Antigen-specific immunotherapy and influenza vaccination in type 1 diabetes: timing is everything
KeywordsAntigen-specific immunotherapy Autoimmunity GAD-alum Influenza Pandemrix Timing Type 1 diabetes Vaccination
Human Immune Response Dynamics
Antigen-specific immunotherapy (ASI) has long been hailed as a promising strategy to induce or restore immune tolerance to beta cells, the loss of which underlies the development of type 1 diabetes. The approach is based on the observation that, in preclinical models, repeated administration of an antigen over days or weeks leads to a state of immunological unresponsiveness. When the antigen used is a major target of the immune response in an autoimmune disease, the result is tolerance and disease prevention.
In the type 1 diabetes space, there are several efforts under development to realise the potential benefits of ASI, including daily oral insulin, peptide antigen immunotherapy (e.g. proinsulin peptide) and GAD conjugated to alum [1, 2, 3, 4]. GAD is well established as an important autoantigen in type 1 diabetes, and anti-GAD antibodies and GAD-reactive T cells are early features of the disease process. GAD-alum is an ASI strategy designed to induce T helper (Th) 2 counter-regulation of the pathological Th1 autoimmune process that is thought to underlie beta cell destruction. Manipulation of such autoimmune networks, which may be finely balanced between health and disease, presents an important opportunity to learn about self-reactivity and how it may be influenced. In the current issue of Diabetologia, Tavira et al describe such an opening .
GAD-alum for the prevention of beta cell loss in type 1 diabetes
The potential therapeutic effect of GAD-alum in preventing beta cell loss in individuals with a recent diagnosis of type 1 diabetes has been evaluated in several studies, from Phase I to Phase III. Whilst there is clear evidence that GAD-alum increases GAD autoantibody levels and induces GAD-specific Th2 responses, the clinical results have been somewhat equivocal [1, 6, 7, 8]. The pivotal study is a Phase III GAD-alum trial in individuals with new-onset type 1 diabetes, which failed to meet its primary efficacy endpoint of preserving C-peptide production . However, important subgroup analyses of this study revealed that beneficial effects of GAD-alum use were indeed observed in individuals from non-Nordic countries, but not in the Nordic participants who comprised nearly half of the study cohort. The study by Tavira et al  provides data to argue that this discrepant outcome is not a result of geographical differences but of different influenza protection strategies in the various countries participating in the clinical trial.
Influenza vaccination and GAD-alum efficacy
The impact of influenza vaccination on GAD-specific immune responses
Intriguingly, differences in the GAD-specific immune response and C-peptide in participants vaccinated with the H1N1 vaccine and GAD-alum <5 months or >5 months apart were only observed in the individuals who received two doses but not four doses of GAD-alum. It is plausible that four doses of GAD-alum are sufficient to overcome any modulatory effects of Pandemrix that were observed in the two-dose group. The authors also noted that they had previously observed greater C-peptide retention in individuals who received two doses vs four doses of GAD-alum . Furthermore, levels of H1N1 antibodies were significantly reduced in individuals vaccinated with Pandemrix close to receiving GAD-alum in the group receiving four doses of the influenza vaccine, implying that responses to both GAD and influenza were reduced as a result of the close proximity of the two interventions.
Cellular mechanisms for GAD-alum/influenza vaccine interference
This study raises the possibility that influenza vaccination interferes with ASI, a phenomenon that is both intriguing from a mechanistic standpoint and worrying from a clinical and drug development vantage. It is known that the co-administration of vaccines can result in immunological interference  and it will be of interest to understand how such interference mechanisms apply in the setting of ASI. Sequential administration of vaccines can result in the suppression of a response to one antigen at the expense of another or reduced responses to both antigens. Such effects depend on the relative amounts of antigens administered, the relative sites of vaccination and the time interval between antigen administration. For example, an impaired response to one antigen over another upon co-administration of vaccines may arise as a result of competition between multiple antigen responses for limited resources within the lymph nodes, such as access to antigen, chemokines, activation signals, follicular dendritic cells and follicular Th cells, thereby affecting the level of T cell help, B cell activation and antibody production directed towards a given antigen. Alternatively, co-administered vaccines may change the balance of Th cell subsets, since different antigens may induce mutually antagonistic Th1 or Th2 cytokine responses, or induce regulatory T cell mechanisms that could inhibit the immune response to one of the antigens administered.
Tavira and colleagues speculate that the interference between influenza vaccination and GAD-alum administration observed in this particular setting could be attributable to the potent immunomodulatory effects of the adjuvant used in Pandemrix, AS03 . AS03 has been shown to enhance the vaccine’s antigen-specific adaptive response by locally activating the innate immune system and by increasing antigen uptake and presentation in draining lymph nodes . Since adjuvants act directly as immune stimulants, there is a possibility that they induce undesirable immune responses that could trigger the onset of immune-mediated disease in susceptible individuals. Indeed, following vaccination with Pandemrix during the 2009 H1N1 pandemic, some European countries, including Sweden and the UK, reported the emergence in a small number of cases of narcolepsy, an immune-mediated destruction of hypocretin-secreting neurons in the hypothalamus. Pandemrix elicits a transient, rapid and expansive activation of myeloid cells and effector cells, similar to the responses induced by other vaccines, including non-adjuvanted influenza vaccines [15, 16, 17, 18, 19, 20]. However, it also differs from other vaccines; the Human Immune Response Dynamics (HIRD) study in our laboratory showed that Pandemrix provokes an overt early-lymphoid response within 24 h of vaccination, with prominent upregulation of IFN-γ transcription, which has not been observed in most other virus vaccine studies . Early-lymphoid- and myeloid-dominated responses were followed by a plasmablast-dominated response at day 7. Therefore, in participants vaccinated with Pandemrix soon before or after starting GAD-alum treatment, the notion that Pandemrix could affect GAD-associated immunity is, perhaps, not surprising, given that the outcome of auto-immunisation depends on the pre-existing immune background. For example, in the absence of inflammatory stimuli, an increased production of TGF-β and other immunomodulatory cytokines in response to Pandemrix may favour the generation of a tolerogenic response to administered autoantigen. On the other hand, a pre-existing pool of activated effector T cells and an aggressive proinflammatory response initiated by Pandemrix might instead aggravate autoimmunity and even accelerate autoimmune beta cell destruction.
Intriguingly, the HIRD study also found that adverse events arising from Pandemrix vaccination were associated with an atypical pre-vaccination B lymphocyte phenotype, with an expanded transitional B lymphocyte pool (which is seen in autoimmunity) and higher levels of autoantibodies . This observation might suggest that non-specific immune activation is enhanced in individuals with an autoimmune background, and a similar phenomenon might have contributed to the effects of Pandemrix in the GAD-alum study.
Application of findings
An important issue will be how this set of serendipitous observations from Tavira et al can and should be built upon. The first question is whether greater mechanistic insights can be gained in future studies, and the second is how it will impact upon the design of ASI trials. It is likely that new studies in man to address the many arising mechanistic questions are precluded by safety and feasibility issues and preclinical models may not be helpful in this setting. To further complicate matters, Pandemrix is no longer in use. Perhaps the best that can be achieved is to obtain robust vaccine record data in ASI studies and consider relevant sub-analyses in relation to study outcomes. What are the other implications for the design of ASI trials in type 1 diabetes? Perhaps it is reasonable for routine infectious-disease vaccines to be avoided for a period of 6 months before or after the ASI treatment period. However, in some settings (e.g. The Diabetes Prevention Trial of Type 1 Diabetes [DPT-1], an oral insulin study; ClinicalTrial.gov registration no. NCT00004984) antigen administration is continuous over months/years and withholding vaccines is neither practical nor desirable. In addition, it is probable that natural exposures to infectious agents could have adverse effects on ASI and these remain beyond easy control. The study by Tavira et al highlights the need to consider the immune background on which ASI is administered more generally and emphasises the yin and yang of immune responsiveness in seeking to balance aggression and regulation in the face of complex external stimuli.
Duality of interest
Related work in our laboratory receives funding from the Innovative Medicines Initiative 2 Joint Undertaking under grant agreement No 115797 INNODIA. This Joint Undertaking receives support from the European Union’s Horizon 2020 research and innovation programme and ‘EFPIA’, ‘JDRF International’ and ‘The Leona M. and Harry B. Helmsley Charitable Trust’. Support from the National Institute of Health Research Biomedical Research Centre Award to Guy’s and St Thomas National Health Service Foundation Trust and King’s College London is also acknowledged.
The authors declare that there is no duality of interest associated with this manuscript.
Both authors were responsible for drafting the article and revising it critically for important intellectual content. Both authors approved the version to be published.