Toward Personalized Medicine in Type 1 Diabetes: Understanding How Patient Heterogeneity Influences Therapeutic Efficacy

ABSTRACT

Pharmacologic interventions for type 1 diabetes (T1D) have advanced significantly in recent years with the advent of the first FDA approved therapy teplizumab for delaying symptomatic disease onset in 2022. Despite this progress, major hurdles remain in moving toward personalized medicine approaches for T1D. Here, we highlight the examples of heterogeneity in therapeutic responses to recent beta cell and immune interventions and what these studies can teach us about how to tailor therapy for maximizing benefit to patients at risk of or living with T1D. We examine the differences between proposed endotypes, such as childhood‐onset versus adult‐onset disease, and how these distinctions may inform the use of different therapies. We also explore the importance of disease staging in determining therapeutic windows, as early interventions appear most effective before extensive beta cell loss. Emerging biomarkers including autoantibody profiles, metabolic indices, and circulating nucleic acids offer additional tools for stratifying patients and predicting responses. Ultimately, recognizing and leveraging patient heterogeneity provides an opportunity to align therapies with underlying physiology, moving beyond ‘one‐size‐fits all’ approaches. The promise of personalized T1D therapy will be realized by surmounting the barriers to implementation, including trial design, paediatric underrepresentation, and cost.

1. Introduction

Type 1 diabetes (T1D) is a chronic autoimmune disease marked by the destruction of insulin‐producing pancreatic beta (β)‐cells, resulting in lifelong dependence on exogenous insulin. While insulin therapy remains essential, it has its limitations, leaving many patients vulnerable to complications ranging from difficulty maintaining euglycemia to hypoglycemia and ketoacidosis, and even long‐term vascular complications [1, 2]. Central to T1D pathogenesis is the activation of autoreactive T cells that infiltrate the pancreas and destroy β‐cells. Early clinical trials in the 1980s using nonspecific immunosuppressants like cyclosporine confirmed that lymphocytes play a pivotal role in β‐cell destruction and disease progression [3]. However, the transient efficacy and toxicity of these agents curtailed further development, prompting a decades‐long shift toward safer, targeted immunotherapies. Recently, the approval of teplizumab (Tzield) by the U.S. FDA in 2022 marked a historic step forward, offering the first disease‐modifying therapy that delays the onset of clinical T1D by a median time of 2–3 years [4, 5].

Current insights have expanded our understanding of T1D beyond the classical model of immune‐mediated β‐cell destruction. Progressive loss of functional β‐cell mass occurs in three stages in genetically susceptible individuals starting from stage 1 with ≥ 2 autoantibodies, to stage 2 with ≥ 2 autoantibodies and dysglycemia, and finally stage 3 with the onset of clinical disease [6]. Intrinsic β‐cell stressors including endoplasmic reticulum (ER) stress, oxidative stress, aberrant prohormone processing, neoepitope formation, autophagy defects and senescence may precede overt autoimmunity and/or amplify the failure of β‐cells [7, 8, 9, 10, 11, 12]. These mechanisms may contribute to the heterogeneity of T1D and ultimately complicate the clinical presentation of T1D in paediatric and adult populations. Indeed, T1D is increasingly recognized as a heterogeneous disease, with variation in age of onset, genetic predisposition, immune profiles, and rates of β‐cell loss [13, 14]. This heterogeneity poses major challenges for diagnosis, staging, and treatment selection, making the need for therapies that preserve β‐cell function and target disease mechanisms more pressing. Given the variability in disease presentation and response to therapy, there is growing emphasis on developing personalized approaches to T1D treatment. Stratifying patients based on age, disease stage, endotype, and even biomarkers could enable more precise and effective interventions.

In this review, we examine how T1D patient heterogeneity (as defined by the diverse demographic, genetic, environmental, immune, and metabolic factors that combine to shape disease risk, progression, and treatment response [15]) relates to the effectiveness of immune and non‐immune based drug therapies as determined from available clinical trial evidence. We explore the potential clinical utility of endotype classification and disease staging, assess emerging biomarkers that may help inform therapeutic stratification, and evaluate the practical challenges in implementing precision medicine into clinical practice. Ultimately, our goal is to join others to advocate for a patient‐centered framework for managing T1D [16], aligning therapeutic strategies with the unique characteristics of each patient.

2. Methods

This review was informed by literature searches conducted in PubMed, focusing on studies published between 2019 and 2025. Search terms included combinations of type 1 diabetes, endotypes, heterogeneity, immunotherapy, beta cell stress, clinical trials, biomarkers, and disease staging. Abstracts were screened for relevance to inter‐individual heterogeneity in T1D and its implications for therapeutic response and disease progression. Priority was given to randomized controlled trials and well‐characterized longitudinal studies, while preclinical and exploratory studies were included to provide mechanistic context. Clinical trials discussed in detail were selected to represent therapies targeting distinct mechanisms implicated in T1D progression, including immune modulation, β‐cell stress pathways, and metabolic support, allowing therapeutic outcomes to be interpreted within staging and endotype informed frameworks. Statements regarding therapy suitability by disease stage or endotype are presented as hypothesis‐generating interpretations rather than clinical recommendations.

3. How Do Age and Disease Endotypes Shape the Progression of T1D and Treatment Response?

T1D is a heterogeneous disease with significant variability in genetic predisposition, immune responses, and disease progression, which directly impacts the effectiveness of different therapies. Recognizing this heterogeneity is critical for developing precision medicine approaches that optimize treatment outcomes for different patient subgroups [17]. Two key frameworks have emerged to categorize this variability: the classification of T1D into distinct endotypes and the staging of disease progression. Both approaches provide insight into how patients may respond to specific therapies.

Recent studies have identified distinct endotypes of T1D based on age at onset and the relationships with immune infiltration patterns, genetic risk factors, and β‐cells preservation [18, 19]. However, these endotypes are primarily defined through histological and molecular analyses of pancreatic tissue and are not yet directly measurable in living patients. Currently, no validated circulating or imaging biomarkers are available to classify individuals into endotypes in clinical settings. The T1D endotype 1 (T1DE1) is characterized by childhood‐onset with rapid and aggressive β‐cell destruction. Patients diagnosed at younger ages (typically ≤ 7 years) often exhibit a CD20^Hi insulitic profile, marked by a high presence of B cells (CD20⁺), cytotoxic CD8⁺ T cells, and extensive immune‐mediated β‐cell destruction [15]. This phenotype is also associated with lower C‐peptide levels at diagnosis [20]. Consequently, patients within this endotype may benefit most from early immune‐modulating therapies that target key pathways in T1D autoimmune pathogenesis to delay progression. In contrast, T1D endotype 2 (T1DE2) represents a later‐onset form of T1D with a more indolent course of β‐cell decline. Adolescents and adults with this endotype typically retain more functional β‐cell and exhibit a CD20^Lo insulitic profile with fewer infiltrating immune cells [19]. Further studies show that individuals diagnosed ≥ 13 years of age exhibit more preserved insulin processing, which was associated with milder insulitis compared to younger children (≤ 7 years). This milder immune activation suggests that patients with T1DE2 may be better suited to β‐cell supportive therapies which aim to preserve β‐cells and limit functional decline. A growing number of studies support that additional heterogeneity may exist even within the T1DE2 endotype, as metabolic and immunologic differences are emerging in T1D diagnosed in younger versus elderly (> 60 years of age) adults [21]. It should be noted that while helpful in terms of pathology, the T1DE1 and T1DE2 frameworks are unlikely to be completely dichotomous. Inter‐individual variation will likely blur the distinctions between these proposed endotypes as larger sample sizes representing more geographically‐diverse populations are studied.

Age is also a major determinant of β‐cell decline in stage 3 T1D and is incorporated into quantitative response (QR) models used in many clinical trials. Because age at diagnosis and baseline C‐peptide strongly predict subsequent C‐peptide trajectories [22], apparent age‐related differences in therapeutic response may partly reflect differences in baseline metabolic reserve in addition to underlying endotype phenotypes. Accordingly, while age‐associated patterns are informative, the extent to which age alone can deterministically define T1DE1 versus T1DE2 remains uncertain. Nevertheless, the proposed endotype framework provides a useful conceptual lens for interpreting disease heterogeneity.

While many clinical trials in T1D have historically used broad inclusion criteria, emerging insights into endotypes and disease staging offer an opportunity to reinterpret therapeutic outcomes through a precision‐medicine perspective. Rather than proposing T1DE1 and T1DE2 as definitive classifiers, we use this framework pragmatically to explore why therapies may succeed or fail across differences in age, immune activity, and disease stage. In the following sections, we consider whether current clinical trial evidence aligns with this emerging endotype‐informed model of therapeutic stratification.

Emerging transcriptomic analyses have further refined the immunological differences between T1DE1 and T1DE2. A recent study by Torabi et al. identified distinct gene expression patterns associated with lymphocyte differentiation and migration between these endotypes. T1DE1 exhibited upregulation of genes such as IKZF3, IL7R, ZAP70, LAT, and CD8A, which are associated with enhanced lymphocyte development and activation [23]. These transcriptomic signatures provide mechanistic insight into the aggressive β‐cell destruction observed in children diagnosed with T1D at a younger age. This heightened T cell activation and gene expression profiles support the rationale for T cell immunotherapies, such as teplizumab, which could be particularly effective in patients with the T1DE1 endotype.

Otelixizumab, a non‐mitogenic anti‐CD3 antibody (ChAglyCD3) has demonstrated durable metabolic benefits in adolescents and young adults with recent onset T1D (≤ 90 days between the initial diagnosis and the first dose of study drug). Patients < 27 years experienced greater preservation of β‐cell function, lower insulin requirements, and improved metabolic control over 4 years compared to older individuals [24]. These findings reinforce the concept that early anti‐CD3 therapy may be particularly effective in the T1DE1 endotype, where aggressive T cell driven autoimmunity predominates. However, otelixizumab has not been directly tested in immunologically‐defined T1DE1 individuals and these studies have been completed with relatively small sample sizes. Furthermore, transient Epstein–Barr virus reactivation following treatment has raised safety concerns that should be carefully considered when evaluating the use of this anti‐CD3 therapy in paediatric populations [24]. Otelixizumab may benefit younger patients with more aggressive T1D, but safety and trial size remain to be established.

Rituximab, an anti‐CD20 monoclonal antibody, has been studied for its ability to selectively deplete B‐lymphocytes [25], thereby disrupting antigen presentation and modulating autoimmune responses. Unlike therapies that target T cells directly, rituximab aims to slow β‐cell destruction by reducing the availability of B cells that interact with autoreactive T cells [26]. Despite the presentation of T1DE1 with relatively higher B and T cells in islets relative to T1DE2, it is unlikely that rituximab alone would be the most beneficial as a therapy for patients with T1DE1 for two reasons. First, the TrialNet Anti‐CD20 Study evaluated rituximab in individuals aged 8–40 years with newly diagnosed T1D, with a mean age of 19 years. Given this age distribution, the study provided limited evidence regarding efficacy in younger paediatric populations. Second, although participants who received rituximab had about 20% higher C‐peptide area under the curve (AUC) compared to the placebo group (p = 0.03), indicating delayed functional β‐cell loss [27], therapeutic responses to rituximab vary, and a study by Linsley et al. reported that individuals with higher baseline T cell levels showed a poorer response to rituximab. Further, RNA sequencing analysis of clinical trial blood samples revealed increased expression of genes involved in T‐cell activation and suggested reduced B‐cell depletion efficiency in non‐responders [28]. This heterogeneity in responses suggests that rituximab alone may not be sufficient for durable β‐cell preservation and thus is it plausible that combination therapies targeting both B and T cells will be more effective when treating T1DE1 patients [28]. Rituximab therapy delays β‐cell loss in newly diagnosed T1D patients, but its effects were not sustained beyond 2 years, likely due to the resurgence of B‐lymphocytes and persistent T‐cell–mediated autoimmunity.

As T1DE1 is defined by childhood‐onset (≤ 7 years of age), it is important to test and identify therapies that have demonstrated efficacy in paediatric populations. However, this presents a significant challenge, as children are typically underrepresented in T1D clinical trials, limiting the generalizability of therapeutic findings from adults to this age group. One study that directly focuses on children and young adults explored golimumab, a tumour necrosis factor‐alpha (TNF‐α) inhibitor, as a potential disease‐modifying therapy for T1D due to its ability to modulate inflammatory pathways implicated in β‐cell destruction [29]. Unlike traditional immunotherapies that broadly suppress the immune system, golimumab selectively neutralizes TNF‐α, a cytokine known to contribute to β‐cell dysfunction and autoimmune activation [29]. The T1GER study evaluated the effects of golimumab in children and young adults (ages 6–21) with new‐onset T1D. Over a 52‐week treatment period, participants receiving golimumab showed significantly better preservation of endogenous insulin production, indicated by higher C‐peptide AUC. Additionally, insulin requirements were lower in the golimumab‐treated group, with 43% achieving partial remission (defined as an insulin dose‐adjusted HbA1c ≤ 9), compared to only 7% in the placebo group [29]. A two‐year follow‐up study assessed whether these benefits persisted after treatment cessation at week 52. Notably, golimumab‐treated participants continued to exhibit better β‐cell function and metabolic stability even 1 year after stopping therapy. Interestingly, a subset of “responders”—individuals who experienced minimal decline in insulin production—demonstrated even greater metabolic improvements [30], suggesting that inter‐individual variability may influence therapeutic outcomes and should be considered when designing clinical trials and personalized treatment strategies.

Imatinib, a tyrosine kinase inhibitor originally developed for chronic myeloid leukaemia (CML), has emerged as a potential disease‐modifying therapy in T1D due to its effects on immune modulation and β‐cell stress. Imatinib acts on multiple pathways, including inhibiting c‐Abl in pancreatic β‐cells and reducing ER stress, preventing β‐cell apoptosis, and modulating immune responses through platelet‐derived growth factor receptor (PDGFR) [31]. Some evidence from nonobese diabetic (NOD) mice suggests that imatinib acts on B cells to protect β‐cells in T1D [32], although this has not been explored in humans. The efficacy of imatinib in preserving β‐cell function was assessed in adults with recent‐onset stage 3 T1D (≤ 100 days from diagnosis, aged 18–45 years), thus more representative of the T1DE2 endotype. The study demonstrated that imatinib‐treated patients had significantly higher C‐peptide AUC levels at 12 months compared to placebo, although this effect was not sustained at 24 months [31], suggesting that prolonged therapy may be necessary. In addition to modest β‐cell preservation, imatinib reduced insulin requirements and lowered HbA1c levels during treatment, although these benefits also diminished after therapy cessation [31]. A notable case study in a 14‐year‐old patient with T1D and a platelet‐derived growth factor receptor beta (PDGFRB) mutation revealed sustained partial remission with imatinib [33]. This patient, initially diagnosed with diabetic ketoacidosis (DKA), started imatinib alongside exogenous insulin. Over the next year, his HbA1c decreased, insulin requirements dropped, and he maintained > 90% time‐in‐range glucose levels without severe hypoglycemia. This case suggests that imatinib may be particularly effective in a subset of T1D patients with specific genetic or immune profiles [33]. Imatinib therapy demonstrates short‐term β‐cell preservation and metabolic benefits in newly diagnosed T1D patients, with some evidence of sustained remission in select cases. On the other hand, significant gastrointestinal side effects occurred in patients treated with this agent [31], suggesting that doses must be closely monitored to minimize side effects. Future research should explore if the role of imatinib is successful in patients stratified by endotypes based on immune profiles or β‐cell stress markers.

Baricitinib, a selective janus kinase 1 and 2 (JAK1/2) inhibitor, also holds promise in treating stage 3 T1D through suppression of pro‐inflammatory cytokine signalling via the JAK–STAT pathway that plays a critical role in the immune‐mediated attack on pancreatic β‐cells. Preclinical studies demonstrated that JAK inhibition can prevent diabetes onset and even reverse newly diagnosed diabetes in the NOD mouse model, supporting its exploration in clinical settings [34]. The BANDIT trial involving 91 participants (ages 10–30) treated within 100 days of diagnosis found significantly higher C‐peptide AUC at 48 weeks in the baricitinib group (p = 0.001), suggesting that baricitinib helped preserve endogenous insulin production [35]. Baricitinib treated patients exhibited reduced insulin requirements, lower glycemic variability as assessed by CGM data, although HbA1c levels remained unchanged [35]. Patients in the baricitinib group also spent more time in the target glucose range (70–180 mg/dL), particularly in the first 24 weeks of treatment [34]. Mechanistically, baricitinib reduced CD8⁺ effector memory T cell populations and STAT3 phosphorylation, indicative of suppressed T cell activity, supporting the hypothesis that baricitinib works by dampening immune responses that drive β‐cell loss [35]. Participant stratification into age at diagnosis would be helpful in determining whether there were differences in the efficacy of baricitinib in T1DE1 versus T1DE2 individuals.

Verapamil, a calcium channel blocker traditionally used for hypertension [36], has demonstrated potential to reduce pancreatic β‐cell apoptosis and enhance β‐cell function by downregulating thioredoxin‐interacting protein (TXNIP), a key regulator of oxidative stress and β‐cell death [37]. Participants with new‐onset T1D receiving verapamil for 12 months had significantly higher C‐peptide AUC (p = 0.0186), and lower insulin requirements than placebo [37]. A meta‐analysis further confirmed improved β‐cell preservation at 1 year, although HbA1c remained unchanged [38]. In paediatric patients, verapamil preserved C‐peptide and reduced progression of β‐cell decline, with 95% of the verapamil treated patients maintaining clinically meaningful insulin production [39]. An exploratory study showed that verapamil can reduce pro‐inflammatory markers and lower serum chromogranin A (CHGA), an autoantigen associated with T1D [40]. The Ver‐A‐T1D trial, a large ongoing multicenter phase 3 trial, aims to further validate the efficacy of verapamil on a larger scale, which will provide more evidence on whether verapamil should be incorporated into clinical practice for T1D management [41]. Verapamil therapy demonstrates significant β‐cell preservation in both adults and children with recent‐onset T1D, with reduced insulin requirements and fewer hypoglycemic episodes. Thus, data are supportive of use of this agent in T1DE1 or T1DE2, where it could be combined with immunotherapy to halt disease progression and protect residual β‐cell mass.

Evidence supports alefacept, a CD2‐targeting fusion protein, in preserving β‐cell function in new‐onset stage 3 type 1 diabetes (T1D). This agent selectively depletes effector memory (T_EM) and central memory (T_CM) T cells while sparing regulatory T cells (T_REG), leading to improved clinical outcomes such as preserved C‐peptide secretion and reduced insulin use [42, 43]. Beyond depletion, alefacept expanded exhausted‐like TIGIT⁺PD‐1⁺ CD4 Tems with reduced inflammatory function, suggesting that therapeutic efficacy may depend more on altering T cell phenotypes than on cell removal alone, although variability in treatment response has been observed in these studies [44]. Importantly, baseline immune endotypes can shape alefacept treatment response. For instance, new‐onset T1D patients with a higher frequency of proinflammatory islet‐autoreactive memory CD4⁺ T cells had poorer C‐peptide preservation following alefacept, indicating that inflammatory T cell signatures may act as biomarkers to predict which individuals will respond [45], although confirmation in larger cohorts is needed. Therefore, alefacept's effectiveness may depend on baseline immune profiles, making biomarker guided patient selection essential. Taken together, these studies support that distinct therapeutic approaches could be tailored for slowing disease during early clinical onset stage 3 depending on age at onset T1D endotypes.

When considered collectively, clinical trial data suggest that therapeutic responses in T1D are biologically heterogeneous across patient populations. Immune‐modulating therapies, including anti‐CD3 antibodies, co‐stimulation blockade, and cytokine‐targeting approaches, may be most relevant during periods of active immune‐mediated β‐cell destruction, consistent with features associated with the proposed T1DE1 endotype. In contrast, therapies that reduce β‐cell stress or support residual β‐cell function, such as verapamil and incretin‐based approaches, may align with clinical features associated with slower disease progression and preserved β‐cell function, consistent with T1DE2 (Figure 1). However, these interpretations remain conceptual. To date, most clinical trials have been designed using age and disease duration rather than prospectively defined endotypes, which limits direct validation of endotype specific treatment effects. In addition, baseline C‐peptide and metabolic reserve may independently influence β‐cell decline and therapeutic response. As a result, linking therapeutic efficacy to proposed endotypes should be considered hypothesis‐generating. Future studies integrating pancreatic pathology, biomarkers, and clinical stratification will be required to establish clinical relevance.

FIGURE 1.

Clinically validated therapeutic targets in T1D. Immune‐mediated β‐cell destruction in T1D involves interactions between T cells, B cells, and inflammatory cytokine signalling pathways. Immune‐targeted therapies act by modulating T‐cell activation (teplizumab, otelixizumab), costimulatory signalling (abatacept), B‐cell function (rituximab), cytokine pathways (golimumab, baricitinib), or memory T‐cell activity (alefacept). In parallel, β‐cell supportive therapies target intrinsic β‐cell stress pathways, including calcium signalling and TXNIP regulation (verapamil), tyrosine kinase signalling (imatinib), and metabolic support via incretin‐based therapies. Together, these approaches illustrate how therapeutic strategies in type 1 diabetes can target both immune and β‐cell mechanisms.

4. What Lessons Have We Learned From Clinical Trials About the Timing and Combination of Therapies to Preserve β‐Cell Function?

Beyond endotypes, the staging model of T1D progression provides another framework for tailoring therapy. T1D develops over multiple stages, beginning with a preclinical phase before the onset of symptomatic hyperglycemia [46]. Stage 1 T1D is defined by the presence of 2 or more autoantibodies typically directed against some combination of islet antigen‐2 (IA‐2), insulin, glutamic acid decarboxylase (GAD), and/or zinc transporter 8 (ZNT8) with normoglycemia. Stage 2 involves continued autoantibody positivity but with dysglycemia, and stage 3 is defined by clinical diabetes with varying β‐cell loss. The timing of intervention is critical, as therapeutic efficacy often varies by stage. Immune therapies such as teplizumab and abatacept have demonstrated efficacy in delaying disease progression when administered before stage 3 T1D onset, suggesting that early intervention may be key to preserving β‐cell function. Conversely, once patients progress to stage 3 and experience significant β‐cell loss, the therapeutic focus shifts toward β‐cell replacement approaches, such as stem cell therapy or islet transplantation, though concurrent immune modulation is still required to prevent rejection and recurrent autoimmunity. Notably, a subset of individuals with long‐standing (> 10 years) T1D retain residual C‐peptide production [47], indicating that some therapeutic strategies, particularly those aimed at β‐cell support, may still provide benefits even after clinical diagnosis. Teplizumab modulates autoreactive T cells and has been extensively studied for its ability to delay disease progression and preserve β‐cell function in individuals at risk for or newly diagnosed with T1D [4, 48]. In the pivotal TrialNet TN‐10 study, a single 14‐day course of teplizumab delayed the onset of clinical T1D by a median of 32.5 months in high‐risk relatives of individuals with T1D. Only 43% of the teplizumab‐treated patients progressed to clinical diabetes within 5 years, compared to 72% in the placebo group (Hazard Ratio, HR = 0.41, p = 0.006) [4]. The PROTECT phase 3 trial further demonstrated that teplizumab preserved C‐peptide responses at 78 weeks (p < 0.001), although effects on insulin use, HbA1c levels, or time in the target glucose range were not statistically significant [48]. The approval of teplizumab (Tzield) in 2022 by the FDA marked a historic milestone, becoming the first therapy to delay progression to clinical T1D. Nevertheless, questions remain regarding optimizing patient selection, determining the potential for repeated dosing, and integrating teplizumab with other immunotherapies for enhanced efficacy. Combination therapy with AG019 and teplizumab demonstrated complementary effects in a small cohort of adults and adolescents with recent‐onset T1D. AG019 is an oral therapy that uses genetically modified Lactococcus lactis bacteria to deliver human proinsulin and IL‐10 directly to the gut to promote antigen‐specific immune tolerance. Participants who received both anti‐CD3 immunotherapy and AG019 developed partially exhausted CD8⁺ T cells, showed reductions in preproinsulin‐specific CD8⁺ T cells, and had increases in Type 1 regulatory T (Tr1) cells [49]. These preliminary findings support the potential of combination strategies, particularly those building on the already approved anti‐CD3 therapy teplizumab.

Abatacept, a selective costimulation modulator that inhibits T‐cell activation by blocking CD80/CD86‐CD28 interactions [50, 51], has shown promise in delaying disease progression. In a phase II clinical trial of 112 participants aged 6–36 years with recent‐onset T1D, abatacept treatment resulted in higher C‐peptide AUC at 2 years (p = 0.0029), and lower HbA1c levels compared to placebo, although insulin requirements remained similar [52]. A follow‐up study demonstrated sustained C‐peptide levels one year after cessation (p = 0.046), indicating a prolonged immunologic benefit [53]. In a more recent phase 2 trial, abatacept was evaluated at stage 1 T1D. While C‐peptide responses were improved (p < 0.03), the therapy did not significantly delay progression to stage 2 or stage 3 T1D (HR = 0.702, p = 0.11) [54]. Nevertheless, abatacept induced measurable immune modulation, including reductions in inducible T‐cell costimulatory (ICOS)⁺PD1⁺ T‐follicular helper (T_FH) cells and transient increases in naive CD4⁺ T cells. Collectively, these studies highlight that the clinical impact of abatacept is moderate and transient, and further research is needed to refine its role in earlier stage intervention.

GAD antigen therapy targets GAD65, a major autoantigen in T1D, and aims to induce antigen‐specific tolerance [55, 56, 57]. While initial subcutaneous GAD‐alum (Diamyd) trials in new‐onset stage 3 T1D patients showed no significant C‐peptide preservation 12 months (p = 0.98) or 24 months (p = 0.50), newer studies testing intralymphatic GAD‐alum injections combined with oral vitamin D have shown promise [58]. The DIAGNODE‐2 trial, focusing on HLA‐DR3‐DQ2‐positive patients, found GAD‐alum‐treated patients had significantly better glycemic control, including higher time‐in‐range values (p = 0.0075) and reduced glycemic variability (p = 0.0219). While the therapy improved metabolic parameters, the effect on β‐cell function was modest [56], indicating that GAD‐alum may provide benefits for select genetic subgroups rather than across the broad T1D population. Additional trials explored different dosing regimens and patient populations, but results remained inconsistent. Overall, while intralymphatic GAD‐alum therapy shows potential, particularly HLA‐DR3‐DQ2 carriers, the broader application of GAD65‐based immunotherapy remains uncertain [59]. The variable efficacy across trials suggests that GAD‐based immunotherapy may require biomarker driven patient stratification and may be most promising as a component in combination therapy.

In addition to immune therapies, incretin‐based therapies such as glucagon‐like peptide 1 (GLP‐1) receptor agonists (GLP‐1 RAs) semaglutide and dual glucose‐dependent insulinotropic polypeptide receptor (GIPR)/GLP‐1R agonist tirzepatide have shown potential as adjuncts in T1D and latent autoimmune diabetes in adults (LADA). These agents, already approved for type 2 diabetes (T2D) and weight management, may offer metabolic benefits in T1D improving glucose control, reducing insulin requirements, and preserving residual β‐cell function. Evidence suggests that these therapies may help address metabolic challenges, including glycemic variability and increased risk of cardiovascular complications associated with T1D. Importantly, studies indicate that tirzepatide could modify disease progression by preserving stimulated C‐peptide secretion, reducing insulin requirements, and extending the remission period, though long‐term randomized trials are needed to validate these effects [60, 61]. Additionally, findings from recent trials suggest that strategic combination therapy, such as pairing SGLT2 inhibitors with glucagon receptor antagonists, can optimize metabolic outcomes while minimizing risks [62]. In adults with established T1D (> 5 years), this approach enabled reductions in insulin dosing and ketogenesis during insulinopenia, underscoring the importance of timing and complementary mechanisms in β‐cell function and metabolic stability. The findings support the inclusion of incretin‐based therapies in personalized treatment strategies for T1D as adjuncts within weeks of diagnosis to support residual β‐cell function.

Available evidence suggests that therapeutic efficacy in T1D may vary by disease stage and heterogeneity, although direct endotype or stage stratified proof remains limited because most trials were not designed for these comparisons. Patients with aggressive autoimmune destruction, such as those with the T1DE1 endotype, may require more intensive immunomodulatory strategies to prevent rapid β‐cell loss, while individuals with slower disease progression, including those with T1DE2, may respond better to therapies that support β‐cell function. The T1D staging model provides a rationale for early, pre‐symptomatic intervention, as therapies aimed at delaying disease onset are more effective when initiated before significant β‐cell destruction occurs. Importantly, findings across recent trials suggest that combination therapies addressing both autoimmunity and β‐cell support may be necessary to achieve durable clinical benefits. Moving forward, personalized therapeutic approaches that integrate stage of disease and combination therapies should become standard considerations in trial design and clinical decision making. Table 1 summarizes selected immune and non‐immune drug therapy strategies currently under investigation or clinical use in T1D, highlighting their mechanisms of action, potential endotype suitability, and optimal disease stage for intervention. Therapeutic strategy selection in T1D should consider both endotype and disease stage to optimize outcomes.

TABLE 1.

Representative therapies in T1D and their feasibility in precision medicine questions.

Therapy	Mechanism	Stage(s) tested	Trial population tested (key trials)	Primary outcome met?	Hypothesized endotype relevance a	Evidence strength	Feasibility in routine care b	Notes/Reference(s)
Abatacept	CTLA‐4–Ig; blocks CD28 co‐stimulation	Stage 1 Stage 3	Stage 1: Aab+; Paediatric/adults (6–45y) Stage 3: Recent‐onset T1D; Paediatric/young adults (6‐45y)	Stage 1: No (delay not met); Stage 3: Yes (C‐peptide preservation)	T1DE1	Moderate (modest and transient C‐peptide preservation)	IV infusion + longitudinal monitoring; feasible in specialty outpatient settings	[52, 53, 54]
Alefacept	Targets CD2; depletes memory T cells	Stage 3	Recent‐onset T1D; adolescents/young adults (12–35y)	Mixed/partial (primary endpoint not clearly met at 12 mo; metabolic + longer‐term benefits reported)	T1DE1	Moderate (immunologic effects with variable metabolic benefit)	Weekly IM injections; in repeated dosing + immune monitoring; scalability may be limited	[42, 43, 44, 45]
Baricitinib	JAK1/2 inhibitor	Stage 3	Recent‐onset T1D; adolescents/young adults (10‐30y)	Yes (C‐peptide preservation reported)	T1DE1	Moderate (promising early phase preservation data)	Oral; requires monitoring; feasible with safeguards	[34, 35]
GAD‐alum	Antigen‐specific immunotherapy (GAD65)	Stage 1 Stage 2 Stage 3	Recent‐onset T1D; subgrouping includes HLA‐DR3‐DQ2 exploration	Variable (signal in genetic subgroup; heterogeneous across approaches)	N/A (may depend more on HLA context)	Limited (heterogenous outcomes; genotype dependent)	SC or intralymphatic injections + follow‐up; feasible in specialized clinics	[55, 56, 57, 59]
GLP‐1 receptor agonist	Metabolic adjunct; slows gastric emptying, appetite; insulin sparing	Stage 3	Adults with T1D (adjunct to insulin)	No (disease‐modifying endpoints not established); metabolic benefits variable	T1DE2	Limited (metabolic benefits without clear disease modification)	SC injection; routine care feasible; GI side effects + hypoglycemia considerations	[60, 61, 62]
Golimumab	Anti–TNF‐α	Stage 3	Recent‐onset T1D; Paediatric/young adult (6‐21y)	Yes (preservation of β‐cell function in treated group reported in main trial publication)	T1DE1	Moderate (positive paediatric signal; durability uncertain)	SC biologic (or IV formulation in some settings); monitoring for infection; feasible in specialty outpatient care	[29, 30]
Imatinib	Tyrosine kinase inhibitor; impacts immune + β‐cell stress pathways	Stage 3	Recent‐onset T1D; Adults	Yes (primary endpoint met at 12 months in phase 2 report)	T1DE2	Moderate (short‐term preservation, durability uncertain)	Oral; tolerability and safety monitoring may limit broad use	[31, 33]
Otelixizumab	Anti‐CD3 mAb	Stage 3	Recent‐onset T1D; Adolescents/adults	No (primary efficacy endpoint not met in pivotal program)	T1DE1	Limited (inconsistent efficacy and safety constraints)	IV infusion; monitoring immune AEs/viral reactivation; routine use limited	[24]
Rituximab	Anti‐CD20; B cell depletion	Stage 3	Recent‐onset T1D, ages ~8–40	Yes (transient preservation of C‐peptide)	T1DE1 (in combination with other therapies)	Moderate (transient β‐cell preservation)	IV infusion; infusion reactions + infection risk; feasible in specialty outpatient infusion settings	[27, 28, 63]
Teplizumab	Anti‐CD3 mAb	Stage 2 Stage 3	Stage 2 at‐risk relatives; Stage 3 recent onset	Stage 2: Yes (delays onset); Stage 3: Yes (C‐peptide preservation in trials)	T1DE1	Strong (Stage 2 delay; stage 3 C‐peptide preservation)	14‐day IV course; monitoring lymphopenia; implementable in specialized centers	[4, 48]
Verapamil	Ca2+ channel blockade/TXNIP modulation	Stage 3	Recent‐onset T1D; Adolescents (7‐17y and 8.5–17.9y) and Adults (18‐45y)	Yes (β‐cell preservation reported in prior trials; ongoing confirmatory work)	T1DE1 and T1DE2	Moderate (β‐cell preservation, larger studies ongoing)	Oral; highly feasible; standard CV monitoring	[37, 38, 39, 41]

Abbreviations: SC, subcutaneous; IV, intravenous; IM, intramuscular; CV, cardiovascular; GI, gastrointestinal; AE, adverse events.

^a Hypothesized endotype relevance reflects conceptual alignment between therapy mechanism, age‐associated pathology, and disease staging. Clinical trials to date have not prospectively stratified participants by T1DE1 or T1DE2 endotypes.

^b Feasibility in routine care reflects anticipated considerations for outpatient implementation. Assessments are qualitative and intended for contextual interpretation.

A subset of individuals diagnosed with insulin‐dependent diabetes do not present with detectable islet autoantibodies, which is frequently referred to as ‘idiopathic’ or autoantibody‐negative type 1 diabetes (historically classified as type 1B diabetes) [64]. These cases appear to be more common in individuals of non‐European ancestry and remain poorly understood in terms of disease mechanism and progression [65, 66]. Current staging frameworks for T1D rely heavily on genetic susceptibility and the presence of islet autoantibodies, which limits their direct applicability to autoantibody‐negative patients. In clinical practice, evaluation of these individuals often includes genetic testing to exclude monogenic diabetes, such as maturity‐onset diabetes of the young (MODY), using next‐generation sequencing approaches [67, 68]. Even after monogenic diabetes is ruled out, biomarkers that reliably predict disease progression or therapeutic response in autoantibody‐negative diabetes are lacking. These limitations highlight an important gap in current precision‐medicine frameworks for T1D and underscore the need for biomarker discovery efforts that extend beyond autoimmunity‐based classification and for genetic risk scores for individuals of non‐European ancestry.

5. How Can We Align Biomarker Profiles With Therapeutic Strategies for T1D?

The identification and validation of biomarkers in T1D have become critical for early diagnosis, disease monitoring, and predicting responses to different therapies. Current biomarker research extends beyond traditional HLA genotyping and autoantibody screening to include metabolic markers, immune profiling, circulating nucleic acids, and proteomic signatures to enable a more refined approach to risk assessment and treatment strategies [69]. Table 2 provides an overview of emerging biomarker categories and their potential use in guiding therapeutic strategies for T1D, which are discussed throughout this review. It is important to distinguish biomarkers that predict disease risk or stage from those that may guide therapeutic selection or predict treatment response. Biomarkers can also be considered according to their level of validation: analytical validity refers to the reliability of measurement, clinical validity reflects the ability to identify disease risk relevant to T1D progression, and clinical utility describes whether biomarker information can meaningfully guide clinical decision making [88, 89]. While several biomarkers in T1D demonstrate strong analytical and clinical validity for staging and disease monitoring, fewer currently demonstrate clinical utility for predicting therapeutic response (Figure 2).

TABLE 2.

Biomarkers in T1D: Current roles and potential implications.

Biomarker type	Specific example	Utility a	Evidence strength	Disease stage	Immune focused therapy suitability b	β‐cell supportive therapy suitability b	Key references
Genetic risk	HLA‐DR3, DR4, Polygenic Risk Scores	Risk prediction and screening	Established for risk prediction	Pre‐autoimmunity (Stage 0)	No (indicates susceptibility rather than active disease)	No (indicates susceptibility rather than active disease)	[69, 70, 71, 72]
Autoantibodies	IAA, GADA, IA‐2A, ZnT8A	Risk prediction and disease staging	Established for staging and risk	Stage 1–3	Yes (supports early immune intervention)	Indirect	[46, 73]
Metabolic and Stress Markers	C‐peptide, Proinsulin: C‐peptide ratio, HSP90, SASP	Assessment of β‐cell stress, function and progression	Established (C‐peptide, PI:C); Emerging (HSP90, SASP)	Stage 2–3	Yes	Yes	[5, 74, 75, 76, 77, 78]
Immune Profiling	T cell exhaustion markers (PD‐1, TIGIT, LAG‐3, KLRG1), autoreactive T cells	Identification of immune activation states	Emerging	Stage 3	Yes	Indirect	[70]
Circulating miRNAs	miR‐375, miR‐21‐5p	Immune activity, β‐cell stress	Emerging	Stages 1–3	Potential	Potential	[70, 79, 80, 81]
Circulating DNA	Unmethylated INS DNA	β‐cell death monitoring	Emerging	Stages 2 and 3	Yes	Yes	[82, 83, 84, 85]
CGM Metrics	Time Above Range > 7.8 mmol/L	Glycemic control, staging	Established for glycemia; Emerging for prediction	Stages 2 and 3	No	Yes	[73, 86, 87]

^a Utility reflects the current or potential clinical relevance of each biomarker based on available evidence; for several markers, application to therapy selection remains preliminary.

^b Therapy suitability indicates whether the biomarker may help inform selection of immune‐directed or β‐cell‐supportive strategies based on current evidence; prospective validation is limited.

FIGURE 2.

Biomarker categories and validation framework in T1D. Biomarkers in T1D can be broadly categorized into established markers used for disease risk assessment and staging (blue box; genetics, autoantibodies, metabolic testing, and C‐peptide) and emerging biomarkers that may support patient stratification and therapeutic monitoring (orange box; immune profiling, circulating nucleic acids, and stress‐related metabolic markers). The biomarker validation framework distinguishes analytical validity, clinical validity, and clinical utility (pink box). Mapping biomarker classes across disease stages (green box) highlights how different tools may be applied at distinct phases of T1D progression, while emphasizing that biomarkers predicting therapeutic response remain under development.

T1D has a strong genetic component, with HLA genotypes such as HLA‐DR3‐DQ2 and HLA‐DR4‐DQ8 conferring the greatest risk, although additional loci contribute to disease heterogeneity [70]. Beyond HLA Class II gene polymorphisms, the INS gene and > 40 other loci expressed in β‐cells/islets also confer T1D risk [71, 72], potentially via post‐transcriptional mechanisms such as RNA editing [90] or UPR‐mediated mRNA decay in β‐cells [91]. Multi‐ancestry genome‐wide association studies have identified population‐specific variants such as NRP1 in African and Hispanic/Latino individuals, reinforcing the need for ancestry‐informed risk assessment [92]. Despite advances in genetic risk scores, these tools remain an imperfect predictor of disease onset [79]. Fine‐mapping studies have also identified novel loci, such as BACH2, a regulator of immune tolerance, as a potential therapeutic target for immune intervention [93]. Integration of genetic and single‐cell epigenomic data has revealed that many T1D‐associated variants reside in non‐coding regions, particularly in CD4⁺ T cells and pancreatic ductal cells, implicating dysfunction in both immune and exocrine pancreas compartments [94]. Genetic data provide foundational insight into disease susceptibility but must be integrated with functional and immunological markers for targeted intervention.

Currently, islet autoantibodies (IAA, GADA, IA‐2A, and ZnT8A) serve as the earliest detectable biomarkers and remain the gold standard for identifying at‐risk individuals [79]. The presence of multiple autoantibodies significantly increases the likelihood of disease progression. Twin studies further emphasize the heritability of T1D, with a 69% risk of progression to clinical T1D in identical twins with multiple autoantibodies, underscoring the predictive value of combining genetic and serologic screening [82]. Notably, IA‐2A positivity has been shown to confer a particularly high risk. A longitudinal analysis from TrialNet revealed that IA‐2A⁺ individuals had a 5.3‐fold increased risk of clinical progression within 5 years when presenting as single‐autoantibody positive, a 2.2‐fold increase at stage 1 and a 1.3‐fold increase at stage 2 [46]. IA‐2A positivity also correlated with greater metabolic dysfunction and genetic risk, with higher glucose levels, lower C‐peptide response, and an association with HLA‐DR4 [46, 73]. Autoantibody profiles, particularly IA‐2A, can enhance patient stratification and timing for immune intervention across T1D stages.

Metabolic and protein‐based biomarkers offer complementary insight into β‐cell stress and function. For instance, an elevated proinsulin‐to‐C‐peptide (PI:C) ratio is a sensitive marker of β‐cell dysfunction and impending diabetes [74]. Elevated fasting PI:C ratios precede the onset of T1D, particularly in younger children, suggesting their potential for early intervention strategies [74, 75]. On the other hand, this marker may not always reflect therapeutic efficacy, since patients responding to teplizumab did not show significantly reduced PI:C ratios compared to placebo in the 2‐year follow up study [5]. Heat shock protein 90 (HSP90) has been identified as a stress marker in β‐cells, with increased levels correlating with disease progression [76]. Proteomic studies and targeted marker panels have indicated that protein signatures may provide useful biomarkers for β‐cell stress during disease progression [95]. Indeed, we recently found that a 7‐marker panel of human islet senescence‐associated secretory phenotype (SASP) factors showed higher levels in plasma samples of paediatric and young adult donors at stage 2 and 3 recent onset T1D (within 1.5 years) compared to autoantibody‐negative controls, which could represent a biomarker for senescent cell accumulation in T1D [78]. Oral glucose tolerance tests (OGTT) in preclinical T1D have revealed that individuals with stage 2 T1D exhibit lower insulin secretion and increased insulin clearance, differentiating them from stage 1 individuals [77]. Building from the OGTT used to classify stage 2 T1D, metabolic measures such as the oral minimal model (OMM) have been applied to discriminate progressors from non‐progressors following teplizumab treatment in stage 2 T1D [96]. These findings suggest that metabolic testing can be a useful tool for tracking disease progression and offer refinements on when and how to intervene.

Circulating microRNAs (miRNAs) have emerged as promising non‐invasive biomarkers of T1D onset and progression. For instance, differential expression of miRNAs at the 14q32 locus, including miR‐409‐3p, miR‐127‐3p, and miR‐382‐5p, distinguishes two subgroups of recent‐onset T1D patients, with one subgroup showing improved glycemic control and different immune profiles [70]. Extracellular vesicle (EV)‐associated miR‐21‐5p increases in response to inflammatory cytokines and serves as a marker of β‐cell stress [80]. Additionally, miR‐200c‐3p, miR‐301a‐3p, and miR‐382‐5p have been implicated in immune dysregulation and β‐cell apoptosis, further supporting their role as predictive markers [79]. Studies indicate that miR‐375 levels were significantly downregulated in children with T1D and further reduced in their siblings compared to healthy controls. This suggests that miR‐375 could serve as an early biomarker of β‐cell function and glycemic control, potentially reflecting the number of functional β‐cells under immune attack. Notably, miR‐375 levels showed an inverse correlation with HbA1c, emphasizing its role in metabolic regulation, although it did not correlate with antibodies or total insulin dosage [81]. MicroRNA signatures like miR‐375 provide a window into immune and metabolic regulation and may serve as a potential biomarker, although larger, longitudinal studies to validate the utility of miR‐375 are clearly needed.

Circulating cell‐free methylated and unmethylated preproinsulin (INS) DNA has been proposed as a real‐time biomarker of β‐cell death, offering insights into disease progression and residual β‐cell activity in T1D. The promoter region of INS typically lacks DNA methylation in pancreatic beta cells which express this gene, and hence the ratio of unmethylated‐to‐methylated INS cell‐free DNA reflects the extent of passive release of beta cell genomic DNA versus other cell‐types DNA and is a measure of recent beta cell death. Elevated levels of both unmethylated and methylated INS DNA were detected at stage 3 disease onset, with unmethylated DNA levels declining after diagnosis, indicating active β‐cell destruction [83]. Others have shown that both unmethylated and methylated INS DNA remain elevated in individuals with longstanding T1D where C‐peptide was undetectable, suggesting β‐cell loss even at late stages in the disease [84]. A multi‐laboratory comparison of assays measuring unmethylated INS DNA confirmed its potential as a reproducible and quantifiable biomarker, emphasizing the need for standardized assays to enhance its clinical application [82]. A recent systematic review identified unmethylated INS DNA among one of the most consistent markers of β‐cell loss across multiple studies [85]. These findings suggest circulating DNA as a biomarker serves as an in vivo way to measure β‐cell death, which could be important in identifying individuals who could benefit from therapies that could target residual β‐cells.

Complementary to these molecular tools, continuous glucose monitoring (CGM) has emerged as a powerful method for tracking progression in individuals with islet autoantibodies. A pooled analysis from five cohorts showed that time spent above 7.8 mmol/L (> 140 mg/dL) on CGM, when combined with HbA1c levels, IA‐2A positivity, and family history of T1D, predicted progression to clinical T1D with high accuracy (C‐statistic = 0.74) [73]. High‐risk individuals had a 48% probability of developing stage 3 T1D within two years [86, 87]. These findings highlight how the utility of CGM for disease staging and for identifying candidates for immunotherapies such as teplizumab and abatacept, which are most effective early in disease. Furthermore, this approach may facilitate more efficient clinical trial recruitment by selecting participants most likely to benefit from disease‐modifying interventions [97]. CGM‐derived metrics enhance clinical care by enabling earlier, more targeted treatment aligned with patient risk.

While autoantibodies and metabolic testing remain foundational for T1D risk assessment, emerging biomarkers including miRNAs, circulating DNA, novel metabolic markers, and CGM metrics offer enhanced resolution for staging and may aid treatment selection. Their integration into clinical decision‐making provides additional layers of precision for monitoring β‐cell stress and immune activation and stratifying patients for tailored therapies.

As biomarker discovery advances, an important step is determining how these tools can be used to identify individuals at risk for T1D before clinical onset. Screening strategies are increasingly being explored to detect islet autoantibodies and metabolic abnormalities in presymptomatic populations. Population‐based screening programs in children have demonstrated feasibility in identifying presymptomatic T1D. For example, the Fr1da study in Germany [98, 99] successfully identified children with early‐stage T1D through autoantibody screening conducted during routine paediatric care, while the Finnish Type 1 Diabetes Prediction and Prevention (DIPP) study [100] demonstrated the utility of combining genetic risk screening with longitudinal autoantibody monitoring to predict progression to clinical disease. Although universal screening programs have not yet been widely implemented in North America, these initiatives highlight the potential for earlier diagnosis and intervention.

Importantly, biomarker testing will need to be tailored to disease stage. In early presymptomatic stages, genetic risk scores and autoantibody profiles are more informative for identifying individuals at risk of progression. As T1D advances to stage 2, metabolic markers such as proinsulin to C‐peptide ratio, OGTT data, and CGM metrics may provide greater insight into β‐cell dysfunction and timing of intervention. In stage 3 T1D, immune profiling and circulating biomarkers of β‐cell stress or death may help stratify patients and guide adjunctive therapies aimed at preserving residual β‐cell function (Figure 3). Integrating biomarker panels within each disease stage could enable more precise patient stratification and therapeutic selection, moving screening programs from risk identification toward actionable precision medicine frameworks.

FIGURE 3.

Potential impacts of patient heterogeneity on T1D treatment approaches. Precision medicine in T1D will require three inter‐dependent avenues. First, an appreciation and knowledge of specific factors mediating disease heterogeneity (e.g., endotypes, such as T1D endotype 1 and endotype 2). Second, refinements of T1D staging will enable appropriate timing of therapies and help to maximize the potential of combination therapies, bringing together immunotherapies and beta cell support therapies. Third, leveraging of current and novel genetic, immune and metabolic biomarkers, with implementation of AI and ML approaches will enable improvements in patient stratification with the goal of maximizing beneficial effects of therapies and targeting them to patients most likely to respond.

6. What Are Some of the Biggest Challenges in Implementing Personalized T1D Therapies Into Clinical Practice?

While significant progress has been made in understanding the mechanisms and therapeutic targets of T1D, translating these insights into clinical care remains a formidable challenge. Several barriers persist, including heterogeneity in treatment response, difficulties in identifying optimal therapeutic windows, limitations in patient selection criteria, and practical concerns around cost and accessibility. Clinical consensus advisory statements and guidelines around autoantibody screening [101] and monitoring autoantibody‐positive individuals [102] are still only just being developed and represent key pieces to solving the puzzle of implementation. One of the most significant barriers is the variable response observed in clinical trials, where only a subset of individuals demonstrates durable benefit from the intervention. Successfully translating T1D therapies into clinical use hinges on addressing heterogeneity through more personalized patient selection and stratification.

Another subject is the integration of new technologies and data science into clinical practice. Artificial intelligence (AI) approaches and machine learning (ML) offer promising solutions to these challenges by enabling predictive modelling of disease trajectories, identifying novel therapeutic targets, and optimizing clinical trial design. For instance, AI‐driven “digital twin” models are being developed to simulate patient‐specific disease progression and forecast responses to various interventions [97]. These tools could significantly reduce trial duration and cost while enhancing precision in treatment allocation. Moreover, AI facilitates the repurposing of existing drugs, potentially enabling development of combination therapies. ML algorithms applied to electronic health records (EHRs) have demonstrated the ability to predict T1D onset prior to symptomatic presentation, offering new avenues for early intervention. For example, a recent study showed that ML models could anticipate T1D diagnosis up to 9.3 days before clinical recognition, potentially reducing the incidence of complications such as DKA [103]. Such predictive models are particularly relevant for stratifying patients into stage‐specific treatment pathways and could help accelerate precision medicine. Despite the promise, the clinical application of AI and ML‐based approaches in T1D remains limited. Many predictive models require external validation across diverse populations, and concerns remain regarding algorithmic bias, data quality, and reproducibility. In addition, regulatory oversight, data privacy considerations, and integration into clinical workflows will be essential before these tools can be implemented in routine diabetes care [104, 105]. Therefore, AI‐driven stratification and digital twin approaches should be viewed as emerging tools that may complement, rather than replace, established clinical and biomarker decision making frameworks.

Despite these advances, financial and logistical barriers must be addressed. Most emerging therapies, such as teplizumab, are prohibitively expensive (currently an estimated $200 000 USD for the full 14‐day course) [106], and require long term monitoring. Health systems must evaluate how to ensure equitable access to these treatments, especially in under‐resourced settings. Additionally, regulatory and ethical frameworks may need to progress to include data‐intensive, algorithm‐driven approaches to care.

In summary, realizing the promise of precision medicine in T1D will require more than scientific progress. It will demand coordinated efforts in public health, digital infrastructure, and clinical trial innovation. Future steps toward this goal should prioritize the validation of biomarker panels, the refinement of patient selection algorithms, and the development of combination therapies that reflect the disease complexity of T1D. Addressing these challenges and incorporating AI and ML into the T1D therapeutic landscape may enable a shift from population‐based protocols to personalized treatment strategies.

7. Conclusions

T1D is no longer considered to be a uniform autoimmune disease, but is rather being appreciated as a complex, multifaceted disease defined by a spectrum of immunologic, genetic, and metabolic features involving the pancreas and immune system. The recognition of disease heterogeneity through endotypes, staging models, and biomarker profiles has reshaped how we understand disease progression and affords new opportunities for personalized therapeutic intervention. Current clinical evidence suggests that immune modulating agents like teplizumab and rituximab are likely to be most effective when applied at early disease stages, and may provide greater benefit in specific endotypes, such as the “T1DE1” aggressive immune and early childhood presentation of disease. Conversely, patients with more indolent disease courses or substantial residual β‐cell function may benefit more from metabolic and β‐cell supportive strategies, such as verapamil or incretin‐based therapies. Stratifying patients by features such as endotype, disease stage, metabolic and immunological markers enables these therapies to be used more precisely and effectively and provides an opportunity to test novel combination therapies. The expanding toolkit of biomarkers including circulating miRNAs, methylated DNA, metabolic and protein stress markers, and CGM‐derived metrics offers new opportunities for risk prediction and therapeutic tailoring. Integrating these with AI‐powered predictive models could reform trial design, improve patient selection, and streamline clinical decision making. Despite these promising advances, substantial challenges remain. Many therapies show only transient benefit, clinical trials often exclude paediatric populations making it difficult to directly study endotypes, and cost of the therapies themselves forms a barrier to widespread clinical implementation. Addressing these issues will require sustained investment in prospective cohort studies and inclusive data collection across diverse populations.

Author Contributions

J.P. wrote and edited the first draft of the manuscript with guidance and input from P.J.T. P.J.T. revised and edited the manuscript. Both authors approved the final manuscript.

Funding

This work was supported by Canadian Institutes of Health Research (CIHR) (PJT‐485691, PJT‐485915), Institute of Nutrition, Metabolism and Diabetes (TDP‐485691, TDP‐485915), Breakthrough T1D (4‐SRA‐2023‐1182‐S‐N, 4‐SRA‐2023‐1184‐S‐N), Research Manitoba PhD Health research studentship, and the Manitoba Medical Service Foundation, 2023 Allen Rouse Basic Science Career Development.

Conflicts of Interest

The authors declare no conflicts of interest.

Data Availability Statement

The authors have nothing to report.