The Pathogenic “Symphony” in Type 1 Diabetes – A Disorder of the Immune System, Beta Cells, and Exocrine Pancreas

Abstract

Type 1 diabetes (T1D) is widely considered to result from the autoimmune destruction of insulin-producing β-cells. This concept has been a central tenet for decades of attempts seeking to decipher the disorder’s pathogenesis and prevent/reverse the disease. Recently, this and many other disease related notions have come under increasing question, particularly given knowledge gains from analyses of human T1D pancreas. Perhaps most crucial are findings suggesting that a collective of cellular constituents — immune, endocrine, and exocrine in origin — mechanistically coalesce to facilitate T1D. This Review considers these emerging concepts, from basic science to clinical research, and identifies several key remaining knowledge voids.

The Pathological Basis of Type 1 Diabetes

From a historical perspective, the autoimmune basis for type 1 diabetes (T1D) has largely been centered, in terms of its pathogenic nature, on three seminal observations including an inflammatory infiltrate of pancreatic islets (i.e., insulitis), genetic susceptibility associated with the Major Histocompatibility Complex (MHC), and autoantibodies reactive against β cell antigens¹. These concepts formed key components for a natural history model of T1D, developed in the mid-1980s², that has served as a multi-generational roadmap for efforts seeking to understand the disorder’s pathogenesis, determine disease risk, and uncover methods for preventing β cell loss and preserving endocrine function. While subject to modification over the years, the benefits of this model have been numerous, including the ability to stage pre-diabetes (i.e., stage 1 and 2 disease) for therapeutic interventions based on risk for stage 3 T1D diagnosis (Figure 1), developing predictive biomarkers, guiding studies examining environmental constituents influencing the disorder’s development, and more^3–8. However, despite such knowledge advances, a major therapeutic shortcoming remains: pragmatic challenges allowing for preventing T1D in those at increased disease risk or providing for stable, long-term preservation of β cell function in those with recently diagnosed T1D. Indeed, to date only one agent, an anti-CD3 antibody (i.e., teplizumab), has received regulatory approval in the United States for attenuating loss of β cell function and delaying progression to stage 3 T1D by approximately two years in those at the highest risk for the disorder^9,10. Further advances towards this end are needed and must be based, optimally, on an improved understanding of the disorder’s pathogenesis.

Figure 1. A re-depiction of the current “staging” model for type 1 diabetes.

Following decades of natural history studies of type 1 diabetes (i.e., monitoring the period prior to and after traditional disease diagnosis), a consensus group within the community of type 1 diabetes developed a model whose intention was to better define risk for disease development. This model was largely developed based on analysis of both immunological (autoantibody type, number) and metabolic markers (response to glucose stimulation). Other considerations not strictly considered in this published model are those of having a relative with the disease, relative relationship, age, and more. Here we depict emerging concepts related to recent studies of pancreatic pathology to improve the staging of type 1 diabetes, as described herein. Adapted from Insel, et al.³ with insets from Figures 2–3.

Effectively addressing this therapeutic challenge for T1D has not been for lack of effort in performing clinical trials attempting to meet these goals. Indeed, since the inception of the cyclosporin trials in the early 1980s, hundreds of trials, both small and large, have attempted to avert autoimmunity and preserve existing β cell function^11–13. Multiple factors contributed to limited therapeutic successes, including limited knowledge regarding mechanisms of β cell death, an inability to safely biopsy the human pancreas, a failure to understand the role for environment in T1D, the absence of a means to image pancreatic islets in vivo, issues of therapeutic equipoise, defining the relationship between the disorder’s polygenic nature with specific mechanistic outcomes, and the relative rarity of the disease in the general population, among others^14–17. An additional and likely major contributor includes the “Streetlight Effect” for T1D¹⁸, a concept that posits advances in T1D research have, to a certain degree, been restrained by the failure to address truly novel if not controversial hypotheses and perhaps, more importantly, accept pedagogical dogma with limited experimental questioning.

Thankfully, this situation appears to be changing as many long-standing dogmas in T1D have recently come under question and are becoming subject to intellectual change (Table 1), creating a situation where multiple aspects of research involving this disease, including clinical research and trials, will benefit. Examples include, but are not limited to, improvements in diagnostic testing, seeing an appreciation for disease heterogeneity (albeit still somewhat an undefined notion in T1D)^19–21, and movement toward precision-based therapies (e.g., based on age, HLA-DR diplotype, etc.)^22–25, which ultimately will require definitive data linking disease phenotypes with treatment response. Such knowledge gains have emanated from a series of sources including large-scale efforts involving longitudinal analysis of blood samples from living subjects with T1D and those at varying levels of risk for the disease, as well as genome wide association studies (GWAS), implementation of emerging technologies (e.g., continuous glucose monitoring, single cell analyses), and identification of improved disease biomarkers⁷. However, among the most transformative findings challenging conventional thought regarding the pathogenesis of T1D have emerged from studies of the human pancreas from persons with this disease, non-diabetic islet autoantibody positive subjects at increased risk for the disorder (including those with two or more autoantibodies indicative of early-stage disease), and those without diabetes^26–28.

Table 1.

Dogmas or long-standing beliefs that have come under change, or seen correction in recent years, largely if not significantly due to studies of the human pancreas.

Long-Standing Dogma	Reality	References
T1D occurs when 85–95% of β-cells are destroyed	The degree of β-cell loss resulting in symptomatic disease is quite variable (i.e., 40–95% loss).	33,104,215
Individuals with longstanding T1D are devoid of insulin positive β-cells	Proinsulin and insulin expression can be detected within scattered single cells as well as small clusters of islets in pancreas from donors with longstanding T1D.	33,104,121,138,215–217
T1D represents a singular disease	T1D likely represents a collective of distinct disorders (i.e., endotypes) with commonalities of insulinopenia and loss of β-cells. Heterogeneity exists in the degree of β-cell cell loss at onset, rate of insulitis, age and onset, and more.	22,77,218,219
β-cells are subject to replication	In those without T1D, β-cell replication rates are highest in the neonatal period and decline exponentially throughout childhood with little to no replication observed in human adults. Limited evidence exists that such replication occurs in settings of T1D.	144,220–222
T1D is an autoimmune disease, with limited pathogenic contributions of β-cells	An extensive number of β-cell alterations preceed T1D onset and may contribute toward autoimmunity (e.g., ER stress, oxidative stress, protein misfolding, neoepitope formation, impaired autophagy, aberrant prohormone processing, senescence, and increased expression of MHC-I and PD-L1).	15,35,53,74,87,99,104,111,113,115,116,120,216,223–237
The exocrine pancreas is unaffected by T1D	Reduced pancreas organ mass and volume, as well as reduced exocrine pancreatic enzyme levels in serum, are characteristics of recent-onset and pre- T1D. T1D donors also exhibit reduced acinar tissue area and increase acinar cell density.	40,104,175,178,180,183,187,188,238,239
Insulitis is a chronic and predominant feature in the natural history of T1D	Insulitis is restricted to insulin containing islets and has been primarily reported in individuals with recent-onset T1D, with rare cases of insulitis observed in pancreata from persons with single autoantibody positivity, with slightly more in those with multiple islet autoantibodies.	33,34,40,43,44,89,104,106,215,240,241
T1D and T2D are distinct in terms of pancreatic features	Reductions in islet number, pancreas size, β-cell mass, and β-cell function have been observed in both T1D and T2D. Genetic analyses have provided new insight into the contribution of specific polymorphisms, including how the presence of variants associated with T2D can influence the phenotype of donors with T1D.	27,192–194
Insulitis, as reflected in NOD mice, is representative of human disease	Compared to NOD mice, human insulitis is far less severe (fewer immune cells per inflammed islet), less pervasive (fewer islets affected) and occurs in a patchy “mosaic” pattern	37,40,89,111,242–244

This Review article considers how studies of the human pancreas, largely conducted over the past 15 years, have resulted in dramatic changes with respect to our understanding of mechanisms underlying T1D pathogenesis, the disorder’s heterogeneity, and identification of cellular contributors to disease formation. Importantly, revelations garnered from these studies have also led to a major rethinking of cause versus effect in terms of the damage observed throughout the T1D pancreas and apparent pathogenic contributions of cells beyond those purely autoimmune in nature, providing for a more expansive view in terms of the roles that β cells, islets, and the exocrine pancreas provide toward defining the development of this disorder. It seems evident that an improved understanding of such notions will have major implications for efforts to improve disease prediction and prevention, alongside of those seeking to preserve metabolic function in individuals with T1D^29–32.

T1D is an Autoimmune Disorder

Insulitis

From a perspective of pathology, T1D is characterized by a selective loss of β cells within the pancreatic islets^33–36. While exceptions exist, the majority of literature suggests that insulitis is more predominant in individuals with youth onset T1D (≤14 years of age) versus those with an older age at onset³⁷. This finding is consistent with the notion that a more vigorous autoimmune response occurs when disease develops in young children compared to adults^38,39. However, in contrast to the NOD mouse model of the disease, the presence of insulitis in human T1D is more limited quantitatively³³, with an expert panel establishing a standardized histological definition for this lesion as three islets, each containing more than 15 CD45⁺ cells, within a pancreas³⁴. This standard has, however, recently come under question alongside a call for its replacement using a “30–30 rule,” whereby the presence of 30 white blood cells within a 30mm region defines insulitis⁴⁰. Hence, while acknowledging that islet inflammation in T1D remains a central tenant, its definition continues to be a subject of debate.

Beyond simply the definition of insulitis is the finding that its occurrence (relative to the NOD mouse model of T1D) prior to disease diagnosis in humans is very limited. Prior to symptomatic onset, insulitis is rare if not totally absent in single glutamic acid decarboxylase autoantibody (GADA)-positive subjects and, surprisingly, has only been observed in a subset of multiple autoantibody-positive subjects — persons at a high 10-year risk of T1D^33,41–44. This observation, often overlooked or underappreciated, should be given consideration for clinical interventions that are based on targeting insulitis in the early (i.e., pre-symptomatic) stages of the disease versus therapies whose presumed mechanism involves immunological activities acting in the periphery.

The immunophenotype of insulitic lesions shows a predominance of CD8⁺ T lymphocytes and macrophages (CD68⁺), but cells of other phenotypes are also observed (e.g., CD4⁺, CD20⁺)^45,46. Recent studies indicate that younger age of onset is associated with greater frequencies of CD20⁺ B cells, CD45⁺ cells, and CD8⁺ T cells in insulitis lesions, alongside fewer insulin-positive islets^46,47. With respect to B-lymphocytes, these efforts reported a CD20^Hi profile in the insulitis lesion of donors diagnosed before 7 years of age versus a CD20^Lo profile in the those diagnosed after 13 years of age. As a result, it has been suggested that the cellular composition of insulitis or, more specifically, the quantitative presence of insulitis, serves as one form of an “endotype” of T1D, a concept to explain potential diversity in T1D disease pathogenesis²². This said, not all studies have supported the notion of a predominance of B-lymphocytes in individuals with an early onset of disease³³, and limited progress has been made in forming a consensus for defining T1D endotypes in the research and care community²⁰. This information void regarding age-associated insulitis could be tested for association with circulating biomarkers (e.g., TCR and BCR repertoire, comprehensive immunophenotyping, serological markers) measurable in clinical cohorts and organ donors, allowing for specific hypothesis testing based on the observed differences in islets.

An abundance of studies suggest that islet-invading T cells are largely directed against known β cell antigens (e.g., insulin, proinsulin, GAD), post-translationally modified peptides, viral antigens, and so-called hybrid peptides^48–58. CD8⁺ T cells expressing T cell receptors (TCRs) that bind MHC class I tetramers loaded with the β cell autoantigens resident to insulitis lesions are also detectable in the peripheral blood of individuals with T1D^59,60. Toward this latter notion, a series of investigations have undertaken the task of evaluating TCR utilization within the insulitis lesion and of addressing clone publicity (i.e., shared amongst those with T1D versus unique to an individual), abundancy, and disease-specificity⁶¹. Notably, most such efforts to date have suggested a bias in terms of receptor utilization⁵⁵. Hope exists that these TCR datasets will eventually form novel biomarkers of T1D development as well as provide an approach for optimizing therapeutic intervention^61,62. For example, such discoveries have facilitated exciting functional studies using primary human T cells engineered to express an islet antigen-specific TCR or chimeric antigen receptor (CAR) to model T1D^63–68. Finally, the frequency of CD8 T cells directed against β cell antigens in the pancreas uniquely discriminates subjects with T1D versus those without the disease, unlike what is often observed in analysis of peripheral blood⁶⁹. Taken as a collective, this body of information, regarding both peripheral and islet invading T cells, is highly supportive for the notion that analysis of the human pancreas in T1D is of importance.

Potential pathogenic “drivers” of T cell infiltration into pancreatic islets are quite numerous (e.g., chemokines, β cell abnormalities, extracellular matrix [ECM] expression)^70,71. However, among the most often discussed, insulin-containing islets of T1D subjects as well as single and double autoantibody-positive non-diabetic subjects overexpress class I HLA, purportedly unmasking β cells to immune-mediated killing^35,72. To be clear, this MHC class I hyperexpression is not specific for islet β cells and, as discussed later, the exocrine pancreas of those with T1D also shows higher levels of exocrine immune infiltration⁷³. Finally, while its pathogenic importance remains unclear, class II HLA upregulation was also recently reported on β cells within insulin containing islets from T1D donors⁷⁴.

The Natural History of Insulitis in T1D

The non–insulin-producing cells of the islets (i.e., α-, δ-, ε-, and pancreatic polypeptide cells) persist in patients with long-standing T1D, with these remaining islets, lacking insulitis and β cells, termed “pseudoatrophic”³⁴. This information sets the stage for another remarkable feature of the T1D pancreas, namely, heterogeneity of islet lesions (Figure 2). Histopathologic examination of the pancreas in settings of multi-autoantibody positivity without T1D, recent-onset disease, or established (long duration) T1D note that within the same tissue section, a “normal” islet with no immune infiltrate can coexist with an islet containing β cells with intense infiltration as well as a pseudoatrophic islet that has no infiltrate^33,75,76. Such heterogeneity of lesions may underlie the variable natural history of T1D, whereby studies suggest the time period between initial β cell autoimmunity and overt symptoms can be one of months, years, or even decades^41,77. Interestingly and as mentioned previously, contrary to studies of the NOD mouse model of T1D, analysis of pancreata from non-diabetic subjects positive for one or multiple islet autoantibodies often show limited to no insulitis^43,78,79. Specifically, the presence of insulitis using the aforementioned criteria³³ is extremely rare in single GADA-positive patients whereas this lesion is observed in approximately one-half of those with multiple autoantibodies. Such information is consistent with data from longitudinally followed cohorts and the staging of T1D⁸⁰. Detection of insulitis in multiple autoantibody positive pancreata may be further impacted by the cross-sectional nature of these studies. This finding has been considered by some to support the notion that T1D represents a relapsing/remitting disease, with waves of immunologic destruction^81–83. However, this concept remains speculative, in part, due to the inability to perform pancreatic biopsy, lack of in vivo imaging, and absence of a circulating biomarker of insulitis.

Figure 2. Histopathologic progression of type 1 diabetes within the human pancreatic islet.

Within an initially healthy islet, a portion of β-cells develop phenotypic and functional impairments. This precipitates immune cell infiltration around (peri-insulitic) and within (insulitic) the islet, coupled with degradation of the peri-islet basement membrane. Following the immune-mediated destruction of functional β-cells, the immune cells exit leaving behind an irregularly shaped, insulin-negative, pseudoatrophic islet characterized by presence of the remaining endocrine cell types (alpha, delta, PP, epsilon), reduced vessel diameter, increased microvascular density, and restored peri-islet basement membrane integrity. PP, pancreatic polypeptide.

Mechanisms of Immune Mediated β Cell Death

Whereas the exact mechanisms of β cell death remain subject to debate, considerable enthusiasm exists for the notion that certain cytokines (e.g., interleukin 1 beta [IL-1β], tumor necrosis factor [TNF], type 1 interferons (T1-IFN)) and CD8⁺ cytotoxic T lymphocytes are likely to be major contributors to β cell destruction⁸⁴. Indeed, that T1-IFN potentiates CD8⁺ T cell cytotoxicity as well as HLA class I hyperexpression on β cells representing a “catastrophic feature” of the T1D pancreas is a popular notion^67,85. However, to be clear, HLA class I hyperexpression is not limited to β cells and has been observed in other islet endocrine cells as well as additional pancreatic constituents. With this, the question becomes one of why only β cells would be subject to destruction. One can speculate that specificity of autoimmunity for β cell-specific antigens, increased sensitivity of β cells to inflammatory based insults or metabolic stress, and/or an ill-defined contribution of environmental agent(s) may underlie such differences. The demonstration that human gene-edited pseudoislets lacking HLA class I/II can escape rejection and autoimmunity in a humanized mouse emphasizes the potential catastrophic role of HLA expression on β cells⁸⁶. Additionally, analysis of pancreata from multi-autoantibody-positive and recent-onset T1D donors note that β cells express TET2, which has been suggested to augment inflammation and autoimmune killing⁸⁷. T follicular regulatory cells (Tfr) are reduced in frequency within the pancreas draining lymph nodes and spleen, but not blood, of T1D donors⁸⁸, and tertiary lymphoid organs (TLOs) were recently detected in the T1D pancreas⁸⁹ — TLO presence was most commonly associated with residual insulin containing islets and younger onset age⁸⁹. CD8⁺ T cells specific for the islet antigen ZnT8 are also reportedly enriched in the pancreas but not blood of T1D donors, altogether supporting local defects in immunoregulation as potentiating cytotoxic β cell killing⁶⁹.

In contrast, several pathways active within the human T1D pancreas are suggested to represent attempts to constrain the autoimmune destruction of islets during T1D. For example and perhaps counterintuitively to the aforementioned notion, exposure to the proinflammatory cytokine γ-interferon (IFN-γ) during activation reduces cytotoxic β cell killing by CD8⁺ T cells engineered to express an islet antigen-specific TCR⁶⁶. In situ, IFN-γ and IFN-α drive programmed death ligand-1 (PD-L1) expression on β cells in order to limit T cell function during T1D progression^90,91. Indeed, the PD-1/PD-L1 pathway has been demonstrated as a key checkpoint for T1D development in murine models of the disease and in cancer patients with T1D-associated HLA genotypes. Specifically, treatment involving a PD-1 checkpoint inhibitor has been shown to precipitate the development of islet autoantibodies and T1D onset⁹². Studies of residual β cells in T1D donors demonstrate the expression of PD-L1 on the surface of these β cells, suggesting that these residual cells may have escaped T cell-mediated killing⁹¹. The signal regulatory protein gamma (SIRPg)/CD47 co-inhibitory pathway has raised similar interest where signaling is speculated to potentiate protection of β cells⁹³. Additionally, murine studies have demonstrated MERTK expressing mononuclear phagocytes to regulate T cell activation in both T1D islets and melanoma tumors⁹⁴. With this, increased numbers of MERTK⁺ cells observed within insulin containing islets from pancreata with longstanding T1D may represent an intrinsic mechanism to preserve residual β cells⁹⁴.

T1D is a Disease of Islets and β Cells

Despite clear evidence for an autoimmune etiology in T1D, a growing minority in the T1D research and care community have increasingly questioned the sole or primary role for this concept⁹⁵, resulting in an expanding debate regarding key contributors to the disorder’s pathogenesis⁹⁶. This controversy has largely occurred based on a mounting appreciation for an active participation of β cells and islets in human T1D beyond processes of autoimmunity.

The concept that β cells themselves influence disease pathogenesis and participate in their own demise is not new, having been proposed in the mid-1980s by a seminal publication questioning whether β cell loss in T1D occurs via a “homicide” or “suicide” like mechanism⁹⁷. The rationale underlying this reemerging notion is largely tied to recent morphologic and functional studies of rodent and human islets, both within and outside settings of inflammation. These efforts suggest a host of metabolic and cellular aberrations occur in the natural history of T1D involving β cells⁹⁸ (Figure 3A), as well as other islet cell types (Figure 3B). Among the many cellular aberrations, we would highlight stress and dysfunction of various organelles within β-cells (as detailed below), β-cell senescence and its release of senescence-related secretory cytokines that attract monocytes⁹⁹, aberrant accumulation of immunogenic GAD65 in β-cells¹⁰⁰, the loss of regulated glucagon secretion in α cells¹⁰¹, as well as alterations in cells that maintain the integrity of islet ECM and its isolation from invading immune cells^71,102. It is important to note that many of these abnormalities are observed in single GADA-positive patients who do not have detectable insulitis. As we will discuss below, research efforts need to be directed at further understanding the dynamic, interactive, and complex dance between β cells, islets, exocrine tissue and the immune system in T1D development. Also, while the potential impact of such abnormalities on disease formation remains unclear, it is conceivable that these islet-immune cell interactions contribute toward the heterogeneous progression of human T1D^23,103. Indeed, as will be discussed throughout this article, we believe that a contemporaneous view of this question would suggest both mechanisms (i.e., β cell homicide and suicide) are crucial to T1D development.

Figure 3. Alterations associated with type 1 diabetes at the β-cell and whole islet level.

(A) Functional and phenotypic changes within the pancreatic β-cell during type 1 diabetes development include protein misfolding reflective of ER stress, neoepitope formation, impairments in autophagy and prohormone processing, secretion of SASP molecules, accumulation of GAD65 in the Golgi apparatus, mitochondrial oxidative stress, and increased expression of MHC class I and PD-L1 molecules, which respectively serve opposing roles related to T cell activation through antigen presentation and co-inhibitory receptor stimulation (Figure 2). Straight arrows depict increases (up) or decreases (down) in type 1 diabetes as compared to healthy β-cells. Curved arrows depict the functional consequences of those changes. (B) Pancreatic islet alterations during type 1 diabetes development include loss of insulin secretion and IDO1 expression within dysfunctional β-cells, dysfunctional glucagon secretion from alpha cells, and degradation of extracellular matrix comprising the peri-islet basement membrane. Gray dashed lines depict cellular dysfunction. Pink dashed lines depict basement membrane degradation. ER, endoplasmic reticulum; PC1/3, proprotein convertase 1/3; SASP, senescence-associated secretory phenotype; GAD65, glutamic acid decarboxylase 65; MHC, major histocompatibility complex; PD-L1, programmed death ligand 1; IDO1, indoleamine 2, 3-dioxygenase 1; HS, heparan sulphate; HA, hyaluronic acid.

β Cell Stress

The β cell is a professional secretory unit that is predominantly tasked with the production of insulin to maintain glucose homeostasis. As part of this function, the biosynthetic machinery, from gene transcription and mRNA translation to protein folding and secretion, is under tenuous homeostasis. In the pre-diabetic state, stress signals (e.g., viral infection, exposure to proinflammatory cytokines, increased metabolic demand) may pose a threat to these homeostatic processes^104–110. The resultant effects on RNA processing, protein production and folding, as well as protein clearance and release would provide for disruption of cellular function and survival independent of direct immune-mediated damage^111–116.

Studies utilizing NOD mice, human islets, and human pancreas from T1D donors have identified several molecular responses that likely contribute to β cell impairment in T1D and the pre-diabetic state (Figure 3). These include altered mRNA splicing^115,117, endoplasmic reticulum (ER) and oxidative stress^118,119, reductions in proinsulin to insulin processing (resulting from reductions in PC1/3 processing enzyme)^120,121, impaired autophagy¹¹⁴, and cellular senescence⁹⁹. These processes may not only impair the function of β cells but, in addition, promote their enhanced visibility to the immune system through formation of neoepitopes, augmented expression of MHC class I molecules, and other processes capable of immune stimulation^59,122–124.

Beyond offering new insights into T1D pathophysiology, the identification of early β cell stress in the natural history of disease also offers an opportunity for developing a novel means of therapeutic intervention, especially if reliable serological biomarkers of these processes can be identified¹²⁵. Among biomarkers having such potential, increasing attention has been directed at the proinsulin-to-C-peptide ratio (PI:C); an indicator reflecting defects in proinsulin processing resulting from ER stress and loss of PC1/3 production. Elevated PI:C ratios have been observed in multiple autoantibody-positive youth who are more likely to progress to T1D¹²⁶. A second biomarker example identifies epigenetically modified DNA fragments uniquely emanating from dying, stressed β cells^127,128. These biomarkers appear to identify an increase in β cell death in recently diagnosed T1D individuals, those receiving islet transplants, and in those with high risk for T1D (i.e., first-degree relatives)^{127–130,131}. Importantly, unlike T1D associated autoantibodies, biomarkers of β cell stress remain in their infancy and uncertainty exists regarding their sensitivity and utility to reliably identify early-stage T1D¹³².

β Cell Function Versus β-Cell Mass

Loss of insulin secretion has historically been viewed as a hallmark of T1D. Yet, the pathophysiology of this loss may not be as simple as the loss of β cells themselves, as it occurs many months to years before the onset of clinically apparent diabetes, and may therefore emanate from some combination of a functional deficit and a deficit in β cell mass⁷. In early studies of autoantibody-positive individuals at increased-risk for T1D, intravenous glucose infusion studies indicated a loss of early phase insulin secretion that preceded alterations in glucose homeostasis¹³³. Similarly, in later studies of high-risk co-twins¹³⁴, those progressing to T1D exhibited similar defects in early phase insulin secretion compared to those who did not progress; moreover, these individuals had impaired insulin responses to different levels of glucose infusion, suggesting a defect in β cell glucose sensitivity. Whereas this early, prediabetic defect in insulin secretion might justifiably be attributed to a loss of β-cell mass, analysis of pancreas pathology suggests a greater complexity. Studies of human pancreas have demonstrated the presence of normal β cell mass¹³⁵, but impaired proinsulin processing in single and multi-islet autoantibody positive donors without diabetes¹¹¹, implying that glycemic dysregulation in the early stages of T1D may reflect impairments in β cell function prior to a catastrophic decline in β cell mass near the time of disease onset. Interestingly, a small number of studies of islets isolated from individuals with T1D have suggested that insulin secretion can be restored ex vivo with prolonged culture under non-diabetogenic conditions; however, these findings remain controversial given the observations from the Diabetes Virus Detection (DiViD) study showing only partial recovery of islet function following 3–6 days of culture^136,137.

The original Eisenbarth model depicting the loss in β cell mass or number being associated with T1D has also come under recent question², with the notion that β cell function, with or without concurrent loss in mass, may be just as relevant. Whether or not β cell dysfunction in T1D arises from a loss in mass or a loss in β cell function (or both) has potential therapeutic implications, since approaches to avert or reverse either concept would differ substantially¹³⁸. Seemingly, the aforementioned studies documenting the loss in early-phase insulin secretion, reduced glucose sensitivity, and preservation of β cell mass in the pre-T1D period support the loss in function model, wherein underlying cellular dysregulation precludes glucose sensing and/or prohormone processing resulting in dysfunctional early docking and release of insulin-containing granules^96,111. However, analysis of subjects undergoing partial (~50%) pancreatectomy revealed that the decline in glucose sensitivity and early-phase insulin release can accompany this procedure and thereby predict the likelihood of developing diabetes¹³⁹. Thus, loss of β cell mass can also manifest as defects in glucose sensing and insulin release kinetics.

Estimating residual β cell mass during disease progression represents a challenge, since histological examination of pancreatic tissue from living persons with T1D is not considered feasible or ethical¹⁴⁰. Nevertheless, several key observations from living subjects note that C-peptide and proinsulin (measures of β cell metabolic function) remain detectable in many subjects, even after decades of T1D, although these levels decline with disease duration^141–143. Furthermore, histological studies of β cell numbers in whole-organ cross-sections (i.e., a measure of mass) are highly variable (60–95%loss) at the time of T1D diagnosis. β cell number is also variable in those with longstanding disease (i.e., decades), albeit with significant reductions (80–98%) relative to non-diabetic controls^144–146. Finally, when measures of β cell function (intravenous glucose tolerance test) and mass (glucose-potentiated arginine stimulation) are simultaneously measured in living subjects at low or high risk of T1D, there is significant variability in the two measures suggesting heterogeneity in the relative loss of mass versus function¹⁴⁷, and more importantly, that the two variables may coexist in any given individual. Collectively, these studies suggest that although the loss of β cell mass predominates in T1D, much less is known regarding the pre-diabetic period when disease pathogenesis is likely most active and when therapies may be most impactful.

Contributions of Islet Microvasculature, Extracellular Matrix, and Sympathetic Innervation

The islets receive a disproportionately large fraction of the total pancreatic blood flow supplied by a dense network of fenestrated capillaries¹⁴⁸, which are subject to neurovascular control through activation of pericytes that densely cover the capillary endothelial cells¹⁴⁹. Indeed, nearly every β cell is supplied by an intra-islet or peri-islet capillary facilitating systemic metabolic control through rapid nutrient sensing and responsive secretion of endocrine hormones¹⁵⁰. In shorter duration T1D (<10 years), islet microvascular density is increased, and vessel diameter is reduced^151–153. Interestingly, these alterations are most pronounced in pseudoatrophic islets (i.e., insulin-negative) while vessel densities and diameters within residual insulin containing islets from T1D donors are comparable to those of non-diabetic controls¹⁵¹. Moreover, islet capillary density appears to return to normal levels in longstanding T1D^152,153.

Recent efforts have demonstrated the structural and functional polarization of human β cells in relation to islet capillaries, but the implications of this observed enrichment of scaffold proteins along the β cell-vascular interface remain subject to investigation, both as it relates to diabetes pathogenesis and the development of β cell-targeted therapeutics¹⁵⁴. Efferent blood is generally thought to travel through venules from the islet to the surrounding acini, forming what is known as the insulo-acinar portal system¹⁴⁸. However, recent work supports the existence of open circulation characterized by bidirectional blood flow between the endocrine and exocrine pancreas compartments¹⁵⁵, providing a potential mechanistic link to exocrine pancreas alterations associated with T1D, as discussed below.

The islet microvasculature is largely responsible for the production of ECM linking capillaries with β cells¹⁵⁰, and the passage of immune cells through the islet ECM represents a key step in forming the insulitis lesion (Figure 2)⁷¹. ECM is comprised of a complex network of molecules (e.g., heparan sulfate/heparan sulfate proteoglycans, hyaluronan, laminins, collagens) both within and surrounding the islets, constituting the interstitial matrix and basement membrane (BM). The importance of islet ECM in T1D found early evidence from in vitro analyses suggesting hyaluronan and the hyaluronan binding protein (HBP) are capable of modulating T-cell adhesion and migration¹⁵⁶. These concepts gained further support by histological examination of human T1D pancreata where hyaluronan and HBP, including inter-α-inhibitor (IαI) and versican, not only accumulated in T1D islets but were also associated with the degree of inflammatory cell accumulation (i.e., insulitis)^71,157. Specifically, hyaluronan was dramatically increased both within the islet and outside the islet endocrine cells, juxtaposed against islet microvessels in T1D pancreata. Moreover, when extending such models to additional vascular constituents, the loss of the peri-islet basement membrane was noted to only occur at sites of leukocyte infiltration into an islet, suggesting that leukocyte penetration of the peri-islet BM is a critical step in the process of insulitis development¹⁰². Importantly, hyaluronan accumulation in islets has been observed in the pancreas of islet autoantibody-positive individuals prior to immune cell infiltration and T1D onset with the degree of hyaluronan staining positively associating with the number of autoantibodies¹⁵⁸, and treatment with a hyaluronan synthesis inhibitor (4-methylumbelliferone, 4-MU) prevented diabetes onset in NOD mice and rendered insulitis non-destructive in the TCR transgenic DORmO mouse model of autoimmune diabetes¹⁵⁹, altogether supporting a potential role for such ECM alterations early in the disease process.

Another ECM constituent, the glycoaminoglycan heparan sulfate (HS), was demonstrated in studies of NOD mice to protect β cells from oxidative damage¹⁶⁰. Examination of insulin-positive islets in T1D pancreases noted significant loss of HS, whereas heparanase expression was pronounced in islet-infiltrating leukocytes¹⁶¹, further supporting the concept that loss of HS could contribute to β cell death in T1D. Taken together, it would appear that in settings of T1D, islet ECM composition represents a significant factor in maintaining immunological tolerance, activation of immune cell populations, islet invasion, and eventual destruction of β cells.

The sympathetic nervous system, a key constituent of the autonomic nervous system, plays a major role in organ homeostasis. With respect to the pancreas, it is intensively innervated with sympathetic efferent nerves originating from the superior mesenteric and celiac ganglia¹⁶². This innervation is considered a key contributor to establishing islet architecture¹⁶³ and with respect to function, sympathetic neural activity inhibits insulin secretion while promoting glucagon release, together increasing systemic glucose levels. Emerging technologies combined with the availability of pancreatic and islet tissues from persons at various stages of T1D have allowed for recent studies that have amplified a potential contributory role for disordered innervation in metabolic disorders¹⁶⁴. While sometimes conflicting, analysis of humans with T1D do suggest that islets undergo a series of key modifications to their sympathetic innervation as part of the pathogenic processes underlying this disease and in particular, the formation of insulitis¹⁶⁴.

At one extreme, a sustained and pronounced loss of sympathetic innervation in islets from pancreas autopsies has been noted for those with T1D, both at disease diagnosis and in those with established disease¹⁶⁵. A second and more recent effort suggested both recent-onset and longstanding T1D donors had greater islet nerve fiber density compared to non-diabetic donors, whereas no changes in sympathetic axons were observed with diabetes of any duration¹⁵³. Sympathetic tyrosine hydroxylase (TH) axon numbers and density were decreased in islet autoantibody-positive non-T1D subjects in comparison to those with T1D¹⁶⁶. Through a series of molecular investigations, this latter study also identified variances in noradrenalin degradation, α-adrenergic signaling, cardiac β-adrenergic signaling, catecholamine biosynthesis, and additional neuropathology pathways.

T1D is a Disorder of the Exocrine Pancreas

Pancreas Size/Weight/Volume

In the absence of T1D, the human pancreas increases in size with age, plateauing in the mid- to late-20s, followed by a decrease after 60 years of age (in most), with its size also reflective of overall BMI¹⁶⁷. Over a century ago, pancreas autopsy studies of individuals with diabetes (albeit undefined by today’s standards) showed reduced pancreas size¹⁶⁸. Later, a series of efforts published between 1975 and 1985 demonstrated low serum levels of exocrine pancreatic enzymes in patients with established T1D, an observation that was initially presumed a consequence of long-term insulinopenia^169–172. Unfortunately, the potential pathogenic significance of these observations remained underappreciated¹⁷³ for decades until recent studies, discussed below, confirmed these observations in individuals definitively diagnosed with stage 1–2 and recent-onset stage 3 T1D, revealing a variety of exocrine changes across the natural history of T1D (Figure 4).

Figure 4. Alterations to pancreas size and organization during type 1 diabetes progression.

Disease progression, from AAb seroconversion (stage 1–2) to recent-onset and established type 1 diabetes (stage 3), is associated with increasingly smaller pancreas organ size. Insets show highly vascularized islets surrounded by exocrine cell bundles arranged in acini. Asynchronous islet infiltration and β-cell destruction result in a heterogeneous mixture of normal, insulitic, and pseudoatrophic islets in AAb+ and recent onset type 1 diabetes pancreas while the majority of islets are pseudoatrophic in longstanding type 1 diabetes. Insulin negative pseudoatrophic islets have distinct microvascular alterations including reduced vessel diameter and increase vessel density. Within the exocrine compartment, recent onset type 1 diabetes is associated with smaller exocrine cells, eventually leading to reduced acinar cell number in established disease. The exocrine pancreas is commonly characterized by periductal and interlobular fibrosis after type 1 diabetes onset. Arrows depict direction of change. AAb+, islet autoantibody positive; T1D, type 1 diabetes.

Although patient entry criteria vary amongst reported studies, reductions in pancreatic volume of approximately 25–30% have been noted in recent-onset T1D cases and reductions of 30–50% in long-standing disease in comparison to control subjects, usually matched for age and/or normalized for BMI^174–177. Importantly, the notion of reduced pancreatic volume in T1D has found support irrespective of the methodology employed and has been confirmed in living subjects, as assessed by magnetic resonance imaging (MRI)^176,178,179. However, from a mechanistic perspective, perhaps the strongest support for a pathogenic link to T1D arises from findings by MRI that pancreas volume is reduced by approximately 25% in those with one or multiple T1D-associated autoantibodies (in the absence of overt disease) with reduced pancreas volume also observed in autoantibody-negative first degree relatives of a T1D proband (compared to unrelated control subjects)^178,180. These observations are especially salient, as they suggest that pancreatic size likely reflects a genetic contribution toward disease and potentially parallels the progressive loss of functional β cell mass.

Considering that islets comprise only 1–2% of pancreas volume, the mechanistic relationship between loss in pancreatic size with T1D also suggests that any reduction in volume or weight should be attributed to a loss in exocrine tissue. Histological evidence suggests that acinar cells are reduced in size and number, with loss of characteristic peri-islet amylase-negative cell clusters in T1D pancreata^{181–183,184}. The mechanisms underlying loss of pancreatic size in T1D are not well understood but have largely been hypothesized to result from loss of insulinotrophic effects on the exocrine pancreas¹⁸⁵.

Given the reductions in exocrine volume, a feature that has gone largely underappreciated in T1D is that of diminished exocrine function. Decreases in exocrine function in T1D patients were identified years ago through analysis of sera^7,186. More recent efforts suggest that subjects with T1D are characterized by diminished levels of serum trypsinogen and lipase as well as stool chymotrypsinogen and elastase^178,187,188. Moreover, serological levels of the exocrine pancreas enzymes lipase and trypsinogen are directly associated with pancreas size in subjects with T1D as well as single and multiple autoantibody positive relatives¹⁸⁸.

Pancreas Morphology

Both the endocrine and exocrine compartments of the normal human pancreas are defined by notable morphological heterogeneity. This heterogeneity includes variations in islet cell composition, size, and density. Similarly, acinar tissues display regionally distinct physiological and morphological variations. Such findings raise the question of their potential impact on health and disease, including T1D¹⁸. Specifically, in T1D, loss of β-cell mass is less severe in the pancreatic head and more dominant in body and tail^174,175,189. In concert, the aforementioned reductions in pancreas weight and volume are largely related to reductions of the ventral portion of the organ, especially in the tail region^174,175,189. Beyond these variations, pancreatic islet infiltration is often “patchy,” akin to the loss of melanocytes as seen in vitiligo and often segregated to neighboring lobules⁷⁵. β-cell loss is accordingly patchy and it remains unclear why β-cell destruction proceeds asynchronously, even within the same lobule. Pancreatic duct gland proliferation is observed in T1D¹⁹⁰ and exhibits a unique exocrine proteomic pattern as determined by LC/MS¹⁹¹. Despite the lack of morphological heterogeneity in the healthy exocrine and endocrine pancreas when compared to the heterogeneous pancreas pathology in T1D, it remains unclear how the three-dimensionally organized cellular environment and morphology might direct pancreatic function and contribute toward T1D pathogenies. Finally, it is also worth noting that while many of these features distinguish the pancreas of T1D from type 2 diabetes (T2D), there is a growing appreciation for how specific polymorphisms, including certain loci classically associated with T2D (e.g., TCF7L2), can influence the phenotype of donors with T1D^27,192–194 (Table 1).

Exocrine:Immune and Exocrine:Islet Interactions within the Pancreas

Beyond size and function, many additional pancreatic abnormalities in T1D are associated with innate and adaptive immunity. Indeed, T1D is often noted to be a disorder where inflammation is limited to the pancreatic islets (i.e., insulitis). Nevertheless, a curious increase in immune cell types has been observed in the exocrine compartment of the pancreas of donors with T1D, including neutrophils³⁹, CD8⁺ and CD4⁺ T cells, and CD11c⁺ cells⁷³. Additionally, enhanced levels of complement deposition, specifically C4d, has been observed¹⁹⁵. The potential exists that these features are the result of what has long been termed “gut leakiness” in T1D¹⁹⁶. Curiously, however, preproinsulin-reactive CD8⁺ T cells are present within the exocrine pancreas at comparable frequencies from T1D, single GADA+ positive and control donors¹⁹⁷ raising questions surrounding the mechanisms that allow for their migration into the islets and β-cell destruction during T1D. The T1D in Acute Pancreatitis Consortium (T1DAPC)^198,199 is poised to address many of the questions, potentially linking acute pancreatitis and exocrine pancreatic inflammation with T1D.

The Chicken and Egg Scenario for T1D Pathogenesis

With the aforementioned evidence in place (i.e., aberrations in the immune response, β-cells, islet cells, and exocrine pancreas in T1D), the classic “Chicken or the Egg” scenario arises for the disorder’s pathogenesis. Specifically, does the initiation of T1D pathogenesis begin via the conventional autoimmune model (i.e., a failure to recognize self-versus non-self in terms of β-cell antigens)? Multiple contributors are key to this concept including genetic susceptibility, largely related to the ability for a given HLA type (class I and class II) to present β-cell self-antigens; failure of central tolerance (i.e., inability to delete autoreactive T-cells in the thymus); together with inefficient peripheral tolerance (i.e., dysregulated immune regulation)⁹⁶. Each of these features has been well established in various research settings and/or experimental models of T1D. More speculative and long unproven is the concept of “molecular mimicry”²⁰⁰ and the identification of environmental factors capable of initiating anti-β-cell autoimmunity^200,201. Toward this latter notion, a number of studies have implicated the gut microbiome^202,203 and prolonged infection with enterovirus B²⁰⁴ as potential modulators of T1D development.

Or is it possible that an alternative model represents actuality — one where β-cells, through specific features (e.g., stress, number), or perhaps the pancreas itself, drive the immune response to function as would be expected? In this setting, the role of the β-cell may be to augment versus initiate the pathogenic processes (including but perhaps not limited to autoimmunity), with pancreatic alterations being subject to the influences of β-cell number and function. Indeed, despite providing an unparalleled opportunity to visualize the site of disease pathogenesis, organ donor tissue studies are inherently limited in that longitudinal analyses are not possible, and unlike studies of mouse models or cell lines, they are not readily poised to make focused/mechanistic manipulations that would address key features of disease development.

Evidence in support for each notion exists, especially when one considers information beyond studies highlighted in this Review⁹⁶. Perhaps both scenarios occur in T1D pathogenesis, with a sliding scale between autoimmune first versus β-cell first forms of T1D (Figure 1); a concept that might shed light onto the notion of disease heterogeneity. It is now becoming clear that the seminal notion of β-cell “homicide vs. suicide”⁹⁷ likely represents a false choice as a growing body of literature supports the notion that islet-specific, whole-pancreas, and autoimmune mechanisms contribute altogether toward the multifactorial pathogenesis of T1D. And, while a detailed discussion of genetics is beyond the scope of this Review, the balance between autoimmune, exocrine, and β-cell drivers of this disease may be subject to polygenic control^205–208, including the HLA class II, insulin, and 77 additional regions identified by genome-wide association studies (GWAS), which are associated with genes thought to impact both leukocyte and pancreatic β-cell function^209,210.

Questions and Future Directions

Recent revelations derived from studies of the human T1D pancreas have clearly improved our understanding of the disorder’s pathogenesis, yet knowledge voids in the autoimmune and pancreatic pathogeneses of the disorder remain. With respect to autoimmunity, questions remain regarding its contributions to disease heterogeneity (including endotypes²¹¹) and the role for environmental “triggers” vs. environmental factors that exacerbate autoimmunity. One important consideration for studies on organ donor tissues is that of age. The reality is, only a limited number of young people (thankfully) become organ donors, resulting in the greatest percentage of organ donors being between 50–64 years of age in the United States²¹². That said, nPOD is amazingly efficient at procuring cases at younger ages when they are available. Beyond this, while insulin autoantibodies (IAA) are measured for accepted cases, nPOD does not perform rapid screens for IAA due to technical limitations of the available validated assays. Hence, studies of organ donor pancreas are biased to older ages and likely miss an important subgroup of IAA+ donors, representing key areas for improvement and advance of our current knowledge. As noted above, studies of pancreata from at risk individuals suggest that T1D is preceded by a period of “waxing and waning” insulitis and β-cell loss²¹³. If true, what factors stimulate or reduce this autoimmune response and might it be chronic in certain individuals?

As for pancreatic contributions, the rapidly gaining notion of “functional β-cell mass” still remains conceptual, at least until the ability to image the endocrine pancreas in vivo becomes a reality²¹⁴. Such a technology would also address other key endocrine issues, including the contribution of islet cell number in the first years of life to T1D risk, the number of islets present at various stages of T1D, and the issue of whether loss of β cells occurs in a linear² versus waxing and waning fashion²¹³. Furthermore, while a role for β-cell stressors has become increasingly appreciated, their exact identification and mechanisms of action remains elusive and largely relies on in vitro models to mimic human disease. Finally, we need see clearer information regarding the question of why some β-cells are destroyed while others are spared. Future efforts investigating the role of the exocrine pancreas in T1D should include whether the many features described herein contribute to the disorder’s pathogenesis. In addition, studies should evaluate how these exocrine features might serve as a biomarker of disease stage as a function of its natural history³. Finally, exocrine features may serve in the pre-diabetic period, as well as post-onset, as a therapeutic sentinel of a positive therapeutic intervention, through preservation of exocrine volume or function.

For each of these areas, hope exists that advances in multi-omics, imaging technologies, novel biomarkers, screening programs defining those at risk for T1D (including in the general population), artificial intelligence systems, investigations across historically underrepresented/minoritized groups in biomedical research, and more will lead to improvements in our understanding of the symphony underscoring the pathogenesis of T1D.