SEARCH

SEARCH BY CITATION

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References

NBI-6024 is an altered peptide ligand (APL) corresponding to the 9–23 amino acid region of the insulin B chain (B(9−23)), an epitope recognized by inflammatory interferon-γ-producing T helper (Th)1 lymphocytes in type 1 diabetic patients. Immunomodulatory effects of NBI-6024 administration in recent-onset diabetic patients in a phase I clinical trial (NBI-6024-0003) were measured in peripheral blood mononuclear cells using the enzyme-linked immunosorbent spot assay. Analysis of the mean magnitude of cytokine responses to B(9−23) and NBI-6024 for each cohort showed significant increases in interleukin-5 responses (a Th2 regulatory phenotype) in cohorts that received APL relative to those receiving placebo. A responder analysis showed that Th1 responses to B(9−23) and NBI-6024 were observed almost exclusively in the placebo-treated diabetic population but not in nondiabetic control subjects and that APL administration (five biweekly subcutaneous injections) significantly and dose-dependently reduced the percentage of patients with these Th1 responses. The results of this phase I clinical study strongly suggest that NBI-6024 treatment shifted the Th1 pathogenic responses in recent-onset type 1 diabetic patients to a protective Th2 regulatory phenotype. The significance of these findings on the clinical outcome of disease is currently under investigation in a phase II multidose study.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References

Type 1 diabetes is a spontaneous organ-specific autoimmune disease of the pancreas in which destruction of insulin-producing β cells leads to impaired glucose metabolism and its attendant complications [1]. The autoimmune mechanism of islet β-cell destruction has been extensively studied in the nonobese diabetic (NOD) mouse, in which autoreactive T helper (Th)1 cells that produce interferon (IFN)-γ appear critical to the pathogenic process and recognize several islet β-cell target antigens, including insulin, glutamic acid decarboxylase (GAD) and heat shock protein 60 [1–3]. The majority of pathogenic CD4+ T-cell clones derived from pancreata of NOD mice with insulitis or frank diabetes react specifically with the 9–23 peptide region of the B chain of insulin (B(9−23); [4–6]), demonstrating that insulin itself is a dominant autoantigen. Likewise, screening of an NOD islet β-cell antigen cDNA library demonstrated that as much as 87% of CD8+ T cells in the pancreata from young NOD mice recognized the 15–23 region of the insulin B chain [7]. The B(9−23) epitope appears to be involved in the human disease as well, in which the majority of type 1 diabetic patients showed positive Th1 responses to B(9−23) in enzyme-linked immunosorbent spot (ELISPOT) and proliferation assays [8]. Thus, the B(9−23) region of insulin appears to contain critical epitopes recognized by autoreactive T cells in type 1 diabetes.

We have produced an altered peptide ligand (APL) of the insulin B(9−23) epitope, designated NBI-6024, that contains alanine substitutions at positions 16 (one of the T-cell contact sites) and 19 (a potentially reactive cysteine residue) of the B(9−23) peptide that has demonstrated efficacy in both preventing and suppressing diabetes in the NOD mouse [9]. The therapeutic mechanism of NBI-6024 is associated with the induction of a B(9−23)-cross-reactive regulatory population of Th2 cells [i.e. interleukin (IL)-4-, IL-5- and IL-10-producing cells][9], consistent with the activity of APL in murine models of multiple sclerosis and experimental autoimmune encephalomyelitis [10–12]. The protective activity of these APL-stimulated Th2 cells appears to be, in part, based on their cross-reactivity with the native antigen [13, 14], which enables their re-activation at sites of pathology and leads to dampening of the pathogenic Th1-cell response against the analogous native epitope. Although it is unclear how APL induce Th2 phenotypes, there appears to be both qualitative and quantitative differences in the ability of native peptides and their APL to induce T-cell receptor-associated signals during antigen presentation [15, 16].

Here, we investigated the immunomodulatory activity of NBI-6024 in recent-onset type 1 diabetic patients enrolled in a phase 1 clinical trial, NBI-6024-0003, in which safety and tolerability of the drug were primary endpoints. We tested whether APL administration could suppress the pre-existing pathogenic Th1 (IFN-γ) response and induce a protective Th2 (IL-5) response to B(9−23) (and NBI-6024) in peripheral blood mononuclear cells (PBMC) using the ELISPOT assay [8], which is able to detect low frequencies of antigen-reactive T cells in peripheral blood (1:104 to 1:106 T cells; [17]). This assay offers unique advantages over other cytokine detection systems in that it displays an extremely high degree of sensitivity and can be used to quantify the number of antigen-specific cytokine-producing cells in a heterogeneous T-cell receptor population [18]. Our results indicate that APL therapy significantly modulated cytokine responses to B(9−23) and NBI-6024 during a 6-month period, which included a shift from a Th1 to a Th2 phenotype.

Materials and methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References

Patients and control subjects.  The study protocol was approved by institutional review boards at each clinical site and was conducted according to the Declaration of Helsinki principles. Written and informed consent was obtained from 32 patients (P1–P32; Table 1) who had recent-onset insulin-dependent type 1 diabetes mellitus for enrolment into a phase 1 safety and tolerability trial with NBI-6024. The average age of adolescent and adult type 1 diabetic patient was 14.1 ± 0.4 years (range: 12–16 years; n = 16) and 25.9 ± 2.2 years (range: 18–43 years; n = 16), respectively. The male to female ratio for the entire patient population was 3:1 (Table 1). Criteria for diagnosis of type 1 diabetes were diabetic ketosis and ketoacidosis, polyuria, polydipsia and weight loss, followed by the assessment of serum autoantibody levels [i.e. anti-Protein Tyrosine Phosphatase IAZ (ANTI-IAZ), -GAD or -insulin autoantibodies (IAA), provided that the patient was not on insulin therapy for greater than 2 weeks]. Patients were considered to have a recent-onset status when the onset of symptoms occurred within 3 and 6 months of diagnosis for adolescents and adults, respectively. Patients who had received systemic immunosuppressive agents (e.g. chronic regular use of inhaled and oral steroids), a thiazolidinedione or an investigational drug within 30 days prior to first dosing were excluded from the study. All recent-onset diabetic patients received human insulin therapy following diagnosis, which ranged from 12 to 86 days (mean ± SEM was 48.9 ± 4.1 days) before initial dosing with NBI-6024 (Table 2).

Table 1.  Type 1 diabetic patients in a phase 1 trial with the altered peptide ligand NBI-6024*
    HLA haplotypes
Patient numberAdolescent (a)/adult (A)SexDosing cohort (mg)DR allelesDQ alleles
  • HLA, human leucocyte antigen.

  • *

    Recent-onset type 1 diabetic patients (P) received five injections (0.1, 1 and 5 mg biweekly) of either placebo or NBI-6024 in a phase 1 clinical trial (NBI-6024-0003). Nondiabetic control subjects (C) were included.

  • HLA typing was done by polymerase chain reaction amplification with sequence-specific primers for DR and DQ alleles.

P1aFemalePlaceboNo sampleNo sample
P2aMalePlaceboDRB1*03xx/*04xxDQB1*0201/*0302
P3aFemalePlaceboDRB1*03xx/*04xxDQB1*0201/*0302
P4aMalePlaceboDRB1*0101/*0401DQB1*0501/*0302
P5AMalePlaceboDRB1*04xx/*14xxDQB1*0302/*0501
P6AMalePlaceboDRB1*14xx/*0401DQB1*0501/*0302
P7AMalePlaceboDRB1*0701/*0403DQB1*0201/*0302
P8AMalePlaceboDRB1*0402/*0402/3DQB1*0302/*0302
P9aMale0.1DRB1*0301/*0401DQB1*0201/*0302
P10aMale0.1DRB1*0405/*0405DQB1*0302/*0302
P11aMale0.1DRB1*0301/*0401DQB1*02xx/*03xx
P12aFemale0.1DRB1*1301/*0301DQB1*0603/*0201
P13AMale0.1DRB1*0401/*14xxDQB1*0302/*0501
P14AMale0.1DRB1*0301/*1327DQB1*0201/*0604
P15AMale0.1DRB1*0301/*0401DQB1*0201/*0301
P16AMale0.1DRB1*0409/*07xxDQB1*0302/*0201
P17aFemale1DRB1*1301/*04xxDQB1*0604/*0302
P18aMale1DRB1*03xx/*04xxDQB1*0201/*0302
P19aFemale1DRB1*1301/*0403DQB1*0201/*0302
P20aMale1DRB1*0404/*04xxDQB1*0302/*0302
P21AMale1DRB1*0101/*0401DQB1*0501/*0302
P22AMale1DRB1*13xx/*13xxDQB1*0604/*0603
P23AMale1DRB1*1301/*0401DQB1*0604/*0301
P24AMale1DRB1*0401/*0304DQB1*0302/*0402
P25aFemale5DRB1*0103/*03xxDQB1*0501/*0201
P26aFemale5DRB1*03xx/*07xxDQB1*0201/*0201
P27aMale5DRB1*13xx/*04xxDQB1*0604/*0302
P28aMale5DRB1*03xx/*0401DQB1*0201/*0302
P29AFemale5DRB1*1101/*1101DQB1*0301/*0301
P30AMale5DRB1*0301/*0901DQB1*0201/*0303
P31AMale5DRB1*1301/*0801DQB1*0302/*0402
P32AMale5DRB1*03xx/*0401DQB1*0201/*0302
C1AFemaleControlDR4/DR4DQB1*0302/*0302
C2AMaleControlDR4/DR4DQB1*0302/*0302
C3AMaleControlDR4/xDQB1*0302/x
C4AFemaleControlDR4/DR4DQB1*0302/*0302
C5aFemaleControlDR4/xDQB1*0302/x
C6aMaleControlDR4/xDQB1*0302/x
C7aMaleControlDR4/xDQB1*0302/x
C8aMaleControlDR4/xDQB1*0302/x
C9AFemaleControlDR2/DR2DQB1*0602/*0601
C10AFemaleControlDR2/DR8DQB1*0602/*0402
C11AMaleControlDR1/DR7DQB1*0501/*0201
C12AMaleControlDR2/DR2DQB1*0502/*0602
C13AFemaleControlDR1/DR2DQB1*0501/*0602
Table 2.  Serum autoantibody levels and insulin in type 1 diabetic patients
  Serum levels of anti-β-islet cell antibodies at baseline* 
Patient numberAdolescent (a)/adult (A)IA2GAD65InsulinInsulin usage (days)
  • *

    Serum levels of anti-GAD65 and anti-IA2 antibody from type 1 diabetic patients were measured by a liquid-phase competitive radioimmunoassay and anti-insulin levels were measured by a protein A microassay. Positive and negative control sera were included which were used to calculate an index for antibody levels as described by the following equation: (unknown sample value − negative control value)/(positive control value − negative control value). The upper normal limit for each autoantibody was established as three times the 100th percentile in 241 healthy controls using the Receiver Operator Curve analysis.

  • Sera were considered positive for the respective antibodies at the following index values: >0.071 for IA2; >0.032 for GAD65 and >0.01 for insulin. BLQ, below level of quantitation; NA, not available.

  • Days of insulin therapy from diagnosis to initiation of dosing.

P1a0.8110.0520.95461
P2a0.0401.2130.51857
P3a0.9180.0120.02639
P4a1.049BLQ0.32048
P5A1.0150.0770.05639
P6A0.5520.1540.06651
P7A0.0230.9280.00862
P8A0.8430.1980.02973
P9a0.8780.0730.17746
P10a0.751BLQ0.15656
P11a0.8480.4300.05630
P12a0.8450.2810.34162
P13A0.0100.3710.07126
P14A0.005BLQ0.00443
P15A0.820BLQ0.00354
P16A1.4941.2090.11352
P17a0.7010.0660.24362
P18a0.2840.1220.05948
P19a0.9100.5620.06141
P20a0.8240.0530.00269
P21A0.7740.0510.00521
P22A0.0500.0300.10649
P23A0.0800.4460.05151
P24A0.0170.0430.08237
P25aNANANA47
P26a0.7400.4570.11143
P27a0.8680.8410.37786
P28a0.2971.0381.66072
P29ABLQ0.1560.01136
P30A0.9630.0090.27969
P31A0.0341.1350.06023
P32A0.3681.2330.00412

NBI-6024 dosing was initiated within 3 months of diagnosis. Sixteen adolescent and 16 adult patients from six centres in the US were randomized into groups of five patients, in which four received drug and one received placebo in a double-blind fashion. Three of the adolescent and three of the adult groups received five subcutaneous injections of 0.1, 1 or 5 mg NBI-6024 in an aqueous vehicle (10 mmol/l acetate buffer with 30 mg/ml d-mannitol pH 6.0 solution), respectively, at biweekly intervals at weeks 0, 2, 4, 6 and 8 (Table 1). Two additional adolescent and adult patients received placebo, which yielded a total of eight patients receiving placebo for the study.

Thirteen nondiabetic healthy control subjects (C1–C13; Table 3) were included in this study in which four adolescent (C5–C8) and four adults (C1–C4) were human leucocyte antigen (HLA)-matched controls (i.e. DR4/DQ8) with no familial association with type 1 diabetes and therefore were not at risk for developing disease. Five adult non-HLA-matched nondiabetic control subjects (C9–C13) were also included (Table 1). The average ages of HLA-matched adolescent and adult healthy control subjects were 14.5 ± 0.6 years (range: 13–16 years) and 31.0 ± 3.8 years (range: 20–41 years), respectively, and that of non-HLA-matched control adults was 34.4 ± 3.6 years (range: 24–45 years).

Table 3.  Frequency and stimulation index of positive responses for type 1 diabetic and nondiabetic subjects
   B(9−23)NBI-6024Tetanus toxoid
Patient numberAdolescent (a)/adult (A)Dosing cohort (mg)IFN-γIL-5IFN-γIL-5IFN-γIL-5
  1. IFN, interferon; IL, interleukin.

  2. Values represent the number of positive responses during a 6-month assessment period (stimulation index ≥2, difference in spots between medium control and antigen wells ≥10, Student's t-test P < 0.05 between the mean spots of medium and antigen wells). Numbers in parentheses represent the sum of stimulation indexes (i.e. mean number of cells in antigen wells/mean number of cells in background wells) for all positive responses per patient (‘–’ denotes no positive response observed).

P1aPlacebo1 (3)1 (4)2 (12)
P2aPlacebo1 (18)1 (3)1 (4)4 (60)
P3aPlacebo3 (47)1 (2)1 (29)2 (17)
P4aPlacebo1 (6)1 (2)1 (12)
P5APlacebo1 (15)2 (40)
P6APlacebo1 (5)2 (100)
P7APlacebo4 (3)2 (12)1 (24)
P8APlacebo1 (15)1 (6)1 (25)
P9a0.11 (2)1 (43)2 (6)1 (4)3 (16)
P10a0.12 (5)1 (20)2 (10)3 (24)3 (38)
P11a0.12 (6)1 (17)3 (9)2 (35)
P12a0.11 (2)1 (28)3 (7)4 (115)
P13A0.11 (24)1 (20)1 (6)
P14A0.12 (9)2 (14)1 (3)
P15A0.11 (12)2 (5)
P16A0.13 (7)1 (9)1 (2)3 (10)
P17a11 (8)1 (3)1 (25)
P18a11 (2)1 (40)3 (19)1 (36)1 (3)1 (60)
P19a11 (9)3 (11)1 (34)1 (32)1 (18)
P20a11 (6)1 (2)
P21A11 (2)1 (8)2 (5)3 (42)
P22A12 (9)1 (3)1 (7)1 (8)
P23A11 (5)2 (13)1 (3)
P24A11 (3)2 (11)1 (15)2 (6)1 (78)
P25a51 (3)1 (3)
P26a51 (3)1 (3)1 (5)
P27a51 (10)
P28a52 (8)1 (15)2 (8)1 (21)3 (14)2 (45)
P29A5
P30A51 (3)
P31A51 (10)1 (63)1 (5)2 (401)1 (13)
P32A53 (135)1 (3)3 (9)2 (28)2 (41)2 (34)
C01AControl2 (12)2 (57)
C02AControl3 (19)1 (14)
C03AControl1 (88)
C04AControl1 (8)
C05aControl
C06aControl1 (39)
C07aControl1 (184)2 (26)1 (13)
C08aControl1 (22)3 (36)
C09AControl1 (5)
C10AControl1 (96)1 (123)1 (175)
C11AControl1 (8)1 (11)
C12AControl1 (7)1 (16)2 (20)
C13AControl2 (37)

Serum autoantibody measurement.  Serum levels of anti-GAD65 and anti-IA2 antibodies were measured by a liquid-phase competitive radioimmunoassay as previously described [19, 20], and IAA levels were measured by a protein A microassay [21] at the University of Colorado Health Sciences Center, Barbara Davis Center for Childhood Diabetes, Denver, CO, USA. Positive and negative control sera were included which were used to calculate an index for antibody levels as described by the following equation: (unknown sample value − negative control value)/(positive control value − negative control value). The upper normal limits for each autoantibody serum level were established as three times the 100th percentile in 241 healthy controls, using the receiver operator curve analysis, which were >0.049 (anti-IA2), >0.032 (anti-GAD65) and >0.011 (IAA) [22]. Measurement of serum levels of autoantibodies to islet β-cell antigens from the 32 type 1 diabetic patients showed that 24 patients were positive for anti-GAD65 antibodies, 24 were positive for anti-IA-2 antibodies and 25 were positive for anti-insulin antibodies (Table 2). NBI-6024 administration had no significant effect on anti-IA2, -GAD65 or -insulin antibody levels within any of the cohorts during the course of the study (data not shown) and did not induce any detectable plasma levels of anti-NBI-6024-specific antibody (data not shown).

HLA typing.  HLA typing was done by PCR amplification with sequence-specific primers for DR and DQ alleles [23] at the University of Colorado Health Sciences Center, Barbara Davis Center for Childhood Diabetes. Twenty-nine of the 31 patients genotyped for HLA alleles expressed the high-risk DR3 (DRB1*0301 and 0304), DR4 (DRB1*0401–0405), DQ2 (DQB1*0201) or DQ8 (DQB1*0302) alleles, in which 21 patients expressed the DR4/DQ8 high-risk combination, 13 expressed the DR3/DQ2 high-risk combination and seven expressed both high-risk combinations (Table 1). Eight of the 13 nondiabetic control subjects expressed the DR4/DQ8 high-risk combination (i.e. HLA-matched subjects), while none expressed DR3/DQ2 (Table 1).

ELISPOT assay.  The number of cytokine-producing cells that recognize specific antigens in PBMC from type 1 diabetic patients and control subjects was quantified by the ELISPOT assay as described [18]. Fresh blood was drawn from each patient directly into Vacutainer CPT tubes (Becton Dickinson, Inc., Franklin Lakes, NJ, USA) at each of six centers in the US and shipped immediately to the central laboratory at the University of Colorado Health Sciences Center, Barbara Davis Center for Childhood Diabetes. Blood samples were received no later than 24 h after the blood was drawn and PBMC were isolated by centrifugation via Vacutainer CPT Ficoll–Paque density gradient immediately upon receipt, and lymphocytes were washed and stored frozen in a dimethyl sulfoxide solution in liquid nitrogen. 3 × 105 PBMC obtained from frozen samples were seeded in triplicate wells of 96-well Unifilter 350 filter-backed plates (Polyfiltronics; Fisher, Tustin, CA, USA) coated with antihuman IFN-γ monoclonal antibody (clone 2G1; Endogen, Inc., Cambridge, MA, USA) or anti-IL-5 monoclonal antibody (clone TRFK5; BD PharMingen, Inc., San Diego, CA, USA). Cells were cultured in triplicate in the presence or absence of 5 µg/ml Tetanus toxoid (Accurate Chemicals, Westbury, NY, USA), 10 µg/ml phytohaemagglutinin (PHA-M; Sigma, St. Louis, MO, USA), 1 µg/ml Staphylococcus enterotoxin B (SEB; BD PharMingen), insulin B(9−23) (50 µm) or the APL, NBI-6024 (10 and 50 µm). These peptides were synthesized by a solid-phase method as described [24]; human insulin B-chain (9–23) amide (B(9−23)) [SHLVEALYLVCGERG] and NBI-6024 [SHLVEALALVAGERG]. RPMI 1640 medium containing 1% N-2-Hydroxyethylpiperazine-N′-2-ethane (HEPES), l-glutamine, Na Pyruvate, Non-essential amino acid NEAA, Penicillin/Streptomycin, 10−5mβ-mercaptoethanol and 1% heat-inactivated fetal bovine serum (HyClone, Logan, UT, USA; endotoxin <10 pg/ml) was used. After 24 h (for IFN-γ assessment) or 48 h (for IL-5 assessment) of incubation at 37 °C in the presence of 5% CO2, cells were washed away and cytokines were detected with antihuman IFN-γ (clone B133.5; Endogen) or antihuman IL-5 (clone JES1-5A10; PharMingen) secondary biotinylated monoclonal antibody plus avidin-peroxidase (DAKO, Philadelphia, PA, USA). 3-Amino-9-ethyl carbazole substrate solution (Pierce, Inc., Dallas, TX, USA) was used to develop the reaction which was stopped by washing the plate with water. Spots derived from cytokine-producing cells were quantified using the Series-1 Immunospot and Satellite Analyzers (Autoimmun Diagnostika, Inc., Strassberg, Germany). We demonstrated that this ELISPOT assay showed good sensitivity and reproducibility by participating in a workshop that compared five different assay formats [25]. At least three of the seven longitudinal samples collected per patient were analysed simultaneously to reduce interassay variability. The majority of samples showed strong responses to the positive control mitogens, SEB and PHA (see Results). Note that the analytical and statistical approach for defining positive ELISPOT responses is described in the Results.

Statistical analyses.  To analyse whether NBI-6024 administration affected the magnitude of ELISPOT responses among cohorts, we calculated the stimulation index (SI; mean antigen-induced spots/mean medium spots) for each sample per patient at each study week. A change from baseline (week 0) SI was calculated for each subject and used as the dependent variable in a Mixed Effects Repeated Measures model. This model included each dosing cohort (including placebo), study week and a dosing cohort-by-week interaction as fixed effects in the model with baseline SI used as a covariate. From this model, P-values were calculated using the Least Squares Means pair-wise differences between each NBI-6024 dosing cohort versus placebo cohort at each study week. A responder analysis was also conducted, where each patient was scored as a responder or nonresponder to each antigen and cytokine (see criteria in Results below), and response rates were analysed using a Fisher's exact test comparing separately each dosing cohort receiving NBI-6024 (or the untreated nondiabetic control group) versus those receiving placebo.

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References

Longitudinal analysis of responses to B(9−23) and NBI-6024

We have reported that IFN-γ responses to the insulin B(9−23) peptide were associated with recent-onset type 1 diabetic patients and with those at high-risk for type 1 diabetes [8]. We have also produced the APL, NBI-6024, that differs from the insulin B(9−23) 15 amino acid peptide by alanine substitutions at position 16 and 19 [9]. NBI-6024 suppressed diabetes in diabetes-prone NOD mice with an associated induction of a regulatory cell population of B(9−23)-cross-reactive Th2 cells [9] that suppresses pathogenic IFN-γ responses. Here, we sought to determine whether NBI-6024 could reduce IFN-γ and enhance IL-5 responses in 32 recent-onset type 1 diabetic patients enrolled in a phase 1 clinical trial. PBMC were obtained from the recent-onset diabetic patients (Table 1) at weeks 0 (baseline), 2, 4, 6, 8, 14 and 26 after initiation of dosing with placebo or NBI-6024 (i.e. 0.1, 1 and 5 mg), and IFN-γ and IL-5 responses to B(9−23) and NBI-6024 were assessed via the ELISPOT assay.

We first determined whether the magnitude (SI) of B(9−23) and NBI-6024 cytokine responses for each dosing group receiving APL were significantly different from those receiving placebo at each time point. The SI is the mean of antigen-induced spots divided by the mean background (medium alone) spots for each patient's sample, and the mean patient SI per dosing group (adolescents and adults cohorts combined; n = 8) at each time point was calculated. While the mean SI of IFN-γ and IL-5 responses to B(9−23) and NBI-6024 for each cohort was variable during the 6-month assessment period, some SI responses of cohorts that received APL were significantly greater than those of the placebo group. Using a Mixed Effects Repeated Measures statistical analysis, there was one IFN-γ response to B(9−23) that was significantly above (P = 0.0002) the respective placebo cohort response which was in the 5 mg cohort at week 14 (Fig. 1A). In addition, there were pronounced trends of elevated IL-5 responses to B(9−23) and NBI-6024 and those that were significant were to B(9−23) by the 1 mg cohort at week 4 (P = 0.003; Fig. 1B) and to NBI-6024 by the 0.1 mg cohort at week 26 (P = 0.046) and by the 5 mg cohort at week 2 (P = 0.032; Fig. 1D). IFN-γ responses to NBI-6024 by all cohorts that received APL were similar to those that received placebo (Fig. 1C). In spite of the longitudinal variability in these ELISPOT responses, this analysis strongly suggests that APL administration induced, predominantly, Th2 responses to both the endogenous B(9−23) epitope and NBI-6024.

image

Figure 1. Repeated Measures analysis of the mean stimulation index of ELISPOT responses between diabetic cohorts receiving NBI-6024 and placebo. ELISPOT analysis was performed on peripheral blood mononuclear cell samples from type 1 diabetic patients before (week 0) and after subcutaneous administration of the altered peptide ligand of insulin B(9−23), NBI-6024 (five doses each of 0.1, 1 or 5 mg administered biweekly), obtained six times during a 6-month period. The mean and SEM of the ELISPOT stimulation indexes (SI; mean antigen-induced spots/mean medium spots per sample) from each patient within a dosing group (i.e. adolescents and adults cohorts combined; n = 8) for each cytokine response to insulin B(9−23) and NBI-6024 is reported. The mean SI of each cohort receiving NBI-6024 at each time point was compared with those of cohorts receiving placebo using a Mixed Effects Repeated Measures analysis (*P = 0.05).

Download figure to PowerPoint

This longitudinal variability was also observed among most individual patients in this study (data not shown). Longitudinal variability was not due to interassay variability, which we have previously evaluated for this assay format [25]. Differences in cell viability among samples could not have accounted for such variability in responsiveness because 92% of all cell samples (i.e. 194 of 210) showed robust responses to the control mitogens, SEB and PHA, which were within an SD of the mean spots per well for each mitogen and cytokine (i.e. PHA and SEB, mean ± SD spots/well for IFN-γ were 935 ± 825 and 1224 ± 804, respectively, and for IL-5 were 92.1 ± 79.6 and 118.5 ± 112.1, respectively).

Responder analysis in type 1 diabetic patients versus nondiabetic control subjects

The longitudinal variability in the magnitude of antigenic responsiveness observed among the cohorts and among individual subjects made it difficult to evaluate whether any particular subject responded to NBI-6024 administration. Therefore, a responder analysis was performed to evaluate each subject's response status at postdosing visits, in which each subject was scored as a ‘responder’ or ‘nonresponder’. This was performed by evaluating whether the subject had at least one significant response to the respective antigen during the 6-month assessment using the following criteria, similar to that described by others [26–28]: the mean number of antigen-responding cells was greater than twofold above (i.e. SI = 2) and at least 10 cells greater than the respective background mean cell number (i.e. medium), and these two means were significantly different using the Student's t-test (P ≤ 0.05). The number of positive responses (i.e. frequency of responsiveness) and the sum of all SI for these positive responses (i.e. magnitude of responsiveness) were reported per subject (Table 3). Each subject was scored positive or negative as an IFN-γ or IL-5 responder to each antigen and the proportion (i.e. percentage) of responding patients within each population was reported (see results below).

In addition to the diabetic patients enrolled in the study, 13 untreated nondiabetic healthy control subjects were also assessed in a similar longitudinal fashion (four adolescent and four adult HLA-matched, and five adult non-HLA-matched; Table 1). We first investigated whether Th1 responses were associated with diabetes by determining the proportion of subjects with positive cytokine responses to each antigen in the untreated type 1 diabetic (placebo) and nondiabetic (control) groups (Fig. 2). The placebo-treated diabetic population had significantly (P = 0.0475, Fisher's exact test) greater proportions of patients that responded to B(9−23) with IFN-γ than did the nondiabetic control population (adolescents and adults combined); 37.5% of the placebo population (i.e. three of eight) responded with IFN-γ only (Th1) whereas only 8% (i.e. one of 13) of the control population responded with this Th1 phenotype (Fig. 2A). Similar proportions of both populations responded to B(9−23) with IL-5 production (Fig. 2A). Note that the reported IL-5 responding populations (i.e. second bar of each group) show responders that produced either IL-5 alone (hatched portion) or IL-5 in combination with IFN-γ (shaded portion), consistent with the idea that a regulatory IL-5 response to B(9−23) is desired despite the presence of IFN-γ. Similar to IFN-γ B(9−23) responses, there was a significantly greater proportion of the diabetic placebo group responding to NBI-6024 with IFN-γ only relative to the nondiabetic control group (50 versus 0%, respectively; P = 0.0117; Fig. 2B). No significant differences in IL-5 responses to NBI-6024 between these two groups were observed. The proportions of the placebo and nondiabetic control populations responding to the control recall antigen, Tetanus toxoid, were similar (Fig. 2C), which underscores the association of the antigen-specific B(9−23) and NBI-6024 Th1 responses with the diabetic population.

image

Figure 2. Percentage of antigen/cytokine-responding placebo-treated type 1 diabetic patients and nondiabetic control subjects. The number of B(9−23), NBI-6024 and Tetanus toxoid reactive cells that produce interferon (IFN)-γ and interleukin (IL)-5 was assessed via the ELISPOT assay in peripheral blood mononuclear cells from eight type 1 diabetic patients (four adolescent and four adult) that received placebo and 13 age- and human leucocyte antigen-matched control subjects during the assessment periods described in Fig. 1. Individuals were scored as positive responders to an antigen if they had at least one significant response during the assessment period using the following criteria: the mean value of antigen-responding cells was greater than or equal to twofold and ≥10 cells above the respective background mean and significantly different using the Student's t-test (P ≤ 0.05). The percentage of patients per cohort responding per antigen per cytokine is shown. Note that patients responding with either IL-5 alone (hatched) or IL-5 plus IFN-γ (solid gray) are denoted as such in the same bar. *Significantly different from the respective cytokine response of the placebo group (*P ≤ 0.05, Fisher's exact test).

Download figure to PowerPoint

NBI-6024 modulation of cytokine responses in type 1 diabetic patients

To determine whether APL therapy modulated cytokine responses to B(9−23) and NBI-6024, we compared IFN-γ and IL-5 responses in PBMC samples, obtained during postdosing visits, with those of the placebo and nondiabetic control groups. No significant differences in the proportions of patients responding to B(9−23) with IFN-γ and IL-5 production were observed among the adolescent cohorts. Note that none of the adolescent patients responded to B(9−23) with IFN-γ only (Fig. 3A), which precluded evaluation of the downregulatory effects of APL administration on this Th1 pathogenic phenotype. In contrast, the adolescent cohort that received placebo had significantly greater NBI-6024-specific IFN-γ only responses relative to nondiabetic controls (i.e. 75 versus 0%, respectively; P = 0.1429, Fisher's exact test; Fig. 3B), and there was a dose-dependent decrease in the proportion of NBI-6024-specific IFN-γ responders so that the highest dosing group, 5 mg, showed no IFN-γ only responders (P = 0.1429; Fig. 3B). In addition, there was a concomitant dose-dependent trend of an increase in the proportion of NBI-6024-specific IL-5 responders with APL administration (Fig. 3B).

image

Figure 3. NBI-6024 modulation of the percentage of adolescent type 1 diabetic patients with ELISPOT responses to insulin B(9−23) and the altered peptide ligand, NBI-6024. The number of B(9−23) and NBI-6024 reactive cells that produce interferon (IFN)-γ and interleukin (IL)-5 were assessed via the ELISPOT assay in peripheral blood mononuclear cells from adolescent type 1 diabetic patients (n = 4 per cohort) that received either NBI-6024 (five doses each of 0.1, 1 or 5 mg administered biweekly) or placebo and from four age- and human leucocyte antigen-matched control subjects during the assessment periods described in Fig. 1. Individuals were scored as positive responders to an antigen if they had at least one significant response during the assessment period using the following criteria: the mean value of antigen-responding cells was greater than or equal to twofold and ≥10 cells above the respective background mean and significantly different using the Student's t-test (P ≤ 0.05). The percentage of patients per cohort responding per antigen per cytokine is shown. Note that patients responding with either IL-5 alone (hatched) or IL-5 plus IFN-γ (solid gray) are denoted as such in the same bar. *Significantly different from the respective cytokine response of the placebo group (*P ≤ 0.1, Fisher's exact test).

Download figure to PowerPoint

A significantly greater proportion of the adult placebo population responded to B(9−23) with IFN-γ only relative to the adult control population (75 versus 8%, respectively; P = 0.0517; Fig. 4A), and this response was completely suppressed by NBI-6024 administration in the highest dosing group (5 mg; Fig. 4A; P = 0.1429). Note that reduced proportions of the adult placebo population responded to NBI-6024 with IFN-γ only, which precluded analyses of the downregulatory effects of NBI-6024 administration on this pathogenic response (Fig. 4B). These results indicate that NBI-6024 administration significantly decreased the proportion of adolescent and adult IFN-γ responders to either B(9−23) or NBI-6024, demonstrating a suppression of the Th1 pathogenic response to this autoantigenic epitope.

image

Figure 4. NBI-6024 modulation of the percentage of adult type 1 diabetic patients with ELISPOT responses to insulin B(9−23) and the altered peptide ligand, NBI-6024. The number of B(9−23) and NBI-6024 reactive cells that produce interferon (IFN)-γ and interleukin (IL)-5 were assessed via the ELISPOT assay in peripheral blood mononuclear cells from adult type 1 diabetic patients (n = 4 per cohort) that received either NBI-6024 (five doses each of 0.1, 1 or 5 mg administered biweekly) or placebo and from nine age- and human leucocyte antigen-matched control subjects during the assessment periods described in Fig. 1. Individuals were scored as positive responders to an antigen as described in Figs 2 and 3, and the percentage of patients per cohort responding to antigen and cytokine is shown. *Significantly different from the respective cytokine response of the placebo group (*P ≤ 0.1, Fisher's exact test).

Download figure to PowerPoint

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References

We have previously reported that IFN-γ responses to the insulin B(9−23) peptide were unique to recent-onset type 1 diabetic patients and to those at high-risk for type 1 diabetes [8]. Here, we aimed to confirm and extend our previous study [8] and to determine whether administration of the APL, NBI-6024, to recent-onset type 1 diabetic patients could modulate these B(9−23) cytokine responses in a phase 1 clinical trial (NBI-6024-0003). In addition, we evaluated, for the first time, specific cytokine responses to NBI-6024. It was apparent from our initial analysis of the ELISPOT data that the nature of longitudinal antigenic responses in peripheral blood was variable for each patient and cohort. This inconsistent responsiveness to B(9−23) and NBI-6024 could be in part due to fluctuations in the disease state, as autoimmune diseases are known to progress in series of relapses and remissions in which autoreactive T cells may show a cyclic migration throughout the body passing between the pathologically involved organs and blood. In addition, the endogenous B(9−23) (and APL) responses may vary with the regulation of insulin secretion. Note that this variation in longitudinal responsiveness per patient was not due to any variability in the quality of sampling, as responses to the positive control activation agents, SEB and PHA, were consistently robust among samples. Moreover, we have previously validated this assay format and showed good reproducibility [25], thus excluding the possibility of interassay variability as a major factor underlying the longitudinal variability observed within a subject's sample set. By performing a Mixed Effects Repeated Measures analysis on the mean magnitude of responses between cohorts, we observed significant increases predominantly in the regulatory IL-5 response to B(9−23) and NBI-6024 in cohorts receiving APL relative to those receiving placebo. This observation suggests that NBI-6024 is able to induce Th2 regulatory responses in type 1 diabetic patients.

Because the Repeated Measures analysis showed a pronounced longitudinal variability that hindered evaluation of responses on a per patient basis, we used a Responder Analysis to more accurately assess antigenic responsiveness per patient. This analysis entailed scoring each patient as a positive or negative cytokine responder to each antigen which allowed us to determine whether NBI-6024 modulated antigenic responses on a per-cohort basis. Using this method, we were able to confirm our previous results [8] indicating that B(9−23)-specific IFN-γ responses were predominantly in the type 1 diabetic population. Surprisingly, a substantial proportion of placebo-treated diabetic subjects showed NBI-6024-specific IFN-γ responses (i.e. four of eight adolescent and adult) with a smaller proportion showing IL-5 responses (i.e. two of eight adolescent and adult), whereas none of the control subjects showed an IFN-γ response and only one had a significant IL-5 response to NBI-6024. Hence, IFN-γ responses to NBI-6024 were found in untreated type 1 diabetic patients but not in nondiabetic control subjects. One explanation for APL responses in untreated (placebo) patients could be due to a strong cross-reactivity with the spontaneous and endogenously arising B(9−23) responses, as has been shown with vaccination of other APL in mice [13, 14]. Although B(9−23) cross-reactivity with NBI-6024 was not evaluated in this study, it has been demonstrated with T-cell lines in NOD mice [9].

The responder analysis also showed that NBI-6024 administration significantly and dose-dependently down-modulated the proportions of Th1 (IFN-γ only) responders to B(9−23) in the adult population and to NBI-6024 in the adolescent population, suggesting a favourable activity of the drug. Furthermore, there was a trend suggesting that the pathogenic Th1 responses to NBI-6024 were dose-dependently converted into regulatory Th2 (IL-5) responses by APL administration. This was especially apparent in the adolescent population. Differences between the adolescent and adult populations were also observed in cohorts receiving placebo, where most adolescents showed responses to both antigens (i.e. IFN-γ and IL-5 to B(9−23) and IFN-γ only to NBI-6024), whereas the adults showed a dominant Th1 response to B(9−23) only. These observations suggest that juvenile diabetes may be comprised of an unfocused antigenic response to the two closely related epitopes (B(9−23) and NBI-6024) with a less mature mixed-cytokine response, whereas the adult (late-onset) disease may be more focused to the B(9−23) epitope with an associated Th1 response. Although more studies are required to further define such differences between juvenile and late-onset disease, these results may suggest differences in the underlying pathologies.

Although cytokine responses correlated with age, they did not correlate with other variables. Cytokine responses measured by the cumulative SI per patient (Table 3) did not correlate with gender (Table 1), HLA haplotype (Table 1), baseline levels of autoantibodies or the duration of insulin therapy prior to initiation of dosing (Table 2). The individual regression analyses between cytokine SI values and autoantibody levels or insulin usage showed correlation coefficient R2 values <0.1 (data not shown). We previously demonstrated that the duration of insulin therapy did not correlate with B(9−23) IFN-γ responses in recent-onset type 1 diabetic patients [8]. The ability of HLA-DQ8+ patients in this study to respond to insulin B(9−23) and NBI-6024 is consistent with the ability of these peptides to bind the HLA-DQ8 haplotype ([29] and unpublished observation, respectively). In addition, the predicted binding of B(9−23) to the DQ2 haplotype [29] is consistent with the responsiveness to B(9−23) and NBI-6024 of the six DQ2+/DQ8 patients in this study. (It has not been reported whether the DR antigens bind this epitope.) One of the three patients (i.e. P22, P23 and P29; Table 1) that did not express the DQ2 or the DQ8 haplotype did not respond to either B(9−23) or NBI-6024 (i.e. P29; Table 3).

The ELISPOT assay has been previously used to monitor responses in patients with diabetes and other autoimmune diseases [17, 25, 30, 31]. While the ELISPOT assay has proven useful in monitoring patient responses to vaccines designed for immunizations against cancer and infectious agents [32, 33], these types of responses tend to have a strong IFN-γ phenotype which may be more feasible to detect compared with vaccines that downregulate IFN-γ or induce the weaker Th2 regulatory responses. Indeed, IFN-γ ELISPOT responses to the recall antigens, purified protein derivative of Mycobacteria and Tetanus toxoid, were more robust than those of IL-5 [34]. However, others have successfully used IL-5 ELISPOT responses to measure vaccine-induced responses to viral peptides [35] and effectiveness of regulatory cell-inducing vaccines such as glatiramir acetate (Copaxone) for multiple sclerosis [30, 31] and DiaPep277 for type 1 diabetes [36]. Although neither of the latter two vaccines are APL, they share APL-like qualities in that they have demonstrated induction of Th2 responses against autoantigens. However, these previous studies [30, 31, 36] have reported ELISPOT assessments at only one visit prior to dosing and at one or two visits after dosing, in contrast to the more comprehensive longitudinal assessment of responses performed in this study.

In this phase I study, NBI-6024 administration was evaluated for both its safety and tolerability and for its ability to modulate B(9−23) responses. Indeed, there were no serious adverse side effects due to NBI-6024 administration (unpublished observation). This study represents the most comprehensive ELISPOT cytokine analysis to date from a longitudinal study of 45 subjects in a regulatory cell-inducing vaccine trial. Our approach of using Responder analyses revealed a correlation between APL therapy and the expected modulation of antigen-specific cytokine responses, and additional clinical studies with regulatory cell-inducing vaccines should confirm these analytical approaches. In conclusion, we showed promising immunomodulatory effects of NBI-6024 treatment, and the significance of such findings on the clinical outcome of disease is under investigation in a phase II multidose study.

Acknowledgments

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References

The authors thank the following contributors from Neurocrine Biosciences, Inc. (San Diego, CA, USA), Dr Albert Zlotnik and Dr Peter Hevezi for reviewing the manuscript, Ms. Donna DeBoer and Dr Henry Pan for recruiting patients and managing the clinical trial and Dr Nick Ling for synthesizing peptides. This study was funded by Neurocrine Biosciences, Inc.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References