Nutritional risk factors in the development of type 1 and type 2 diabetes

Type 1 and type 2 diabetes have been recognised as two different disease entities since the early 1970s. Quite different nutritional factors are thought to influence the development of these two types of diabetes. Type 1 diabetes is characterised by progressive beta-cell destruction, which leads to complete insulin deficiency. In children there is epidemiological evidence that an early introduction of cow’s milk and/or cereals, high intake of nitrites and N-nitroso compounds or high linear height and weight gain increase the risk of type 1 diabetes, whereas breastfeeding, vitamins C, D and E, and zinc may protect from this disease. Type 2 diabetes is characterised by insulin resistance and impaired insulin secretion at the time of appearance of hyperglycaemia and clinical diabetes. Obesity, particularly of abdominal type, and sedentary lifestyle are well-known risk predictors of type 2 diabetes. High (saturated) fat intake seems to be related to insulin resistance, obesity and increased risk of type 2 diabetes, whereas diet high in fibre may protect from insulin resistance, glucose intolerance and diabetes.

Type 1 diabetes is considered an immune-mediated disease, in which signs of beta-cell autoimmunity can be detected at variable times before the diagnosis of linical disease (Knip 2002). Large geographical differences in incidence and linearly increasing incidence seen in many countries during the last five decades cannot be explained solely by genetic factors (Onkamo et al. 1999; Karvonen et al. 2000; Green et al. 2001). The relatively low concordance of identical twins also confirm the important role of environmental factors in the aetiology of this disease (Barnett et al. 1981). So far there is little firm evidence on the role of nutritional factors. Breastfeeding, vitamins C, D and E, nicotinamide and zinc have been reported as possibly protecting from type 1 diabetes, whereas N­nitroso compounds, cow’s milk, some cereals, increased linear growth, and obesity may increase the risk (Virtanen & Knip 2003).

Birth height and weight of children who later developed type 1 diabetes have been similar to the controls in most of the studies (Virtanen & Knip 2003), although in some studies cases have been longer and weighted more than controls at the time of birth (e.g. Dahlquist et al. 1999). In the only cohort study available, higher birth weight was related to an increased risk of type 1 diabetes (Stene et al. 2001). Increased height gain during childhood seems to be related to greater risk of type 1 diabetes (e.g. Blom et al. 1992; Price et al. 1992). Higher weight gain in infancy is consistently related to greater risk of type 1 diabetes according to case-control evidence (e.g. Baum et al. 1975; HyppoÈnen et al. 1999), whereas the findings on the role of weight gain after infancy are inconsistent (e.g. Blom et al. 1992; HyppoÈnen et al. 2000). Clearly results from cohort studies are awaited to settle the putative importance of height and weight gain in the development of this disease.

Maternal diet and composition of breast milk may play a role in the development of immune-mediated diseases. It has been shown that small amounts of cow’s milk proteins may be carried over to breast milk from the maternal diet (Axelsson et al. 1984), and sensitive infants may develop cow’s milk allergy on exclusive breastfeeding (Hùst 1994). Per capita coffee consumption correlated positively with incidence of type 1 diabetes in an international ecological comparison (Tuomilehto et al. 1990). However, maternal coffee or tea consump­tion during pregnancy was not related to the risk of type 1 diabetes in the offspring in two case-control series (Virtanen et al. 1994a; SolteÂsz et al. 1994). A positive association was seen between maternal nitrite intake and the risk of diabetes in the child independently of the child’s own intake and when adjusted for several sociodemographic factors (Virtanen et al. 1994b). Paternal use of coffee or tea or intake of nitrate or nitrite at the time of conception was not related to the risk of diabetes in the offspring (Virtanen et al. 1994a,b). In a case-control study maternal supplementation of cod liver oil during pregnancy was inversely related to the risk of type 1 diabetes in the offspring, suggesting that either vitamin D, vitamin A or n-3 fatty acids, which are all abundant in cod liver oil, play a role in the development of this disease (Stene et al. 2000).

Whether breastfeeding protects from or early introduction of supplementary foods causes type 1 diabetes, remains unsolved, although these aspects of the diet have received more research attention than many other areas in the etiology of type 1 diabetes (Virtanen & Knip 2003). Findings from prospective studies with enough statistical power are awaited.

Putative protecting effects of breastfeeding could be due to protection against infections provided by breast milk through, for example, secretory IgA antibodies and enhancement of the infant’s own immune responses, increased beta-cell proliferation (Juto 1985), or delayed exposure to foreign food antigens. Breast milk contains cytokines and growth factors, which affect the maturation of the gut-associated lymphoid tissue (GALT) (Srivastava et al. 1996).

An early introduction of cow’s milk-based infant formulas and other cow’s milk products may increase the risk of type 1 diabetes according to case-control evidence, although the results remain inconclusive (Virtanen et al. 1991; Virtanen & Knip 2003). An early introduction of cow’s milk or a short exclusive breastfeeding were not related to early stages of beta-cell autoimmunity in birth cohort studies of individuals with increased genetic risk of type 1 diabetes (Norris et al. 1996; Couper et al. 1999; Hummel et al. 2000; KimpimaÈki et al. 2001; Norris et al. 2003; Ziegler et al. 2003), but inversely to the development of four type 1 diabetes-associated autoantibodies out of the four studied (KimpimaÈki et al. 2001). The findings from a pilot study of the only randomised trial available suggest that beta-cell autoimmunity can be prevented or delayed by giving hydrolysed infant formula instead of regular cow’s milk-based one (AÊ kerblom et al. 1999).

Several theories try to explain the putative diabetogenicity of cow’s milk (Knip & AÊ kerblom 1998). Early immunisation to bovine insulin may be related to the development of beta-cell autoimmunity (Vaarala et al. 1999). Abnormal tolerance development has been observed in those infants who develop early signs of beta-cell autoimmunity (Vaarala et al. 1999). The putative diabetes-promoting effects of dietary antigens may be mediated through GALT. Food proteins may induce beta-cell autoimmunity also because of changes in gut permeability due to microbial infections. Greater weight gain related to greater intake of energy has been observed in infant formula-fed compared with breast-fed infants from 3 months of age (Heinig et al. 1993). By increasing insulin demand, increased weight gain caused by supplementary feeding could be a contributory factor in the development of diabetes. An early exposure to cow’s milk and rapid growth in infancy were both independent risk factors of childhood type 1 diabetes in one case-control series (HyppoÈnen et al. 1999).

Recently, it was suggested that exposure to gluten-containing cereals and rice at the age of 4 to 6 months would protect from development of early beta-cell autoimmunity compared to earlier or later exposure (Norris et al. 2003). German birth cohort findings related early gluten exposure to development of early beta-cell autoimmunity (Ziegler et al. 2003).

Cow’s milk consumption may be diabetogenic also during childhood according to case-control and cohort findings (Verge et al. 1994; Virtanen et al. 1998, 2000). In an Australian case-control study, cereal consumption was positively related to the risk of diabetes, although the association disappeared after adjust­ment for other dietary factors (Verge et al. 1994). The cell-mediated immune response to gluten was detected more frequently among newly diagnosed children with type 1 diabetes than among controls (Klemetti et al. 1998).

Case-control studies suggest that dietary N-nitroso compounds (Dahlquist et al. 1990) and nitrite (Dahlquist et al. 1990; Virtanen et al. 1994b) increase the risk of type 1 diabetes in children. Also mother’s intake of nitrite at the time of pregnancy was positively related to the risk of type 1 diabetes in children independently of child’s nitrite intake (Virtanen et al. 1994b). Nitrate is a naturally occurring compound in vegetables. Nitrate and nitrite are both used as food additives in the processing of meat products. In food and the human gastrointestinal tract nitrate is reduced to nitrite by bacteria, and N-nitroso compounds are formed from nitrite in the chemical or bacterial nitrosation reaction with amino compounds (Slorach 1981). Vitamin C and alpha-tocopherol inhibit and thiocyanate ions accelerate the formation of N-nitroso compounds (Leaf et al. 1989).

Recently Bafilomycin A1, a toxin produced by Streptomyces species in soil, was shown to induce glucose intolerance and reduction in pancreatic islet size in mice (Myers et al. 2001). Streptomyces species can infest tuberous vegetables such as potatoes and beet.

Low groundwater zinc was associated with an increased risk of type 1 diabetes in a case-control series (Haglund et al. 1996). A low concentration of alfa-tocopherol in serum was related to increased risk of type 1 diabetes in a nested case-control study in adults (Knekt et al. 1999). Vitamin C intake from diet was not associated with type 1 diabetes in a Swedish case-control series (Dahlquist et al. 1990), whereas in an Australian one there were fewer users of vitamin C supplements among cases than among controls (Glatthaar et al. 1988). Vitamin D has been shown to prevent the development of insulitis and autoimmune diabetes in non-obese diabetic (NOD) mice (Mathieu et al. 1992, 1994). Vitamin D has immunosuppressive effects. Vitamin D supplementation during early infancy may protect from type 1 diabetes according to a recent case-control study from several European countries (EURODIAB Substudy 2 Study Group 1999) and fish liver oil use during pregnancy may protect the child from type 1 diabetes (Stene et al. 2000). In the prospective Northern Finland mother±child cohort, both the use of vitamin D supplementation during infancy and the dose of supplementation were inversely associated with the risk of type 1 diabetes, whereas a positive association was observed between suspected rickets and risk of diabetes (HyppoÈnen et al. 2001). It should be noted that the recommended dose of vitamin D supplementation was high at that time (in 1966), 2500 IU i.e. five times the current recommendation.

Obesity and sedentary lifestyle are well-established risk determinants of type 2 diabetes (Costacou & Mayer-Davis 2003). Recently two randomised trials provided evidence that type 2 diabetes can be prevented by changes in diet and in exercise pattern in men and women at high risk of the disease (Tuomilehto et al. 2001; Diabetes Prevention Program Research Group 2002). The beneficial role of dietary fibre and a harmful one of saturated fatty acids seem to be rather well shown, whereas evidence is contradictory as to whether dietary or supplementary antioxidant vitamins or minerals prevent the disease. Discussion on optimal proportions of fat and carbohydrate in the diet for the prevention of type 2 diabetes continues. Attention is increasingly paid to different types of fat and carbohydrate. Also, attempts to individualise the advice according to metabolic status have decreased the discrepancies in views (Grundy et al. 2002).

Both overall and abdominal obesity have deleterious effects on insulin sensitivity and insulin secretion and are risk predictors of impaired glucose tolerance and clinical type 2 diabetes (reviewed in Feskens 1992; Virtanen & Aro 1994; Costacou & Mayer-Davis 2003). Already a small, sustained decrease in weight improves insulin sensitivity and decreases the risk of type 2 diabetes (Tuomilehto et al. 2001; Diabetes Prevention Program Research Group 2002). The putative effects of dietary factors other than energy overload (such as fibre, glycaemic load, proportion of fat, type of dietary fatty acids) in the aetiology of overall and abdominal obesity remain to be settled.

Overweight and obesity have during the recent decades increased alarmingly in various populations worldwide, also among children and adolescents (e.g. Kautiainen et al. 2002; Jolliffe 2004).

High total and saturated fat intake has been linked to the development of impaired glucose tolerance and type 2 diabetes by the findings of several long-term cohort studies (Marshall et al. 1994; Feskens et al. 1995; van Dam et al. 2002). In the Health Professionals’ follow-up study, the significance disappeared after adjustment for body mass index (van Dam et al. 2002). However, obesity may well be in the causal pathway between fat intake and development of type 2 diabetes (Bray & Popkin 1998). In two large women cohorts, with adjustment for body mass index, total and saturated fat intake was not associated with the development of type 2 diabetes (Meyer et al. 2001; SalmeroÂn et al. 2001). In US women with a high intake of vegetable fat the risk of developing type 2 diabetes was reduced (Colditz et al. 1992; Meyer et al. 2001). Monounsaturated fat has beneficial effects on glucose tolerance and insulin resistance. The putative protective effect of fish/fish oils on the risk of developing glucose intolerance remains to be settled (Feskens et al. 1991; Bhathena et al. 1991).

Compared with fat carbohydrate feeding consisting of the same amount of energy may induce higher postprandial plasma glucose, insulin and triglyceride and lower HDL cholesterol concentrations (Costacou & Mayer-Davis 2003). However, the effects of various types of carbohydrate are different and difficult to separate. The proportions of monounsaturated fat and carbohydrate in the diet can depend on an individual’s cultural as well as on metabolic status, e.g. hyperlipidaemic individuals should receive more energy from monounsaturated fat than from carbohydrates, whereas carbohydrate intake should be increased and fat intake decreased in individuals attempting to lose weight (Costacou & Mayer-Davis 2003).

There is increasing evidence suggesting a beneficial effect of dietary fibre on insulin sensitivity (e.g. Manolio et al. 1991; Feskens et al. 1994; Vitelli et al. 1996; YloÈnen et al. 2003a), whereas the findings on the effect of dietary fibre on glucose tolerance and risk of type 2 diabetes remain somewhat controversial (Marshall et al. 1991; Colditz et al. 1992; Feskens et al. 1995; SalmeroÂn et al. 1997a,b). However, some recent cohort studies suggest that type 2 diabetes could be prevented by increasing intake of dietary fibre, especially non-soluble cereal fibre (SalmeroÂn et al. 1997a,b; Meyer et al. 2000; Montonen et al. 2003).

The glycaemic responses of foods and meals have been suggested to play a role in the development of obesity and type 2 diabetes, although not consistently so (SalmeroÂn et al. 1997a,b; Meyer et al. 2000; Pi-Sunyer 2002).

Oxidative stress may be implicated in the aetiology of diabetes (Ho & Bray 1999). Carotenoids and vitamins C and E are important components of the human defence system against oxidative stress (Stahl & Sies 1997). Studies in humans on the effects of vitamin E on glucose or insulin metabolism are controversial (e.g. Paolisso et al. 1993; Facchini et al. 2000; Sanchez-Lugo et al. 1997; YloÈnen et al. 2003b). Two prospective cohorts have provided evidence for a protective effect of vitamin E in the development of type 2 diabetes in non-supplement users (Salonen et al. 1995; Mayer-Davis et al. 2002). Fruit and vegetable intake have been inversely related to the development of type 2 diabetes in three cohort studies (Colditz et al. 1992; Feskens et al. 1995; Ford et al. 2001), providing indirect evidence on the putative protective effect of carotenoids. A long-term randomised trial did not show any effect of beta-carotene supplementation on the incidence of type 2 diabetes (Liu et al. 1999). In a long-term cohort serum beta-carotene was not related to the incidence of type 2 diabetes (Reunanen et al. 1998).

Cross-sectional findings on the associations between dietary and serum carotenoids and measures of glucose and insulin metabolism are inconsistent (e.g. Ford et al. 1999; Facchini et al. 2000; YloÈnen et al. 2003b). High dietary intake of vitamin C was related to a lower incidence of type 2 diabetes in a long­term cohort study (Feskens et al. 1995). Some minerals such as potassium, magnesium, chromium and zinc may affect glucose and insulin metabolism, but their putative effects on the development of glucose intolerance and type 2 diabetes remain to be elucidated (Costacou & Mayer-Davis 2003).

According to the so-called thrifty phenotype hypothesis, conditions such as disturbed nutrition during foetal time or early infancy could cause structural or functional changes in muscles, liver or pancreas and therefore predispose to later disorders of glucose and insulin metabolism (Hales & Baker 1992). Low birth weight and birth thinness reflect foetal growth disturbance and have been related to an adverse profile of later glucose and insulin metabolism and to an increased risk of type 2 diabetes according to most of the studies (Hales & Baker 1992; Newsome et al. 2003). Catch-up growth may be detrimental to long-term survival and may increase the risk of type 2 diabetes (Forsen et al. 2000; Hales & Ozanne 2003).

Short, exclusive breastfeeding or an early age at introduction of supplementary feeding has been suggested as a risk determinant of both type 1 and type 2 diabetes (Pettitt et al. 1997). Increased weight gain may be a risk predictor of not only type 2, but also of type 1 diabetes (e.g., HyppoÈnen et al. 2000; Wilkin 2001; EURODIAB Substudy 2 Study Group 2002). Overweight was associated with the presence of autoantibodies to glutamate decarboxylase (GAD) antibodies in unaffected male first-degree relatives of subjects with type 1 diabetes (Weets et al. 2001) and among glucose-intolerant men and women (Rolandsson et al. 1999). Evidence is inconclusive whether type 1 and type 2 diabetes overlap within families (e.g. Dahlquist et al. 1989; Quatraro et al. 1990; Douek et al. 2002).

Both types of diabetes are increasing rapidly in many countries in various parts of the world. For type 2 diabetes we already have well-established means to prevent or at least to delay the onset of the disease. Large health programmes are needed to promote exercise and diets high in fibre and low in saturated fatty acids and to prevent and decrease obesity in high-risk groups as well as in the total population. The preventive measures should start from childhood.

We cannot yet prevent type 1 diabetes. Research resources need to be allocated increasingly to study environmental risk determinants of this disease with high economical and human costs. If a dietary factor/s turned out to either prevent or cause type 1 diabetes, a safe and cost-effective way to prevent diabetes at the population level would be available.