Insulitis in the pancreas of non-diabetic organ donors under age 25 years with multiple circulating autoantibodies against islet cell antigens

Insulitis is an inflammatory lesion of the islets of Langerhans characteristic for patients with recent-onset type 1 diabetes. The infiltrate predominantly consists of CD8+ T-lymphocytes of which a fraction is thought to mediate beta cell specific cytotoxicity [1‐4]. It is now widely accepted that clinical onset of type 1 diabetes is preceded by a usually long hidden disease process as judged by the appearance of circulating antibodies against beta cell antigens including islet cell cytoplasm (ICA) [5], insulin (IAA) [6], glutamic acid decarboxylase (GADA) [7], insulinoma-associated protein 2 (IA-2A) [8], and zinc transporter 8 (ZnT8A) [9], months to decades before diagnosis [10, 11]. These autoantibodies tend to develop sequentially rather than simultaneously [12], with IA-2A and ZnT8A usually appearing closer to clinical onset [13]. Both in children and in adults up to 40 years of age, presence of at least two molecularly defined autoantibodies confers 90% risk of developing clinical onset within 20 years, and is now termed asymptomatic type 1 diabetes [10, 11, 14]. However, the relationship between the development of autoantibodies and the development of histopathological hallmarks of type 1 diabetes during the natural history of asymptomatic disease is poorly understood. In a now classical model for the development of type 1 diabetes, autoantibodies are considered to be indicators for ongoing autoimmune beta cell destruction in genetically susceptible individuals, triggered, and promoted by as yet unidentified environmental factors [15, 16]. At clinical onset, functional beta cell mass has dropped to 10–40% of normal [17], but the timing and kinetics of its decline during the asymptomatic phase remain a matter of debate [15, 16]. The classical model remains to a large extent hypothetical since firm evidence supporting a causal link between insulitis and circulating autoantibodies is lacking. In previous histopathological studies, donor pancreata from autoantibody-positive non-diabetic organ donors—used as a model for at risk individuals [3, 4, 18‐22]—showed no evidence of decreased relative beta cell area [4, 18, 20, 21]. Only four autoantibody-positive individuals in total were identified with insulitis according to international consensus criteria [4, 18], half of them in subjects >45 years of age, an age group that is substantially older than the group of 10–14 years in which the incidence of type 1 diabetes is at its highest level [1]. Insight into early histopathological changes in islet tissue in autoantibody-positive individuals under the age of 25 years is therefore limited and formed the rationale for studying a large cohort of non-diabetic organ donors below the age of 25 years. We compared histopathological changes in islet tissue from autoantibody-positive organ donors to a matched control group of autoantibody-negative organ donors from the same cohort.

Screening for five islet autoantibody types in sera from 556 pancreas donors aged under 25 years without symptomatic type 1 diabetes allowed to identify 25 organs from single marker-positive individuals and two from double positive ones. This substantially expands the limited number of previous single- or multiple autoantibody-positive (n=7 and n=3, respectively) organ donors under age 25 years and for whom the individual age was reported [3, 4, 19, 21]. This illustrates the considerable effort needed to study pancreas tissue from persons with apparent immune activation against islet cells or, in case of multiple antibody-positivity, who are allegedly in an asymptomatic disease stage [1, 4, 14, 18]. Histopathological alterations were discrete and restricted to both double positive individuals, hereby confirming previous observations [4, 18]. In the present study, histological analysis of the pancreata from both double positive donors showed discrete peri-insulitis in a small fraction of islets, as well as an intact relative beta cell area despite the presence of numerous pseudo-atrophic islets predominantly in a lobular area. A relatively high islet cell replication rate was also noted. An intriguing novel observation is the presence of >30 small focal areas of 3–5 pseudo-atrophic islets scattered throughout the parenchyma of an individual with alleged asymptomatic type 1 diabetes on the basis of positivity for IA-2A, ZnT8A, and the presence of HLA class I- and II-inferred susceptibility [10, 11, 14].

Strengths of the study include the procurement of donor organs, systematic screening for five autoantibody markers, and the use of international consensus criteria to define insulitis. The fact that for most donors only a single usually small tissue sample was received is a limitation. It can thus not be completely excluded that focal histopathological alterations may have been missed in some organs due to the non-random distribution of lesions throughout the organ. Combined screening for both insulitis and pseudo-atrophic islets, as performed in the present study, may decrease but not avoid such a risk. As molecularly defined autoantibodies, including GADA, may contribute to ICA reactivity, joint presence of ICA, and GADA and may not necessarily imply multiple autoantibody positivity [31]. However, high-titer ICA levels, such as present in donor DBB-3504, were associated with increased risk of progression to clinical onset [32]. Likewise in latent autoimmune diabetes in adult patients, the time to insulin-dependence was shorter in joint presence of ICA and GADA than in presence of GADA alone [33]. Part of the ICA positivity in in donor DBB-3504 may have been caused by an autoantibody specificity for which we did not test [13, 23, 34, 35].

When combining our findings with available histological and biological information on multiple autoantibody-positive organ donors (all ages) outside the context of gestational diabetes, polyendocrinopathy or exocrine co-morbidity, and for whom the exact age was reported, it is remarkable that so few histopathological hallmarks of type 1 diabetes were found in the pancreata from these individuals, who are allegedly in an asymptomatic disease stage (Table S3) [3, 4, 18, 20]. Discrete insulitis and foci of pseudo-atrophic islets were observed in 4/4 donors under age 25 years with two or more autoantibody markers and in 2/11 multiple marker-positive donors above that age (Table S3), in line with observations on a higher prevalence of insulitis with younger age at diagnosis in recent-onset type 1 diabetes [2, 36]. The composition of the islet infiltrates was similar in all cases, with a preponderance of CD8+ lymphocytes. The presence of HLA class I alleles reported to confer susceptibility independently of HLA class II-inferred risk [27, 28, 30, 37], and of a high-risk autoantibody profile (IA-2A or ZnT8A plus ≥1 other marker) [11, 13, 34] further support the probability of progressive subclinical disease in (many of) these donors, despite the presence of protective HLA class II haplotypes in some (Table S3); indeed, the latter plays no longer a major role once multiple antibodies have developed [11, 28, 38, 39].

In contrast, neither insulitis nor pseudo-atrophic islets were reported in pancreata from single autoantibody-positive organ donors screened for all antibody markers and reported individually or as a group [3, 4, 18‐22]. Conceivably, some of the donors with low-level single autoantibody positivity may represent “statistical” positives due to the use of the 99th percentile of large numbers of healthy controls as cut-off in all assays [23, 24], conferring a risk up to 5% of a false-positive result when testing for five markers. Moreover, single autoantibody positivity is often a low-level transient phenomenon of uncertain significance [40]. Multiple autoantibody positivity, on the other hand, is associated with 90% 20-year risk of progression to clinical onset in children and adults [10, 11, 28].

We found no indication that autoantibody-positivity, be it for one or multiple specificities, was associated with a decreased relative beta cell area in the identified donor organs, in line with previous studies [4, 18, 20, 21, 41, 42]. Interestingly, even in presence of clear insulitic lesions and pseudo-atrophic islets devoid of beta cells, the relative beta cell area was not found to be significantly different from that in matched controls.

Of special interest is donor DBB-A096 with insulitis and multiple small foci of beta cell loss scattered throughout the gland. Such large numbers of small lesions, consisting of 3–5 pseudoatrophic islets, have not previously been described in recent-onset patients, although conversely, a lobular pattern of beta cell survival is relatively common in type 1 diabetes patients, especially those with older age at onset [43‐45]. If the small focal lesions are indicative of a pre-diabetic state in this subject, it could be hypothesized that during disease progression, such focal lesions grow in number or size and finally fuse, leading to the characteristic presentation of a pancreatic gland at disease onset with predominantly pseudo-atrophic islets being present, but also with regions in which islets show no histopathological changes. The early disease processes leading to such (multi)focal lesions are unknown. Whether they represent focal points of (auto)immunity with the autoreactive cells migrating to nearby islets, possibly via the connective tissue septa, as observed in mouse models of the disease [46], or represent focal points of (possibly viral) infection spreading outwards [47], can only be speculated.

In the absence of longitudinal studies on islet pathology in high-risk individuals, it cannot be excluded that the observed histopathological changes in multiple autoantibody-positive donor pancreata do not represent true pre-diabetic lesions. This is, however, unlikely since multiple autoantibody-positivity is rarely transient, and confers a high probability of progression to clinically overt disease [11, 28, 38, 39]. The seemingly preserved beta cell mass in the identified multiple autoantibody-positive donors should not be taken as argument against a high probability of later disease progression. Indeed, data by us and others [48, 49] indicate that the functional beta cell mass, as assessed by (stimulated) C-peptide release, only starts to sharply decline within 2 years from diagnosis. The age range of the multiple autoantibody-positive donors in Table S3 indicates an underrepresentation of children in all series of donor pancreata and may point to a selection bias towards slow progressors [50]; hence, the observed histopathological alterations may not be fully representative for asymptomatic childhood-onset type 1 diabetes. The described lesions may also reflect remnants of a low-level immune process during which islets are still able to compensate for any immune-mediated loss as indicated by the presence of islet cell replication in some of these donors. We can also not fully discount the possibility that the relatively minor level of insulitis that is observed is a consequence rather than a cause of the pathogenetic events taking place.

The finding of relatively high levels of replication in the islets of Langerhans of two donors with insulitis is of interest, especially because of the virtual absence of replication in most adult organ donors [51]. However, the interpretation of these observations and the increased CD68 macrophage infiltration in one donor is complicated by studies showing that organ donors with an increased duration of stay in intensive care show evidence of tissue repair, including increased numbers of M2 macrophages, increased vascular density, and increased replication of all pancreatic cell types [52]. As both donors in the present study are characterized by a relatively long duration of stay in intensive care, it cannot be excluded that the increase in beta cell replication found in these two donors is caused by a repair process and is not related to the presumed autoimmune process itself. The percentage of Ki67-positive beta cells was not correlated with relative insulin-positive area, both in our autoantibody-positive and in autoantibody-negative donors (data not shown). This does, however, not exclude compensatory beta cell regeneration via, e.g., transdifferentiation.

In conclusion, evidence for early type 1 diabetes-related histopathological lesions in the pancreas of donors under age 25 years appears restricted to organs from multiple autoantibody-positive individuals at high risk of developing symptomatic disease, in line with observations in older age groups [13]. Donors with single autoantibody-positivity presented no histopathological evidence of immune-mediated beta cell destruction. Single autoantibody-positivity may thus not be a sufficient indicator of islet lesions and not a suitable parameter to select patients for clinical intervention studies. More stringent inclusion criteria are necessary to identify individuals at high risk for the development of type 1 diabetes, such as those proposed in a recent successful immune intervention study where inclusion was limited to relatives of patients with type 1 diabetes with multiple autoantibodies and evidence of dysglycemia [53]. A better insight into the relationship between autoantibody-positivity, insulitis, and beta cell destruction will help in devising better therapies aimed at preventing or curing the disease. However, the fact that none of the donors with multiple autoantibodies and insulitis reported in the present study showed evidence of a decreased relative beta cell area indicates that the relationship between T-cell infiltration and beta cell destruction is more complex than proposed in the classic models of the disease. Alternative models have recently been proposed in which both islet autoimmunity and beta cell dysfunction are suggested to play equally essential roles [54]. Histopathological studies of multiple autoantibody positive donors may help to characterize such pathogenetic pathways, investigating to what extent islet autoimmunity is accompanied by evidence of beta cell dysfunction and stress.