Type 1 diabetes (T1D) is a chronic disorder requiring lifelong treatment. Despite intensive treatment, the disease causes substantial morbidity and mortality. Several clinical intervention trials with different approaches to delay or halt disease progression have shown limited efficacy, sometimes accompanied with adverse events (1–6). As an alternative approach to target and limit the systemic side effects of T1D treatments, autoantigens therapies to induce immunologic tolerance has been considered (7, 8).

Use of 65-kD isoform of glutamic acid decarboxylase (GAD₆₅) formulated in alum did show encouraging results in a Phase II trial (9), but did not accomplished the expected efficacy in subsequent studies (10, 11). To improve the treatment, GAD-alum was administrated directly into the lymph nodes of patients with recent-onset T1D in an open-label pilot study (DIAGNODE-1) (12). The patients showed decreased C-peptide depletion and HbA1c, less insulin requirement, as well as immunological changes (12). Comparison of participants in DIAGNODE-1 with previous studies, where GAD-alum was administrated subcutaneously at higher doses, showed a somewhat better clinical course and clear differences in the immunological response (13). After these encouraging results, a larger randomized, double-blind, placebo-controlled phase II trial (DIAGNODE-2) was consequently performed. The treatment did not change disease progression in the entire cohort but appeared to be beneficial in patients with DR3DQ2 haplotype, whose preservation of β-cell function was significantly improved compared to placebo treated patients (14).

Here we describe treatment effect on the immune response of patients who participated in the DIAGNODE-2 trial, comparing the effect in the GAD-alum treated patients in relation to the placebo group, as well as focusing on the immunomodulatory effect of GAD-alum in the patients both according to the presence or absence of the DR3DQ2 haplotype and according to the preservation of beta cell function.

The DIAGNODE-2 study constituted the first placebo-controlled trial in which an autoantigen was administrated into the lymph nodes of patients with Type 1 diabetes, and immunological changes observed in GAD-alum treated patients with respect to placebo have not been previously described. Modifications following therapy included an increment of GADA, GADA subclasses, and GAD₆₅-induced proliferation and cytokine secretion in GAD-alum treated individuals, with higher titers than the placebo group. GAD₆₅-induced secretion of a broad range of cytokines supports the idea of immunomodulation of GAD-specific immune responses following administration of low doses of GAD-alum directly into the lymph nodes.

Although no significant treatment effect was seen in the whole study cohort during the clinical trial, a statistically significant preservation of beta cell function was found in the prespecified subgroup of GAD-alum patients with DR3DQ2 haplotype (14). This is in line with previous findings indicating the influence of HLA genes in T1D (16, 21). Thus, this prompted us to focus on the analysis of the immune responses in GAD-alum treated individuals according to this classification. Strikingly, higher GAD₆₅-induced proliferation and secretion of several cytokines was observed in patients with DR3DQ2, compared to those without DR3DQ2 haplotype after the second injection of GAD-alum. Notably IL4, IL5 and IL13, which are major effector cytokines produced by Th2 cells, were among them. It is possible that the generation of an early predominant Th2 and anti-inflammatory response to GAD₆₅ might counteract proinflammatory factors and generate an environment where autoreactive Th1 effectors cells could be suppressed, thereby restoring immunological balance. For instances, IL4 mediates signaling to promote Th2 differentiation (22), while IL13 has potent anti-inflammatory activities, both in vitro and in vivo, and possesses several unique effector functions (23). Together with IL13, IL5 have a central role in immune regulation and differentiation, governing the onset of inflammatory responses and providing signals that help turn off chronic inflammation and protect tissues from ongoing damage (24, 25). Interleukin 10, well known for its potent anti-inflammatory effects, inhibits activation, proliferation, and production of pro-inflammatory cytokines on T cells, in addition to promoting survival, proliferation, differentiation (26). Other cytokines as IFNγ, GM-CSF and IL7 were also part of the earlier cytokine secretion induced by GAD-alum in the DR3DQ2 positive patients. In line with results from our studies, samples from individuals treated with GAD-alum from another clinical trial (10) secreted IL4, IL5 and IL13 Th2-associated cytokines, together with IFN and IL17 (27). The same study also showed that the majority of expanded GAD-specific cells exhibited a hybrid phenotype and were both IL13⁺ and IFN-γ⁺ (27). Thus, it might be argued that responses generated by the treatment represents a non-desirable effect. However, the immune cell network is complex, and cytokines have paradoxical functions. For instances, among the wide-ranging effect of IFNγ on the innate and the adaptive immune system (28), this cytokine can exert different effects depending on their concentrations and microenvironment (29). It has been suggested that low levels of IFNγ may protect animal models from autoimmune diabetes (30, 31). Thus, under the right circumstances, cytokines can exhibit either Th1 or Th2-promoting activities. IL7 is important for B and T cell development and proliferation (32–34), and that has been connected to immune recovery following stem cells transplantation (35). Higher levels of GM-CSF were also detected in the same supernatants, a cytokine that might alter the Th1/Th2 cytokine balance in both directions (36). GM-CSF is known to play an important role in the differentiation of dendritic cells (DCs) and induction of tolerance (37–40) and has a suppressive effect on autoimmune diabetes in mice (41). Indeed, the quality of T cell response induced to vaccination has been connected to multiple cytokines (42), and protective vaccines induced CD4 T cells able to secrete several cytokines while non-protective mostly trigger T cells that produce one or two cytokines (43, 44). Measurement of multiple cytokines simultaneously in cell supernatants by Luminex enabled us monitoring samples from all participants throughout the study, getting a comprehensive depiction of changes induced by intra-lymphatic injections of GAD-alum. In that way, we were able to identify cytokine secretion early after treatment in samples from individuals carrying the DR3DQ2 haplotype with better C-peptide preservation as main finding. However, Luminex does not distinguish the source and extension of cytokines at the single cell level as intracellular cytokine staining does. Thus, as next step we will perform high-dimensional phenotyping in PBMCs selected according to the presence or not of DR3DQ2 haplotype and preservation of C-peptide secretion versus placebo, with focus on the response early after treatment.

The increasing consensus on the heterogenicity of T1D, that can reflect diverse immunological pathways to disease, brings focus to the matter that, as in many other autoimmune diseases, many of the patients participating in T1D clinical trials benefit from the treatments, while others do not (9, 45–47). Most clinical trials evaluating autoantigen immunotherapy have so far failed or shown inconclusive results (9–11, 48, 49). One reason might be that studies often fail to consider T1D as a heterogenous disease (50), and subtypes presenting distinct underlying pathobiological mechanisms, disease endotypes, should be considered in the design and evaluation of clinical trials (51). In that sense, identifying individuals who are most likely to have benefit from a certain treatment will help to develop experimental therapies and personalized approaches. Another big challenge in human studies with autoantigens is optimization of the treatment, including among others the definition of administration routes, doses, frequency and the use or not of adjuvants. The use of additional strategies as DNA-based delivery of autoantigens (52) or the use of tolerogenic dendritic cells pulsed with islet antigen, are also promising strategies as they might boost the efficacy of antigen specific immunotherapy (53, 54).

In an attempt to find biomarkers for clinical response, GAD-alum treated individuals were classified according to their preservation of C-peptide. Comparison of the immunological response revealed that patients displaying a better beta cell preservation had a more pronounced GAD₆₅-specific immune response at 3 months. Moreover, this early immune response was more pronounced in individuals with the DR3DQ2 haplotype. Intriguingly, better preservation of residual beta cell function was associated with increased GAD₆₅-induced IL13 secretion and proliferation. Thus, it is feasible that early immune responses in individuals with DR3DQ2 haplotype determine the shape of the memory responses to GAD-alum. To generate adequate immunity after vaccination, early innate immune responses are crucial for subsequent signaling for T cell activation and adaptive immune development. Antigen‐specific T cell frequencies peak between days 7 and 14, with only a small percentage of cells surviving as long‐lived memory cells, and their repertoire is determined by selection mechanisms during expansion (55). HLA is an essential checkpoint in antigen presentation and in the shaping of the adaptive immune repertoire (56, 57). The influence of HLA genotype in the immune responses following vaccination has been reported for other vaccines as rubella (58) and anthrax vaccine (59). Although activation of regulatory mechanisms capable of suppressing harmful autoreactive responses is a desired effect of antigen-specific immunotherapies (60), this hypothesis has been generated from experimental animal models but has not yet been demonstrated in humans so far. Deep phenotyping of major immune cell populations in a previous pilot trial showed that the immunomodulatory effect of GAD-alum injections into the lymph nodes did no induced any effect on Tregs, in line with results from previous studies (12, 61–63). Changes induced by the treatment included instead expansion of follicular helper T cells (Tfh), and enhancement of CD8+ T cells with Th2 phenotype in parallel to increased exhausted CD8+T cells (61). Characterization of GAD-specific CD4 T cell clones from 4 patients who received GAD-alum (10) showed that the majority of the clones had characteristics of bifunctional Th1/Th2 cells, leading to the speculation that they might lack immune-modulating properties and hence may not be capable of modifying autoimmunity in a clinically favorable way (27). Thus, it is possible that the therapy does not necessarily induce tolerance or unresponsiveness but rather modulates the quality of the existing immune response to the antigen, resulting in the reduction of deleterious effector functions.

In a previous study we have shown that administration of GAD-alum subcutaneously induced an early Th2-skewed immune response associated to clinical benefit (64). Inconsistent results from following trials led to the search of an alternative approach to improve treatment efficacy, by direct administration of GAD-alum into the lymph nodes (12, 14, 20). The immunological impact of the antigen can be improved by this approach, but it is known that administration route of antigens may not only govern the strength but also the type of the immune response (65). Intra-lymphatic administration delivers more antigen to the site of immune response induction, and difference in antigen dose available for stimulation of antigen-specific T cells may also lead to a predominant Th2 response in responder individuals (55).

Although better preservation of C-peptide and an improvement in several clinical parameters was observed in GAD-alum treated patients with DR3DQ2 haplotype (14), our exploratory analysis showed that some of them has a faster loss of c-peptide secretion. For the analysis, GAD-alum treated individuals were classified as “Good responders” and “Poor responders” according to their preservation of C-peptide (14, 20). However, we should keep in mind that the differences in preservation of beta cell function can sometimes be explained by faster disease progression and baseline characteristics in some individuals, and not directly related to the treatment. It is also likely that pre-existing characteristics other than the HLA genotype might determine the immunological and clinical impact of GAD-alum treatment. Indeed, a possible manifestation of endotypes in T1D was evidenced by birth cohort studies wherein the emergence of autoantibodies towards insulin and GAD₆₅ appears to be linked with the HLA DR4DQ8 and DR3DQ2 haplotypes respectively (51, 66). It is feasible that the quality and size of specific immune responses triggered by GAD-alum is determined both by the direct administration into the lymph node and pre-existing phenotypes and genotypes in individuals who benefit from the treatment. For instance, the influence of genetics on the enhancement of effector response to antigens might be related to polymorphisms in the IL-2 and/or TCR signaling pathway (67), rendering T cells more responsive to low levels of stimuli and resulting in the increase of the effective pool size of potentially antigen-responsive precursors. Given the functional role of MHC molecules in the initiation of the immune response, the DR3DQ2 haplotype may also influence the number of GAD₆₅ peptides presented by antigen presenting cells. It might also be possible that antigen-specific memory T cells have to compete with preexisting cells for survival factors. Thus, higher frequency of effector T cells when autoantigens are administered might not be desirable, as shown in a in a previous study where individuals with poor response to GAD-alum had a predominant proportion of CD4+ TEM pre-treatment (12). These are important questions that should be addressed in future studies. Our results highlight the importance of incorporating biomarkers into clinical trials. Immune surrogates associated to the response to interventions can be useful in the design of sequential trials and accelerate the development of effective immunotherapies for T1D (60, 68).

In conclusion, here we show that patients with DR3DQ2 haplotype had an early immune response to GAD-alum injections into the lymph node, and predominant GAD₆₅-induced IL13 secretion and proliferation that seems to be associated to a better clinical outcome. Better preservation of residual beta cell function was associated with increased GAD₆₅-induced IL13 secretion and proliferation. Our results require confirmation in the ongoing larger randomized double-blind placebo-controlled clinical trial (DIAGNODE-3) including only patients carrying DR3DQ2 haplotype. If confirmed, our findings might be used as early surrogate markers for clinical efficacy. It is possible that further immunological or genetic parameters can improve patient selection in future T1D therapies with autoantigens.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

The studies involving human participants were reviewed and approved by the relevant regulatory authorities and research ethics boards of the participating sites and countries. Written informed consent to participate in this study was provided by the participants’ legal guardian/next of kin.

RC designed the experiments. SP-M, FD, and PA performed experiments. SP-M and HB analyzed the data. SP-M and RC interpreted the results and wrote the manuscript. JL conceived DIAGNODE-2 study, recruited, and followed patients. All authors contributed to the article and approved the submitted version.