Type 1 diabetes results from autoimmune-mediated destruction of islet beta cells due to complex interactions between genetic and environmental factors [13]. The pathology of what we now consider type 1 diabetes was reported over 100 years ago, based on autopsy findings from individuals at the onset of the disease (reviewed previously [14]). These studies established the defining feature of the disease as a significant loss of islet beta cells, with 50–90% of islets having no beta cells (depending on disease duration) despite having other islet endocrine cells present at expected numbers. Loss of islet beta cells has also been observed in islet-autoantibody-positive non-diabetic organ donors [14]. Significant heterogeneity exists in the numbers of islets with residual beta cells (insulin⁺) vs those with partial or complete loss of beta cells (insulin⁻). The insulin⁺ islets also show a high degree of heterogeneity, ranging from normal to greatly reduced beta cell numbers. A summary of histopathological features between non-diabetic control and type 1 and type 2 diabetic organ donors is presented in Table 1.

In 2013, a consensus statement was published for islet infiltration (insulitis), to provide standardisation of histopathological investigations [15]. The consensus statement defines insulitis as lesions present in a minimum of three islets, with a threshold level of ≥15 CD45⁺ cells immediately adjacent to islet endocrine cells. Insulitis primarily affects insulin⁺ islets, yet <10% of insulin⁺ islets are typically infiltrated [14]. The reasons for this remain unclear. Insulitic islets were rarely observed after 10 years of type 1 diabetes duration, coinciding with loss of islet beta cells [14]. Insulitis in multiple-autoantibody-positive donors was similar to that observed in donors with type 1 diabetes, with a highly variable lobular pattern and similar immune-cell phenotype and numbers [14]. The insulitic mononuclear cell infiltrate is dominated by CD3⁺ T lymphocytes, comprised of CD8⁺ T cells with lesser numbers of CD4⁺ T cells, and variable numbers of CD20⁺ B lymphocytes. Among the CD3⁺ T cells, CD8⁺ T cells are clearly activated, as indicated by the dramatically elevated proportion of proliferating CD8⁺ T cells, even in long-standing type 1 diabetes [16]. Interestingly, numbers of CD3⁺ T and CD20⁺ B lymphocytes parallel each other, as well as total immune-cell numbers (CD45⁺) [14]. The same patterns were observed in two young adult organ donors with islet-associated autoantibodies but no clinical history of diabetes, suggesting infiltration may precede clinical onset of the disease. Likewise, the presence of both CD8⁺ T cells and CD20⁺ B cells was higher in islets with residual beta cells vs those without in a very recent-onset type 1 diabetes case [16]. Other studies in younger individuals at type 1 diabetes onset were conducted through retrieval of autopsy samples and reported that the highest numbers of CD20⁺ B cells were found in the youngest patients, suggestive of potential disease heterogeneity. However, high numbers of B cells were also observed with elevated numbers of CD8⁺ T cells, such that younger patients appeared to have greater numbers of islet-infiltrating cells than older patients at disease onset [17]. These findings may have implications for treatment choices. The number of islets with insulitis has been demonstrated to inversely correlate with type 1 diabetes duration (e.g., higher insulitis frequency at diabetes onset) and to positively associate with beta cell mass; age of onset was not associated with insulitis frequency [14]. In donors with insulin⁺ islets, insulitis was detected many years post onset of disease. Indeed, studies in the Joslin Medalist Program found residual beta cells in all donor pancreases, despite ≥50 years of type 1 diabetes [18]. Other immune cells identified within insulitic islets include CD68⁺ macrophages, CD11c⁺ dendritic cells, forkhead box P3 (FOXP3)⁺ regulatory T cells and mast cells [14, 19, 20]. The monocytic infiltrate in insulitic islets can be comprised of diffusely scattered cells, and as aggregates within the islet interior or in the immediately adjacent exterior.

Until recently, islet amyloidosis, which develops from extracellular deposition of islet amyloid polypeptide (IAPP), was widely considered pathognomonic for type 2 diabetes [21]. Islet IAPP is a 37 amino acid polypeptide that is co-produced/co-secreted with insulin from beta cells; hypersecretion of insulin with insulin resistance may result in IAPP deposition [22]. Two recent studies report the presence of islet amyloid in people with type 1 diabetes, including a 12-year-old individual [23, 24]. This observation provides an intriguing parallel between the pathological processes in type 1 and type 2 diabetes affecting at least a subset of patients, adding support to the concept that common metabolic derangements affecting beta cells are likely to occur in both diseases.

In comparison with control islets, islets from individuals with type 1 diabetes have smaller islet vessels that are similar in size to vessels of the exocrine region [25]. Of note, islets with residual beta cells had a similar vascular phenotype to control islets, suggesting that beta cells within islets maintain the normal microvasculature. Recent studies using optical clearing show these vessels are more tortuous in islets without beta cells [26]. In addition, the sympathetic innervation of islets from individuals with type 1 diabetes is preserved and could promote vasoconstriction, as seen with smaller vessels in these individuals by two-dimensional (2D) studies. Parasympathetic innervation of human islets is important for both insulin and glucagon responses to a meal or during fasting. Optical-clearing three-dimensional (3D) studies show abundant cholinergic innervation of human islets; however, no studies are yet available regarding parasympathetic innervation in type 1 diabetes [26].

The exocrine pancreas size is significantly reduced in both individuals with type 1 and type 2 diabetes, with the former exhibiting the largest difference [27–29]. Indeed, smaller pancreases in individuals with type 1 diabetes have been reported by several groups by either autopsy or radiographic studies (reviewed previously [30]). These studies show that the pancreas is 20–50% smaller in children and adults with type 1 diabetes compared with unrelated non-diabetic control participants. Potential mechanisms underlying reduced pancreas size in type 1 diabetes could include atrophy, impaired organ growth rate during fetal or postnatal life, or a combination of both. Loss of functional beta cell mass and, thereby, loss of insulinotropic effects on acinar cells, has also been proposed as a primary mechanism [31]. Since the studies reported to date are cross-sectional in nature, it is not known if people with type 1 diabetes are born with a smaller pancreas or if the organ shrinks during the disease process. In addition, several groups reported no effect of diabetes duration on pancreas weight or size, suggesting insulin therapy may not reverse exocrine changes (previously reviewed [32]).

Insulin-secreting beta cells are remarkable for their ability to adapt to metabolic demand. In fact, trained athletes secrete up to three times less insulin to achieve euglycaemia than untrained individuals; conversely, non-diabetic obese people can secrete five times more insulin than control participants in response to a glucose challenge [33]. However, the adaptive response of beta cells is not limitless and when it fails, type 2 diabetes ensues. In contrast to the dramatic changes in islet morphology and immune infiltration described for type 1 diabetes above, there is no stereotyped histology of the pancreas in type 2 diabetes (Table 1). One histological feature that has historically garnered interest is the deposition of amyloid, an extracellular protein aggregate derived from IAPP (Fig. 2). While islet amyloid is present in some islets of the majority of individuals with type 2 diabetes, its causal role in diabetes pathogenesis has not been established. In addition, it is not a definitive histological marker since a significant fraction of individuals with type 2 diabetes do not have amyloid in their islets, while these deposits can be present in the islets of euglycaemic individuals and, as mentioned earlier, of those with type 1 diabetes [34, 35].

Multiple autopsy studies have found pancreatic beta cell mass in individuals with type 2 diabetes to be about 60% of normal beta cell mass [34]. Similar findings have been obtained from organ-donor pancreatic samples [35]. Since islet and beta cell mass at birth and in childhood are highly variable, and because longitudinal determination of beta cell or, even, whole-islet mass in humans is still impossible, cause and effect cannot be determined. In other words, we do not know if beta cell-mass decline during the course of type 2 diabetes is due to genetic predisposition, or lifestyle choices and associated glucotoxicity, or if individuals born with a small islet mass are simply more likely to develop type 2 diabetes. It is clear, however, that the beta cells present in individuals with type 2 diabetes do not function normally, as was already established more than 30 years ago [36].

One frequently, though not consistently, reported feature of the pancreas in type 2 diabetes is its increased fat content, as determined by computed tomography (CT) [2, 37] and MRI [38,39]. While Saisho and colleagues found pancreatic fat content increased with age, but not further with type 2 diabetes, multiple other studies documented additional lipid accumulation in individuals with type 2 diabetes and suggested that intra-organ fat might contribute to beta cell dysfunction [2, 38]. Hepatic steatosis is, of course, a common feature in obesity and insulin resistance. Therefore, it is not surprising that steatosis also occurs in the pancreas; in fact, pancreatic steatosis co-occurs in more than two thirds of individuals with type 2 diabetes [39]. But does it play a role in islet dysfunction? This question is much more difficult to answer; although, increased pancreatic fat content precedes the development of diabetes in some animal models [39]. One possibility is that, in the initial stages of the disease, pancreatic steatosis may contribute to a decline in function, but once overt diabetes ensues, multiple deleterious factors, such as increased inflammation and oxidative stress, play a much more determinative role than pancreatic lipid content [40].

At present, no clear histological marker specific to type 2 diabetes has emerged. Currently, large-scale efforts directed at the analysis of the human pancreas from deceased organ donors using multiple experimental modalities include the European RHAPSODY, INNODIA and Innovative Medicines Initiative for Diabetes: Improving beta-cell function and identification of diagnostic biomarkers for treatment monitoring in Diabetes (IMIDIA) projects (summarised previously [41]), and the NIH-funded Human Pancreas Analysis Program (HPAP) [42], which was recently extended to the analysis of type 2 diabetes. It can be hoped that these projects will shed new light on the pathogenic events that occur in the pancreas in diabetes. For instance, until recently, insulin resistance, particularly of the liver, was considered by many to be the most relevant cause of type 2 diabetes [43, 44]. Now, the concept of islet failure as key to both forms of diabetes is gaining acceptance among diabetes researchers worldwide and represents a substantial step forward.