Population based study of prevalence of islet cell autoantibodies in monozygotic and dizygotic Danish twin pairs with insulin dependent diabetes mellitusBMJ 1997; 314 doi: https://doi.org/10.1136/bmj.314.7094.1575 (Published 31 May 1997) Cite this as: BMJ 1997;314:1575
- Jacob S Petersen, scientista,
- Kirsten O Kyvik, assistant lecturerc,
- Polly J Bingley, professore,
- Edwin A M Gale, head of departmente,
- Anders Green, assistant lecturerc,
- Thomas Dyrberg, immunology managerb,
- Henning Beck-Nielsen, head of departmentd
- a Hagedorn Research Institute, DK-2820 Gentofte, Denmark
- b Novo Nordisk A/S, DK-2880 Bagsvaerd, Denmark
- c Genetic Epidemiology Research Unit, Institute of Community Health, Odense University, DK-5000 Odense, Denmark
- d Department of Endocrinology, Clinical Research Institute, Odense University Hospital, Odense
- e Diabetes and Metabolism, St Bartholomew's Hospital, London EC1A 7AB
- Correspondence and requests for reprints to: Dr J S Petersen ZymoGenetics, 1201 Eastlake Avenue East, Seattle, WA 98102, USA
- Accepted 17 February 1997
Objective: To study the comparative importance of environment and genes in the development of islet cell autoimmunity associated with insulin dependent diabetes mellitus.
Design: Population based study of diabetic twins.
Setting: Danish population.
Subjects: 18 monozygotic and 36 dizygotic twin pairs with one or both partners having insulin dependent diabetes.
Main outcome measures: Presence of islet cell antibodies, insulin autoantibodies, and autoantibodies to glutamic acid decarboxylase (GAD65) in serum samples from twin pairs 10 years (range 0-30 years) and 9.5 years (2-30 years) after onset of disease.
Results: In those with diabetes the prevalence of islet cell antibodies, insulin autoantibodies, and autoantibodies to glutamic acid decarboxylase in the 26 monozygotic twins was 38%, 85%, and 92%, respectively, and in the dizygotic twins was 57%, 70%, and 57%, respectively. In those without diabetes the proportions were 20%, 50%, and 40% in the 10 monozygotic twins and 26%, 49%, and 40% in the 35 dizygotic twins.
Conclusion: There is no difference between the prevalence of islet cell autoantibodies in dizygotic and monozygotic twins without diabetes, suggesting that islet cell autoimmunity is environmentally rather than genetically determined. Furthermore, the prevalence of islet cell antibodies was higher in the non-diabetic twins than in other first degree relatives of patients with insulin dependent diabetes. This implies that the prenatal or early postnatal period during which twins are exposed to the same environment, in contrast with that experienced by first degree relatives, is of aetiological importance.
Autoantibodies against several islet cell antigens greatly increase the risk of developing insulin dependent diabetes
Development of islet cell autoimmunity is determined by enviromental rather than genetic factors
Fetal life is aetiologically important for induction of islet cell autoimmunity
Progression to clinical insulin dependent diabetes depends on genetically controlled responses to enviromental exposures after fetal life
Much is known about the autoimmune phenomena associated with insulin dependent diabetes mellitus, but little is known about the nature or the time of initiating aetiological events. The crude concordance rates of insulin dependent diabetes are no higher than 23-53% in monozygotic twin pairs,1 2 3 4 indicating that exogenous factors are important in the pathogenesis of the disease. Such putative exogenous factors could have considerable impact during fetal life, when the immune system is immature and tolerance to various antigens is induced.5 Several findings suggest that exposure to a virus during fetal life may stimulate the development of islet cell autoimmunity.6 7 8 Thus, the rubella embryopathy syndrome is associated with an increased risk for later development of insulin dependent diabetes, and maternal enteroviral infection during pregnancy is also a risk factor for development of childhood insulin dependent diabetes.8 The risk of developing insulin dependent diabetes later in life also seems to be influenced by diet in infancy, especially early exposure to cows' milk products.9 10 11 Several environmental factors present early in life thus seem to be potentially associated with an increased risk of developing insulin dependent diabetes.
Studies in first degree relatives of patients with insulin dependent diabetes have shown that islet cell antibodies–that is, insulin autoantibodies, islet cell antibodies, and glutamic acid decarboxylase autoantibodies–can be detected several years before clinical insulin dependent diabetes,12 13 14 15 16 suggesting a long disease prodrome. Patients become positive for islet cell antibodies before 5 years of age in 85% of cases,17 suggesting that the initial lesion could occur early in life.
To investigate the contribution of environmental and genetic factors we analysed the presence and concentration of the three antibodies in a Danish population based cohort of dizygotic and monozygotic twins in which at least one twin had insulin dependent diabetes.
Patients and methods
The diabetic twin pairs were identified through the population based Danish twin register. This cohort consists of 20 888 twin pairs born from 1 January 1953 to 31 December 1982, representing 74.4% of all twin pairs born in Denmark during 1953-67 and 97.4% of those born during 1968-82.18
The diabetic twin pairs (table 1) were identified by means of a questionnaire to the total cohort, with an individual response rate of 92.6%, and through record linkage with the Danish register of causes of death. The total number of twin pairs in which one or both partner had insulin dependent diabetes according to criteria of the World Health Organisation was 95: 26 monozygotic and 69 dizygotic pairs. The cumulative concordance, which is comparable with risk of recurrence in other relatives, estimated by life table analysis from birth to age 40 was 70% in monozygotic and 13% in dizygotic twins.1 A total of 54 pairs (18 monozygotic and 36 dizygotic) were available for detailed studies; the remainder did not participate because of death or emigration of one of the twins, disability in the diabetic twin, or unwillingness to participate. The absence of diabetes in the unaffected twin was confirmed by means of an oral glucose tolerance test performed immediately before or after the clinical examination.1
For the twin pairs participating in the clinical investigation, zygosity was established by serological analysis of 11 blood and enzyme type systems by using the same approach as in paternity testing. Twin pairs with complete concordance for all systems were regarded as monozygotic; the incidence of misclassification is below 1% for dizygotic twin pairs.19
Glutamic acid decarboxylase autoantibodies
Glutamic acid decarboxylase antibodies were measured as previously described.20 Briefly, the human islet glutamic acid decarboxylase cDNA21 22 was transcribed and translated in vitro according to the manufacturer's instructions (Promega, Madison, United States) in the presence of methionine labelled with sulphur-35 (Amersham). Serum samples were incubated with the tracer overnight at 4°C, with or without the addition of 1 μg of unlabelled, affinity purified, recombinant human glutamic acid decarboxylase as competitor. The immune complexes were isolated on protein A Sepharose (Pharmacia, Sweden), and the amount of immunoprecipitated glutamic acid decarboxylase was measured by scintillation counting. All samples were tested in duplicate. Concentrations of glutamic acid decarboxylase antibody were expressed as index values: (cpm of sample-cpm of negative control sample)/(cpm of positive control sample-cpm of negative control sample), where cpm is counts per minute.
Serum samples were regarded as positive when index values exceeded the mean plus 3 SD of glutamic acid decarboxylase antibody indices in 50 healthy controls (data not shown).
Islet cell antibodies
Undiluted serum samples were screened for conventional IgG islet cell antibody by means of indirect immunofluorescence on 4 μm cryostat sections of blood group O human pancreas.14 Positive samples were then titred by doubling dilutions in phosphate buffered saline on tissue obtained from a single pancreas under standard incubation conditions.23 Local standard serum samples calibrated to 2, 4, 8, 16, 32, and 80 JDF units were included in each assay. End point titres were converted to JDF units, which are based on a standard curve of known serum samples from patients with insulin dependent diabetes mellitus who are positive for islet cell antibodies. The threshold of islet cell antibody detection was 4 JDF units; thus values reported as 0 indicate <5 JDF units.
Insulin autoantibodies were assayed as previously described.14 Serum samples were extracted by using acid washed, dextran coated charcoal to remove endogenous insulin; 80 μl of serum was then incubated for 48 hours at 4°C with 80 μl of 40 mmol/1 phosphate buffer and 5.3x10-3 pmol radiolabelled human insulin (specific activity 2000 Ci/mmol; Amersham), with and without excess (2.55 pmol/tube) non-labelled insulin (Actrapid, Novo-Nordisk, Bagsvaerd, Denmark). The immunoglobulin fraction was precipitated with polyethylene glycol 6000 (12.9% weight/volume) and washed. The specific binding was calculated by subtracting the counts in the presence of cold insulin from the counts without the cold insulin. Results were expressed as percentage displaced binding. Subjects were classified as positive for insulin autoantibody if the corrected binding was >3SD above the mean of 172 adult blood donors.
Statistical analyses included the Mann-Whitney, χ2, and Fisher's exact tests; and 5% was chosen as the level of significance.
Antibodies in monozygotic and dizygotic diabetic twins
Of the individual diabetic monozygotic twins 85% (22/26) were positive for insulin autoantibodies, 38% (10/26) were positive for islet cell antibodies, and 92% (24/26) were positive for glutamic acid decarboxylase autoantibodies, comparable with prevalences found in individual diabetic dizygotic twins–that is, 70% (26/37), 57% (21/37), and 60% (21/37), respectively (table 2) and figure 1).
The level of insulin autoantibodies in both the diabetic monozygotic and dizygotic twins was, as expected, high–median 4.4 (95% confidence interval 4.4 to 13.7) and 3.9 (6.2 to 19.0) units, respectively–as all had been treated with insulin, sometimes for many years (fig 1). The levels of islet cell antibodies and glutamic acid decarboxylase autoantibodies were also comparable between the diabetic monozygotic and dizygotic twins–that is, the median titre was 13.5 (6.0 to 49.1) and 7 (7.6 to 29.0) JDF units, respectively, and the median glutamic acid decarboxylase autoantibody titre was 0.22 (0.07 to 0.23) and 0.47 (0.11 to 0.33), respectively (fig 1).
Antibodies in monozygotic and dizygotic non-diabetic twins
Of the non-diabetic monozygotic twins, 5 out of 10 were positive for insulin autoantibodies, 2 out of 10 were positive for islet cell antibodies, and 4 out of 10 were positive for glutamic acid decarboxylase autoantibodies (table 2) and fig 1). Similar high prevalences were also found in the non-diabetic dizygotic twins–that is, 17 out of 35, 9 out of 35, and 14 out of 35, respectively (table 2) and fig 1).
The levels of the three antibodies in non-diabetic monozygotic and dizygotic twins were also similar–that is, the median concentrations of insulin autoantibodies were 1.30 (-7.44 to 19.80) and 1.28 (1.23 to 7.71) units; median islet cell titres were 6.5 (0.14 to 12.85) and 7 (4.56 to 12.19); and median levels of glutamic acid decarboxylase autoantibody were 0.15 (-0.14 to 0.60) and 0.06 (0.1 to 0.26), respectively (fig 1).
Combinations of the three antibodies in monozygotic and dizygotic twins
The presence of more than one kind of islet cell autoantibody was more common in the diabetic monozygotic and dizygotic twins than in non-diabetic twins (fig 2). Both glutamic acid decarboxylase and islet cell antibodies were present in 38% (14/37) and 38% (10/26) in the diabetic twins, respectively, compared with 12% (4/35) (P=0.014) and 10% (1/10) (P=0.13) in non-diabetic twins (fig 2). In non-diabetic twins 31% (11/35) and 40% (4/10), respectively, were positive for more than one autoantibody (fig 2). A similar calculation for the diabetic twins was not done because they had been treated with insulin for many years. Overall, 77% (27/35) and 70% (7/10) of the non-diabetic dizygotic and monozygotic twins were positive for at least one type of islet cell autoantibodies, respectively, compared with 87% (32/37) (P=0.367) and 100% (26/26) (P=0.0168) of the diabetic dizygotic and monozygotic twins, respectively (fig 2).
Twin studies are a powerful tool to investigate the environmental and genetic contribution to the development of islet cell autoimmunity. Most twin studies conducted to date comprise mostly monozygotic twin pairs and have not been population based.2 3 4 Furthermore, non-population based studies are subject to potential selection bias in terms of clinical symptoms or pairwise concordance. In contrast, the twins in this study were ascertained from a population based twin register and independently of zygosity and diabetes status in the twin partner.1 Selection bias should thus have been avoided, and this is supported by the fact that the distribution of the monozygotic and dizygotic twin pairs was as expected within the general population.1
In the general population the prevalence of autoantibodies to islet cell antoantigens is only a low percentage20 24–for instance, the prevalences in a recent study that analysed a large number of children (n=415) were 4%, 3%, and 4% for islet cell antibodies, insulin autoantibodies, and glutamic acid decarboxylase autoantibodies, respectively.24 In this study, however, as many as 70% (7/10) of the non-diabetic monozygotic twins had at least one of the three autoantibodies (fig 2), despite a rather long onset of disease in the affected twin, thus confirming the results from a recent report.4 In this study 11 monozygotic twins discordant for insulin dependent diabetes (discordance for 8-39 years) with normal ß cell function were analysed for islet cell autoantibodies–that is, islet cell antibodies, insulin autoantibodies, glutamic acid decarboxylase autoantibodies, and ICA512. Thirty six per cent were persistently positive for one or more autoantibodies in addition to 27% being transiently positive, thus overall 63% showed sign of islet cell autoimmunity. The prevalence of autoantibodies is thus much higher than that in first degree relatives of patients with insulin dependent diabetes, in whom these autoantibodies are found in prevalences of about 5%12 13 14 15 16; even combined they will be less than 15%,16 suggesting that the presence of islet cell autoimmunity might be genetically determined. Surprisingly, our new finding that the non-diabetic dizygotic twins have a similar high prevalence of islet cell autoantibodies (77% (27/35), fig 2) indicates that the presence of islet cell autoimmunity is probably determined by environmental rather than genetic factors. The observation that the presence of antibodies is high in unaffected twin partners and independent of zygosity suggests that shared environment rather than shared genes initiates autoimmunity. Accordingly, the period over which the twin partners within a pair are equally exposed to the same environment–that is, during fetal life or soon after birth–may be very important for the subsequent risk of developing insulin dependent diabetes.
Environmental exposure leading to islet cell autoimmunity can take place during fetal life or soon after birth. If the environmental exposure takes place after birth a negative correlation would be expected between the presence of islet cell autoantibodies and the time between the delivery of siblings who develop diabetes and their non-affected siblings. Thus, the longer the siblings are exposed to the same environment (because of a short time interval between deliveries) the higher the prevalence of autoantibodies. In a large study of first degree relatives of patients with insulin dependent diabetes (n=752) less than 15% were positive for islet cell autoantibodies, but there was no correlation between the time interval between deliveries and the presence of these autoantibodies (M Knip, personal communication). This might suggest that the period of shared environment after birth is less important for the development of islet cell autoimmunity and that the events resulting in the high prevalence of islet cell autoantibodies in both diabetic and non-diabetic twins probably take place during fetal life. Not only is the prevalence of islet cell autoantibodies higher in the non-diabetic dizygotic twins than in first degree relatives of diabetic patients12 13 14 15 16 but the cumulative concordance or recurrence risk of insulin dependent diabetes up to the age of 40 years in dizygotic twin pairs is twice as high as in ordinary first degree relatives of patients up to the same age.24 The prevalence of islet cell autoantibodies is thus much higher than the prevalence of insulin dependent diabetes,17 20 21 25 26 27 indicating that islet cell autoimmunity may occur without progression to clinical diabetes. We suggest that a continuum of events is required for islet cell autoimmunity to progress to ß cell destruction and clinical insulin dependent diabetes, such that a non-pathogenic immune response against islet cell antigens could be promoted by, for example, a viral infection, to a more pathogenic cellular immune response against islet cell antigens. This speculation is supported in humans by epidemiological studies showing that viral infections are correlated to subsequent development of insulin dependent diabetes.28 29 As the initiation of autoimmunity seems to be environmentally determined, promotion of the autoimmune process leading to clinical insulin dependent diabetes must be genetically controlled to account for the well established strong associations with markers from the HLA complex and other loci.
It should be noted that, although the non-diabetic twins had a high prevalence of islet cell autoimmunity, the levels were generally lower than in the diabetic twins (fig 1). Combinations of more than one autoantibody–for instance, glutamic acid decarboxylase autoantibodies and islet cell antibodies–were also more common in twins with insulin dependent diabetes than in those without (fig 2), an observation similar to that in studies in siblings, in which the presence of multiple autoantibodies greatly increases the risk of developing insulin dependent diabetes.14 15
We conclude that the development of autoantibodies is determined by environmental rather than genetic factors. Furthermore, the higher prevalence of autoantibodies in non-diabetic monozygotic and dizygotic twins compared with other first degree relatives of patients with insulin dependent diabetes suggests that the period over which the twins are exposed to the same environment–that is, during fetal life–is aetiologically important for induction of islet cell autoimmunity. Progression to clinical insulin dependent diabetes must, however, depend on genetically controlled responses to further exposures.
We are indebted to Alistair Williams, Dorte Svendson, Else Jost Jensen, and Elise Beck-Nielsen for technical assistance.
Funding: KOK received grants from the Danish Diabetes Association, the Danish Medical Association Research Fund, the Poul and Erna Sehested Hansen Foundation, the Foundation for Promotion of Medical Sciences, the Nordic Insulin Foundation Committee, the P Carl Petersens Foundation, and the Novo Foundation. JSP is supported by the Juvenile Diabetes Foundation.
Conflict of interest: None.