Who is at risk of developing type 1 diabetes?

If one child in a family has type 1 diabetes, we know that other family members have a slightly higher risk of also developing diabetes than people with no family history of the condition. By long term study of the very large number of families taking part in the BOX study we are able to look at the differences between the small number of relatives who do go on to develop diabetes, and the large number who do not.

Researchers from the BOX study and its predecessor, the Bart’s Windsor family study, were one of the first to recognise that the detection of Islet autoantibodies is a sign of the activation of the immune system against insulin-making cells. These autoantibodies can be detected many years before a person develops the typical symptoms of diabetes. The current record time between first detection of autoantibodies and the onset of diabetes in the BOX study is 25 years.

We have been able to work out ways to combine tests to predict very accurately which family members are likely to develop diabetes based on a blood test. We showed that the number of antibodies gives an excellent measure of risk. Someone with no antibodies has a very low risk of getting diabetes, while finding three or more of the four markers indicates very high risk, Diabetes 1997. 46:1701-10.

We have shown that these markers are useful for predicting disease many years in advance, Diabetes Care 1999. 22: 2049-54. These combined tests are now in use all over the world to identify people to take part in prevention trials. We have confirmed the value of this approach in the large number of participants in the European Nicotinamide Diabetes Intervention Trial (ENDIT) and have shown how antibodies with can be combined with tests of insulin secretion and glucose tolerance to pick out the small number of relatives who are at extremely high risk, Diabetologia 2006. 49: 881-90.

We are also collaborating with other researchers across the world to develop better ways to measure antibodies and to combine results to improve prediction even further. As yet, there is no effective way to prevent type 1 diabetes, so these tests are still research tools.

Genetics

Type 1 diabetes is most strongly associated with genes in the Human Leukocyte Antigen (HLA) region on chromosome 6 that are concerned with the way the immune system works.

In the BOX study, we have shown that 52 per 1,000 children with the highest risk combination of HLA class genes would be expected to develop diabetes by age 15, compared with three per 1,000 who do not carry the high-risk genes. This information helps us prepare the way for screening if an effective prevention is identified in the future, J Clin Endocrinol Metab. 2004. 898: 4037-43.

We have compared the genes from people with type 1 diabetes in the BOX study with genes from people who had developed type 1 diabetes more than 50 years ago, when diabetes was much less common. We showed that a ‘lower dose’ of the diabetes-related genes is required to develop type 1 diabetes today than in the first half of the twentieth century. The results show that the current increase in type 1 diabetes is due to exposure to changing environmental factors, Lancet. 2004; 364:1699-700. Initial analysis of genetic material from our more recent mouth brush collection suggests that the trend for people developing diabetes with a ‘lower dose’ of diabetes-related genes is continuing.

One of the theories proposed to explain the recent rise in childhood asthma is that babies now live in a very clean environment thereby being less exposed to childhood infections and more likely to develop allergies and autoimmune conditions, Diabetologia. 2002. 45: 588-94.

We have looked at whether alterations or changes in the frequency of Natural Killer (NK) cell activity is altered in people with type 1 diabetes compared with healthy controls. NK cells are the first cells to detect viral infection. We have found that certain combinations of NK cells receptors are more common in children who develop type 1 diabetes early in life, Diabetologia. 2011. 54: 3062-70.

We looked at a group of children who have high susceptibility genetic markers HLA DR3-DQ2/DR4-DQ8 within BOX. We found that their chance of developing type 1 diabetes in the future was further increased if their sibling with type 1 diabetes had been diagnosed at a young age. Early onset of diabetes in the proband is the major determinant of risk in HLA DR3-DQ2/DR4-DQ8 siblings’, Diabetes 2014. 63: 1041-7.

Incidence of type 1 diabetes in BOX

We have monitored how many children in the Oxford region have developed diabetes since 1985.

‌In 1985, 17.4 per 100,000 children aged less than 15 were diagnosed and their average age at diagnosis was 10. By 2004, the incidence was 26.5 per 100,000 and their average age was nine years. The increase seen in all age bands under 15 years was equivalent to 2.3 per cent per year.

Over the twenty years, the cumulative incidence of diabetes by age five increased five-fold from 0.2 to 1.0 cases per 1,000, and by age 15 from 2.5 to 4.2 cases per 1,000. In the first 10 years of the study, the increase was most rapid in children under five, BMJ 1997 vol 315, 713-717, (12 per cent per year) but this slowed to 0.8 per cent per year during 1995-2004. In contrast, in those aged 10-14 there was no rise in incidence from 1985-94 but an increase of 3 per cent per year from 1995-2004.

There are now twice as many children starting insulin under the age of five in the study as there were in the late 1980s.

The European Perspective

Childhood diabetes is much more common in some countries than others. For example, in Sweden more than 36 children per 100,000 develop diabetes each year, while in Macedonia it is only 5-6 per 100,000. The BOX study has worked with other groups to map the frequency of childhood diabetes right across Europe. The rate is generally very high in Scandinavia and low in Southern Europe, but there are some very interesting exceptions such as the Italian island of Sardinia, where diabetes is almost as common as it is in Finland. Further investigation of these differences should teach us more about the causes of diabetes.

We have also shown that childhood diabetes is becoming more common in almost every country, particularly in very young children. Lots of people have interesting theories as to why this could be, but nobody really knows, BMJ 1997. 20: 315:713-7 and Diabetologia 2012. 55: 2142-7.

Maturity-onset diabetes of the young (MODY) in BOX

This relatively rare form of diabetes is inherited and usually family members in several generations have developed diabetes as children or young adults. Several forms of MODY occur as a result of genetic mutations. Diagnosing MODY is important in diabetes treatment. In some cases it may be possible for an individual to control their diabetes with tablets rather than insulin. Some forms of MODY are caused by mutations that are transmitted from parent to child so that each child has a 50 per cent chance of inheriting the mutation. The children of people with MODY have therefore a much higher risk of diabetes than the children people with type 1 diabetes.

MODY can be missed at diagnosis and thought to be type 1 diabetes, as the two forms of diabetes often present at a similar age in childhood. We researched BOX data for any such cases within the study. We checked for evidence of a strong family history of type 1 diabetes, i.e. three or more generations on the same side, either all maternal or all paternal. MODY3 results from mutations in a gene called HNF-1 alpha and in the BOX study we found one family with this form of MODY. This resulted in two members of the family stopping insulin treatment and starting oral tablet treatment. Therefore even in families with such a strong family history, MODY is very uncommon, Diabetes Care 2003. 26: 333-7.

Diagnosis of Diabetes in the very young

We have initiated the first national collection of children diagnosed with diabetes under the age of two years including many children participating in BOX. We also re-established the British Diabetic Association (BDA) 1972-1981 cohort, a collection of individuals diagnosed under the age of two years and obtained samples from 58 individuals who have also a cohort of children diagnosed under the age of two. In collaboration with Professor Andrew Hattersley’s group in Exeter, which has developed very accurate tests for non-autoimmune forms of diabetes infants, we have shown that children under the age of six months diagnosed with diabetes are very unlikely to have type 1 diabetes. These children almost definitely have other genetic forms of diabetes that can often be treated with tablets. All children diagnosed with diabetes under the age of six months should be screened for non-autoimmune forms of diabetes, Diabetes. 2006. 55:1895-8.