Editor—Two twin studies of type 1 diabetes have reached opposite conclusions. In one, a population based cohort of Danish twins in which one or both cotwins had type 1 diabetes was studied for the presence of islet autoantibodies.1 High rates of autoantibody positivity were identified in twins with and without diabetes. Since positivity did not differ between the monozygotic and dizygotic twins it was suggested that a shared intrauterine or early postnatal environment might be more important than genetic factors. The second study found that the prevalence of islet autoantibodies was lower in initially unaffected dizygotic twins than monozygotic twins and did not differ from that found in unaffected non-twin siblings. The authors concluded that islet autoimmunity is determined predominantly by genetic rather than environmental factors.2
These conflicting findings might be due to differences in study design, differences between the two populations, or differences between the assays. The design of the studies differed greatly: the Danish study was population based and cross sectional, while the American one was clinic based and longitudinal. The American study also contained many more monozygotic twin pairs. The Danish study included all twin pairs regardless of diabetes status, whereas the American study included only those initially discordant for diabetes and did not analyse samples drawn after the development of diabetes. Both studies were comparatively small.
To resolve differences due to assay performance we exchanged serum samples; blinded samples were measured for insulin autoantibodies and glutamic acid decarboxylase and IA-2/ICA512 antibodies. Samples from all the Danish cases were exchanged, and samples from 34 of 53 monozygotic twins and 16 of 30 dizygotic twins from the American series were also measured. The assays used and their performance in international workshops have been reported elsewhere.3,4
Concordance between the assays was good. Since several American samples were not available for testing in both laboratories, the American assay was used for the comparison shown in the table, using data already published. The analysis was restricted to samples collected from unaffected twins, since affected twins were not analysed in the American study. Tests for islet cell antibodies were not repeated for this comparison, but IA-2/ICA512 antibodies are reported here for the first time in the Danish cohort.
The prevalence of autoantibodies was much higher in American than Danish discordant monozygotic twins. The likely explanation is that 12 of 13 antibody positive American twins subsequently developed diabetes and would be expected to have had autoantibodies before diagnosis. These individuals were classified as non-diabetic in the American series but would have been classified as diabetic in the Danish series. In contrast, the prevalence of autoantibodies was similar in the two groups of dizygotic twins, and the prevalence of autoantibodies and glutamic acid decarboxylase antibodies in the Danish twins was considerably lower than that in the previous report. The original analyses for islet autoantibodies and glutamic acid decarboxylase antibodies were performed blind in different laboratories. We attribute the discrepancy to improvements in assay performance and standardisation.
Reanalysis of the Danish cohort still shows no difference in the prevalence of islet autoantibodies between monozygotic and dizygotic twins, but the prevalence in both groups is much lower than that reported previously. Other differences, such as the contrast between autoantibody positivity in the monozygotic twin groups, can be attributed to study design and sampling procedures. Reanalysis of the Danish cohort does not exclude the shared uterine environment hypothesis, but the observation from the American study that discordant twins are no more likely than singleton infants to carry islet autoantibodies favours the genetic view.
Table.
Numbers of discordant twins positive for certain antibodies divided by numbers of samples tested
| Islet autoantibodies | Glutamic acid decarboxylase antibodies | IA-2/ICA512 antibodies | |
|---|---|---|---|
| Monozygotic: | |||
| Denmark | 0/10 | 1/10 | 0/10 |
| USA | 13/52 | 14/38 | 6/42 |
| Dizygotic: | |||
| Denmark | 1/35 | 2/35 | 0/35 |
| USA | 2/28 | 3/23 | 1/23 |
References
- 1.Petersen JS, Kyvik KO, Bingley PJ, Gale EAM, Green A, Dyrberg T, et al. Population based study of prevalence of islet cell autoantibodies in monozygotic and dizygotic Danish twin pairs with insulin-dependent diabetes. BMJ. 1997;314:1575–1579. doi: 10.1136/bmj.314.7094.1575. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Redondo MJ, Rewers M, Yu L, Garg S, Pilcher C, Elliott RB, et al. Genetic determination of islet cell autoimmunity in monozygotic twin, dizygotic twin, and non-twin siblings of patients with type 1 diabetes: prospective twin study. BMJ. 1999;318:698–702. doi: 10.1136/bmj.318.7185.698. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Yu L, Robles D, Abiru N, Kaur P, Rewers M, Kelemen K, et al. Early expression of antiinsulin autoantibodies of humans and the NOD mouse: evidence for early determination of subsequent diabetes. Proc Natl Acad Sci USA. 2000;97:1701–1706. doi: 10.1073/pnas.040556697. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Bingley PJ, Williams AJK, Gale EAM. Optimized autoantibody-based risk assessment in family members. Implications for future intervention trials. Diabetes Care. 1999;22:1796–1801. doi: 10.2337/diacare.22.11.1796. [DOI] [PubMed] [Google Scholar]