Main

Early studies of the concordance rate of IDDM in monozygotic twin pairs clearly demonstrate the importance of nongenetic risk factors for the disease(1, 2). More recently, a study of monozygotic and dizygotic twin pairs suggested that as much as 79% of the liability to contract diabetes if one twin already had the disease was due to a shared environment(3). Nongenetic risk factors may contribute to the causation of diabetes by initiating autoimmunity or by accelerating the autoimmune destruction of the β cell during the long prediabetic period(4). Physical stress such as high growth rate(5) and psychologic stress(6) are thought to accelerate or precipitate diabetes onset in children, probably through an increased peripheral insulin need.

Events that may initiate the autoimmune destruction of β cells most probably occur many years before the clinical onset of the disease(7–9). In childhood onset diabetic cases the initiating events may very well occur during the perinatal period of life, and it was recently suggested that IDDM occurring even late in life may be initiated very early(10). This would be an example of the popular theory that pre- or perinatal events may affect later disease susceptibility(11). In a population-based epidemiologic study we were able to show that a number of perinatal events were associated with an increased risk for childhood onset IDDM(12). Thus, a diagnosis of blood group incompatibility in the child was associated with a strong risk increase, whereas weaker associations were found for more unspecific events such as low gestational age, delivery by cesarean section, or diagnosis of any neonatal respiratory disease. It was not possible to disentangle the effects of the blood group incompatibility syndrome itself from treatment, such as blood transfusion and/or phototherapy. Common for all indicated perinatal risk factors is the treatment in the neonatal ward, usually in incubators.

In the present study, we have used NOD mice as a model for autoimmune diabetes to study whether treatments in the neonatal period including phototherapy will affect the incidence of diabetes in this genetically susceptible inbred mouse model.

METHODS

NOD mice were obtained form the Uppsala colony which had a relatively low cumulative incidence of diabetes(13) by courtesy of Decio Eizirik. The animals were kept under standard conditions and were fed standard laboratory food and water ad libitum, and strict brother-sister mating was maintained. Newborn litters were treated during the first few days of life in the following way: the newborn pups were taken from their mother for two 4-h periods during each of 5 successive days. Between the two periods, the pups were allowed to feed for 1.5 h. Some of the litters were treated with strong light (an ordinary phototherapy lamp borrowed from the neonatal ward at the Department of Pediatrics, Lund University Hospital), others did not receive such treatment (sham). Treatment resulted in a high mortality, at an average of 30%. Control litters were left with their mother. The mice were weaned at the age of 3 wk. To detect a possible effect on early weight development a group of litters similarly treated and untreated, respectively, were weighed at the time of weaning. Beginning at the age of 3 mo, each mouse was followed with urine analysis for glucose and, when positive, a blood sample was taken and blood glucose was measured using the ExacTech system. When definite diabetes had developed or, if no diabetes developed, at the age of 8 mo, the animal was killed, and the pancreas was dissected and fixed in Bouin's fluid. The pancreas was embedded in paraffin wax, and histologic sections were prepared and stained with hematoxylin and eosin.

The sections were read without knowledge of their origin, and the degree of insulinitis was scored with a 0-3 score. Ideally, 10 different islets were scored from each mouse, but in some instances, fewer islets were identified. The histologic evaluation was made only when six or more islets had been scored, and then an average score was calculated. When this was above 1, histology was regarded as definitely positive.

Three comparisons were made: the proportion of animals developing diabetes before the age of 8 mo, the average blood glucose level at the age of 8 mo among those who did not develop diabetes, and the mean histology score as just described.

Statistical comparisons were made using the StatExact program. Exact 95% confidence limits of odds ratios were calculated. Homogeneity between groups was tested with the Breslow and Day method, differences in rates withχ2 or exact tests, and means of glucose levels with t tests.

RESULTS

Effect on survival and growth. Treatment resulted in a high mortality, at an average of about 30%, similar in the two treatment groups. Control litters showed little mortality. The mean pup weight at the time of weaning was 8.9 ± 0.8 (n = 122) in the control group equal to 9.0 ± 0.7 g (n = 66) (mean ± SEM) in the experimental group.

Clinical diabetes. Table 1 shows the number of mice that developed clinical diabetes and were killed before 8 mo age. In males, the percentage of diabetic mice was highest in the sham group, in females in the light-exposed group. There is no heterogeneity between sexes, however (p = 0.27), and the joint odds ratio for sham versus light, stratified for sex, is 0.99 (95% C.I. 0.34-2.87). There is thus no difference between the two experimental groups and they were therefore added and compared with the control group.

Table 1 Clinical diabetes in mice according to sex and treatment
Full size table

The odds ratio for experimental versus control, stratifying for sex, is 3.42 (95% confidence interval, 1.57-7.74). The treatment, with or without strong light, thus increased the rate of overt clinical diabetes with clear statistical significance.

Histologic insulitis. Table 2 shows a comparison of positive and negative histology according to sex and experimental versus control group. As is obvious from Table 2, no effect on the histologic score is seen from the treatment. An odds ratio for treatment, startifying for sex, gives an odds ratio = 0.86 (0.45-1.64).

Table 2 Presence of definite histologic insulitis(“positive”) in mice according to sex and treatment
Full size table

Blood glucose level at mo 8. The blood glucose level was compared between animals killed at 8 mo of age without developing clinical diabetes. In untreated males, the mean blood glucose level was 4.73 ± 0.20, and in untreated females 3.83 ± 0.48. This difference is statistically significant (t = 2.08 at 89 d.f., p = 0.046). The blood glucose level in experimental males was nearly identical with that of control males (4.72 ± 0.43), and that of experimental females (2.89 ± 0.41) was lower than that of males (t = 3.31 at 86 d.f., p = 0.002), but there was no statistically significant difference in glucose level between experimental and control females(t = 1.27 at 85 d.f., p = 0.18).

Surviving females thus seem to have a lower glucose blood level at 8 mo than do males, but treatment did not increase it in either sex.

DISCUSSION

Studies in recent onset type 1 diabetes cases have indicated that psychologic stress can precipitate onset of the disease(14–16). A similar accelerating effect has been shown in the BB rat that had a significantly earlier onset of diabetes when exposed to different environmental stress situations from the age of 55 d(17). The NOD mice have been considered a good model of human autoimmune type 1 diabetes, because this strain develops insulitis spontaneously with progressive β cell destruction with diabetes onset between the 12th and 30th wk of age(18). The cumulative incidence of diabetes in these mice varies in different colonies but it is always much higher in females than in males. This sex difference has been shown to be partly due to sex steroids because ovariectomy induced a reduction in the percentage of diabetes, whereas orchidectomy increased the incidence of diabetes in these mice(19). In the NOD mice it has also been shown that serum glucocorticoid levels under basal conditions are higher in females than in males and immobilization stress at 2 mo of age also results in higher glucocorticoid values in females than in males(20). Thus environmental stress through higher glucocorticoid levels might be associated with a higher risk for IDDM incidence in adult NOD mice.

In the present study we used a colony of NOD mice with a relatively low percentage of overt diabetes in the untreated group but still with a significant sex difference. Only 1% of males and 22% of females developed clinical diabetes even though histologic insulitis existed in 53% of males and in 75% of females. The use of such a strain increases the possibility to demonstrate an enhancing effect of environmental factors on diabetes development. The present study demonstrates, for the first time, that separation from the mother in early life can cause a significantly increased rate of overt diabetes in the NOD mice. The exposure of the newborn mouse to phototherapy in the same form as used for treatment of neonatal jaundice in children did not further increase the risk. Because the frequency of insulitis was similar in treated and untreated animals, the effect of the treatment seems to be affecting either long-term peripheral insulin need or the sensitivity of the β cell to an immune attack rather than the immune response per se. The former possibility accords with the hypotheses proposed by Barker et al.(11) saying that the early events already from intrauterine life might affect metabolic programming including insulin sensitivity. Because weight at weaning was similar between control pups and a group of pups treated exactly as the experimental group, the effect is probably not due to starvation. It is also possible that the treatment may have resulted in more severe histologic changes, but that the histologic scoring method may be too crude to identify this, due to the large variation in insulitis between different insula in the same animal.