Introduction
In type 1 diabetes, leucocyte infiltration of islets (insulitis) has a presumed central role in beta cell destruction and has been regarded evidence for an autoimmune aetiology of the disease [
1,
2]. Different criteria have been used to define insulitis [
2], making it difficult to compare and analyse results from different studies. A consensus opinion was recently reached on the criteria for insulitis: a patient is diagnosed with insulitis when a lesion is established in a minimum of three islets with a threshold level of ≥ 15 CD45
+ cells within the islet parenchyma or in the islet–exocrine interface [
1]. CD45, also known as protein tyrosine phosphatase receptor type C, is an enzyme specifically expressed on haematopoietic cells. Macrophages, neutrophils, T cells and B cells have been reported to be part of the insulitic infiltrate in type 1 diabetes [
2‐
5]. At the time when the consensus definition was established, seven was the highest number of CD45
+ cells found in islets from 61 non-diabetic control pancreases examined in the Network for Pancreatic Organ Donors with Diabetes (nPOD) [
1,
6], and thus the consensus definition adequately separated non-diabetic controls from individuals with type 1 diabetes.
Inflammation of the pancreas has been suggested to be part of the aetiopathology of type 2 diabetes [
7]. Supporting this theory, an elevated number of macrophages have been detected in islets of type 2 diabetes patients in conjunction with increased levels of cytokines and chemokines [
8]; CD8
+ cells and macrophages have also been found to be elevated in the exocrine compartment in type 2 as well as type 1 diabetes [
9,
10]. Furthermore, IL-1 receptor antagonist treatment of type 2 diabetes patients reduces HbA
1c and enhances C-peptide levels, but does not alter insulin resistance [
11], suggesting a direct role of inflammation in type 2 diabetes islet dysfunction. Collectively, these observations suggest that inflammation of the pancreas plays a prominent pathogenic role in type 2 diabetes.
In this study we determined the frequency of individuals diagnosed with type 2 diabetes fulfilling the definition of insulitis according to the consensus criteria developed for type 1 diabetes. Further, in pancreases with insulitic lesions, the number of macrophages, neutrophils, T cells and B cells was determined. In addition, other morphological hallmarks for type 1 diabetes were evaluated in the type 2 diabetic pancreases, such as islets without beta cells [
2,
12,
13] or the presence of HLA class I hyperstaining in the islets [
14,
15]. Finally, potential correlations between the presence of insulitic lesions and glucose-stimulated insulin secretion in vitro or clinical characteristics such as age, BMI, HbA
1c and autoantibody positivity of the donor were examined.
Discussion
The current definition of insulitis cannot be used to distinguish pancreases retrieved from individuals with type 1 diabetes from those with type 2 diabetes, as in this study insulitis was found in 31% and 28% of individuals, respectively. The biopsies analysed in this study were of high quality as the pancreases were procured from heart-beating organ donors. A limitation of our study is that the medical records of the donors were not made available in order to protect the integrity of the deceased person. However, most individuals had been diagnosed with type 2 diabetes for several years. In a meta-analysis, insulitis defined by the less stringent criteria of ≥ 5 CD45
+ cells in ≥ 3 islets was present in 56% of individuals with type 1 diabetes examined within the first month of diagnosis and in only four of 132 individuals (3%) examined > 1 year after diagnosis [
2].
The consensus definition of insulitis was adopted at the nPOD meeting in February 2013 and published in November 2013 in an effort to enhance understanding of the pathogenesis of type 1 diabetes and to allow comparisons between histopathological studies [
1]. CD45 was chosen as a marker for immune cells because the insulitic infiltrates in type 1 diabetes include T cells (predominantly CD8
+ but also CD4
+ cells), B cells, macrophages and granulocytes [
1,
3,
5]. This definition was designed to distinguish pancreases from individuals with type 1 diabetes from pancreases with background inflammatory infiltrates, as observed in non-diabetic individuals [
1]. Since then, few publications have adopted this more stringent definition of insulitis. The findings presented in this study call for a revision of the definition in order to allow pancreases from individuals with type 1 diabetes to be distinguished from those with type 2 diabetes. Furthermore, in a recent study of non-diabetic autoantibody-negative donors, four of eight individuals fulfilled the criteria for insulitis [
12], further reinforcing the need to revise the current definition of insulitis.
The highest number of CD3
+ cells within or adjacent to an islet from an individual with type 2 diabetes in the present study was 18 and we found multiple islets containing 6–10 CD3
+ cells. We therefore suggest that an insulitic islet should be redefined so that a lesion can be established when ≥ 15 CD3
+ cells are present in one section, within the islet parenchyma or in the islet–exocrine interface. The disease process in type 1 diabetes is heterogeneously distributed within the pancreas [
13]. Therefore, the presence of insulitis should preferentially be examined in multiple lobes of the pancreas. However, the limited availability of well-preserved human pancreases means this may not always be possible. The human pancreas contains several millions of islets of hugely varying size; small islets composed of only a few endocrine cells dominate and islets with a diameter of more than 250 μm are only rarely found [
18‐
20]. Obviously, the size of an islet will influence the number of immune cells present per islet. Based on these considerations, specifying the number of islets in the definition of insulitis (i.e. ≥ 3 islets) should, if possible, be avoided. Instead, the proportion of islets with infiltrating CD3
+ cells provides a better diagnostic criterion. In addition, the variation in islet size suggests that at least 50, but preferably 100, islets should be examined. In an effort to adhere to the consensus definition, it is proposed that if < 100 islets are available for evaluation, ≥ 3 islets should contain ≥ 15 CD3
+ cells. However, if ≥100 islets are available for evaluation, ≥ 3% should contain ≥ 15 CD3
+ cells (see text box: proposed definitions of insulitis). Using this definition for type 1 diabetes, both of the acute-onset type 1 diabetic donors included in this study and six out of six individuals with recent-onset type 1 diabetes examined within the Diabetes Virus Detection (DiViD) study fulfil the criteria [
13], i.e. 11% of all examined islets in the DiViD study had ≥ 15 CD3
+ cells. Furthermore, the proposed definition would adequately distinguish between individuals with type 1 diabetes and type 2 diabetes, as none of the 50 type 2 diabetic donors included in this study fulfilled these criteria. An adjusted definition of insulitis may also be applied to type 2 diabetes (see text box). The proposed definition for insulitis in type 2 diabetes was fulfilled by 28% of individuals with type 2 diabetes. This definition is, however, unable to discriminate between type 2 diabetes and type 1 diabetes. Therefore, based on the findings presented here, it is proposed that the discrimination of type 1 diabetes and type 2 diabetes should be based on the presence of insulin-deficient islets.
The recruitment of immune cells in type 2 diabetic pancreases has been proposed to be caused by beta cell stress [
8]. However, the CD45
+ cells found within the pancreases in this study were not preferentially located to the islets, but were seemingly randomly distributed throughout the entire gland. An increased presence of immune cells in the exocrine pancreas has been reported previously in both type 2 and type 1 diabetes [
9,
10]. Accumulations of CD45
+ cells were frequently found in areas of fibrosis, in which islets were densely surrounded by clusters of immune cells. Fibrosis constitutes the end stage of inflammation and it is likely that cellular inflammation maintained several years after diagnosis contributes to the loss of exocrine as well as endocrine parenchyma in individuals with type 2 diabetes [
21]. Inflammation mediated by CD45
+ cells, mainly CD68
+ macrophages, may also directly affect islet function and survival, arguing for a direct role of these cells in the pathogenesis of type 2 diabetes. Supporting this notion, laser capture-microdissected islets from type 2 diabetic pancreases show cytokine and chemokine patterns that suggest a sustained mild inflammation affecting the islets [
22,
23] and possibly contributing to islet dysfunction. This notion was not, however, supported by the functional characterisation of isolated islets in the present study. Even though we found a pronounced decrease in glucose-stimulated insulin release in islets isolated from pancreases obtained from individuals with type 2 diabetes when compared with islets from non-diabetic individuals, no difference in function could be observed between islets from type 2 diabetes donors with and without insulitis. Similarly, insulitis did not correlate with the level of HbA
1c. These functional observations should, however, be interpreted with caution—technical limitations mean we do not have information on insulitis in the actual islets examined in vitro, and HbA
1c critically depends on the intensity of diabetes treatment.
No correlation could be found between insulitis and the presence of islet autoantibodies in the type 2 diabetic donors. This should, however, be interpreted with caution as only 8% of the individuals with type 2 diabetes in our study had one or more autoantibodies. Although markedly higher than in non-diabetic individuals, the frequency of autoantibodies in children or adolescents diagnosed with type 2 diabetes has in larger studies been reported to be as high as 30% [
24]. Lack of insulitis in individuals with autoantibodies has been reported previously [
25], and suggests that autoantibodies may appear in response to beta cell damage without being accompanied by a cellular immune response infiltrating the islets.
In addition to insulitis, other morphological characteristics of type 1 diabetes are also present in the pancreases of individuals with type 2 diabetes, for example HLA class I hyperstaining. However, the frequent finding of islets devoid of beta cells in pancreases obtained from individuals with recent-onset as well as long-standing type 1 diabetes is not found in individuals with type 2 diabetes. Thus, the appearance of insulin-deficient islets constitutes the main morphological criterion able to distinguish type 1 diabetic pancreases from type 2 diabetic pancreases, even though the loss of beta cells in type 1 diabetes varies greatly, especially at the onset of disease [
13]. These observations indicate that leucocyte infiltration of islets has different immunopathological roles in type 1 diabetes and type 2 diabetes, a notion also supported by the higher proportion of T cells in the insulitic lesions in type 1 diabetes [
2].
Insulitis is considered one of the major morphological hallmarks of type 1 diabetes. The findings presented show that the current consensus definition of insulitis [
1] is not sufficient to separate the ongoing inflammation in the pancreas in individuals with type 1 diabetes from that occurring in those with type 2 diabetes. A modification of the definition of insulitis is therefore warranted. Based on our findings in a large cohort of type 2 diabetic donors, we propose that the diagnosis should be made when ≥ 15 CD3
+ cells, not CD45
+ cells, are found in ≥ 3 islets.