Type 1 Diabetes TrialNet: Leading the Charge in Disease Prediction, Prevention, and Immunotherapeutic Mechanistic Understanding Free

Type 1 Diabetes TrialNet (TrialNet) is an international consortium of clinical research centers that studies the development of type 1 diabetes and performs clinical studies to determine whether the disease can be delayed or prevented. This review is part of the National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK) 75th anniversary collection. TrialNet investigations have confirmed the original observation that type 1 diabetes is a chronic autoimmune disease that begins years before the clinical diagnosis. Despite the challenges inherent to a clinically silent prodromal phase, significant progress has been made thanks to the efforts of TrialNet, its investigators, research coordinators, nurses, and, most importantly, its participants over the past 24 years. This progress includes positive results of several trials, including the first approval of a disease-modifying therapy in type 1 diabetes, by the U.S. Food and Drug Administration (FDA), to delay the onset of disease. Of course, progress has not always been linear, and several trials with negative outcomes or small trials were integral to our current understanding of type 1 diabetes or clinical trial design. This review highlights the role of those efforts in paving the way toward the current and future directions for TrialNet.

Until the 1970s and 1980s, type 1 diabetes was considered to occur with the rapid onset of symptomatic hyperglycemia. It was also believed that nearly all insulin-producing β-cells were destroyed by the time of clinical diagnosis (1–9). However, studies of relatives of individuals with type 1 diabetes showed that autoantibodies can be identified years before the clinical diagnosis and that there was deterioration in glucose tolerance (measured with oral glucose tolerance tests [OGTT]) beforehand. These observations led to the conclusion, proposed by George Eisenbarth, Noel MacLaren, and Jerry Palmer, that type 1 diabetes is a chronic, progressive disease, where β-cell function deteriorates over time but is sufficient, during early, presymptomatic phases of disease, to circumvent overt signs and symptoms of hyperglycemia (10–13).

Early trials showing reduced loss of β-cell function with treatment with immune modulators (e.g., cyclosporine [14] and azathioprine with prednisone [15]), together with the evidence that the disease was a chronic immune-mediated process, led to studies to try to prevent its onset. The first attempts were to reduce antigen exposure with β-cell rest (using parenteral insulin) or to induce antigen-specific tolerance (using oral insulin) in the Diabetes Prevention Trial–Type 1 (DPT-1) (16–18).

For the NIDDK-funded DPT-1, launched in 1994, >80,000 relatives of individuals with type 1 diabetes, aged 3–45 years, were screened to identify those who were positive for islet autoantibodies and were postulated to have a high risk of developing type 1 diabetes. The trials concluded in 2001, and while they did not meet their primary end points of delaying symptomatic type 1 diabetes onset, there were valuable lessons gained. First, the screening confirmed predictions that the risk of disease progression for participants positive for multiple islet autoantibodies could be accurately quantified. Participants in the parenteral insulin trial (n = 339) with a low first-phase insulin response on an intravenous glucose tolerance test or abnormal OGTT had at least a 50% 5-year predicted risk of progression to type 1 diabetes (actual risk 65%) (16). Similarly, participants in the oral insulin trial (n = 372) with normal glucose tolerance had a 26%–50% 5-year predicted risk of progression to type 1 diabetes (actual risk 35%) (18). Second, although the overall outcomes were negative, post hoc analyses suggested that some individuals may have shown a clinical response (16,18,19). These observations identified heterogeneity in a population that had previously been considered relatively homogeneous. Third, DPT-1 showed that type 1 diabetes prevention trials could be performed using a large-scale screening program. Finally, it was recognized that further studies to understand the mechanisms of disease would provide data that could be used to identify which treatments should be tested for future studies. Following DPT-1, TrialNet was funded in 2001 by the NIDDK to support investigations to identify and test the most promising therapeutics in clinical studies to prevent or delay type 1 diabetes. Currently, TrialNet includes 17 clinical centers located in the U.S., Canada, and Australia. Clinical centers in other countries have been engaged over TrialNet’s history.

Effective clinical trial design requires identification of individuals at risk and an understanding of disease progression. To meet these needs, TrialNet has several ongoing observational studies. These observational studies include the TrialNet Pathway to Prevention (TrialNet no. 1 [TN01]) study, the Long-term Investigational Follow-up in TrialNet (LIFT) (TN16), and the TrialNet Autoantibody screening Comparative Effectiveness Study (TRACES) (TN34) (Table 1). The effort to identify individuals for whom clinical interventions would be most beneficial was focused on family members, since the risk of type 1 diabetes for relatives of people living with the disease is 15-fold higher than for the general population (4,20,21). A broad screening effort was needed to identify individuals for whom intervention studies would be appropriate.

Hence, in the first study, TrialNet-01 (TN01), launched as a “Natural History Study,” TrialNet screened relatives, collected metabolic data and samples, and identified individuals who might participate in clinical studies. Follow-up metabolic monitoring was offered to those who were autoantibody positive for identification of those who were progressing and enabled investigators to better quantify risk and evaluate immune changes associated with clinical disease over time. Recently published data have shown that metabolic dysregulation can be found in individuals even 5 years before clinical presentation (22). The findings from TN01 have proven to be reproducible across multiple cohorts in different intervention studies.

In TN01 (renamed Pathway to Prevention) >250,000 relatives of individuals with type 1 diabetes have been screened for autoantibodies (Fig. 1), and Pathway to Prevention remains the largest long-running cohort study of both children and adults at risk for disease with 8,937 individuals enrolled in monitoring. Of note, TrialNet is one of the few type 1 diabetes research networks that has historically included adults in screening and trial design. The risk for development of clinical diabetes in at-risk relatives led to the delineation of the stages of disease progression (23): stage 1 was defined as multiple (two or more) positive islet autoantibodies with normal glycemic responses to an OGTT. Individuals at stage 1 have been found to have a 22%–36% risk of progression to clinical disease in 5 years (J. Krischer, unpublished data) and a near 100% lifetime risk. Individuals at stage 2 individuals also have two or more positive autoantibodies but have dysglycemia during an OGTT as well. The risk of clinical disease is 54%–66% in 5 years (J. Krischer, unpublished data). Stage 3 type 1 diabetes is defined according to the American Diabetes Association diagnostic criteria for clinical diabetes with or without symptoms of hyperglycemia. Clinical studies have been conducted at each of these stages within TrialNet (Fig. 2).

Analysis of TN01 data has indicated that β-cell dysfunction begins well before disease diagnosis (22,24). In addition, these observations suggested that TrialNet intervention studies may include use of quantitative outcomes (e.g., preservation of C-peptide) as well as clinical criteria (e.g., diagnosis of stage 3 of type 1 diabetes). The similarities in the progressive loss of β-cell function from before to after clinical diagnosis suggested that studies in individuals after clinical diagnosis can be useful for evaluating potential benefit of agents before the diagnosis. Therefore, TN16, LIFT, was also developed to follow participants, after the diagnosis of stage 3 type 1 diabetes, who had previously been in TrialNet studies. A total of 619 participants have been enrolled in LIFT. For engagement of members of the broader endocrine community, TrialNet has established “affiliate” centers for screening, enrolling, and conducting the clinical observation and intervention studies depending on site capabilities. A Web-based screening tool with home capillary testing with an option for venous blood draws was developed to facilitate engagement of sites and individuals away from the TrialNet clinical centers. Currently, ∼50% of recruitment for the TN01 study is from affiliates and the online screening process. Engaging affiliates not only extends TrialNet’s screening capabilities but also enables individuals who may not live near a clinical center or individuals who have been historically underrepresented to participate in studies.

Most recently, TrialNet has used Web-based technology to extend the ability to engage individuals who are discovered to have islet autoantibodies with screening outside the auspices of TrialNet. A virtual tool developed from TRACES is to be initiated in 2025 and will provide individuals who have been identified as having islet autoantibodies the opportunity to self-report their data and obtain an assessment of diabetes risk and recommended next steps and will bring awareness of TrialNet studies.

Testing for islet autoantibodies has been the cornerstone of type 1 diabetes screening. TrialNet currently tests for four biochemical autoantibodies, glutamic acid decarboxylase 65 (GADA), insulin (IAA), insulinoma-associated antigen-2 (IA2A), and zinc transporter 8 (ZnT8A), using a radioimmunoassay, in addition to indirect immunofluorescence against islet cell antigens (25). Depending on age, relatives may be rescreened if found to be negative on initial autoantibody screening. Relatives with single autoantibodies are retested annually. Follow-up (monitoring) with metabolic testing is focused on those who are positive for at least two biochemical autoantibodies. When the Pathway to Prevention study initiated testing in March 2003, screening included measurement of GADA, IAA, and IA2A. In 2010, an NIDDK-supported effort to harmonize GADA and IA2A assays was fruitful in determining cross study assay comparability (26). Newer autoantibody assays (e.g., electrochemiluminescence [ECL]-IAA and ECL-GADA or novel epitopes to IA2) are also being studied; the concordance and predictive value of these newer assays have been tested in samples from individuals who do or do not show progression (27–29). TrialNet continues to support testing of novel strategies to measure autoantibodies.

For efficiency, initial screening for autoantibodies includes measures of IAA and GADA, which identifies the majority of those at risk for development of multiple autoantibodies and progression of disease. If a participant is positive, the remaining autoantibodies are measured. TrialNet actively screens relatives of individuals with type 1 diabetes, but other individuals who do not have a relative with the disease may join TN01 if they are found to have positive islet autoantibodies in an outside laboratory.

Once autoantibodies are discovered, participants are invited for staging. A standardized OGTT is administered with measurements of glucose, C-peptide, and insulin. Younger participants have progressed faster to clinical type 1 diabetes diagnosis in comparison with with older individuals in the TN01 study (25), and this finding has been confirmed in other cohorts (30). However, while the rate of progression is generally slower in adults, adults with multiple autoantibodies also progress to clinical disease, and as such, adults have been included in all TrialNet studies. This is particularly important as it is estimated that 50% of those with type 1 diabetes develop disease in adulthood (31,32). As one of the few consortia that include adult populations in type 1 diabetes studies, TrialNet provides unique and important data that have contributed to the increasing appreciation of the heterogeneity within a type 1 diabetes diagnosis. Additional measures of β-cell function have proven useful for detecting risk of progression to stage 3 and the effects of therapies. These have been studied extensively with ancillary studies. The investigations have included studies of intravenous glucose tolerance testing (33,34), maximal acute insulin response (AIRmax) (35), oral minimal model (36,37), HbA_1c (38), and continuous glucose monitoring (39,40). Importantly, metabolic measures that only use glucose, the standard measurement in an OGTT, have less predictive value for type 1 diabetes than combined glucose and C-peptide measures. A variety of metabolic risk scores such as the Diabetes Prevention Trial–Type 1 Risk Score (DPTRS) (41), Index60 (33,42), area under the curve (AUC) ratio (42), C-peptide index (33), M₁₂₀ (single blood draw) (43), and others have shown superior predictive value for stage 3 type 1 diabetes in comparison with the glucose measurements from an OGTT alone. Other biomarkers of disease progression have been studied in TrialNet including measures of β-cell stress (e.g., proinsulin–to–C-peptide ratio [44]) and miRNA (45). Importantly, TrialNet data show that once metabolic deterioration has been identified, the effect of age on rate of progression becomes irrelevant (46). An ongoing ancillary study is underway to investigate the value of longitudinal pancreas volume imaging as a predictor of diabetes progression.

In addressing the significance of risk genes in type 1 diabetes progression, the majority of participants in TN01 were genotyped with the Immunochip (47) as part of an NIDDK-funded mechanistic study, and a previously published type 1 diabetes genetic risk score (48) was calculated. Interestingly, once a person is discovered to have multiple autoantibodies, the significance of genetic risk is reduced (49).

The TrialNet Collaborative Mechanistic Studies Panel (CMSP) was developed to evaluate and use mechanistic studies to answer, with each perturbation that was tested in a clinical study, regardless of outcome, questions pertaining to the disease process and immunologic mechanisms and inform clinical trial design (Table 2). Ongoing studies include analysis of gene transcriptional features and proteomics during disease progression. Furthermore, to identify transcriptional signatures in cell subsets that may give insights into pathologic mechanisms in addition to identifying high risk, a series of “Key Questions” was created, focusing effort on understanding these biologic pathways. The focus of Key Question 1 is identifying immune signatures of people at the highest risk of progression from stage 1 or 2 to stage 3 type 1 diabetes within 3 years. RNA sequencing and cytometric analysis of immune cell populations from a matched case-control cohort of ∼180 people who do versus do not progress in this short timeframe allow for comparisons of transcriptomic signatures of cell subsets in progressors and nonprogressors and have recently been completed; serum measures in the same cohort will follow in the coming years.

Since its inception, after the close of DPT-1, TrialNet has performed 19 clinical studies, with an additional 3 ongoing trials, including observational investigations and studies of targeted immune therapies and antigen-specific immune therapies for prevention and treatment in people at risk for and with clinical stage 3 type 1 diabetes. The following is a description of selected studies.

Because TrialNet involves multiple academic clinical centers and affiliates that collaborated with the centers, studies in stage 3 type 1 diabetes could be completed rapidly.

The rationales for the interventions that have been tested were developed from preclinical investigations, largely in murine models of type 1 diabetes, studies of people with and at risk for clinical disease, and evidence of mechanisms of action in other clinical settings. As of December 2024, a total of nine interventional studies were conducted or are ongoing in individuals with stage 3 disease (Table 3).

Studies from nonobese diabetic (NOD) mice showed that T lymphocytes were the primary effectors of the disease and diabetes could be transferred with T cells alone. However, other studies from NOD mice suggested that B cells also play an important role in antigen presentation to pathogenic T cells (50) and that MHC class II deficiency on B cells protected NOD mice from disease development (50,51). The TN05 study was a randomized, double-blind study where participants with new-onset type 1 diabetes were treated with a four-dose course of the B-cell-depleting agent rituximab. This study met its primary end point, with higher C-peptide, reduced HbA_1c, and reduced exogenous insulin usage in comparison with placebo (52). The delay in β-cell dysfunction was transient, however, and by year 2 the differences between the treatment arms were not statistically significant (53). Linsley et al. (54) showed that elevated T-cell levels during B-cell reconstitution were associated with the loss of efficacy, suggesting a strategy for combination therapies.

Parallel studies of pancreata from people who died with type 1 diabetes showed a greater infiltration of B cells into the islets in younger versus older individuals (55,56). These findings, together with other clinical data, led to the notion of a B-cell-associated type 1 diabetes “endotype.” TrialNet contributed to the conceptual framework of type 1 diabetes endotypes and may identify those most likely to respond to particular treatments (57).

The costimulatory signaling of CD28 activates T cells, and its blocking with CTLA4Ig prevented diabetes in NOD mice. Abatacept (CTLA4Ig) was given monthly to newly diagnosed individuals in 27 infusions over 2 years, and significant improvement in C-peptide responses was shown at 1, 2, and even 3 years after enrollment (1 year after the drug was discontinued). However, like with rituximab, C-peptide levels declined over time—in this instance, even during the treatment phase. The mechanistic basis for response was suggested by studies of the impact of abatacept on T follicular helper (Tfh) cells that play a key role in T cell–B cell cross talk. Those with higher levels of activated Tfh cells showed reduced responses to abatacept (58). Furthermore, a B-cell or neutrophil signature in whole blood RNA, even at baseline, identified poor response to therapy. This signature was age related, again suggesting the potential utility of looking at subsets of people, or endotypes/immunotypes, in identifying those most likely to respond (59).

After encouraging results from a pilot study of low-dose antithymocyte globulin (ATG) followed by pegylated granulocyte colony-stimulating factor (GCSF) that showed improved β-cell function in participants even after the new-onset period, the multiarm TN19 trial was developed to test the efficacy of ATG alone or ATG followed by GCSF versus placebo in individuals with new-onset type 1 diabetes (60,61). Low-dose ATG reduced C-peptide decline in treated individuals in comparison with placebo, but the combination therapy did not. Two-year follow-up analysis of the study showed that stimulated C-peptide remained higher in ATG-treated individuals in comparison with but not in the combination arm (62). HbA_1c was lower in both the ATG alone group and the ATG followed by GCSF group in comparison with placebo at 1 and 2 years. Mechanistic studies suggested that ATG induced exhaustion in CD4 T cells, which was associated with better response to therapy, while GCSF appeared to cause increased methylation of the regulatory T cell (Treg)-specific demethylation region, potentially impairing function of Tregs (63).

In the TN08 trial investigators tested whether two or three doses of GAD-Alum would change the phenotype of the immune response to this type 1 diabetes antigen. This study demonstrated an increase in T cell responses to GAD and increased autoantibody titer, but there was no improvement in C-peptide at 1 year.

The primary objective of TrialNet studies is to delay or prevent the progression of type 1 diabetes (Table 4). Because of the effort needed to identify and recruit individuals and the variable timeframe for progression to specific end points (i.e., stage 2 or stage 3), prevention studies are more time-consuming and, at times, have required >8 years for completion. However, because of the greater functional β-cell mass in at-risk individuals than after the clinical diagnosis, the potential for impact on the disease is greater for successful preventative interventions.

Based on the extensive clinical experience in rheumatologic diseases, hydroxychloroquine (HCQ) was tested for its ability to arrest or delay progression to stage 2 type 1 diabetes in individuals with stage 1 type 1 diabetes. This FDA-approved agent is believed to target antigen-presenting cells by decreasing cytokine production and reducing antigen presentation—events thought to be critical in early disease (64). Though the therapy was well tolerated and there were no safety concerns, the trial was stopped early due to futility. Further studies showed that treatment with HCQ decreased the titers of GADA and IAA and reduced the acquisition of additional autoantibodies.

Following the results from the TN09 study in individuals with stage 3 type 1 diabetes, a study of abatacept was developed for individuals with stage 1 type 1 diabetes. In this trial also an intermediate end point was used, stage 2 type 1 diabetes, in the trial design to reduce the time needed for evaluation of the effects of the drug. The duration of treatment was 1 year (14 doses). The clinical end point was not met, but a number of mechanistic findings were of note (65). First, there was an increased frequency of Tfh and T peripheral helper cells in younger versus older individuals, although age was not a determinant of the clinical response. Second, there was an increase in the frequency of naïve CD4⁺ and CD8⁺ T cells but a decline in Tregs after treatment, suggesting that the treatment blocked T-cell development but also may have reduced a subpopulation important for controlling disease progression. Importantly, the trial identified limitations in directly comparing agents in the prediabetes and new-onset period. It remains unclear whether the differences in the trial outcomes between TN18 and TN09 are due to the difference in duration of treatment, differences in disease activity in people with stage 1 versus stage 3 type 1 diabetes, or simply differences between the rates of progression used in trial design and the actual rate observed in study participants.

Studies with anti-CD3 monoclonal antibodies (mAb) began with preclinical investigations in the 1990s. The humanized FcR nonbinding anti-CD3 mAb teplizumab was first trialed in individuals with stage 3 type 1 diabetes. The original clinical trial and a follow-up trial, as well as investigations with another humanized FcR nonbinding anti-CD3 mAb, otelixizumab, confirmed the effects of one or two 12- or 14-day courses in delaying the loss of β-cell function after the clinical diagnosis of type 1 diabetes.

TrialNet developed the TN10 protocol to test whether a single 14-day treatment course of teplizumab would delay or prevent the diagnosis of diabetes in those at the highest risk (stage 2 type 1 diabetes), among whom the diagnosis of stage 3 type 1 diabetes was expected to occur for almost 75% of individuals in 5 years. Strict enrollment criteria (for all participants back-to-back OGTTs indicating dysglycemia were required) were initially developed but modified to allow for a single dysglycemic OGTT for participants age <8 years because younger individuals with an abnormal OGTT progressed to clinical disease regardless of confirmation with a second OGTT.

A total of 76 participants enrolled, with a median age of 13 years (range 8.5–49.5), and were randomized to 14 consecutive days of teplizumab or placebo (66). There was no further therapy given after the single treatment course, and the frequency of and time to progression to stage 3 type 1 diabetes were determined in a time-to-event analysis. After a median follow-up of 745 days, clinical diabetes was diagnosed in 43% of the 44 teplizumab-treated and 72% of the 32 placebo-treated participants. The median time to clinical diagnosis was 48.4 months with teplizumab versus 24.4 months with placebo. In extended follow-up (median 80.5 months), the median time to clinical diagnosis remained significantly longer in the teplizumab arm (52.2 months) versus placebo (27.3 months). After this follow-up, 36% of the teplizumab-treated and 12.5% of the placebo-administered participants were still not diagnosed with stage 3 type 1 diabetes (67).

The delays in progression were accompanied by C-peptide preservation and reduced glucose excursions during OGTTs. Similar to the new-onset setting (68), partially exhausted KLRG1⁺TIGIT⁺ memory CD8⁺ T cells increased after treatment in the prevention trial. These cells from the teplizumab-treated participants secreted less IFNγ and TNFα when activated ex vivo (69). A low expression of the IL-7 receptor (CD127) identified those who were most likely to have extended disease-free intervals (69).

The participants receiving active drug tolerated the treatment well but described low-grade side effects of cytokine release syndrome and rash. There was transient lymphopenia, mechanism driven. Increased EBV loads were detected in 8 of the 30 EBV seropositive participants (all in the teplizumab arm) but resolved spontaneously and caused only mild symptoms in one participant. No difference in severe infections was observed compared with placebo.

Based on these data, and studies in stage 3 type 1 diabetes, Provention Bio, the owner of the drug, filed a Biologics License Application and received approval in November of 2022 for treatment with teplizumab to delay clinical stage 3 type 1 diabetes in patients with stage 2 disease ages 8 years and older. This approval marked the first FDA-approved treatment for delay of the clinical diagnosis of type 1 diabetes—or any autoimmune disease.

TrialNet built on the observations of the oral insulin trial in DPT-1 that suggested there might be a signal of efficacy in the subgroup of individuals with insulin autoantibodies. The TN07 trial was developed and within the trial a secondary stratum of participants was predefined: those with a low first-phase insulin response to intravenous glucose. These individuals had an increased risk for progression to stage 3 type 1 diabetes (70). In the intention-to-treat analysis of this trial, oral insulin again failed to prevent clinical disease, but in the secondary stratum (n = 55) the time to clinical diabetes was significantly delayed in the participants in the oral insulin group, where 48.1% were diagnosed with stage 3 disease, compared with 70.3% who received placebo. Post hoc analysis of the primary stratum has suggested the possibility of additional markers of endotypes with improved response to oral insulin involving those with high IA2A titers or HLA DR4/DQ8 (71).

Lessons learned throughout the last 24 years have and will continue to shape future approaches to prevent and stop type 1 diabetes progression. From mechanistic lessons that have altered approach to clinical trial design to recognition of the participant voice in the network’s priorities, these lessons span scientific and social domains. Several important lessons are highlighted below.

As in other autoimmune diseases where these therapies have long been standard of care, none of the agents tested have led to permanent tolerance. Therefore, efforts to prevent progression of autoimmune diabetes have turned toward repeat dosing or combinations of drugs with complementary mechanisms of action. Mechanistic studies from single-agent trials have suggested how combination studies might be designed to prolong therapeutic efficacy. The results from the TN05 and TN09 studies suggested how combining abatacept and rituximab might prolong efficacy of both agents. In TN09, a poor clinical response to abatacept therapy was associated with an increased response by B cells after treatment in the new-onset setting (59). Conversely, T-cell frequencies were increased after rituximab when C-peptide responses deteriorated (54). Based on this information, a study in patients with new-onset stage 3 type 1 diabetes (Rituximab-pvvr Followed by Abatacept Versus Rituximab-pvvr Alone in New Onset Type 1 Diabetes [T1D-RELAY]) (TN25) was designed with administration of these drugs sequentially, with rituximab for all participants followed by abatacept or placebo. Likewise, the analysis of responders in the TN10 teplizumab trial suggests that blocking IL-7 or its receptor (CD127) might be a mechanistically driven combination for future testing.

Understanding of what is acceptable to patients is also an important consideration for prevention. What risks are patients willing to accept when disease is likely but not guaranteed? How long will patients be willing to be on a treatment? What must be demonstrated before patients are comfortable testing an agent in prevention? How does the FDA approval of a drug to delay the onset of stage 3 disease in those at risk change the acceptability? To address this, a community advisory board has been developed. It engages individuals with or at risk for type 1 diabetes, which will provide important insight into what is tolerable and preferred by participants and their families that can be incorporated into future trial design.

It is critical that autoantibody screening, clinical trials, and therapies are available to all. Enrolling individuals from all racial and ethnic groups provides key insights into differences in disease progression and clinical phenotypes. For example, although as a single demographic group White children have the highest rates of type 1 diabetes, the rate of type 1 diabetes is increasing among all ethnic backgrounds (72). Serving everyone with type 1 diabetes and their families and enabling access to TrialNet’s services are priorities. While TrialNet centers are currently located in North America (U.S., Canada) and Australia, moving forward, the development of a global type 1 diabetes research consortium is of great interest.

The findings presented above, such as the hypothesis-generating data from the oral insulin study, and post hoc observations from the rituximab, teplizumab, and ATG studies, have underscored that not all therapies will be effective in all patients. Further study will help identify the distinguishing features best suited for specific therapies. Several examples of subgroups of responders have been identified. Alongside the now robust finding of development of T-cell exhaustion, work in samples from the teplizumab arm of TN10 indicated that individuals with antibody responses to three commensal bacteria strains had more robust responses to treatment than those without these markers (73). As suggested above, the magnitude of responses in children and adults may differ. The risk-to-benefit ratio of agents may be greatly improved by selection of participants who are most likely to respond to treatment, although further work remains to validate these findings.

Another approach to personalizing treatment is to evaluate responses to therapies compared with the predicted outcome on a per-patient basis. In individuals with stage 3 type 1 diabetes, C-peptide responses to clinical trial agents can be personalized and measured using the quantitative response variable (QR): a model derived from 492 TrialNet participants in stage 3 trials with use of age and baseline C-peptide to determine an expected rate of C-peptide fall over months to years. This measure has now been validated in a larger group of studies and can be applied to several clinical trial situations, e.g., to make go/no go decisions before the time needed for a full evaluation of an agent, to compare responses with prior expected outcomes in a large control group, or to use for a standardized definition for response to therapy across studies (74,75).

Finally, a key objective of TrialNet has been training translational investigators who can apply lessons from preclinical studies to new clinical trials to achieve the consortium’s mission. Investigators who conducted the DPT-1 and other studies are the mentors for the TrialNet Emerging Leaders program and NIDDK’s DiabDocs National K12 Physician Scientist Career Development Program, which support trainees at and outside of TrialNet centers to train a new group of investigators who can build on the successes of the past.

The studies and work by the network have brought the field of type 1 diabetes into a new era. The Pathway to Prevention is the largest screening program in the world and has enabled the conduct of robustly designed and executed clinical studies that have identified promising treatment strategies to accomplish TrialNet’s mission of delay or prevention of the disease. A pivotal TrialNet clinical study has led to the approval of the first drug for delay of type 1 diabetes or any autoimmune disease. With this approval investigators worldwide have begun to ask whether the general population might be screened for risk, and TrialNet has developed tools to engage those individuals in clinical studies where TrialNet will continue to test agents with potential for sustained delay or prevention of disease progression. The outcomes of each trial, whether or not it met its primary end points, have generated data on which future trials can be based. Furthermore, endotypes that define subgroups of individuals for whom interventions might be selected for optimal clinical benefit have been identified for further testing (76). Finally, opportunities have been created for new investigators who are interested in translational studies that already are yielding benefits for patients.

Acknowledgments. The authors give special thanks to all TrialNet participants and families, without whom these discoveries would not be possible.

Funding. The Type 1 Diabetes TrialNet Study Group is a clinical trials network currently funded by the NIH through the NIDDK, the National Institute of Allergy and Infectious Diseases, and the Eunice Kennedy Shriver National Institute of Child Health and Human Development, through cooperative agreements U01 DK061042, U01 DK061058, U01 DK084565, U01 DK085453, U01 DK085461, UC4 DK085466, U01 DK085476, U01 DK085504, U01 DK085509, U01 DK103282, U01 DK106984, UC4 DK106993, U01 DK106994, U01 DK107013, U01 DK107014.

The content is solely the responsibility of the authors and does not necessarily represent the official views of the NIH.

Duality of Interest. L.M.J. has served on an advisory board for Insulet and Sanofi. J.L.F. has served on an advisory board for Sanofi. B.M.N. has been a speaker for Med Learning Group and on an advisory board for RubiconMD. C.S. has received research funding from COUR Pharmaceuticals, consulted for GentiBio, and participated in an advisory board meeting for Sanofi. K.C.H. has consulted for Sanofi, NexImmune, GentiBio, and Dompé. No other potential conflicts of interest relevant to this article were reported.

Author Contributions. L.M.J., J.L.F., B.M.N., and C.S. wrote the manuscript. J.K. reviewed the manuscript and contributed to discussion. K.C.H. conceptualized the manuscript.

Handling Editors. The journal editor responsible for overseeing the review of the manuscript was M. Sue Kirkman.