Prevention of type 1 diabetes mellitus using a novel vaccine

Abstract

Type 1 diabetes mellitus (T1DM) affects 1 in 300 people and the incidence of the disease is rising worldwide. T1DM is caused by chronic autoimmune destruction of the insulin-producing β-cells. The exact etiology and primary auto-antigen are not yet known. The autoimmune, chronic, and progressive nature of the disease raises the possibility of intervention, preferably by slowing down or stopping the destruction of the β-cells as early as the prediabetic stage. Since the 1980s, several attempts have been made to maintain β-cell function using immunosuppressive agents, immune modulation such as plasmapheresis, cytokine therapy, or antibody treatment. These agents were not diabetes specific; the occasionally observed beneficial effect did not compensate for the often very severe side effects. According to the latest assumption, the administration of diabetes-specific auto-antigens can elicit tolerance, which can prevent the destruction of the β-cells, hopefully without serious side effects. The authors summarize current understanding of the immunology of T1DM, review the trials on prevention, and discuss their vaccination study.

Introduction

Type 1 diabetes mellitus (T1DM) affects about 4 million people in North America and Europe, which is approximately 10% of the entire diabetic population in those areas. In contrast to type 2 diabetes mellitus (T2DM), which is characterized by insulin resistance and a relative lack of insulin, in T1DM there is insufficient production of insulin caused by the chronic autoimmune destruction of the insulin-producing β-cells. The incidence of the disease is rising worldwide, and if present trends continue, new cases of T1DM in European children younger than 5 years of age are predicted to double between 2005 and 2020, and the prevalence of patients younger than 15 years of age will rise by 70% [Patterson et al. 2009]. The disease typically develops relatively fast in childhood, and starts with polyuria, polydipsia, and weight loss. There is a subgroup called latent autoimmune diabetes of adults (LADA), which is immunologically similar to T1DM, and usually affects adults, developing slowly. Presumably 10% of T2DM patients are in fact LADA patients [Panczel et al. 2001]. T1DM affects patients for the rest of their lives, and can lead to acute and chronic complications.

Immunological background

The β-cells are destroyed directly by cluster of differentiation (CD) 8+ cytotoxic T cells and macrophages. The death of the insulin-producing islet cells is caused by the cytokine, tumor necrosis factor alpha (TNF-α), which forms pores on the cells (‘kiss of death’). The selective loss of β-cells leads to a predominance of glucagon-secreting α-cells, with an end result of absolute insulin deficiency and secondary hyperglucagonemia [Gianani et al. 2010]. The impaired regulatory T cells are critically important factors in the defective autoimmune response. The regulatory T cells express the interleukin (IL)-2 receptor α-chain (CD25) at a high level, as well as other molecular markers, such as transcription factor Forkhead box P3 (FoxP3), cytotoxic T lymphocyte-associated antigen 4 (CTLA-4), and the glucocorticoid-induced TNF receptor (GITR) [Bluestone et al. 2008; Kis et al. 2007]. These regulatory T cells are mainly CD4+, T-helper (Th) cells, but other types of T cells, such as natural killer T cells (NKT) and CD8+ cells, also have regulatory functions. The CD4 + CD25 + FoxP3 + T cells are able to shift the immune response either to a cellular (Th1) or humoral (Th2) path [Raz et al. 2005; Singh and Palmer, 2005; Winter and Schatz, 2003; Wilson et al. 1998]. The invariant NKT (iNKT) cells are one of the most potent immune regulators, thus these cells are widely studied in the pathogenesis of several diseases like T1DM, multiple sclerosis, and asthma, etc. [Godfrey et al. 2004]. The iNKT cells are a special group of thymus-derived T cells, which express both the natural killer markers and the T-cell receptor (TCR) [Lee et al. 2002]. Most of the NKT cells have an invariant TCR, which means that between the alfa chain 24 (Vα24) variable region and the junction Q (JαQ) region (equivalent to Vα14-Jα18 in mice), there is no nucleotide insertion; these cells are the iNKT cells. Through the expression of CD4 and CD8, these cells can be positive for each or none of them (double negative iNKT cells) [Godfrey et al. 2004]. After TCR stimulation, the iNKT cells can rapidly produce very high amounts of Th1-related cytokines, such as interferon-gamma (IFN-γ) or Th2-related cytokines, such as IL-4 [Kis et al. 2007]. The cytokines produced can influence the differentiation of naïve T cells, the inflammatory responses, and they have a role in either the acquired or innate immunity. In T1DM patients, the cytokine production of CD4–CD8 iNKT cells shifted significantly to a Th1 bias [Wilson et al. 1998]. The iNKT cells derived from the pancreatic lymph nodes of cadavers of T1DM patients produced less IL-4 [Kent et al. 2005b]. Even the percentage of CD4 + iNKT cells is decreased in T1DM patients compared with healthy and T2DM individuals [Kis et al. 2007].

In T1DM patients, not only these special cells, but the entire CD4 + T-cell population is down-regulated specifically affecting the cell cycle, key immune functions, cell surface receptor-linked signal transduction, and electron transport [Orban et al. 2007]. The case history of an agammaglobulinemic T1DM child showed that the humoral immune response is not necessary for the pathogenesis of T1DM [Martin et al. 2001]. Although the islet cell antibodies (ICA), glutamic acid decarboxylase antibody (GADA), insulinoma-associated protein tyrosine phosphatase antibody (IA2-A), and insulin auto-antibody (IAA), are probably not involved directly in the pathogenesis of T1DM, they are clinically important [Kulmala, 2003]. By measuring them we can distinguish between T1DM and T2DM, evaluate the risk, and follow the development of the disease. However, the exact reasons or the trigger antigen are still not known. The genetic predisposing factors, such as some human leukocyte antigen (HLA) types, are very well known, but they are not causative. External triggers, such as virus infection, toxins or cow’s milk, may initiate the autoimmune response in a genetically predisposed individual. There are several candidates for these initiating antigens, such as GADA, IA2-A, a heat shock protein (Hsp277), GAD65, a recently described zinc transporter (Znt8), and insulin and/or proinsulin [Vaziri-Sani et al. 2010; Wenzlau et al. 2009; Agardh et al. 2005; Singh and Palmer, 2005; Wong, 2005; Petrovsky et al. 2003; Winter and Schatz, 2003; Wilson and Buckingham, 2001; Fuchtenbusch et al. 1998]. It is possible that there are more trigger antigens, but probably they differ from patient to patient. The most common animal model of T1DM is the nonobese diabetes (NOD) mouse. Supported by investigations using NOD mice and some human experiments, the primary auto-antigen is very likely the insulin itself or one of its precursors (e.g. preproinsulin or proinsulin) [Kent et al. 2005a; Nakayama et al. 2005; Wilson and Buckingham, 2001]. T lymphocytes isolated from lymph nodes draining the pancreas are autoreactive for the A chain of insulin [Kent et al. 2005a]. During the autoimmune destruction increasing numbers of antigens become auto-antigens, a process called antigen spreading. Different types of ICA appear while the β-cells perish. Clinical T1DM develops when the numbers of β-cells decrease to below 20% of the initial volume [Kulmala, 2003]. The chronic and progressive nature of this autoimmune disease raises the possibility of an intervention that aims to slow down or stop the destruction of the β-cells as early as possible, and preferably in its prediabetic stage [Winter and Schatz, 2003; Wilson and Buckingham, 2001; Fuchtenbusch et al. 1998]. Since the 1980s several clinical trials have been carried out trying to maintain β-cell function with immunosuppressive treatments. Later this treatment was followed by immune modulation intervention, such as plasmapheresis, cytokine or antibody treatments (blocking the cell involved in the autoimmune process). More recently trials trying to achieve immune tolerance using different antigens as an intervention have been published. These diverse methods can have beneficial effects by shifting the immune response from a Th1 bias to the less harmful, possibly protective, Th2 bias. They can also produce antigen-specific regulatory T cells, or inhibit autoreactive T cells, or influence communication between the immune cells [Raz et al. 2005; Monetini et al. 2004; Fuchtenbusch et al. 1998]. The exact mechanism by which any of these interventions arrest the autoimmune process is still not understood.

Intervention studies

There are three different stages at which the process leading to T1DM can be intercepted. The first stage is when autoimmunity has not yet started. This is primary prevention largely targeting individuals genetically at risk, and the goal is to prevent the onset of autoimmunity. The second stage is when the subject has already developed auto-antibodies signaling the presence of ongoing autoimmunity. The goal here is to prevent the onset of the clinical disease. These secondary prevention studies target those individuals at risk of developing diabetes, such as the relatives of T1DM patients carrying genetic markers, and/or positive for one or more ICA, or showing an abnormal handling of glucose (but not yet diabetic). The third stage is when clinical T1DM is already present but there is still a residual β-cell function to preserve. These tertiary prevention studies are often labeled as intervention studies. If the autoimmune process is inhibited, there is a chance of regeneration of the β-cells. It is difficult to use potentially dangerous treatments for primary and secondary prevention, even in intervention studies, and the side effects have to be taken into account, and the costs versus benefits carefully assessed and balanced [Chatenoud, 2010].

Several different types of intervention have been tried for T1DM so far. To date, these drugs have largely been used as single agents. Some of the most important intervention trials grouped by the key characteristic of the agent used are discussed in the following.

Immunosuppressive treatment

Glucocorticoids have been widely used as immunosuppressive treatments. Their anti-inflammatory effect works in different ways, but their main action is the inhibition of the granulocytes and lymphocytes. Unfortunately several serious adverse effects can follow chronic treatment, including weight gain, dyslipidemia, and even steroid-induced diabetes. Glucocorticoids have been used in several clinical trials. In combination with azathioprine β-cell function was improved in half of newly diagnosed T1DM patients. As a consequence of the significant side effects, steroids are now omitted from T1DM intervention protocols [Winter and Schatz, 2003].

Cyclosporine and tacrolimus reduce the secretion of IL-2 by inhibiting calcineurin, so they have an immunosuppressive effect as IL-2 is necessary for the propagation of different cells. Only temporary results can be achieved with these agents; moreover, they cannot be used for prevention because of serious toxic effects on the kidney, liver, and nervous system [Wilson and Buckingham, 2001].

Immune modulation

In some clinical trials carried out in the 1980s, plasmapheresis was used to try to maintain β-cell function in T1DM patients. These attempts did not have any beneficial effects, which is understandable once the fundamentally cellular pathogenesis was recognized [Ludvigsson et al. 1983]. The intravenous immunoglobulin treatment had a positive outcome at the beginning, but this disappeared after a year; furthermore, the treatment had serious side effects [Colagiuri et al. 1996].

The administration of Th2 cytokines (e.g. IL-4, IL-10, IL-13, IFN-α), or the inactivation of Th1 cytokines (e.g. IL-2, IFN-γ) can shift the immune response to a Th2 bias [Wilson et al. 1998]. After treating newly diagnosed patients with human recombinant IFN-α, the honeymoon period was increased [Brod et al. 2001].

A smaller trial in Germany, followed by a larger European study, examined the effect of nicotinamide in T1DM patients. The nicotinamide can protect cells from IL-1-mediated cell destruction, damage due to free oxygen radicals, and can reduce the expression of major histocompatibility complex (MHC) II. In the Deutsche Nicotinamide Intervention Study (DENIS), 55 ICA-positive patients were followed for 5 years. No advantageous effect was experienced on the nicotinamide arm; even after 2 years, the first phase of insulin secretion was diminished, but remained in the placebo arm [Wilson and Buckingham, 2001]. In the European Nicotinamide Diabetes Intervention Trial (ENDIT), 552 children were involved [Schatz and Bingley, 2001], and again nicotinamide did not have any positive effect; however in contrast to DENIS, it was not noxious for the first phase of insulin secretion [Gale et al. 2004].

Monoclonal antibody treatment

Blocking any link of the autoimmune process (e.g. by antigen-presenting cells, certain cytokines, and effector cells) can have a beneficial outcome.

CD3 is a part of the TCR, thus it can be found on every T cell. With humanized anti-CD3 treatments the number of T cells decreased, and favorable clinical effects could be achieved, such as less insulin needed for treatment, and the decline of C-peptide levels lessened [Herold et al. 2005; Keymeulen et al. 2005]. Based on these results, a double-blind, multicenter, phase II placebo-controlled trial was conducted in Europe. Eighty recently diagnosed T1DM patients were randomized to a 6-day treatment with a humanized anti-CD3 antibody called otelixizumab or placebo. The administration of anti-CD3 efficiently preserved β-cell function. At 18 months of follow up, the insulin dose was ≤0.25 IU/kg/day, which means almost insulin independence. Few adverse effects, such as flu-like symptoms, were observed [Keymeulen et al. 2010a, 2010b].

The anti-CD20 (rituximab) is used in the treatment of B-cell lymphomas. It decreases the amount of CD20+ B cells, and thus it can also reduce auto-antibody production. Antigen presentation is also influenced by deceasing numbers of B cells [Perosa et al. 2005]. Pescovitz and colleagues carried out a double-blind study including 87 newly diagnosed T1DM patients, randomly allocated to receive four infusions of rituximab or placebo on days 1, 8, 15, and 22 of the study [Pescovitz et al. 2009]. After 1 year, in the rituximab group, the stimulated C-peptide values were higher, the HbA1c levels were lower, and the insulin needs were reduced compared with the placebo group [Pescovitz et al. 2009].

Antigen-based treatment: trials based on re-establishing antigen-specific self-tolerance

Autoimmunity can develop against every antigen, against which the immune system did not become tolerant, in other words, the auto-antigen was not or not properly presented in the thymus during the maturation of T cells, thus the autoreactive T-cell clones were not sufficiently removed. Previously hidden auto-antigens can be reached by the immune system after infection or trauma. The immune system can also be sensitized by molecular mimicry, which gives an explanation for the infection theory of T1DM pathogenesis [Chatenoud. 2010]. It is well known that the incidence of T1DM increases in spring and autumn. A few cases of recently diagnosed patients had antibodies for Coxsackie virus; moreover these viruses were also isolated from the pancreas. The vaccines for these microorganisms can be useful, but the high number of possible agents makes this idea impossible to carry out [Viskari et al. 2005].

The presentation of auto-antigens to the T cells at a high level, without costimulation, can lead to the inactivation of the autoreactive T cells, which can be used as a means of prevention. There many possible antigens, such as milk protein and bovine serum albumin, which show molecular mimicry with the surface proteins of the β-cells, Hsp, GAD65, and insulin [Singh and Palmer, 2005; Wong, 2005; Petrovsky et al. 2003].

17 amino acid-long bovine serum albumin peptide

One 17 amino acid-long part of the bovine serum albumin peptide is similar to one of the β-cell markers. Early milk consumption has been suspected to be a trigger in the pathogenesis of T1DM. In the Trial to Prevent Diabetes in the Genetically at Risk (TGIR) study the aim was to prevent the development of T1DM by using a milk-free diet. The results were contradictory, so the role of milk protein in T1DM etiology remains unclear [Wilson and Buckingham, 2001].

Hsp

In the 1990s, autoantibodies against GAD65 and hps65 were discovered in the serum of T1DM patients. A modified hsp65 oligopeptide (p277 DiaPep) was used successfully in the treatment of NOD mice. In a small human study in which the p277 was administered, better C-peptide levels and lower insulin requirements were achieved compared with the placebo group [Singh and Palmer, 2005].

GAD65

Diabetes was prevented in NOD mice using the transgenic GAD65 protein [Quinn, 2009; Singh and Palmer, 2005]. There are also studies of GAD65 used in human patients. In a phase II study involving 47 LADA patients, 4, 20, 100, and 500 µg GAD65 were administered in weeks 1 and 4. The participants were followed for 24 weeks. No side effects were observed. The higher C-peptide levels in the 20 µg group were measured, and the percentage of CD4 + CD25+ (presumably regulatory cells) had increased. These results were observed only with the 20 µg dose [Agardh et al. 2005; Singh and Palmer, 2005].

Insulin

The oral and parenteral administration of insulin prevented diabetes in NOD mice. One of the main epitopes of insulin is on the B chain between amino acids 9 and 23. Only this oligopeptide among the proinsulin oligopeptides prevented diabetes in the NOD mice. Of the islet-infiltrating T cells, 90% were reactive to the B9–23 peptide [Liu et al. 2002]. A large phase II clinical trial showed no effect of an altered peptide ligand of the 9-23 insulin B-chain peptide (NBI-6024) [Walter et al. 2009].

Some small human studies showed that subcutaneously administered insulin prevented, or postponed, the onset of T1DM [Fuchtenbusch et al. 1998]. Since the insulin is relatively safe and is not immunosuppressive, it gave the opportunity for a larger human trial. The Diabetes Prevention Trial (DPT) started in 1994, and 103,391 first- and second-degree relatives of T1DM patients were screened. Of these, 3483 were positive for at least one of the islet antibodies, and 2523 people had further genetic, immunological, and metabolic tests [Skyler et al. 2005; Schatz and Bingley, 2001]. The participants were enrolled in intermediate (oral insulin trial; 372 participants with 25–50% risk for T1DM in the next 5 years), and high (parenteral insulin trial; 339 participants with higher then 50% risk for T1DM in the next 5 years) risk groups. Intravenous (0.015 IU/kg/h for 4 days once a year) and subcutaneous (0.25 IU/kg/day, ultralente) insulin were administered to the high-risk patients. Oral insulin (7.5 mg/day) was given to the intermediate-risk participants. Unfortunately none of the treatments achieved beneficial results. However, the oral treatment was advantageous in the subgroup of intermediate-risk participants with high IAA (above 80 nU/ml). In this subgroup, fewer participants developed T1DM (6.2%) in the oral insulin arm, than in the placebo arm (10.4%, p = 0.015) [Orban et al. 2009; Barker et al. 2007; Skyler et al. 2005; Schatz and Bingley, 2001]. The data seem to support the notion that a better, more efficient way to apply antigen-based interventions is required. The combination of antigen plus adjuvant looks more promising than the use of antigen alone [Chatenoud, 2010].

Insulin B chain and incomplete Freund’s adjuvant

We have recently completed a double-blind, placebo-controlled, phase I clinical trial with insulin B chain and incomplete Freund’s adjuvant [Orban et al. 2010; Chatenoud, 2010]. We have hypothesized that intramuscular administration of the putative primary auto-antigen with the combination of adjuvant will re-establish immune tolerance in T1DM patients. The study drug, IBC-VS01, contained 2 mg insulin B-chain peptide in Montanide ISA 51-incomplete Freund’s adjuvant (IFA, Seppic Inc., Fairfield, NJ, USA) in a 50/50 (w/w) emulsion. Twelve subjects who had been diagnosed with T1DM within 3 months of enrollment, between the ages of 18 and 35 years, and who were positive for diabetes auto-antibodies, were enrolled at the Joslin Diabetes Clinic in Boston, MA, USA. The subjects were followed up for 104 weeks. Standard examinations, laboratory tests, questionnaires, metabolic tests, including HbA1c, insulin requirements, stimulated C-peptide, and a wide range of humoral and cellular immunological investigations were performed regularly. Our primary aim was assessment of safety. All patients completed the study as scheduled. Only one study-related adverse event was noted: a patient had a mild transient discomfort/pain at the injection site. There were no significant metabolic differences between the vaccinated arm and the placebo arm. The study was designed to test safety and was not powered to test the therapeutic effect of the study drug.

In addition to the excellent safety results, robust and potentially advantageous immunological changes were observed. The vaccinated patients developed a vigorous insulin-specific humoral and T-cell response. The T-cell response to the insulin B chain peaked at 24 weeks, then slowly declined but remained positive for up to 2 years of follow up. In the supernatant of these insulin B chain-stimulated T cells, an excess of the transforming growth factor beta (TGF-β), a potent regulatory cytokine, was detected. The most prominent response observed was to the full length B chain, a smaller T-cell response was recorded to the proinsulin or some oligopeptide fragments of the B chain, including B9–23. Insulin-specific potent regulatory T cells were isolated from the vaccinated participants, but not from the patients of the placebo arm [Orban et al. 2010]. These insulin-specific regulatory T cells secrete potent regulatory cytokines such as IL-10 and TGF-β. Our limited data also suggest that these insulin-specific potent regulatory T cells are functional. Our results showed that the immune response is more effective to all the B chain, than to the B9–23 fragment, which was considered to be the main epitope of insulin as this is supported by data derived from NOD mice [Liu et al. 2002]. Previously the altered 9–23 insulin B chain (NBI-6024) did not have any beneficial effect in a large phase II clinical trial [Walter et al. 2009]. We anticipate that our novel vaccine, which uses a unique adjuvant (incomplete Freund’s adjuvant) and the full length B chain, will be more successful because it evokes a powerful immune response.

Summary

Patient enrollment during the T1DM primary prevention studies is difficult because only 0.7–0.8% of patients among the relatives of T1DM patients develop T1DM [Wilson and Buckingham, 2001]. Genetic, immunological, and metabolic tests can be helpful in finding patients at higher risk. A further problem is the lack of an appropriate animal model. There are many successful treatments and prevention methods using NOD mice, but those failed to yield the same results in humans [Serreze and Chen, 2005]. The pathogenesis of T1DM is different in humans and NOD mice. In NOD mice the genetic factor is much stronger, there is a female predominance, and the inflammation of the islets is different and more profound than in humans. The T1DM rat model is the BioBreeding (BB) rat, and in this animal the diabetes develops with the lack of CD8+ cells, thus its pathomechanism is rather different [Serreze and Chen, 2005]. Using other animals, such as dogs or monkeys, is not feasible for ethical and financial reasons.

Among the prevention and intervention studies, the most promising are those based on the re-establishment of immune tolerance to diabetes-specific self-antigens. The identity of the initiating auto-antigen is still under debate. However, increasing data indicate that the primary auto-antigen in humans is insulin or proinsulin. It is conceivable that combination therapies may need to be used to tackle autoimmunity; a two-pronged approach. One such combination may use immune suppression to reduce auto-aggressive T-cell populations followed by an antigen-based therapy to boost antigen-specific regulatory T-cell populations. Meanwhile, the instrumentation of insulin administration is constantly developing; probably in the near future an artificial pancreas will be available, which will be able to deliver an optimal treatment for diabetes, but will not cure the disease itself [Hovorka, 2006]. Therefore, a full understanding of the exact pathogenesis and the prevention and cure of T1DM remain the ultimate challenge.