The cyclosporin A and the steroid-azathioprine trials proved that established T1DM was reversible [98
] and, more importantly, that a reserve of beta cell mass was able to restore normoglycemia contingent on suppression of inflammation and autoimmunity [101
]. Nevertheless, once cyclosporin A administration was suspended/terminated, the disease reappeared with a vigor no different than that during pretreatment. From a historical perspective, the first successful immunomodulation-based reversal of T1DM was achieved by lymphocyte-specific serum in BB rats [103
]. It would be 20 years later that this approach would be attempted in humans [104
], although the significant side effects precluded the justification to further explore this approach. Since then, a number of preclinical studies, mainly in NOD mice, have been clinically-translated.
In 1985, Eisenbarth and colleagues reported that steroid-supplemented antithymocyte globulin administration into new-onset T1DM patients reduced insulin requirements. However, this approach was abandoned due to the thrombocytopenia that the procedure was generating [105
]. Since then, antithymocyte globulin has been reconsidered and, in new-onset patients, carefully adjusted dosing slowed C-peptide decline [106
]. Antithymocyte globulin binds to the CD3 complex on T cells as one of its many targets and causes significantly greater T-cell depletion. Thus, its justification is hampered by significantly greater adverse events and immunosuppression compared to anti-CD3 antibodies.
Long before confirmation in genetic models, insulin and proinsulin were strong contenders as the T1DM-initiating autoantigens [107
]. As autoantigens, many proposed that the exposure of the T1DM immune system to exogenous insulin in a manner that could modulate immunity towards insulin-specific tolerance could prevent disease in prediabetic states and perhaps delay the progression to overt clinical hyperglycemia in more advanced, but subclinical, states [115
]. Persistent oral insulin treatment of NOD mice delayed T1DM onset and reduced disease incidence in NOD mice [119
]. The addition of adjuvants provided a practical benefit in reducing the amount of exogenous insulin required to achieve the beneficial effects [120
]. Largely based on the data supporting the view that mucosal delivery of soluble peptides promotes tolerance to them, a number of efforts targeted, in addition to oral, intranasal insulin or insulin-derived peptide, aerosolisation into prediabetic NOD mice. In these studies, diabetes was significantly delayed [71
]. These very promising studies with what essentially was a simple intervention led to the DPT-1 multicenter trial which determined the efficacy of oral insulin in first- and second-degree relatives of T1DM patients deemed to fall into high-risk status based on metabolic, immune, and genetic evidence [73
]. Other than a possible benefit in individuals with the highest autoantibody titers, the DPT-1 study failed to delay or prevent T1DM. A similarly disappointing outcome resulted in another prevention trial where insulin was administered intranasally Näntö-Salonen et al, [124
]. However, a very small subgroup of autoantibody-positive patients was identified in which some effect was shown. The general view of nonpharmacologic insulin therapy is that, if beneficial, it probably will be restricted to very well-defined and characterised subpopulations of patients. It is unclear what can distinguish these patients from a general at-risk population.
A number of other trials were initiated since then without any significant benefits. The IMDIAB trial showed that oral insulin provided no benefit over placebo after a one-year followup in terms of mean C-peptide secretion and insulin requirements. The ORALE trial, comparing low- with high-dose oral insulin (2.5
mg/day versus 7.5
mg/day), after one year could not discern any benefit in decelerating the loss of physiologic beta cell function [125
]. It was recently shown that combined intranasal insulin with CD3 antibody was able to reverse new-onset T1DM in NOD mice, although the effect was quite likely due to the CD3 antibody and the characteristics of the NOD mouse cohorts and not due to the insulin [126
The most recent attempts at using insulin to improve functional beta cell mass involve intramuscular injection of human insulin B chain in incomplete Freund's adjuvant as well as subcutaneous injection of a DNA plasmid vector encoding proinsulin see [127
] and http://www.bayhilltherapeutics.com/
. The outcomes of these proposals rest on the publication of phase II studies which are currently at various stages of implementation.
As strong an argument for GAD can be made in the etiopathogenesis of T1DM as for insulin. The 65
KDa isoform of GAD has been demonstrated to be a target of early-insulitic T cells in vitro
and in vivo
]. Administration of GAD into very young NOD mice suppresses anti-GAD T-cell reactivity as well as disease onset [129
]. GAD is equally effective when administered into older NOD mice [130
]. The relevance of GAD to human T1DM relies on the presence of GAD autoantibodies in prediabetic humans and is one of three reliable markers of susceptibility [131
Two clinical trials have used human GAD65 with alum as adjuvant to determine improvement of physiologic beta cell function. One was conducted in LADA patients [132
] and determined that this formulation increased fasting and stimulated C-peptide at 24 weeks compared to baseline, a benefit that was associated with an increase of CD4+ CD25+ Tregs. The other study considered the effects of GAD65 in alum in recent-onset T1DM individuals (10–18 years of age) on the rate of decline of stimulated C-peptide [133
]. The study revealed a slower rate of decline of stimulated C-peptide in GAD-treated diabetics compared to the placebo group.
Early studies showed that intrathymic administration of Hsp60-derived peptides was prophylactic in NOD mice even though Hsp60 was not considered a bona fide autoantigen involved in T1DM initiation [134
]. More experiments in NOD mice revealed that the p277 peptide of Hsp60 was quite effective in suppressing disease progression [135
]. These data compelled the development of a human equivalent of p277 (DiaPep277; a 24 amino acid synthetic peptide derived from the C terminus of human hsp60).
DiaPep277 has been evaluated in phase II studies [136
], and data demonstrate some degree of preservation of stimulated C-peptide secretion.
Since then, three other studies have been conducted in new-onset T1DM patients [137
]. In two of these studies, adult patients were the study population, and these recipients exhibited better, yet limited, C-peptide production than placebo-treated individuals. In contrast, DiaPep277 did not offer any objective benefits in young new-onset patients [137
T-cell depletion targeting the CD3 complex with the OKT3 monoclonal antibody was successfully employed about two decades ago [91
]. Administration of the antibody into diabetic NOD mice resulted in complete remission of disease [91
]. One caveat of this particular antibody was that, under some situations, it could activate T cells and therefore it was not considered suitable for human use. Soon thereafter, FcR nonbinding embodiments were manufactured which exhibited either IgG1 Fc chain or which eliminated glycosylation sites of the original OKT3 clone. These variants were found to be less activating than OKT3 [140
]. Herold and colleagues initially demonstrated that a short-term administration of the nonglycosylated variant into new-onset T1DM patients suppressed the loss of beta cell function as measured indirectly by surrogate physiologic markers [90
]. Some patients maintained improved physiologic markers of beta cell function for as much as two years following the treatment in the absence of systemic immunosuppression. A second round of anti-CD3 antibody injections was postulated as necessary to maintain the beneficial outcome. However, additional injections of the antibody were shown to confer more serious, previously undocumented clinical complications.
T1DM is unquestionably a T-cell-mediated disease; however, other cell populations have been implicated in the onset, and early progression, of the disease, including B-lymphocytes. Accumulated findings in NOD mice demonstrated that B-cell depletion could be beneficial [143
], even some studies suggested the contrary [146
] where adoptive transfer of T1DM into NOD-SCID mice was possible in the absence of B cells and antibodies. B cells can also participate in disease exacerbation by promoting epitope spreading as shown by Tian and colleagues [147
]. Indeed, capture of autoantigens such as GAD65 by B-cell surface immunoglobulin (Ig) is followed by processing and presentation of T-cell determinants by B cells, a step that is crucial for activation of autoreactive T cells and induction of diabetes [148
]. B-cell-mediated processing of self-Ags may contribute to the generation of an inflammatory microenvironment in the pancreas, which is critical for overcoming the regulatory barrier(s) to initiation of diabetes in NOD mice [151
]. Antigen-specific B cells and their antibodies are essential in catalyzing determinant-spreading reaction via
(a) generation of novel previously cryptic epitopes through altered antigen processing or (b) facilitation of T-cell activation through generation of ligands with higher affinity for TCR and delivery of costimulatory signals. Ag processing through surface receptor-mediated internalization, for example, Fc receptors and soluble Ig, is different from that occurring through pinocytosis and phagocytosis [152
Pescovitz and colleagues demonstrated a small but statistically significant improvement in physiologic markers of beta cell function in new-onset patients [65
]. Nevertheless, the improvement was transient as placebo and rituximab-treated patients eventually (by two years) exhibited an identical decline in beta cell function. One possible mechanism could involve the restriction of epitope spreading by B cells after rituximab-mediated deletion, or the repopulation by B-cells that process and present epitopes that compete with diabetes-relevant epitopes.
Autoreactive T cells have been frequently associated with the production of TH1-type proinflammatory cytokines. It is reasonable, therefore, to use cytokine blockade as a unique or part of a combination therapeutic strategy. IL-1beta, TNFalpha, and type 1 interferons have been historically the first to exhibit direct beta cell cytotoxicity [153
Recently, administration of etanercept, a soluble TNFalpha receptor, in an early-phase efficacy trial resulted in a reduction in the loss of C-peptide in T1DM patients [156
]. It is important to note, however, that TNFalpha blockade may have unwanted outcomes; TNFalpha can prevent diabetes in older NOD mice even when splenocytes from diabetic NOD mice are adoptively transferred [157
]. Furthermore, TNFalpha blockade may require careful consideration of the age of the individuals being treated since blockade in younger NOD mice prevented disease, while in older NOD mice it accelerated T1DM.
Even though IL-1beta is now known to exert early-onset impairment in beta cell function and mediates early-phase recruitment of immune cells into islets [158
], it is unclear if IL-1beta blockade will have any beneficial outcome in prevention of disease or reversal of new-onset disease. In certain populations characterised by metabolic diabetes (LADA, type 2 diabetes), recent trials with the IL-1 receptor antagonist protein (IRAP, Anakinra) improved glucose control [162
], and mechanistically it appears that this benefit is achieved by blockade of IL-1beta impairment of immune cell activity (i.e., inflammation) and blockade of IL-1beta effects on beta cell impairment. It is unclear if administration of the IL-1 receptor antagonist protein (Anakinra) will be of any benefit in new-onset cases, especially as the autoimmunity in new-onset disease is largely IL-1beta independent.