with Tregs in T1DM [
7,
8,
24]. Safety was confirmed in both studies; however, the results are from small cohorts and, therefore, should be interpreted with caution. Still, the results can help design new, improved treatment protocols. As our trial included matched untreated controls, some efficacy could be also confirmed when comparing the data of the treated and untreated subjects. More importantly, we have found factors that most probably affect the efficacy of this therapy. Mainly, it is the advanced stage of the disease, which has a negative impact on the changes in the Tregs compartment. Based on our previous work, as well as this study, we assume that the inflammatory milieu characteristic for the progression of diabetes is responsible for these changes. High dependency of Tregs on IL2 may also influence the results, but this seems to be controlled by the IL2 produced in vivo. Early administration and repetitive doses of Tregs seemed to improve the results. Hence, new trials should take into account these observations in order to improve the efficacy of this approach. These factors were considered in the design of our new currently ongoing trial TregVac2.0 registered as EudraCT: 2014-004319-35 [
5], in which two doses of Tregs are administered in a 3-month interval, and patients are recruited in earlier phases of the disease.
Authors’ contributions
NMT contributed to the study design, protocol writing, cell preparation, data collection, analysis, interpretation, and writing and reviewing of the report; MM contributed to the study design, protocol writing, cell preparation, data collection, analysis, interpretation, and writing and reviewing of the report; DIG, MG, contributed to cell separation and data collection; ID, MŻ, MZ, MG, HZ, KP contributed to data collection and interpretation; AJC contributed to data collection; RO contributed to data collection; AS, KW contributed to data collection and interpretation; PW contributed to data analysis, interpretation and reviewed the report; WM contributed to data collection, analysis and interpretation and reviewed the report; PJCh contributed to data collection and interpretation and reviewed the report; AB contributed to data collection and interpretation and reviewed the report; JS contributed to data collection, interpretation and reviewed the report; PT was a supervisor of the study who contributed to the study design, protocol writing, cell preparation, data collection, analysis, interpretation writing and, reviewing of the report. All authors read and approved the final manuscript.