Dynamic changes in CD4+ CD25+ high T cell apoptosis after the diagnosis of type 1 diabetes

Authors

  • S. Glisic-Milosavljevic,

    1. The Max McGee National Center for Juvenile Diabetes and Human Molecular Genetics Center, Medical College of Wisconsin and the Children's Research Institute of the Children's Hospital of Wisconsin, USA,
    Search for more papers by this author
  • T. Wang,

    1. Division of Biostatistics and Human Molecular Genetics Center, Medical College of Wisconsin, USA, and
    Search for more papers by this author
  • M. Koppen,

    1. The Max McGee National Center for Juvenile Diabetes and Human Molecular Genetics Center, Medical College of Wisconsin and the Children's Research Institute of the Children's Hospital of Wisconsin, USA,
    Search for more papers by this author
  • J. Kramer,

    1. The Max McGee National Center for Juvenile Diabetes and Human Molecular Genetics Center, Medical College of Wisconsin and the Children's Research Institute of the Children's Hospital of Wisconsin, USA,
    Search for more papers by this author
  • S. Ehlenbach,

    1. The Max McGee National Center for Juvenile Diabetes and Human Molecular Genetics Center, Medical College of Wisconsin and the Children's Research Institute of the Children's Hospital of Wisconsin, USA,
    Search for more papers by this author
  • J. Waukau,

    1. The Max McGee National Center for Juvenile Diabetes and Human Molecular Genetics Center, Medical College of Wisconsin and the Children's Research Institute of the Children's Hospital of Wisconsin, USA,
    Search for more papers by this author
  • P. Jailwala,

    1. The Max McGee National Center for Juvenile Diabetes and Human Molecular Genetics Center, Medical College of Wisconsin and the Children's Research Institute of the Children's Hospital of Wisconsin, USA,
    Search for more papers by this author
  • S. Jana,

    1. The Max McGee National Center for Juvenile Diabetes and Human Molecular Genetics Center, Medical College of Wisconsin and the Children's Research Institute of the Children's Hospital of Wisconsin, USA,
    Search for more papers by this author
  • R. Alemzadeh,

    1. Children's Hospital of Wisconsin Diabetes Center, Pediatric Endocrinology and Metabolism, MCW, USA
    Search for more papers by this author
  • S. Ghosh

    Corresponding author
    1. The Max McGee National Center for Juvenile Diabetes and Human Molecular Genetics Center, Medical College of Wisconsin and the Children's Research Institute of the Children's Hospital of Wisconsin, USA,
      Soumitra Ghosh MD, PhD, The Max McGee National Center for Juvenile Diabetes, Medical College of Wisconsin, Department of Pediatrics, PO Box 26509, Milwaukee, WI 53226–0509, USA.
      E-mail: sghosh@mcw.edu
    Search for more papers by this author

Soumitra Ghosh MD, PhD, The Max McGee National Center for Juvenile Diabetes, Medical College of Wisconsin, Department of Pediatrics, PO Box 26509, Milwaukee, WI 53226–0509, USA.
E-mail: sghosh@mcw.edu

Summary

Because type 1 diabetes (T1D) is a chronic, autoimmune, T cell-mediated disease, interventions affecting T cells are expected to modulate the immune cascade and lead to disease remission. We propose that increased CD4+ CD25+high T cell apoptosis, a trait we discovered in recent-onset T1D subjects, reflects T1D partial remission within the first 6 months after diagnosis. Apoptosis of forkhead box P3 (FoxP3)+ CD4+ CD25+high T cells, in addition to total daily doses of insulin (TDD), blood glucose, HbA1c and age, were measured in 45 subjects with T1D at various times after diagnosis. Sixteen healthy control subjects were also recruited to the study. Higher CD4+ CD25+high T cell apoptosis levels were detected within the first 6 months of diagnosis (odds ratio = 1·39, = 0·009), after adjustment for age, TDD and HbA1c. A proportional hazards model confirmed that the decline of apoptosis after diagnosis of T1D was related significantly to survival time (hazards ratio = 1·08, P = 0·014), with TDD and age also contributing to survival. During this time there was an inverse relationship between CD4+ CD25+high T cell apoptosis with TDD (= −0·39, P = 0·008). The CD4+ CD25+high T cell apoptosis levels decline significantly after the first 6 months from diagnosis of T1D and may help in the close monitoring of autoimmunity. In parallel, there is an increase in TDD during this time. We also propose that CD4+ CD25+high T cell apoptosis assay can be used to gauge the efficacy of the several immune tolerance induction protocols, now under way.

Introduction

Type 1 diabetes (T1D) is an autoimmune disorder in which T cells destroy insulin-producing beta cells of the pancreas within the islets of Langerhans [1]. During the period immediately after the onset of T1D, many individuals need a reduced dose of insulin than expected. This period (remission period, also known as the honeymoon period) has been acknowledged as an advantageous time for immune intervention [2,3]. However, not all subjects will go through this remission phase and some might experience it for a longer period of time than others. In order to identify subjects with optimal residual beta-cell function for secondary immune intervention and to monitor severity of autoimmune T1D, clinical studies need exact criteria for recognition of the remission period, although few trials use these criteria today. In most studies, the main criteria used to diagnose the remission phase have been TDD ≤ 50% of the initial insulin dose [4] or < 0·5 U/kg of body weight per day [5]. Recently, the remission phase has been redefined to include HbA1c at or less than 8% [6]. However, these parameters reflect remaining beta-cell function. If appropriate T cell-based assays were available, the extent of ongoing autoimmunity could also be studied during this period. Furthermore, besides potential clinical and therapeutic utility, it would be ideal to have a specific T cell assay that reflects remission, allowing monitoring of changes during immune modulation.

Autoimmune destruction of β cells is the result of a perturbed immune balance between autoreactive T cells and those that regulate their proliferation and function in the periphery. We recently reported high levels of CD4+ 25+high T cell apoptosis in recent-onset T1D patients and in unaffected relatives of patients with T1D, who were positive for multiple antibodies [7]. Regulatory T cells (Treg) within the CD4+ CD25+high T cell population are derived thymically [8] and adoptive transfer of CD4+ CD25+high Treg cells prevents or delays diabetes in non-obese diabetic (NOD) mice [9,10]. Because T1D is a T cell-mediated autoimmune disorder, death of Treg cells that control proliferation of effector cells could contribute to the pathogenesis of diabetes as well as provide a marker for autoimmunity. Furthermore, with autoimmunity subsiding in the face of a diminishing beta-cell mass, we show that apoptosis of CD4+ CD25+high T cells returns to lower levels at the end of the remission period.

In this paper, we propose CD4+ CD25+high T cell apoptosis as a measure of remission in patients with recent-onset of T1D. As T1D is a T cell-mediated disease, a T cell-based criterion has an important aetiological relevance for monitoring the effect of immune intervention, where remission is the outcome of interest.

Methods

Subjects

Forty-five subjects with T1D were ascertained primarily through the diabetes clinic at Children's Hospital of Wisconsin (CHW), and 16 healthy controls were recruited from placement of flyers and notices in online medical campus publications (Table 1). Inclusion criteria for control subjects were a random blood glucose < 110 mg/dl, no personal and family history of T1D and an absence of diabetes-specific autoantibodies [to glutamic acid decarboxylase (GAD), insulin and IA-2][11]. Diabetes was defined according to accepted criteria: (1) symptoms of diabetes plus casual plasma glucose concentration 200 mg/dl (11·1 mmol/l) or (2) fasting plasma glucose (FPG) 126 mg/dl (7·0 mmol/l) [12]. Diagnosis of diabetic ketoacidosis was made based on venous pH < 7·30 and serum bicarbonate < 15 meq/l and the duration of hyperglycaemic symptoms was recorded. The protocol was approved by the CHW Institutional Review Board (IRB) and participants and/or their parents (guardians) provided written informed consent and completed a questionnaire. All subjects had T1D-specific autoantibodies measured at the recruitment visit. Self-monitoring of blood glucose (SMBG) four to six times daily was reviewed and average TDD was calculated at each quarterly visit. Also, HBA1c levels were determined every 3 months. During follow-up visits at 3, 6, 9, 12 and 24 months after diagnosis, symptoms and glucose levels were reviewed and insulin doses were adjusted accordingly. Participants with T1D were divided into three groups at the recruitment visit according to time elapsed after diagnosis: group 1 (n = 17), within the first 6 months from diagnosis (3·8 ± 0·7 months); group 2 (n = 14), between 6·1 and 12 months (8·3 ± 0·7 months); and group 3 (n = 14) after 12·1 months from diagnosis (66·7 ± 13·2 months).

Table 1.  Clinical and biochemical characteristics of studied groups of subjects.
Time after diagnosis for recruitment visitAdult control
(> 18 years)
Age-matched controls
(< 18 years)
Group 1
(1·5–6 months)
Group 2
(6·1–12 months)
Group 3
(> 12.1 months)
P
  1. Data are mean values with standard error. *When groups 1, 2 and 3 were compared, there was no difference in glucose levels (Kruskal–Wallis test, P = 0·24). DKA: diabetic ketoacidosis; HLA: human leucocyte antigen.

n106171414n.s.
Average time after diagnosis (months)  3·8 ± 0·728·3 ± 0·7266·7 ± 13·240·0001
Gender (% female)50·050·070·642·814·30·0001
Autoantibodies (GAD65, IAA, IA2) (%)  50·057·138·5n.s.
Autoantibodies (no GAD65) (%)  31·328·653·8n.s.
HLA haplotypes (DR3 only) (%)20·033·329·435·732·1n.s.
HLA haplotypes (DR4 only) (%)10·0017·6032·132·1n.s.
HLA haplotypes (DR3 + DR4) (%)30·033·347·067·864·2n.s.
BMI SDS at recruitment1·160 ± 1·25−1·82 ± 1·48−2·33 ± 1·28−0·76 ± 1·793·86 ± 1·94n.s.
Age of diagnosis (years)  9·73 ± 1·128·80 ± 1·259·35 ± 0·79n.s.
Age at recruitment (years)33·96 ± 3·1313·75 ± 0·9110·07 ± 1·119·37 ± 1·3712·78 ± 0·90·0001
Glucose (mg/dl) at recruitment84·4 ± 2·6292·33 ± 3·39187·2 ± 17·12223·3 ± 29·38162·3 ± 21·290·0005*
DKA at diagnosis (%)  23·521·414·3n.s.
Serum pH at diagnosis  7·35 ± 0·057·32 ± 0·047·34 ± 0·02n.s.
Duration of symptoms before Dx (weeks)  3·55 ± 0·874·38 ± 0·972·71 ± 0·47n.s.
Insulin requirement (U/kg/day)
 3 months after diagnosis  0·45 ± 0·110·44 ± 0·030·37 ± 0·06n.s.
 6 months after diagnosis  0·65 ± 0·030·55 ± 0·050·43 ± 0·050·055
 9 months after diagnosis  0·64 ± 0·040·56 ± 0·100·51 ± 0·04n.s.
 12 months after diagnosis  0·64 ± 0·050·60 ± 0·110·59 ± 0·05n.s.
 24 months after diagnosis  0·74 ± 0·050·63 ± 0·050·69 ± 0·04n.s.
 At recruitment  0·42 ± 0·040·50 ± 0·100·73 ± 0·060·006
HbA1c (%)
 Diagnosis  9·7 ± 0·5510·6 ± 0·4910·9 ± 0·45n.s.
 3 months after diagnosis  6·7 ± 0·236·7 ± 0·336·4 ± 0·22n.s.
 6 months after diagnosis  7·2 ± 0·307·3 ± 0·337·2 ± 0·32n.s.
 9 months after diagnosis  7·5 ± 0·286·9 ± 0·297·3 ± 0·29n.s.
 12 months after diagnosis  7·2 ± 0·297·7 ± 0·357·8 ± 0·27n.s.
 24 months after diagnosis  7·6 ± 0·507·9 ± 0·407·6 ± 0·28n.s.
 At recruitment  7·0 ± 0·576·73 ± 0·657·6 ± 0·18n.s.
CD4+CD25+ T cell frequency (%)6·33 ± 0·736·84 ± 0·725·33 ± 0·668·21 ± 1·297·11 ± 1·22n.s.
CD4+CD25+high T cell apoptosis (%)3·33 ± 0·803·48 ± 0·9710·7 ± 1·75·0 ± 0·93·9 ± 0·40·0010

Cell isolation and sorting procedure

Peripheral blood mononuclear cells (PBMC) were collected, using vacutainers with acid citrate dextrose (ACD) solution B of trisodium citrate, at the recruitment visit and for a second visit from 18 of 61 subjects. Isolation of PBMCs was performed using Ficoll-Hypaque density gradient centrifugation following the manufacturer's protocol (Amersham Pharmacia, Uppsala, Sweden). Cells were counted and stained with a cocktail of fluorochrome-conjugated monoclonal antibodies in phosphate-buffered saline (PBS) [allophycocyanin (APC)-αCD4 (clone RPA-T4), APC-Cy7-αCD25 (clone M-A251), fluorescein isothiocyanate (FITC)-αCD14 (clone M5E2), FITC-αCD32 (clone FLI8·26), FITC-αCD116 (clone M5D12), phycoerythrin (PE)-Cy7-CD8 (clone RPA-T8), all from BD Pharmingen, San Diego, CA, USA] and sorted on fluorescence activated cell sorter (FACS)Aria (BD Biosciences, San Diego, CA, USA). Cells were gated first for live lymphocytes, eliminating dead cells and debris. Next, two gates were set up to eliminate non-CD4 T cells (FITC- and PE-Cy7-conjugated antibodies). CD4+ T cells were gated further as CD4+ CD25, CD4+ CD25+low and CD4+ CD25+high T cells using the fluorochrome minus one (FMO) method, that allows a more precise definition of cells having fluorescence above the background level [13]. The 1% of cells expressing the highest level of CD25 were collected and defined as CD4+ CD25+high T cells, known to be enriched for Treg cells [14].

Apoptosis assay

Apoptosis in all samples was measured as described previously [15]. Briefly, within 1–2 h after FACS isolation (during which cells were kept at 4°C), cells were stained in the dark with 250 nM YOPRO1 (Molecular Probes, Eugene, OR, USA) for 20 min and 250 ng 7-AAD (BD Biosciences) was added 10 min before acquiring at least 10 000 events. The thresholds for both YOPRO and 7AAD nucleic acid stains were determined according to autologous, freshly sorted CD4+ CD25 T cells used as a negative control. Apoptosis of CD4+ CD25+high T cells was measured as the percentage of apoptotic cells (YOPRO1+/7AAD) among live cells (all 7AAD cells comprising both YOPRO1+ and YOPRO1 cells). Using a similar computation, apoptosis was confirmed by intracellular staining for the active form of caspase 3 (BD Bioscience, data not shown). The intracellular staining procedure, provided by the manufacturer, was followed for the dual staining with caspase 3 and forkhead box P3 (FoxP3) (clone PCH101, eBiosciences, San Diego, CA, USA). Briefly, 50 000 cells were fixed in fixation buffer (eBiosciences) for 30 min at 4°C, and washed in permeabilization buffer (eBiosciences). After resuspending the cells in 50 µl of Perm, 20 µl of anti-FoxP3 and 15 µl of anti-active Cas3 antibodies were used for staining (30 min at 4°C). Cells were washed in Perm buffer, resuspended in 200 µl PBS and acquired on LSRII (BD Biosciences).

Functional (suppression) assay of CD4+ CD25+high T cells

The suppressive capacity of FACS-isolated CD4+ CD25+high T cells was measured via proliferation assays, where responder (CD4+ CD25) and suppressor (within the CD4+ CD25+high population) T cells were stimulated in vitro separately and together, in co-culture, under homogeneous conditions. The proliferation assay was set up with 20 000 autologous CD4+ CD25 T cells as responders isolated at the same visit and 20 000 irradiated (5000 rad) antigen-presenting cells (PBMC) per well. In co-culture, 20 000 responders and 20 000 APCs had additional 2000 of CD4+ CD25+high T cells as suppressors (ratio 10 : 1) seeded in a 96-well plate. Stimulation was with aCD3-coated beads (1 µg/ml, three beads/cell) for 5 days under 5% CO2 and under saturated humidity. Cells were left in culture after pulsing (1 µCi of [3H]-thymidine) for an additional 16 h, then harvested, and after adding scintillation liquid were read for counts per minute (cpm) using a standard plate reader (Perkin Elmer). Suppression (%) of Treg was calculated as [(cpm of responders in single culture – cpm of cells in co-culture)/cpm of responders in single culture] × 100.

Other analyses

Blood glucose was measured using glucose oxidase method and the HbA1c was determined using Bayer DCA 2000 instrument (Bayer Diagnostic Inc., Tarrytown, NY, USA), with a non-diabetic range of 4·5–5·7%.

Statistical analysis

The Mann–Whitney U and Tukey–Kramer tests were used to compare results between clinical groups or at different time-points with P-value ≤ 0·05 considered significant. We also performed univariate analysis using the non-parametric, two-sided Kruskal–Wallis test to check for association of quantitative parameters (age, insulin requirement, HbA1c and CD4+ CD25+high T cell apoptosis) using 6 months from diagnosis as a cut-off, in addition to a one-way analysis of variance (anova). A logistic regression model was performed with the proportion of subjects diagnosed within 6 months as the dependent variable of interest. Finally, a proportional hazards model was implemented on the failure time measured from the date of diagnosis. In this way, we were able to look at remission differences with respect to a specific time-point as well as across a continuous time range. The statistical analysis was implemented using sas software (SAS Institute, Cary, NC, USA). GraphPad and Excel softwares were used for data presentation.

Results

Characterization of FACS-sorted T cells

Figure 1a illustrates the final gating step in sorting of CD4 T cells in which CD4+ CD25, CD4+ CD25+low and CD4+ CD25+high T cell populations were isolated from each subject. CD4+ CD25+high T cells showed the highest frequency of cells expressing FoxP3 transcription factor (59·1 ± 2·7%, Fig. 1b). In vivo activated T cells (CD4+ CD25+low) expressed FoxP3 at intermediate levels (20·2 ± 2·1%), with the lowest percentage of FoxP3+ expression in CD4+ CD25 T cells [3·9 ± 0·6%; Kruskal–Wallis test, P < 0·0001; anovaF =  201·7 (d.f. 2,96), P < 0·0001)]. Levels of FoxP3 showed some variation in CD4+ CD25+high T cells, but the suppressive function of CD4+ CD25+high T cells was significant in a sample of 14 subjects (Fig. 1c, 52·5 ± 6·2%). Therefore, CD4+ CD25+high T cells do not proliferate, have the highest detectable FoxP3 levels and can inhibit proliferation of responder T cells, making them highly likely to be true Treg cells and not activated effector T cells. In addition, there was no suppression seen from CD25+low T cells despite higher FoxP3 levels compared to CD4+ CD25 T cells (data not shown). There were no differences in percentages (frequencies) of CD4+ CD25+ T cells between the groups studied (Table 1).

Figure 1.

(a) Fluorescence activated cell sorting (FACS) of peripheral blood mononuclear cells (PBMCs) with CD4+ CD25, CD4+ CD25+low and CD4+ CD25+high T cells isolated for further analysis. (b) Ex vivo forkhead box P3 frequency in CD4+ CD25, CD4+ CD25+low and CD4+ CD25+high T cells FACS T cells differs significantly (Kruskal–Wallis test, P < 0·0001; analysis of variance F = 201·7 (d.f. 2,96), P < 0·0001). (c) Functional (suppression) confirming isolation of CD4+ CD25+high T cells with regulatory properties. The suppression assay was set up with 20 000 autologous CD4+ CD25 T cells as responders isolated at the same visit and 20 000 irradiated antigen-presenting cells per well. Suppressor : responder ratio was 1 : 10. Cells were stimulated for 5 days with aCD3-coated beads (1 µg/ml, three beads/cell), after which cells were pulsed with 1 µCi of [3H]-thymidine for the next 16 h, harvested and read for cpm; n = 14 subjects). Suppression (%) was calculated as [(cpm of responders in single culture – cpm of cells in co-culture)/cpm of responders in single culture] × 100.

High CD4+ CD25+high T cell apoptosis in most recent-onset T1D subjects recruited within 6 months after diagnosis

Figure 2 shows results of ex vivo CD4+ CD25+high T cell apoptosis levels. Group 1, comprising paediatric recent-onset T1D patients, had the highest apoptosis levels in this cell population (10·7 ± 1·7%). Group 1 subjects were recruited within 6 months after diagnosis and had mean insulin requirements < 0·5 U/kg/day and mean HbA1c, 7% (Table 1). CD4+ CD25+high T cell apoptosis was significantly different across the clinical groups [Kruskal–Wallis, P = 0·0005; anovaF = 9·75 (d.f. 3,57), P < 0·0001, Fig. 2]. High apoptosis levels were not detected in long-standing patients with T1D (group 3). Based on the Tukey–Kramer post-hoc test, CD4+ CD25+high T cell apoptosis in subjects assigned to group 1 differed significantly from the control group (P < 0·001) as well as from groups 2 (P < 0·01) and 3 (P < 0·001). Results from the intracellular staining of active form of caspase 3 of the same sample, as a confirmatory measure for ongoing apoptosis, correlated significantly with the results obtained from the YOPRO1/7-AAD method [Spearman's rank correlation coefficient for n = 16, r = 0·72 (95% CI = 0·34–0·90), P = 0·0015, data not shown]. Simultaneously, the CD4+ CD25 T cell subpopulation was assessed for apoptosis without showing differences among studied groups (data not shown) [7]. The possibility of higher CD4+ CD25+high T cell apoptosis in group 1 due to hyperglycaemia was excluded. Although significant differences in glucose levels were observed between healthy control and subjects with T1D (Table 1, P = 0·0005), there were no differences in glucose levels at recruitment visit when only three groups of subjects with the disease were compared without including control subjects (Kruskal–Wallis test, P = 0·24). However, for the latter T1D groups, a dramatic difference in apoptotic cell percentage was detected (Kruskal–Wallis test, P = 0·002).

Figure 2.

Percentage CD4+ CD25+high T cell apoptosis in control subjects and in three groups of patients. Group 1 comprises paediatric patients with type 1 diabetes, who were recruited within 6 months after diagnosis, had the lowest total daily doses (TDD) and, according to definition, were in the partial remission phase. Patients in group 2 showed the highest variation in insulin requirements (TDD). Group 3 comprised long-standing diabetic patients with stabilized blood glucose levels and TDD. Apoptosis levels show a significant difference across the four groups [analysis of variance F = 9·75 (d.f. 3,57), P < 0·0001; Kruskal–Wallis test, P = 0·0005]. Two-way P-value comparisons between groups presented are the result of a Tukey test.

Dynamic change of CD4+ CD25+high T cell apoptosis levels in subjects with recent-onset T1D

The apoptosis levels of CD4+ CD25+high T cells were very informative in subjects with T1D, as shown in Fig. 3. As opposed to healthy control subjects (Fig. 3a) with similar levels of apoptosis recorded 5·0 ± 0·7 months apart (3·7 ± 0·8 versus 4·1 ± 0·7, P = 0·64), subjects with recent-onset T1D showed significantly decreased apoptosis levels 4·9 ± 0·6 months later, suggesting that almost 5 months later they were outside a ‘remission phase’, as determined by both their apoptosis levels (12·5 ± 1·3% versus 4·2 ± 0·7%, Mann–Whitney U-test, P = 0·0001, Fig. 3b) and TDD (at first visit 0·36 ± 0·06 U/day/kg and at second visit 0·61 ± 0·09 U/day/kg; Mann–Whitney U-test, P = 0·01).

Figure 3.

The variation of percentage CD4+ CD25+high T cell apoptosis with subsequent visits. (a) CD4+ CD25+high T cell apoptosis in eight healthy control subjects in two subsequent visits (5·0 ± 0·7 months apart) showed similar CD4+ CD25+high T cell apoptosis levels (first visit 3·7 ± 0·8 versus 2nd visit 4·1 ± 0·7, Mann–Whitney U-test, P = 0·64); (b) 10 recent-onset type 1 diabetic subjects (first visit 3·9 ± 0·9 months after diagnosis and second visit 8·8 ± 1·9 months after diagnosis, with the two visits 4·9 ± 0·6 months apart) showed significantly lower CD4+ CD25+high T cell apoptosis levels at the second visit compared to the first (12·5 ± 1·3 versus 4·2 ± 0·7, Mann–Whitney U-test, P = 0·0001).

Univariate analysis of insulin requirement, HbA1c and CD4+ CD25+high T cell apoptosis with respect to time after diagnosis (categorized as ≤ 6 months or > 6 months) revealed that CD4+ CD25+high T cell apoptosis (Kruskal–Wallis test, P = 0·0003), insulin requirements (Kruskal–Wallis test, P = 0·005) and HbA1c ((Kruskal–Wallis test, P = 0·03) are all significant factors associated with the partial remission in T1D.

Logistic regression analysis showed that CD4+ CD25+high T cell apoptosis is the most significant factor associated with time after diagnosis categorized as ≤ 6 months and > 6 months [OR = 1·39, 95% CI = (1·09–1·78), and P-value = 0·009] after adjustment for age (= 0·09), TDD (= 0·21) and HbA1c (= 0·47), with an overall adjusted R-square value of 39%. Using the same predictive variables (CD4+ CD25+high T cell apoptosis, age, insulin and HbA1c) in the proportional hazards model, it emerged that CD4+ CD25+high T cell apoptosis, insulin (TDD) and age are all associated significantly with the remission time with hazard ratios (HR) being 1·08 [95% CI = (0·01–1·14), P = 0·014] for apoptosis of CD4+ CD25+high T cells, 0·12 [95% CI = (0·03–0·54), P = 0·006] for TDD; 0·91 [95% CI = (0·83–0·99), P = 0·03] for age, and 0·17 [CI = (0·59,1·27), = 0·47] for HbA1c, respectively.

Inverse relationship of CD4+ CD25+high T cell apoptosis and insulin requirement, independent of age

An inverse relationship between CD4+ CD25+high T cell apoptosis and insulin requirements with respect to the time after diagnosis (< 6 months and > 6 months) and stratification for age (3–5 years, 5·1–10 years and 10·1–16 years) is illustrated in Fig. 4. Overall, high CD4+ CD25+high T cell apoptosis detected after metabolic stabilization correlated significantly with low TDD (Spearman's rank correlation coefficient r = − 0·39, P = 0·008). We also present the linear relationship between insulin doses and CD4+ CD25+high T cell apoptosis (Fig. 5a). However, CD4+ CD25+high T cell apoptosis did not show a linear relationship with HbA1c values (Fig. 5b). Importantly, the percentage of CD4+ CD25+high T cell apoptosis is independent of age at recruitment (Spearman's rank correlation coefficient r = − 0·12, P = 0·44). When all the subjects recruited within 6 months of diagnosis (n = 17) were compared to all the subjects recruited after 6 months (n = 28), a difference in the CD4+ CD25+high T cell apoptosis was more pronounced (Mann–Whitney U-test, P = 0·001) than for TDD (Mann–Whitney U-test, P = 0·01).

Figure 4.

Inverse relationship between CD4+ CD25+high T cell apoptosis and insulin requirements [total daily doses (TDD)] with respect to the age groups. Patients recruited within 6 months from diagnosis showed significantly higher levels of CD4+ CD25+high T cell apoptosis and a lower need for exogenous insulin compared to the patients recruited after 6 months from diagnosis, for each age group. There was an inverse correlation between time after diagnosis and CD4+ CD25+high T cell apoptosis (Spearman's correlation coefficient r = −0·51, P = 0·0003). This was independent of age (Spearman's correlation coefficient r = −0·12, P = 0·44). When all subjects, regardless of age but recruited within and after 6 months from diagnosis, were combined into two groups (17 within 6 months versus 28 subjects recruited after 6 months from the onset), differences in the CD4+ CD25+high T cell apoptosis were more pronounced (Mann–Whitney U-test, P = 0·001) than for TDD (Mann–Whitney U-test, P = 0·01).

Figure 5.

Linear relationship between CD4+ CD25+high T cell apoptosis with insulin requirements [total daily doses (TDD)] and HbA1c. All type 1 diabetes subjects together, regardless of stage of the disease, show a significant inverse linear relationship with insulin doses (5a, r2 = 0·149, P = 0·01). However, no correlation was found between CD4+ CD25+high T cell apoptosis and HbA1c (5b, r2 = 0·00024, P = 0·92).

Discussion

In the present study, high levels of CD4+ CD25+high T cell apoptosis were detected after metabolic stabilization in most patients within 6 months after diagnosis (group 1), and showed a significant negative correlation with TDD. In contrast, high apoptosis was not detected in long-standing patients with T1D (group 3) nor in most paediatric patients who were seen after 6 months of diagnosis (group 2). The apoptosis levels of CD4+ CD25+high T cells were stable in control subjects but transiently high in subjects with recent-onset T1D. During the period immediately after the onset of T1D, many individuals develop partial remission of diabetes characterized by reduction in TDD, as the remaining beta cells, not exposed to excess glucose, continue to produce insulin, while the immune cell attack on the beta cells continues unabated [16]. Age-independent high levels of CD4+ CD25+high T cell apoptosis during the relatively short time after the onset of the disease may parallel the autoimmune destruction of beta cells, and is detected during T1D remission only.

It has already been acknowledged that this period, when most subjects show reduced need for exogenous insulin, is an optimal time for immune intervention in human clinical trials for T1D [2,3,16–18]. However, there are immune-modulating protocols in secondary prevention of T1D (see TRIALNET) that do not take the remission period into consideration, perhaps because of the difficulty in its definition. In order to identify subjects with residual beta-cell function for immune intervention and to monitor severity of autoimmune T1D, clinical studies need exact criteria for recognition of this period. In most studies, the main criteria used to diagnose the remission period has been TDD ≤ 50% of the initial insulin dose [4] or < 0·5 U/kg of body weight per day [5]. The duration of the remission period is under the control of many factors that influence glycaemic control. Depending on age, pubertal status, insulin sensitivity and criteria for partial remission, up to 60% of patients with T1D might experience a partial remission [4,16,18–21]. Furthermore, partial remission has been shown to be a heterogeneous phenomenon with inter- and intraindividual variation [22] and factors such as diabetes care intensity and early identification of hyperglycaemic symptoms prior to diagnosis influencing its course [20]. While TDD, HbA1c and SMBGs have great value in everyday management of the disease, the addition of CD4+ CD25+high T cell apoptosis assay brings a quantitative T cell component to the monitoring process. The assay may help identify candidates for immune intervention and consequent monitoring for prolongation of the partial remission phase. In this study we show that CD4+ CD25+high T cell apoptosis results were not correlated with age, HbA1c, high human leucocyte antigen (HLA) risk (due to higher frequencies of DR3 and DR4 haplotype among T1D subjects; see Table 1) or type and/or number of detected T1D-specific autoantibodies. In contrast, apoptosis of CD4+ CD25+high T cells correlated significantly with insulin doses. We believe that the presence and cessation of T1D remission can be estimated by high CD4+ CD25+high T cell apoptosis.

Furthermore, the current clinical criteria used in different immune-modulation trials could benefit from utilizing this easy-to-use CD4+ CD25+high T cell apoptosis assay. The CD4+ CD25+high T cell apoptosis assay can also be potentially used for primary prevention trials, as elevated CD4+ CD25+high T cell apoptosis levels were confirmed in first-degree relatives who are at higher risk to develop T1D [7]. However, not every high-risk first-degree subject will actually succumb to the disease. Therefore, the CD4+ CD25+high T cells apoptosis assay could potentially be a risk marker for unaffected first-degree relatives and therefore have predictive utility in making a decision as to which unaffected subject to choose for immune intervention. We are beginning longitudinal studies to address these questions formally.

One limitation of the present study is the lack of a direct assessment of beta-cell function in subjects after the onset of T1D. This would have allowed us to correlate our CD4+ CD25+high T cell apoptosis results with beta-cell function. Another is our inability, because of ethical considerations, to measure the apoptosis status of CD4+ CD25+high T cells locally, at the site of the damage. Finally, due to a lack of a specific Treg cell marker we are unable to quantify their absolute numbers.

It is difficult to speculate why induced death of CD4+ CD25+high T cells is associated with partial remission. We can only suggest that two processes occur concurrently. The first is the reduced insulin requirements that occur after metabolic stabilization the latter partially alleviating beta-cell toxicity. The second process is apoptosis of CD4+ CD25+high T cells, due possibly to continued activation of CD4+ CD25+high T cells. Finally, when beta-cell function is diminished, insulin requirement increases and apoptosis of CD4+ CD25+high T cells decreases to the levels detected in control subjects. At that point, the subject is out of the remission phase and is identified as a long-standing diabetic subject.

The apoptosis assay of CD4+ CD25+high T cells offers not only a new insight into the ongoing pathogenesis of T1D, but is also a promising criterion for a more exact determination of the presence and the end of transient remission period. Therefore, we suggest the CD4+ CD25+high T cell apoptosis assay (YOPRO1/7AAD) as a useful marker for determining candidates with the highest chance for a positive response to immune intervention, which can help in assessing efficacy of immune-based therapies.

The present study formulates a connection between high apoptosis levels of CD4+ CD25+high T cells and remission of the T1D. Future clinical studies are needed to determine exactly the remaining beta-cell function corresponding to a certain level of CD4+ CD25+high T cell apoptosis and to verify, with a larger sample size, the utility of the CD4+ CD25+high T cell assay.

Acknowledgements

The authors thank to Karen Zeqiri, Hope Albertz, Jeffrey Woodliff and Corbett Reinbold for technical support. The authors thank Dr Marty Hessner for reviewing the manuscript. This project was funded by the Children's Hospital of Wisconsin, Clinical Research Institute and General Clinical Research Center grant no. M01 RR00058 from the National Center for Research Resources, National Institutes of Health. The funding source had no role in the study design, collection of data, analysis and interpretation of results.

Ancillary