• Open Access

Making progress: preserving beta cells in type 1 diabetes


Carla J. Greenbaum, M.D., Benaroya Research Institute, 1201 9th Ave, Seattle, WA 98101. cjgreen@benaroyaresearch.org


The clinical care of patients with type 1 diabetes (T1D) has greatly improved over the past few decades; however, it remains impossible to completely normalize blood sugar utilizing currently available tools. Research is underway with a goal to improve the care and, ultimately, to cure T1D by preserving beta cells. This review will outline the progress that has been made in trials aimed at preserving insulin secretion in T1D by modifying the immune assault on the pancreatic beta cell. Although not yet ready for clinical use, successful trials have been conducted in new-onset T1D that demonstrated utility of three experimental agents with disparate modes of action (anti-T cell, anti-B cell, and costimulation blockade) to preserve insulin secretion. In contrast, prevention studies have so far failed to produce positive results but have shown that such studies are feasible and have identified new promising agents for study.


Immune-mediated type 1 diabetes (T1D) affects an estimated 900,000 people worldwide, and its frequency is increasing throughout the developed world.1 Several studies point to a particular increase in children under the age of five years, and others demonstrate an increase among those without “high-risk” human leukocyte antigen (HLA) genotypes, suggesting a changing phenotype of disease.2 Despite significant advances in diabetes management over the past two decades, it is not yet possible to completely normalize glucose levels, and episodes of hypoglycemia and hyperglycemia are essentially unavoidable. Although decreasing in frequency with recent improvements in clinical care, the risks for severe hypoglycemia and microvascular and macrovascular complications of diabetes have not been abolished. The successful clinical management of T1D requires daily uncomfortable, time-consuming, expensive, and often frustrating self-monitoring and self-management skills. Management is burdensome at best and often impossible to accomplish for patients and families, particularly those with psychosocial or financial barriers to care. Clearly, new approaches are needed.

The description of islet cell autoantibodies and the presence of insulitis in pancreata from people with T1D led to the consensus that T1D was caused by an immune-mediated assault on the pancreatic beta cells, thus paving the way for clinical trials to determine if the disease course could be altered by modulation of the immune system, preserving beta cell function before or after clinical disease onset. The clinical course of T1D, from genetic risk and development of autoantibodies through years after diagnosis, as illustrated in Figure 1, was conceptualized in print as early as 1984. The initial marker of immune system activation against the islet is the presence of autoantibodies. Subsequently, impaired beta cell function can be identified in the earliest stages via intravenous glucose tolerance tests that demonstrate diminishing insulin responses. More significant impairment can later be identified via an oral glucose tolerance test and subsequently to an oral “mixed meal” stimulus (e.g., “mixed meal” of fat, carbohydrate, and protein). Although earlier models of the natural history of T1D depicted complete absence of C-peptide soon after diagnosis and recent data suggest that detectable C-peptide may be present for many years, there is not yet a full understanding of the clinical relevance of these observations.3 In individuals progressing to T1D, abnormal postprandial glucose values are the first glycemic abnormality detected, and essentially all such individuals eventually cross the threshold to the diagnosis of clinical disease. In the early stages of evolving T1D, postprandial glucose values >200 mg/dL are often present at a time when fasting glucose and HbA1c values are normal. Symptoms of hyperglycemia are usually absent until fasting hyperglycemia occurs. Thus, one could consider that T1D is present in antibody-positive individuals with abnormal glucose tolerance, in essence shifting the line for “diagnosis” earlier in the disease course. Although not illustrated in Figure 1, it is of interest that the underlying cause of the honeymoon period, characterized by a transient diminished requirement for exogenous insulin postdiagnosis, is not known. It may be the result of exogenous insulin relieving the effect of “glucose toxicity” on beta cell function and/or insulin action. Whatever the mechanism, this is a transient event in the course of the disease but does support the concept that beta cell dysfunction rather than death is a component of the early disease process.

Figure 1.

Models of pathogenesis and natural history of type 1 diabetes. Schematic representation of gradual decline in beta cell function over time in T1D. Printed with permission from Ref. 88.

Preserving beta cell mass and/or function before diagnosis would provide a clear clinical benefit in preventing hyperglycemia and clinical disease. Less well known, but important to understand, is the clinical benefit of preserving beta cells after diagnosis. As shown in the Diabetes Control and Complications Trial (DCCT), subjects in the intensively treated group who had endogenous insulin secretion, characterized by stimulated C-peptide values ≥0.2 pmol/mL, had a 50% reduction in retinopathy progression and 62% less severe hypoglycemia than intensively treated subjects without residual insulin secretion (Fig. 2).4 These data echo results from studies in islet cells transplantation in which even a limited amount of endogenous secretion from transplanted cells results in a significant reduction in hypoglycemia even if insulin independence is not achieved.5 Because hypoglycemia is the limiting problem in achieving strict glucose control and reduction in vascular complications, preservation of beta cells is clinically important to those with diabetes as well as to those at risk for disease.

Figure 2.

Subjects receiving intensive insulin therapy in the DCCT had fewer micro- and macrovascular complications compared to those receiving conventional therapy. Within the intensively treated group, subjects with residual insulin secretion (C-peptide > 0.2 pmol/mL) had a marked reduction in severe hypoglycemia. Adapted with permission from Ref. 4.

The first generation of T1D studies, those performed before the year 2001, demonstrated that intervention may alter the disease course, highlighted the potential risks involved, and fostered new considerations in study design. The next generation of studies, largely conducted since 2001, built upon this framework and was also able to use increasingly sophisticated information about the immune system to test new therapies.

In this review, we will highlight the new information gained from clinical trials in T1D performed in the past decade. We will critically examine these results in both a scientific and clinical context, focusing on questions to address for the next generation of studies aimed at prevention or reversal of T1D.

First-generation T1D studies

More than 30 studies testing the concepts of immunosuppression, immunostimulation, beta cell rest, and antigen therapy, as well as other mechanisms on preservation of insulin production in T1D, were conducted and reported from the 1980s to 2001 (reviewed in 2002).6 Most were pilot studies using varying outcome measures, making interpretation somewhat difficult. Nonetheless, lessons from these studies inform the current generation of trials. For example, cyclosporine was the immunosuppressive agent most extensively studied in the 1980s and 1990s. Several trials, which included a total of approximately 500 patients, demonstrated an increase in “clinical remissions” (most often defined as fasting glucose <140 mg/dL, postprandial glucose <200 mg/dL, and HbA1c <7.5% without exogenous insulin) but also an unacceptable incidence of drug-induced nephrotoxicity.7–16 Although not formally tested in these studies, there were no apparent long-lasting effects of therapy after treatment was discontinued. Several other non-specific immunosuppressant drugs, such as anti-thymocyte globulin, prednisone, and azathioprine, were also evaluated with mixed results.17–19 In adequately powered, randomized trials, no effect was seen of immunostimulation (BCG),20 antigen therapy,21,22 or nicotinamide23 in subjects with recently diagnosed T1D. In contrast, aggressive glucose control at the time of diagnosis did seem to preserve beta cell function one year after of diagnosis. Shah et al. randomized subjects immediately after T1D diagnosis to usual care or a bedside closed-loop glucose-controlled insulin infusion system. After one year, stimulated C-peptide was significantly higher in the treatment group.24 The results of this trial were consistent with the observations made in the DCCT. In the DCCT, the group of T1D subjects who received more intensive insulin therapy early in the course of the disease had higher stimulated C-peptide levels than subjects who received less intense metabolic control.4 A trial to further assess whether intensive diabetes control in the immediate post-diagnosis period can preserve beta cell function is currently recruiting patients through the NIH Type 1 Diabetes TrialNet (TrialNet; an NIH-sponsored clinical trial network).25 Although many improvements have been made since 1998 with respect to intensive metabolic control of T1D, the mechanisms by which controlling hyperglycemia leads to better beta cell function are not well understood.

Great strides have been made over the last few decades in identifying individuals at risk for T1D, thus enabling trials to prevent autoimmunity (primary prevention) or to prevent hyperglycemia in subjects in whom beta cell destruction is already underway (secondary prevention). Moreover, although fully powered prevention studies require large screening efforts and a long period of follow-up, the accuracy of prediction estimates and the ability to conduct such studies has been clearly demonstrated in multiple settings (Table 1). Three such fully powered trials were completed by 2001. The European Nicotinamide Diabetes Intervention Trial (ENDIT) tested whether oral nicotinamide would prevent or delay disease onset in antibody-positive individuals. Although the treatment was found to be safe, no effect on disease onset was found (approximately 30% after five years in both groups).26 The North American Diabetes Prevention Trial-Type 1 (DPT-1) evaluated the effect of both parenteral and oral insulin administration in relatives of T1D patients who were at risk for T1D. Parenteral insulin (IV insulin for four continuous days annually and ultralente insulin 0.25 u/kg administered BID) was ineffective in delaying or preventing disease in high-risk subjects; 15% of subjects in both the experimental and control groups developed T1D annually.27 In a separate arm of the DPT-1 study, oral insulin was administered (7.5 mg/day) daily in relatives at intermediate risk for diabetes development. Although the overall results of the oral insulin intervention study were negative (∼7% progressed to diabetes annually in both groups), a post-hoc analysis revealed that subjects with very high titers of antibody to insulin demonstrated up to a four-year delay in diabetes onset in those given oral insulin as compared to placebo (Fig. 3).28 The hypothesis that oral insulin could delay or prevent the onset of disease in individuals with high levels of insulin autoantibodies and moderate risk of T1D is currently being evaluated in a randomized, placebo-controlled trial through TrialNet.29 Lessons learned from the first generation of trials include the following:

Table 1. Summary of fully powered T1D prevention trials to date
StudyEuropean Nicotinamide Diabetes Intervention Trial (ENDIT)Diabetes Prevention Trial-1 (DPT-1)Diabetes Prevention Trial-1 (DPT-1)T1D Prediction and Prevention Study (DIPP)
Study centersEurope, N. AmericaNorth AmericaNorth AmericaFinland
Primary or secondarySecondary; antibody-positive relativesSecondary; antibody-positive relatives with low insulin secretion or abnormal glucose toleranceSecondary; antibody-positive relativesSecondary; antibody-positive children
Experimental therapyNicotinamideParenteral insulinOral insulinNasal insulin
Number screened30,00084,228103,391116,720
Number eligible1,004372388328
Number enrolled552339372224
Primary endpointDiagnosis of diabetesDiagnosis of diabetesDiagnosis of diabetesDiagnosis of diabetes
Outcome∼30% developed T1D; no effect of therapy∼65% developed T1D; no effect of therapy∼35% developed T1D; overall no effect of therapy. Post hoc analysis suggests effect in subgroup with high IAA∼60% developed T1D; no effect of therapy
Figure 3.

Data from post hoc analysis of the DPT-1 Oral Insulin Trial demonstrating the delay in diabetes among those treated with oral insulin who had high levels of insulin autoantibodies. The number of subjects at risk in each group at each year of follow-up is enumerated at the bottom of the figure. The log-rank test was used for comparison between the groups, with the P-values as indicated. Printed with permission from Ref. 28.

  • 1Adverse effects of immunosuppressant therapies may limit their chronic use in T1D (e.g., nephrotoxicity for cyclosporine).
  • 2Immunosuppressant therapies do not seem to have lasting effects on insulin secretion in T1D after treatment is discontinued.
  • 3To effectively study and compare the effects of experimental agents on preservation of insulin secretion in new-onset T1D, standardized study design and outcome measures are important.
  • 4Large numbers of subjects must be screened to identify those at increased T1D risk for potential participation in T1D prevention studies. Although feasible and critically important, these prevention studies are logistically difficult, labor-intensive, and long in duration. Because of these limitations, only very few studies can be conducted at any one time and it is perhaps not surprising that the first efforts were not successful.

Recent trials

Building on previous work, the next generation of trials aimed to exploit new information about the immune response and in most cases, to standardize key aspects of study design. Standardized variables now shared by many new-onset T1D trials include the primary outcome measure of C-peptide after mixed meal stimulation test as the marker for beta cell function; and common subject characteristics (time after diagnosis, subject age, and level of glucose control). Adequately powered, randomized trials have demonstrated three therapies with distinct mechanisms of action (anti-T cell, anti-B cell, and costimulation blockade) that can preserve beta cell function in recently diagnosed subjects for a period of months. Other well-designed studies using other agents have shown no effect on the T1D disease process. However, a key point is that negative trials are nonetheless successful trials when they enable us to answer the question posed. This past decade has seen results from multiple successful trials teaching us about the pathophysiology of T1D and suggesting new avenues to pursue. To that end, selected trials that illustrate important points have been chosen for discussion below.


CD3 is a protein complex located on the surface of T cells and is integral to the initiation of T cell activation. The toxicity of the first anti-human CD3 monoclonal antibody, OKT3, used for the treatment of organ transplant rejection, was reduced by elimination of FcR-binding portion of the molecule in the development of two different humanized antibodies, teplizumab (hOKT31(Ala-Ala)) and otelixizumab (ChAglyCD3). Elegant preclinical studies suggested several mechanisms by which FcR-non-binding CD3-specific antibodies may produce a state of “tolerance” to self. These include “antigenic modulation,” partial phosphorylation of the T cell receptor complex resulting in decreased production of IL-2 and inactivation of T helper 1 (Th1) cells, and promotion of T regulatory cells.30–32

In the first trial with hOKT3(Ala-Ala), 24 subjects with new-onset T1D were randomized to receive 14 days of drug or were assigned to a control group.33 In this open-label study, 9 of the 12 subjects in the treatment group had stable insulin secretion at one year as compared to 2 of 12 controls. Subsequently, additional data from this open-label study including a total of 42 subjects (21 drug-treated) were reported.34 Stimulated C-peptide response (compared to baseline) was greater one year after diagnosis in subjects treated with drug compared to control subjects (97 ± 9.6% vs. 53 ± 7.6% of response at study entry, P < 0.01; Fig. 4A). HbA1c and insulin requirements were both lower in the treatment group as well. During the second year, C-peptide levels declined at a similar rate in treatment and observation groups. Nonetheless, the marked effect in the first year resulted in a significantly greater percentage of drug-treated individuals with stimulated C-peptide of 0.2 pmol/mL at 24 months compared to controls (67% vs. 26%; P= 0.01); recall that C-peptide of 0.2 pmol/mL was identified as a clinically important cutoff in the DCCT study (Fig. 2).4

Figure 4.

(A) C-peptide responses to a MMTT in control and h-OKT3–treated subjects. Total C-peptide during a 4 h MMTT over 24 months is shown. Black, h-OKT3–treated; gray, control. **P < 0.02. Printed with permission from Ref. 34. (B) C-peptide responses to a MMTT in control and anti-CD20–treated subjects. Mean C-peptide AUC during first 2 h of 4 h MMTT over 12 months are shown. Printed with permission from Ref. 32. (C) C-peptide responses to a MMTT in control and abatacept-treated subjects. Mean stimulated C-peptide 2 h AUC during first 2 h of 4 h MMTT over 24 months are shown. Printed with permission from Ref. 44.

Side effects of drug treatment included transient lymphopenia that occurred initially in all subjects. In most subjects, lymphocyte counts returned to >80% of baseline levels within several weeks. In rare cases, more prolonged lymphopenia has been observed.35 Despite these transient laboratory abnormalities, no serious infections were reported, making the clinical importance of these changes uncertain. Additional adverse events included symptoms associated with cytokine release such as fever, headache, and arthralgias. Rash was present in almost all subjects and a transient grade 3 cytopenia resulted in one subject stopping therapy.

The marked effect of anti-CD3 treatment on beta cell function was also seen in a placebo-controlled trial with the other therapeutic, otelixizumab. In this study, 80 subjects with new-onset T1D, ages 12–39, received six days of active drug or placebo.36 Insulin secretion (assessed by C-peptide release during a glucose clamp and after glucagon stimulation) was significantly improved in the drug-treated group at 6, 12, and 18 months as measured by change in C-peptide from baseline and an equivalent HbA1c was observed in the context of less insulin dosing. Side effects of drug treatment included transient reactivation of Epstein Barr virus associated with a “mono-like” illness as well as transient infusion reactions, rash, and lymphopenia. Post hoc analysis suggested that there was a more pronounced effect in subjects with higher C-peptide levels at study entry. Long-term follow-up demonstrated continued effects on insulin dose up to four years after treatment with no long-term adverse effects reported.37

These positive findings in the context of manageable adverse events led to several subsequent studies of anti-CD3 drugs aimed to move toward clinical use. Two Phase 3 trials were initiated in new-onset T1D: the Protégé Study of teplizumab (Macrogenics, Inc.) and the DEFEND Study of otelixizumab (ToleRx, Inc). In addition, the AbATE study, sponsored by the NIH Immune Tolerance Network (ITN; an NIH sponsored clinical trial group), was launched to test whether repeat dosing of teplizumab would prolong its effect on insulin secretion, and the Delay study tested whether treating subjects with teplizumab further out from diagnosis would be effective. Finally, the Anti-CD3 Prevention Study was initiated by TrialNet to study whether teplizumab could delay onset of T1D in high-risk individuals with dysglycemia.

The Protégé Study involved more than 100 sites worldwide and reported no effect of teplizumab as measured by the study's primary endpoint of a HbA1c level of <6.5% with an insulin dose of <0.5 units/kg/day.38 This result was perhaps not surprising considering the diversity of subjects, health care systems, and clinicians encompassed in the study. In addition, C-peptide was not different between treatment and placebo groups overall. Post-hoc analysis suggested an effect of C-peptide secretion in selected subgroups of subjects (younger children and those treated in the United States), but these observations can only be considered as hypothesis generating, not as true evidence of efficacy in these subgroups. The DEFEND-1 study results revealed that there was no difference in C-peptide between subjects treated with otelixizumab or placebo at one year.39 It is likely that this lack of effect was related to the fact that the dose of otelixizumab used in this trial was markedly lower than in the original study. The investigators chose the lower dose with the aim of reducing short-term side effects.

The AbATE study was an open label trial of 83 subjects randomized within three months of diagnosis to 14 days of IV drug administration or observation. As in previous studies, the drug-treated subjects had significantly greater C-peptide production in response to a mixed meal at one and two years compared to those in the observation group. However, it is unclear whether the second dose of drug contributed to the preservation of C-peptide seen at two years. It is worthwhile to note that approximately one-third of subjects were unable to complete their full dose due to pre-specified stopping rules or adverse events.40

Together these results highlight a current conundrum in the clinical development of anti-CD3 therapies to preserve beta cells; too low a dose may be ineffective, yet a higher dose may lead to short-term safety concerns in a significant number of subjects. Carefully administered, however, a clinical role of anti-CD3 remains possible, particularly if a population of individuals with diabetes can be identified who have a prolonged clinical response or if the anti-CD3 (teplizumab) TrialNet Prevention Study is able to demonstrate a clinically significant delay in disease onset (further discussed in section on ongoing trials below).


Initially considered an unlikely candidate for immunotherapy in a T cell–mediated disease such as T1D, the B cell–depleting anti-CD20 monoclonal antibody, rituximab, was evaluated in subjects with recently diagnosed T1D largely on the basis of its effectiveness in other autoimmune disease (the FDA has approved rituximab for use in rheumatoid arthritis and it is commonly used off-label for multiple sclerosis, systemic lupus erythematosus, and autoimmune hemolytic anemia).41 Rituximab is the second agent that has shown positive results in preserving insulin secretion in new-onset T1D.

In a randomized, double blind, Phase 2 study led by the late Mark Pescovitz, M.D., and sponsored by TrialNet, 87 patients with recently diagnosed T1D between 8 and 40 years of age were randomized to receive four weekly infusions of rituximab or placebo. At one year, the stimulated C-peptide was 20% higher in the rituximab group than in the placebo group (or 0.56 pmol/mL vs. 0.47 pmol/mL; Fig. 4B). The rituximab group also had significantly lower levels of HbA1c and required less insulin to achieve better glycemic control.42 Infusion reactions to the first dose of rituximab were common, likely due to the omission of glucocorticoid routinely used as a pre-treatment when rituximab is given in other autoimmune diseases. Nonetheless, infusion reactions were generally mild and rarely prevented further rituximab dosing. Indeed, after the first dose, there was no difference in “reactions” between placebo- and rituximab-treated subjects—thus highlighting the importance of placebo-controlled trials. As reported in other rituximab studies, B cells reached a nadir at about six months after therapy, IgG levels were unaffected, and IgM levels remained lower in treated subjects at one year. No increase in infection or neutropenia was noted in the rituximab treatment group. As with anti-CD3 therapy, however, the drug effect was most apparent early, with the slope of C-peptide declining in parallel with the placebo-treated patients after six months. Importantly, rituximab is in widespread clinical use for other indications; therefore, much is known about adverse effects. Thus it is reasonable to test whether repeated rituximab dosing, as is used in other autoimmune disease, would prolong C-peptide preservation in those with disease, or whether this therapy would delay disease onset in at-risk subjects.


Cytotoxic T lymphocyte antigen 4 (CTLA-4) is a homologue of CD28, a high-affinity receptor that down-regulates T cells. Abatacept, a CTLA-4-immunoglobulin fusion protein that modulates the costimulation of immature T cells and prevents full T lymphocyte activation,43 was also recently found to be effective in preserving insulin secretion in new-onset T1D.

The effect of abatacept in recent-onset T1D was evaluated in a multicenter, double-blinded, randomized controlled trial sponsored by TrialNet.44 The placebo-controlled study recruited 112 subjects with recently diagnosed T1D, ages 6 to 36. The stimulated C-peptide was 59% higher at two years in the abatacept treatment group compared to placebo (Fig. 4C). Hemoglobin A1c was lower in the treatment group but insulin use was not different. Adverse events were minimal and there was no increase in infections or neutropenia. Strikingly, however, as in the anti-CD3 and anti-CD20 studies, the major drug effect was seen early despite continued administration of drug. After six to nine months, the rate of C-peptide fall in the treatment group was parallel to the decline in the placebo group. As in the AbATE trial, this result was unexpected because there was continued infusion of abatacept throughout the two-year study period. This result hints that perhaps a shorter course of drug would have similar benefits. Testing abatacept in T1D prevention is being considered to determine whether treatment during an earlier stage of the disease process can change the duration of therapeutic effect.

Other studies

The DIPP study aimed at secondary prevention, delaying or preventing antibody-positive children from developing clinical diabetes. After screening more than 100,000 infants and children for high-risk HLA types in Finland, 264 individuals with two or more autoantibodies were randomly assigned to 1 u/kg of nasal insulin per day. Presentation of antigen (insulin) to the nasal mucosa was thought to have potential to induce a toleragenic response.45 No safety concerns were identified and compliance was high; however, there was no effect of this therapy on development of clinical disease, which occurred at a rate of about 15% per year in both groups.46 In contrast to testing the immune effect of insulin as an antigen, another group evaluated the effect of twice daily subcutaneous regular insulin titrated to avoid postprandial hyperglycemia in a small study of ICA-positive high-risk relatives of T1D patients and also found no effect.47 Following up epidemiologic data suggesting an impact of gluten-containing foods on the development of diabetes,48 150 infants who had first-degree relatives with T1D and who also had high-risk HLA genes were randomized to initial gluten exposure at 6 months or 12 months in the BABYDIET study, conducted in Germany. This primary dietary prevention trial had no effect on appearance of islet antibodies, with about 12% of children in both groups developing antibodies over three years.49

Two other primary prevention studies also involved nutritional interventions but were designed as pilot studies. The Finnish Trial to Reduce Diabetes in the Genetically at Risk (Finnish TRIGR Trial) was undertaken to test the hypothesis that protein in cow's milk contributes to the development of islet autoimmunity and T1D. A total of 230 infants with high-risk HLA genes were randomly assigned to standard baby formula (with cow's milk proteins) or hydrolyzed protein formula as a supplement to breast feeding and followed for the development of islet antibodies and T1D. The pilot study results suggested that there may be an effect, but the small numbers involved and imbalance in follow-up lessen the strength of the observations reported.50 Importantly, however, the fully powered TRIGR Study, testing this hypothesis, has completed recruitment and results should be available to answer this question more definitively in a few years.51 In the Nutritional Intervention to Prevent (NIP) Diabetes Trial, sponsored by TrialNet, infants with a high genetic risk for T1D and pregnant women carrying a fetus with high genetic risk for T1D were randomized to receive docosahexaenoic acid (DHA) or placebo to see if supplementation in infancy or prenatally might prevent or delay the autoimmunity leading to T1D. As a pilot trial, this study was not powered to determine whether the therapy had an effect on autoimmunity; however, preliminary reports from this study suggest a reduction in inflammatory cytokine levels in infants receiving the DHA, providing a rationale for this approach in future large-scale primary prevention trials.52

Additional studies have yielded negative results in new-onset T1D. The 65-kD isoform of glutamic acid decarboxylase (GAD) is a major autoantigen in T1D. Initial human studies suggested that treatment with GAD might provide a very low-risk approach to help preserve beta cell function.53,54 However, the TrialNet GAD study found no significant differences in stimulated C-peptide levels, HbA1c, or insulin use at one year in GAD or placebo-treated subjects with new-onset T1D.55 Similarly, a Phase 3 study of GAD in Europe failed to meet its primary endpoint of preserving insulin secretion.56 Although disappointing, it is premature to abandon antigen therapy as an approach to alter beta cell destruction. It is possible that using different doses of antigen, different adjuvants, different routes of administration, using peptide fragments instead of whole protein, or using antigens other than GAD would be more effective. The additional data from studies examining the effects of GAD administration on immune markers and other biomarkers may provide some answers. Such data may also support testing GAD in conjunction with other therapies or earlier in the disease process (before clinical onset), particularly because of the good safety record to date even in subjects as young as age three years.

In addition to anti-T cell or anti-B cell therapy, anticytokine therapy is another approach to altering the immune response. Widely used in rheumatoid arthritis, etanercept is a tumor necrosis factor-alpha (TNF-α) antagonist tested in a pilot study of 18 subjects with T1D (ages 7–18 years) in a placebo-controlled study. At week 24, C-peptide had increased by 39% in the etanercept group and had decreased by 20% in the placebo group. HbA1c and insulin dose were also both lower in the etanercept group compared with the placebo group.57 Although the treatment was well tolerated in this study, the FDA recently issued a warning regarding the use of this drug due to risk of tuberculosis and other infections, thus any potential future trials of etanercept in T1D will need careful patient selection and monitoring.

Interleukin 2 (IL-2; proleukin) is a proinflammatory cytokine used to stimulate the immune response in HIV and cancer. By combining this drug, which augments both T effector and T regulatory cells, with rapamycin (sirolimus) to block the proliferation and survival of T effector cells, beta cell destruction might be halted by shifting the balance to T regulatory cells.58 Although effective in mouse models, results recently presented from a Phase 1 trial of IL2 and rapamycin in T1D, sponsored by the NIH Immune Tolerance Network, were perplexing.59 Combining one month of therapy with IL-2 and sirolimus resulted in a marked increase in T regulatory cells. Further, defects in T cell signaling previously described in individuals with T1D were improved. However, there was an unexpected drop in C-peptide at three months. This was transient, as C-peptide subsequently increased in almost all subjects. Although the mechanism of the untoward transient effect on beta cell function is not clear, preliminary data show that both an increase in eosinophils and NK cells occurred.60,61 These data highlight the difficulties in extrapolating from mouse to human and emphasizes the caution needed in evaluation of novel approaches.

The use of a bone marrow reconstitution as a “reset” for the immune system has been considered for autoimmune diseases. However, the risk of morbidity (secondary malignancy, opportunistic infection, and metabolic syndrome) and potential for mortality with even the safest conditioning regimens for autologous hematopoietic stem cell transplant (AHSCT) have limited the use of this therapy. As discussed in a thoughtful review article, expert clinicians and investigators generally agree that at this time, the study of AHSCT in autoimmune disease should be reserved for patients with severe, debilitating disease that is refractory to standard therapies and for whom there is a poor prognosis.62 It is therefore surprising that a group of investigators nevertheless tested this approach in subjects with T1D and reported that 20/23 patients became insulin free for variable periods of time.63,64 This was a small uncontrolled study and the results reported excluded data from some patients. The investigators reported an overall positive effect on C-peptide response at 36 months. Though there were no deaths in this small series, significant toxicities were seen, including severe respiratory disease, hypogonadism, and oligospermia. Although not minimizing the difficulties of living with T1D, given the significant morbidity reported and the fact there is no evidence that AHSCT therapy consistently results in a permanent cure of any autoimmune disease, it is our opinion that such high-risk therapy should not be pursued in the new-onset T1D population at this time.

Ongoing T1D trials: prevention

Individuals at risk for diabetes can now be clearly identified and enrolled in studies according to their degree of risk.

New information gathered from analysis of the DPT-1 has now identified a group of individuals at extremely high risk of T1D. As illustrated in Figure 5, autoantibody-positive relatives of patients with T1D, who have abnormal glucose tolerance, have a greater than 75% risk of diabetes.65 The identification of this subgroup is extremely important. First, because the rate of progression in this group is so high, a trial to delay or prevent the onset of disease can be carried out with smaller numbers of subjects and over a shorter period of time. Second, a therapy that would delay the onset of clinical disease by even a few years would be clinically important. This group is now being enrolled in a TrialNet study using teplizumab in a single 14-day course of drug with the aim of delaying or preventing diabetes onset.66

Figure 5.

Marked risk of progression to diabetes in antibody-positive relatives of T1D patients with indeterminate or impaired glucose tolerance whether above or below age 18. Printed with permission from Ref. 65.

First-degree relatives with two antibodies and normal glucose tolerance have a five-year risk of developing T1D of 30–50%. Two prevention studies in this population are ongoing. The first is the INIT II study in Australia, in which intranasal insulin is delivered in an attempt to induce immune tolerance, as discussed above. Although the Finnish DIPP trial failed to demonstrate a benefit of intranasal insulin in young antibody-positive children, the INIT II study is evaluating different dosing regimens (40 IU and 440 IU IN daily for seven days then weekly for 12 months) and includes older subjects (ages 4–30 years). In subjects with latent autoimmune diabetes in adults (LADA), nasal insulin was associated with a decreased T cell response to insulin, supporting that there may be some immunologic effect of intranasal insulin that warrants further investigation.67

The second ongoing trial for T1D relatives with normal glucose tolerance and positive antibodies is the TrialNet study to determine if insulin delivered orally will delay or prevent the onset of T1D. As discussed above, this is the trial to address the post hoc observation seen in the original oral insulin trial conducted by the Diabetes Prevention Trial (DPT-1). This study population's five-year estimated risk of diabetes is approximately 35% at the time of study entry. Recruitment is ongoing for this study.

Newborn infants with a first-degree relative with T1D and an at-risk HLA type are estimated to have a 10-year risk of developing T1D of ≥ 7.6%.50 As discussed above, the fully powered TRIGR study, testing whether casein hydrolysate formula, rather than cow's milk–based formula, can prevent autoantibodies and diabetes in this group of children with high genetic risk has completed recruitment and results are anticipated within the next several years.51

Ongoing T1D trials: intervention

Several studies are underway that focus on the concept that “inflammation” plays a large role in beta cell dysfunction (Table 2). The IL-1 pathway is known to play a key role in the inflammatory response. Two therapies designed to block this pathway are under investigation. The Anti-Interleukin-1 in Diabetes Action (AIDA) Study (sponsored by the JDRF) is testing Anakinra, a selective antagonist of the IL-1 receptor in 18–35-year-old subjects within 12 weeks of T1D diagnosis.68 A trial of Canakinumab, a human anti-interleukin-1-beta (anti-IL-1β) monoclonal antibody, sponsored by TrialNet, has fully enrolled the planned 66 subjects (ages 6–45) with new-onset T1D within 100 days of diagnosis.69 Canakinumab has been evaluated and shown benefit in patients with Rheumatoid Arthritis and Systemic Juvenile Idiopathic Arthritis (SJIA).70 Results of these trials should be available within a year and will lay important groundwork for consideration of combination therapies.

Table 2. Selected T1D beta cell preservation trials open for recruitment at the time of publication
Experimental agentSponsorSubject agesEntry criteriaPhase of trialWebsite
Intervention trials
 Alefacept (T1DAL Study)NIH/Immune Tolerance Network12–35Within 100 days of diagnosisII http://clinicaltrials.gov/ct2/show/NCT00965458
 Aralast (RETAIN Study)NIH/Immune Tolerance Network8–35Within 100 days of diagnosisII http://clinicaltrials.gov/ct2/show/NCT01183468
 NeulastaJDRF/NIH12–46Within six months of diagnosisI–II http://clinicaltrials.gov/ct2/show/NCT00662519
 Sitagliptin and lansoprazole (REPAIR Study)Sanford Health and JDRF11–45Within six months of diagnosisII http://www.clinicaltrials.gov/ct2/show/NCT01155284?term=NCT01155284&rank=1
 T regulatory cellsJDRF18–35Between three and 24 months of diagnosisI http://www.clinicaltrials.gov/ct2/show/NCT01210664?term=Tregs&cond=Diabe-tes&rank=1
 Thymoglobulin and NeulastaHelmsley Trust16–64Between four months and two years of diagnosisI–II http://www.clinicaltrials.gov/ct2/show/NCT01106157?term=thymoglobulin+and+neulasta&rank=1
Prevention trials      
 Anti-CD3NIH/Type 1 Diabetes TrialNet16–45Antibody positive; dysglycemiaII http://www.diabetestrialnet.org/studies/ACD3.htm
 Oral insulinNIH/Type 1 Diabetes TrialNet3–45Antibody positiveIII http://www.diabetestrialnet.org/studies/oral-insulin.htm
 INIT IIDiabetes Research Centre4–30Antibody positiveIII http://www.diabetestrials.org/initii.html

Another anti-inflammatory approach is the use of Alpha-1 antitrypsin (AAT; Aralast). AAT is a serine proteinase inhibitor found in high concentrations in the serum that decreases cytokine production, complement activation, and immune cell infiltration. The ITN is currently evaluating the safety and efficacy of AAT in patients (ages 8–45) with recently diagnosed T1D.71

A randomized, placebo-controlled trial of thymoglobulin soon after diagnosis sponsored by the ITN is currently underway.72 This study of about 55 subjects has completed enrollment with results expected in the next year. Another phase I/II pilot trial is investigating the safety and efficacy of combination therapy with Thymoglobulin® (ATG) and Neulasta® (GCSF) in a randomized, placebo-controlled trial in subjects with T1D.73 This combination of agents has shown some effect in the NOD mouse model. The effect of ATG may be partially from its ability to deplete T cells but ATG may also induce an immunoregulatory shift from a Th1 to a Th2 cytokine phenotype. Previous small studies using ATG were stopped due to toxicities;17,74 differences in drug, dose, and the use of steroids are hoped to ameliorate most of these concerns.

The REPAIR-T1D (Restore Pancreatic Insulin Response in type 1 diabetes) trial is a randomized, controlled, trial evaluating the effect of combination therapy with sitagliptin (a dipeptidyl peptidase-4 inhibitor [DPP-4]) and lansoprazole (a proton pump inhibitor) in subjects with recently diagnosed T1D, ages 11–45 years.75 Though not yet convincingly demonstrated in humans, several animal studies have suggested that GLP-1 agonists such as sitagliptin or exenatide enhance beta cell growth.76–78 Two studies in the NOD mouse model demonstrated restoration of euglycemia when used in combination with gastrin or proton pump inhibitor therapy, agents that are also purported to support beta cells.79,80

Given that T regulatory lymphocytes (Treg cells) are thought to have a role in shaping the body's immune reactions, the ability to modify Treg cell behavior has been identified as a potential therapy. A phase I trial is currently ongoing to evaluate the safety and potential efficacy of the administration of an expanded pool of Treg cells that have been harvested from the subjects (ages 18–35 years) and then expanded before being re-infused.81

Umbilical cord blood (UCB) is also a potential source of regulatory T lymphocytes, and infusion of autologous-banked UCB in subjects with T1D has been evaluated in a small pilot study demonstrating the feasibility and safety of this approach.82 These investigators are now evaluating the effect of UCB infusion coupled with administration of vitamin D and omega 3 fatty acid supplementation in a randomized, controlled pilot trial.83

Looking toward the future

Although a new therapy has not yet been identified that is ready for clinical use in T1D patients or those at risk for disease, much progress has been made toward the important goal of beta cell preservation. In prevention, stratification of risk groups continues to allow for testing of multiple agents (Fig. 6). As the newer sciences of genomics, proteomics, and metabolomics are applied, we anticipate the development of better tools to help dissect the heterogeneity of the disease and target therapies more selectively in at-risk individuals. Several immunomodulatory agents have been found to slow the rate of beta cell deterioration during the early stages of T1D when compared to placebo. These studies were well designed, well powered, and because similar outcome measures were studied, easily compared. It is not clear why experimental drugs with different mechanisms, including targeting T cells, B cells, and costimulation, all had their major effect on insulin secretion soon after treatment. It is also not clear why the salutary drug effect waned with time so that, with time, the fall in insulin secretion in both drug- and placebo-treated subjects became parallel.

Figure 6.

Prevention trials available for relatives of patients with T1D.

In other autoimmune diseases, such as rheumatoid arthritis or systemic lupus erythematosus, immunomodulatory drugs are often given in repeated doses to control clinical “flares” of the disease. However, as demonstrated in the abatacept trial, despite continued drug dosing for 24 months, the rate of fall in C-peptide in the drug-treated group eventually paralleled the decline seen in the placebo group. It is possible that the immune and inflammatory mechanisms underlying beta cell loss are distinct early in the course of T1D from those that occur later. Studies targeting inflammation as well as immune modulation may address this possibility.

Antigen-based therapy, such as GAD, did not have a significant effect on beta cell function when administered to subjects within three months of diagnosis. It is possible that antigen-based therapy would be more successful earlier in the course of the disease and should be evaluated in the prevention of T1D. Similarly, rituximab and abatacept could be considered in prevention trials, for use at a time when beta cell mass is more substantial than at clinical diagnosis of new-onset T1D. More novel approaches to antigen delivery, including administration via a plasmid, have undergone early trials and await publication.84

The engagement and collaboration of patients, families, funding and regulatory agencies, and industry along with a large number of clinicians and clinical investigators has resulted in marked successes this past decade in that T1D trials have provided clear answers. It takes 10 to 15 years for a typical drug to be developed successfully from discovery to registration with the FDA,85 and in early phases of research, the chance of a drug reaching patients is small—approximately 1 in 10.86 Moreover, at least 45% of all pivotal trials have a negative outcome.87 The stakeholders must not lose sight of the significant progress made. The message to patients and families living with T1D, as well as to the clinicians providing their care, is that carefully conducted and rigorous clinical trials are being performed and are providing critical information that will provide the path forward to new treatments (Table 2). Through their awareness of active clinical trials and their willingness to encourage patients to participate, clinicians play a critical role in the success of moving this field forward.

The autoimmune response will continue to present a threat to the patient's own islets or any new beta cells transplanted into a patient with a history of beta cell autoimmunity. Thus, therapies to stop beta cell destruction will continue to be necessary even when beta cell replacement is a viable clinical option made possible through stem cell research or other advances in transplantation.

Clinical trials testing experimental treatments singly and in combination in T1D are feasible and can be done relatively expeditiously and across sites. In particular, planning and recruitment for such studies can be done rapidly and with careful monitoring in clinical trial networks with experienced research teams. The goal continues to be to identify therapies with clinical benefit to patients with T1D, and building upon the very important data obtained in earlier studies; the next generation of clinical studies will bring us closer to this goal.


We gratefully acknowledge Ellen Greenberg, M.S., for her invaluable assistance with manuscript preparation and review; Daniel Casper, M.D., Ph.D., for his creation of graphics and illustrations; and Kelly Smith for her editorial assistance.

Conflicts of interest

The authors declare no conflicts of interest.