Thus, it would be unreasonable to expect a stronger effect (in other words, after onset of diabetes) in humans. Secondly, no preclinical study ever tested the clinical GAD-Alum preparation, and no efficacy was noted in our recent studies in NOD and B6 diabetes models (Pagni, Boettler and von Herrath, unpublished). 3-deazaneplanocin A Again, it is probably unreasonable to expect an antigenic formulation to work in humans when it does

not even prevent diabetes in otherwise permissive animal models. Several other theories have been proposed to account for the failure of GAD-Alum in humans, including the lack of GAD expression in β cells; this is a controversial area, as many studies have demonstrated expression of GAD-65 and 67 proteins in murine and human β cells [36]. Lastly, one could ask whether the dose of GAD-Alum was sufficient – as most patients mounted a clearly detectable immune response, this appears less likely. However, alum might have been a suboptimal adjuvant for an ASI, as the resulting mixed but T helper type 2 (Th2)-dominated cytokine response of induced GAD-reactive T cells (Arif, Selleckchem Cetuximab Roep and Peakman, unpublished) did not result in protective cell populations. In the absence of a functional mouse model of GAD-Alum preventing diabetes, it will be difficult at this point to clarify these issues. The question of the antigenic dose might have more bearing on the

issue of efficacy with oral insulin [15]. As predicted from animal models [37], prophylactic oral ioxilan insulin given at a daily dose of 7·5 mg had a very marginal effect in preventing diabetes in individuals at high risk (exhibiting multiple autoantibodies [38-41]), but not in any other patient groups. However, as has been evident from multiple studies in different mouse models, oral insulin dosages have to be comparatively

much higher to induce optimal disease preventive effects, which are seen at a dose of 1 mg given twice per week [42]. This dose would equate to approximately 1 g of oral insulin twice per week in humans. In addition, it is likely to be necessary to provide the drug in enteric-coated capsules, without which > 99·99% of the insulin is lost through digestion in the stomach and only minimal amounts of intact antigen or some peptides will reach the lower gut and the Peyer’s patches, the location at which oral insulin has been shown to induce its desired immune-regulatory response. Therefore, more precise dose calculations should have probably preceded the oral insulin trial and its current follow-up study. A further human/mouse mismatch relates to the overall management of expectations when devising trials for ASI. In rodent studies most, if not all, ASI is effective only for early and, at best, late prevention of disease, but never after onset of hyperglycaemia. Thus, we should not expect antigens to reverse human diabetes or even preserve C-peptide after onset (at least with effects detectable in reasonably sized studies); and this has indeed been the case.