Institute of Cell and Molecular
Science
Centre for Diabetes & Metabolic
Medicine
Bart’s & The London, Queen Mary’s
School of Medicine & Dentistry
INSTITUTE OF CELL AND MOLECULAR
SCIENCE
4 Newark Street
London E1 2AT
T: +44 020-7882-2482
F: +44 020-7882-2186
A study of Epigenetics of type 1 diabetes using identical Twins
HOW THE STUDY BEGAN
This study of diabetic identical twins started in 1967, at Kings College Hospital Diabetic clinic. In
1992 the study moved to Royal London and St Bartholomew’s Medical and Dental School, London.
Britain is an ideal place for such a diabetic twin study because: 1) it has a large population in a small
area and people move house infrequently (compared, for instance, with the USA), 2) all patients are
part of the National Health Service so there is no financial bar to patients attending the clinic and 3)
doctors and patients co-operate readily with the clinic. Many doctors and patients around the country
have contributed to this study and are still doing so.
PRESENT STATE OF THE TWIN STUDY
The diabetic twin study is an integrated programme of clinical and scientific research in which the
clinical observations direct the scientific projects. The research team includes one senior lecturer, one
lecturer, three biochemists, a technician and a research co-ordinator. The study is also a collaborative
programme and therefore involves international and British scientists employing the most advanced
technology.
This is the largest on-going twin study in the world. There have been other twin studies in Germany,
Scandinavia and the United States but none is more than a quarter of the size and in most, the twins are
not under continuing study. The continuity of the British twin study has turned out to be one of its more
important features.
Presently 501 pairs of twins are studied of which at least one twin in each pair has diabetes. Of these
331 pairs are identical twins, of which 206 pairs were classified as having Type 1 or Insulin-Dependent
Diabetes Mellitus (IDDM) and 117 pairs as Type 2 or Non-Insulin Dependent Diabetes Mellitus
(NIDDM); the remainder can not be accurately classified. Ten to twelve new twin pairs join each year.
WHY THE STUDY OF DIABETIC TWINS IS IMPORTANT
Identical twins are a pair of individuals with the same germ line genes - no other individuals are exactly
alike; identical twins are. Similarities between identical twins might be due to shared genetic or
environmental factors but differences must be non-genetically determined. Identical twin studies have
therefore been most valuable where they have identified differences between twins for either a disease
or a feature of that disease. Since the unaffected co-twin is at increased risk of developing the disease it
is also possible to study the disease process in some twins before the clinical syndrome is manifested.
The considerable potential offered by the study of identical twins has led to major studies throughout
the world in a number of different diseases.
The causes of Type 1 diabetes may be due to either a factor in the environment, for example a virus, or
because the disease is inherited, or, an interaction of environmental factors and an inherited
predisposition. Identical twins have identical genes. So if one identical twin gets the disease and the
other does not then it is probable that the factor causing the diabetes must lie outside the genes and
probably in the environment. The ability to separate environmental factors from genetic factors is what
makes the twin study of crucial importance in trying to unravel the mystery of diabetes.
IMPORTANT FINDINGS OF THE DIABETIC TWIN STUDY
Of those twins referred to us where one has insulin dependent diabetes and the other is normal, 34% of
the normal twins will eventually develop insulin dependent diabetes, i.e become concordant. By regular
follow-up of these twins we are able to identify early markers of the disease process that lead to the
development of diabetes. Identification of these markers can then be used to screen the population at
risk and ultimately, prevention of the disease.
We have also discovered that some non-diabetic twins develop changes in the blood linked to the
eventual development of insulin dependent diabetes. However, for some reason, as yet unexplained, the
abnormalities found in the blood could disappear and the twins remain normal. If we could discover
why some non-diabetic identical twins can combat the disease process successfully, while others
cannot, then we would be well on the way to understanding the cause of diabetes.
DIABETES IS NOT ONE DISEASE
As diabetes has some tendency to run in families, we expected to find that twin pairs would be
concordant, i.e. both twins of a pair would be diabetic. This was not the case. Initial studies in 1972
demonstrated that of the twin pairs who are children or young adult often only one twin was diabetic
while the other twin was normal and stayed normal (28 of 59 pairs diagnosed under the age of 40
years); in contrast, the adult diabetic identical twins invariably had a similarly affected co-twin (34 of
37 pairs diagnosed over 40 years of age). These observations were seminal in identifying the two major
types of diabetes, Type 1 and Type 2 diabetes.
TYPE 1 DIABETES IS DUE TO NON-GENETIC FACTORS
Actuarial studies of 49 non-diabetic co-twins of recently diagnosed Type 1 diabetics over 21 years
(mean follow-up 9 years) demonstrated in the majority (66%) only one twin of a pair develop diabetes.
This observation indicates the importance of non-genetic factors in causing this disease. Of the 217
pairs under observation 99 are currently discordant for diabetes. If the critical period of exposure or
susceptibility to the environmental factor occurred over a prolonged period we would expect the
disease rate to increase with time. It does not; when both twins of a pair do develop diabetes, 63% do
so within five years of each other and 94% within 12 years. It is probable therefore that the critical
period leading to the onset of the disease process is brief and in the majority occurs in early childhood.
However, there is no evidence that the twin who developed diabetes had been exposed to a virus
infection while their co-twin had not. In addition it is clear that even in the twins who remain
discordant for diabetes these twins have a genetic susceptibility to the disease. Certain genes, called
HLA genes, on the sixth chromosome are associated with Type 1 diabetes. HLA-DR 3 or 4 occurred in
86% of long-standing discordant pairs and 99% of concordant pairs as compared with 59% of controls.
PROCESS LEADING TO TYPE 1 DIABETES
We have studied 17 non-diabetic twins before they developed Type 1 diabetes. These twins showed
extensive abnormalities including changes in immunity (islet cell complement-fixing and cytoplasmic
antibodies, insulin autoantibodies, antibodies to glutamic acid decarboxylase and activation of Tlymphocytes), metabolism (glucose intolerance, increased proinsulin levels and decreased insulin
responses to glucose) and growth (growth delay) in the pre-diabetic period. Characterisation of these
changes has helped us to define the process leading to the disease and even to predict it. Further
definition of these changes at the molecular level are underway.
ABNORMALITIES WHICH DO NOT LEAD TO TYPE 1 DIABETES
Surprisingly some of the non-diabetic twins show changes in immunity, metabolism and growth which
do not lead to diabetes. Of twins who are now unlikely to develop diabetes because they are more than
12 years from the diagnosis of the index twin transient changes were observed in 13 with islet cell
antibodies, 4 with impaired glucose tolerance and one with growth delay. Thus a process which in
some twins leads to Type 1 diabetes can in others remit spontaneously without diabetes developing. If
we could understand how this remission occurs then we might be able to prevent the disease. Persistent
changes in beta cell function including increased proinsulin levels were observed in non-diabetic twins
of long-standing, consistent with a biochemical "scar". We are currently designing studies with the aim
of preventing type 1 diabetes and initial trials are underway.
ABNORMALITIES IN TYPE 1 DIABETES WHICH ARE INHERITED
Non-diabetic identical twins, who remain unaffected more than 12 years from the diagnosis of the
index twin are very unlikely to ever become diabetic. We can therefore identify a unique cohort of
individuals (56 non-diabetic twins currently) who are genetically susceptible to Type 1 diabetes but
unlikely to develop the disease. If an abnormality in Type 1 diabetes is inherited, then it should also be
found in non-diabetic twins. We have used this technique to demonstrate that a number of immune
changes in Type 1 diabetes may be inherited while other changes are acquired; thus identical twins tend
to be similar for low complement 4 levels and haemolytic function, decreased T-lymmphocyte
cytotoxic/suppressor phenotype, decreased natural killer cell numbers irrespective of whether they are
diabetic or not. In contrast, identical twins can be discordant for islet cell antibodies and activated Tlymphocytes expressing the HLA-DR antigen and consistent with these changes being non-genetically
determined.
TYPE 2 DIABETES IS INHERITED
In contrast to Type 1 diabetic identical twin pairs, the majority of pairs in which one twin has Type 1
diabetes are concordant for the disease. Prospective studies of these twins demonstrated that all pairs
became concordant within 15 years of the diagnosis of the index twins, 94% become concordant within
10 years and 65% within 5 years of the index twin. This striking concordance rate contrasts with
diabetic non-identical twins in that only 10% of those co-twins also have diabetes. At present 107 twin
pairs in which the index twin has Type 1 diabetes are under observation of which 22 are discordant for
diabetes. These observations suggest that Type 2 diabetes is largely inherited. The likelihood that the
non-diabetic co-twin of a Type 2 diabetic will eventually develop diabetes has enabled us to study
these twin pairs in order to ascertain whether abnormalities of insulin secretion or insulin sensitivity are
the primary abnormality in this disease
COMPLICATIONS OF DIABETES ARE NOT INHERITED
The twin pairs are a perfect test bed to ascertain whether a complication is due to diabetes and, if so,
whether it is non-genetically determined. We have used the diabetic twins to demonstrate that
glycosylated haemoglobin is not an inherited phenomenon since in 16 twin pairs elevated levels were
only found in the diabetic twin; diabetic retinopathy is non-genetically determined since 5 twins had
severe proliferative retinopathy while their co-twin had no retinopathy; that diabetes can cause growth
reduction with diabetics tending to be shorter than their non-diabetic co-twins after many years
diabetes (mean difference in 12 twin pairs 2 and 1/4 inches) but also at diagnosis suggesting an effect
on growth during the pre-diabetic period. The absence of basement membrane thickening in all of eight
non-diabetic twins of long-standing diabetics is consistent with evidence that diabetic microangiopathy
is not inherited. We have recently demonstrated that many of the risk factors for cardiovascular disease
are strongly inherited in Type 1 diabetes and that Type 1 diabetes itself only adversely affects systolic
blood pressure and fibrinogen levels. Finally we have demonstrated changes in heart function, in
particular left ventricular diastolic function, associated with diabetes and consistent with the existence
of a specific diabetes induced disorder of the heart muscle.