Screening for Insulitis in Adult Autoantibody-Positive Organ Donors

Pancreas biopsies were obtained from the Beta Cell Bank, which operates for a clinical trial on islet cell transplantation in Belgium (15,16). They were taken as part of a quality control procedure that was approved by the ethics committees of the Belgian Diabetes Registry and participating hospitals. Tissue (∼0.5 cm³) was excised from the body region of cold-preserved (UW flushed) donor organs that were provided by Eurotransplant Foundation (Leiden, The Netherlands). It was fixed in 4% (vol/vol) phosphate-buffered formaldehyde, pH 7.4, or Bouin's fixative; embedded in paraffin; and then histologically analyzed. Between 1989 and 2004, a total of 1,507 biopsies were collected from patients aged 25–60 years for whom serum or plasma was also available for islet cell antibody assays. For none of these donors was diabetes mentioned in the donor information sheets.

Serum samples were prospectively tested for the presence of ICA, IA-2A, and GADA and retrospectively for IAA (17). ICAs were assessed by indirect immunofluorescence, and end-point titers were expressed as Juvenile Diabetes Foundation units (JDFU). IA-2A, GADA, and IAA were measured by liquid-phase radiobinding assay and expressed as the percentage of tracer bound in hemolysis-free sera. Cutoff values for antibody positivity were calculated as the 99th percentile of antibody levels in 790 nondiabetic controls after omission of outlying values (minimally 12 JDFU for ICA, 0.4% for IA-2A, 2.6% for GADA, and 0.6% for IAA). The aAb assays were validated in successive Immunology of Diabetes Workshops and international proficiency testing programs; all positive results were confirmed in a separate subsequent assay (17). Whole blood was haplotyped for DNA polymorphisms at the HLA-DQA1 and DQB1 gene loci, and DQ-associated risk was stratified as reported (18).

For detection of insulitis, paraffin sections were immunohistochemically double stained for leukocyte common antigen (LCA) (using mouse anti-LCA from Dako [Glostrup, Denmark]) and the pan-neuroendocrine marker synaptophysin (rabbit anti-synaptophysin; Dako). Binding was detected with biotinylated anti-mouse or anti-rabbit Ig (Amersham, Little Chalfont, U.K.) and, respectively, streptavidin horseradish peroxidase or alkaline phosphatase complex (Dako), using diaminobenzidine or new fuchsin as substrate. Sections were also double stained for insulin and glucagon (guinea pig anti-insulin and rabbit anti-glucagon were a gift of Dr. Van Schravendijk [Brussels Free University, Brussels, Belgium]). For each case, an average of 180 ± 16 (means ± SE) islets were screened. Insulitis was arbitrarily defined as an infiltrate of ≥15 LCA⁺ cells within the islet (central insulitis) or directly surrounding the islet (peri-insulitis). This number was set after determining mean and range of the number of LCA⁺ cells per islet in 62 islet aAb⁻ controls: 0.35 ± 0.04 (range 0–7) LCA⁺ cells per islet (1,550 islets investigated); the “insulitis” level was set at twice the maximum number encountered in these controls. Although arbitrary and therefore susceptible to discussion, this definition provides a more quantitative basis than that used in other studies for comparing the occurrence of insulitis. Only one study on human insulitis has defined insulitis on the basis of the number of leukocytes per islet (Gianani et al, [14] examined 14 normal controls to determine the number of leukocytes in the 10 islets with the largest mononuclear infiltrate).

Leukocytic infiltrates were immunophenotyped on paraffin sections using immunofluorescent double and triple staining with the following antibodies: rabbit anti-CD3 and anti-LCA (Dako), mouse anti-CD4 and anti-CD8 (Novocastra Laboratories, Newcastle upon Tyne, U.K.), and mouse anti-CD20 and anti-CD68 (Dako). Binding was visualized with anti-rabbit fluorescein isothiocyanate or anti-mouse Cy3 (Jackson Immunoresearch, Soham, U.K.) and examined in an Axioskop M fluorescence microscope (Zeiss, Oberkochen, Germany) equipped with an Orca AG camera (Hamamatsu, Japan) and Smartcapture imaging software (Digital Scientific, Cambridge, U.K.). For negative controls, primary antibodies were omitted; positive controls were conducted on paraffin-embedded human tonsils.

Relative insulin⁺ cell area was measured according to Rahier et al. (19) on coded slides using a 266-point counting grid in 10 randomly chosen microscope fields at a final magnification of ×140. The number of points hitting insulin immunoreactive cells (N_i) and pancreatic parenchyma (N_p) were counted in 10 randomly chosen microscope fields per case. Relative β-cell area was expressed as a percentage and calculated as (N_i/N_p) × 100. All morphometric analyses were carried out blinded on coded slides. Two-color immunohistochemistry using antibodies for Ki67 (Dako) and insulin was used to determine the percentage of proliferating β-cells. Reproducibility of the point-counting technique was evaluated with the formula of Weibel (19); the calculated relative error in each section was ∼11% for a mean relative β-cell area of 1.2%.

Testing for the four IAAs resulted in 62 positive cases (aAb⁺) of 1,507 donors in the age-group of 25–60 years (4.1%). Most cases were positive for a single aAb (n = 55), with only four double, two triple, and one quadruple positive case(s) (Table 1). The three donors with ≥3 aAbs displayed a susceptible HLA-DQ genotype, whereas the four cases with only two aAbs exhibited a neutral or protective HLA-DQ genotype (Table 1). We observed lower aAb titers in the single aAb cases than in the multiple aAb cases (results not shown), probably as a result of their high fraction of “statistical positives ” (1% cutoff). A control group of 62 aAb⁻ donors was selected from the total donor group by matching for age, sex, and BMI.

Of the 62 antibody⁺ and 62 antibody⁻ donors, only two cases presented islets with insulitis as defined under research design and methods; they belonged to the small subgroup (n = 3) with positivity for ≥3 aAbs (Table 1). Case subject 1 (M-59y) died 10 h after hospitalization for a subdural hematoma (plasma glycemia 6.4 mmol/l at admission), and case subject 2 (F-46y) died 43 h after hospitalization for subarachnoidal hemorrhage (plasma glycemia 8.2 mmol/l at admission).

In case subject 1, 5 of the 58 examined islets (9%) showed peri- or central insulitis. Four of these islets contained both insulin⁺ and glucagon⁺ cells, whereas one was insulin negative and mainly composed of glucagon⁺ cells. No other insulin⁻/glucagon⁺ islets were detected in this donor. In case subject 2, 27 of 917 islets (3%) presented insulitis (Fig. 1A and B), all islets containing insulin⁺ and glucagon⁺ cells. Another 3% of the islets were insulin negative and mainly composed of glucagon⁺ cells; these islets did not present signs of insulitis (Fig. 1C).

In both cases, the infiltrating cells predominantly corresponded to CD3⁺CD8⁺ T-cells (Fig. 1G–I) and CD68⁺ macrophages (Fig. 1F), with a few CD20⁺ B-cells and CD3⁺CD4⁺ T-cells detected (Table 2).

When the average β-cell surface area in antibody⁺ donors with ≥2 aAbs was compared with that in donors with a single aAb or that in antibody⁻ controls, no significant difference was noted (Table 3). Individual values in case subjects with multiple aAbs fell within the range of the control group.

The average percentage of insulin⁺ cells that were also positive for the proliferation marker Ki67 was very low (<1‰) in the three groups. The range in aAb negatives was 0–7‰. Only one aAb⁺ case presented a value outside this range (49‰). This case was characterized by insulitis and positivity for the four aAbs (case subject 1); the Ki67⁺ cells were only noticed in islets with a leukocytic infiltrate (Fig. 1D and E; Table 3).

In the present study, we have screened for insulitis in pancreatic tissue from 62 adult organ donors carrying aAb risk markers for type 1 diabetes. Insulitis was detected in two of three cases with at least three aAbs, be it in <10% of the islets. It was not found in any of the 59 case subjects with only one or two aAbs or in any of the 62 antibody⁻ control subjects. Both case subjects also presented a susceptible HLA-DQ genotype. There were no signs of a reduced β-cell mass in the two insulitis case subjects nor in the group of aAb⁺ donors. These data demonstrate that insulitis is a rare phenomenon in aAb⁺ nondiabetic adults.

Insulitis is also a rare finding at clinical onset of type 1 diabetes after 30 years of age, which contrasts with its detection in all onset patients younger than 7 years of age (4). Although type 1 diabetes in adults is, by definition, characterized by the presence of aAbs, a combination of three or more of such aAbs in association with a high-risk genotype is infrequent, especially in latent autoimmune diabetes in adults (20). The absence of insulitis in most adult aAb⁺ donors does therefore not exclude progression to the disease.

The present observations should, however, be interpreted with caution in terms of their possible significance for the development of type 1 diabetes. The seven donors with more than one antibody risk marker were older than 38 years, an age that is not typical for the development of classical type 1 diabetes. Moreover, four of them presented with a protective HLA-DQ genotype that may have modulated an autoimmune response. It is nevertheless conceivable that the presence of insulitis in two subjects with four risk markers illustrates a stage in the disease process that might eventually lead toward a sufficient β-cell loss such that diabetes develops. However, no decrease in β-cell mass became apparent after measuring relative β-cell surface area. There were also no substantial numbers of pseudo-atrophic islets (1) as a remnant of prior destructions and a sign of self-limiting insulitis at an earlier age. It is conceivable that both subjects exhibit a low-intensity autoimmune process affecting only a small percentage of the islets. The occurrence of Ki67⁺ β-cells in some of the infiltrated islets raises the possibility that β-cell proliferation can compensate for any losses. Such a subclinical autoimmune process may at a later stage result in slowly progressive type 1 diabetes or latent autoimmune diabetes in adults (21,22). The histopathology of late-onset type 1 diabetes has not been well studied. One case has been described: A 65-year-old female with positivity for two aAbs (GADA and IA-2A) and a HLA-DQ/DR risk profile was initially diagnosed with type 2 diabetes and then shown to present several islets with predominantly CD4⁺ T-cell infiltrates, without signs of β-cell destruction (23). In the presently described cases, infiltrating leukocytes mainly corresponded to CD8⁺ T-cells and CD68⁺ macrophages, as was also the case in type 1 diabetic patients with insulitis (2,24–26).

Our data do not strengthen or weaken the significance of the detected circulating markers as predictors for type 1 diabetes. They indicate an association between triple antibody positivity with a high-risk genotype and an insulitis process in the pancreas. In elder individuals, this insulitis process appears limited to <10% of the islets and may thus not lead to type 1 diabetes or may lead only to a mild form. We cannot exclude that cases with a low percentage of infiltrated islets were missed as a result of our sampling in one region (body) and of the relatively small number of analyzed islet sections (averaging 180 islets per organ); more extensive sampling was precluded by the islet isolation procedure for which these organs were harvested. For the same reason, we may have missed differences in β-cell mass if these would have occurred in other regions.

Despite the limitations imposed by the small tissue specimen, the nature and extent of our study provide information with respect to the use of organs from adult aAb⁺ donors for transplantation. Absence of histopathological changes in all 59 donors with one or two aAbs questions exclusion of these organs, while the detection of insulitis in triple antibody⁺ donors can be seen as an exclusion criterion.

In conclusion, we have screened 62 nondiabetic aAb⁺ organ donors older than 25 years for the presence of insulitis. Insulitis was found in two of the three case subjects who presented at least three antibodies but in none of the other 59 antibody⁺ subjects or 62 matched controls. Presence of one or more antibodies was not related to a decrease in β-cell mass. These observations can be used to include or exclude organs from aAb⁺ donors for transplantation in diabetic recipients. They also need consideration when recruiting adult aAb⁺ subjects for prevention trials.

This study has received grants from the Research Foundation Flanders (G035803 and G051704) and a Center Grant from the Juvenile Diabetes Research Foundation (4-2001-434).

We thank Nicole Buelens, Violeta Ardelean, Kristien Van Belle, Sylvie Duys, and Patrick Goubert for expert technical assistance. Crystalline humulin was a gift from Dr. H. Schmitt (Eli Lilly and Co.), cDNA for GAD65 was a gift from Dr. Å. Lernmark (University of Washington, Seattle, WA), and cDNA for IA-2_ic was a gift from Dr. M. Christie (King's College, London, U.K.).