Intake of antioxidant vitamins and trace elements during pregnancy and risk of advanced cell autoimmunity in the child 1 3

Transcription

1 Intake of antioxidant vitamins and trace elements during pregnancy and risk of advanced cell autoimmunity in the child 1 3 Liisa Uusitalo, Mike G Kenward, Suvi M Virtanen, Ulla Uusitalo, Jaakko Nevalainen, Sari Niinistö, Carina Kronberg-Kippilä, Marja-Leena Ovaskainen, Liisa Marjamäki, Olli Simell, Jorma Ilonen, Riitta Veijola, and Mikael Knip ABSTRACT Background: Type 1 diabetes may have its origins in the fetal period of life. Free radicals were implicated in the cause of type 1 diabetes. It was hypothesized that antioxidant nutrients could protect against type 1 diabetes. Objective: We assessed whether high maternal intake of selected dietary antioxidants during pregnancy is associated with a reduced risk of advanced cell autoimmunity in the child, defined as repeated positivity for islet cell antibodies plus 1 other antibody, overt type 1 diabetes, or both. Design: The study was carried out as part of the population-based birth cohort of the Type 1 Diabetes Prediction and Prevention Project. The data comprised 4297 children with increased genetic susceptibility to type 1 diabetes, born at the University Hospital of Oulu or Tampere, Finland, between October 1997 and December The children were monitored for diabetes-associated autoantibodies from samples obtained at 3 12-mo intervals. Maternal antioxidant intake during pregnancy was assessed postnatally with a self-administered food-frequency questionnaire, which contained a question about consumption of dietary supplements. Results: Maternal intake of none of the studied antioxidant nutrients showed association with the risk of advanced cell autoimmunity in the child. The hazard ratios, indicating the change in risk per a 2-fold increase in the intake of each antioxidant, were nonsignificant and close to 1. Conclusion: High maternal intake of retinol, -carotene, vitamin C, vitamin E, selenium, zinc, or manganese does not protect the child from development of advanced cell autoimmunity in early childhood. Am J Clin Nutr 2008;88: INTRODUCTION Since the fetal origins hypothesis was widely introduced in the 1980s and 1990s (1), increased attention has been paid to the fetal period in search for the causes of chronic illnesses, including type 1 diabetes (2). Maternal dietary nitrite intake was associated with increased risk of type 1 diabetes in the offspring (3), and a strong inverse association was reported between maternal use of cod liver oil during pregnancy and the risk of type 1 diabetes in the child (4). Perinatal events were related to an increased risk of childhood type 1 diabetes (5). Oxidative stress increases during pregnancy and birth. The high metabolic rate of the placenta leads to increased generation of free radicals, biomarkers of which were observed in maternal circulation (6). The concentration of lipoperoxides in cord blood is only 30% of that in maternal blood, suggesting that the placenta suppresses lipoperoxide formation or transplacental passage, protecting the fetus from the action of these free radicals (7). At birth, the neonate faces an increase in oxidative aggression and presents high concentrations of hydroperoxides in erythrocyte membranes, indicating oxidative stress (8). Free radicals were implicated in the cause of type 1 diabetes (9), and it was hypothesized that antioxidant nutrients could protect against type 1 diabetes. Animal models of type 1 diabetes have indicated that dietary antioxidants may have protective effects in the disease process (10 12). Epidemiologic studies about associations with type 1 diabetes were published on vitamin C (13, 14), vitamin E (15, 16), selenium (15), and zinc (17), but the evidence was insufficient to draw conclusions about the possible protective role of dietary antioxidants against the disease. 1 From the Tampere School of Public Health, University of Tampere, Tampere, Finland (LU, SMV, JN, and LM); the Department of Health Promotion and Chronic Disease Prevention, National Public Health Institute, Helsinki, Finland (LU, SMV, UU, JN, SN, CK-K, and M-LO); the Department of Epidemiology and Population Health, Medical Statistics Unit, London School of Hygiene & Tropical Medicine, London, United Kingdom (MGK); the Research Unit (SMV) and Department of Pediatrics (MK), Tampere University Hospital, Tampere, Finland; the Department of Public Health (SN) and the Hospital for Children and Adolescents (MK), University of Helsinki, Helsinki, Finland; the Department of Pediatrics (OS) and the Immunogenetics Laboratory (JI), University of Turku, Turku, Finland; the Juvenile Diabetes Research Foundation (JDRF) Center for Prevention of Type 1 Diabetes in Finland, Turku, Oulu, and Tampere, Finland (OS and MK); the Department of Clinical Microbiology, University of Kuopio, Kuopio, Finland (JI); and the Department of Pediatrics, University of Oulu, Oulu, Finland (RV). 2 Supported by Academy of Finland (grants 63672, 79685,79686, 80846, , ); Finnish Diabetes Association; Finnish Diabetes Research Foundation, Finnish Pediatric Research Foundation; Häme Foundation of the Finnish Culture Fund; Juho Vainio Foundation; Yrjö Jahnsson Foundation; Medical Research Funds; Turku, Oulu, and Tampere University Hospitals; JDRF (grants , , , ); Novo Nordisk Foundation; EU Biomed 2 (BMH4-CT ), Doctoral Programs for Public Health; Jalmari and Rauha Ahokas Foundation; and Orion-Farmos Research Foundation. 3 Address reprint requests to L Uusitalo, Department of Health Promotion and Chronic Disease Prevention, National Public Health Institute, Mannerheimintie 166, Helsinki, Finland. liisa.uusitalo@uta.fi. Received February 14, Accepted for publication May 12, Am J Clin Nutr 2008;88: Printed in USA American Society for Nutrition

2 MATERNAL ANTIOXIDANT INTAKE AND PREDIABETES 459 Circulating concentrations of most dietary antioxidants in the newborn are correlated with respective maternal values and are thus likely to be influenced by maternal dietary intake. We hypothesized that high maternal intake of dietary antioxidants during pregnancy leads to high antioxidant status in the fetus and newborn, thereby reducing his or her risk of type 1 diabetes in early childhood. To test this hypothesis, we assessed whether maternal intake of selected dietary antioxidants during pregnancy was associated with a reduced risk of advanced cell autoimmunity in the offspring, defined as repeated positivity for islet cell antibodies (ICAs) plus 1 other antibody, overt type 1 diabetes, or both. SUBJECTS AND METHODS Subjects In the Type 1 Diabetes Prediction and Prevention (DIPP) Project, a prospective population-based cohort study, newborns at 3 university hospitals in Finland were screened for major histocompatibility complex, class II, DQ beta 1 (HLA-DQB1) conferred susceptibility to type 1 diabetes with the use of cord blood (18). Infants carrying increased genetic susceptibility (19) were monitored for diabetes-associated autoantibodies, growth, and viral infections at intervals of 3 12 mo. Study design and procedures were approved by the respective ethics committees, and the families signed an informed written consent. The DIPP Nutrition Study falls within the framework of DIPP (20). The present series comprised 4297 children (79.4% of invited) born between 20 October 1997 and 31 December 2002 at the University Hospitals of Oulu and Tampere. Dietary data were available from mothers of 3723 of these children (86.6%), including 38 twins or triplets. Descriptive dietary information is based on a cohort of 3730 mothers, including 45 mothers whose child did not participate in the follow-up. Of the 4 type 1 diabetes-associated autoantibodies analyzed, ICAs were used as the primary screening tool for cell autoimmunity (18). When a child seroconverted to positivity for ICA for the first time, all his or her preceding and subsequent samples were analyzed for antibodies against insulin, 65-kDa isoform of glutamic acid decarboxylase, and islet antigen 2 antibodies. For the present analyses, autoantibodies were analyzed until 31 August Dietary methods Dietary intake during pregnancy was assessed postnatally by a self-administered, semiquantitative 181-item food-frequency questionnaire (FFQ), which was validated specifically for the present study against 10-d food records in a sample of pregnant Finnish women (21). Also in the validation study the questionnaire was completed after the delivery. The validity of the questionnaire for antioxidant nutrients was, in general, good; correlation coefficients between the 2 methods ranged from 0.45 for zinc to 0.71 for retinol. The only exception was vitamin E, for which the correlation was only The FFQ was designed to assess the entire diet during the past month before the beginning of the maternity leave, ie, the 8th month of pregnancy. The respondents were asked to report the frequency of use in terms of natural units (eg, 1 apple), common household measures (eg, 1 glass of milk), or portions (eg, a portion of lasagna), for which typical portion sizes identified in earlier Finnish dietary studies were used in dietary calculations. The mothers received the FFQ by mail and returned it at the first visit to the study center when the infant was 3 mo old. A trained study doctor or nurse checked the FFQ and, if required, completed it in collaboration with the mother. All the returned FFQ forms were checked by a trained nutritionist. If there were 10 food items with missing frequency, the FFQ was rejected (n 52, 1.4% of all FFQs). Other missing items were imputed with null (22). FFQ data were entered into a data file at a commercial data entry service. Daily energy intake and the intake of 7 selected antioxidant nutrients from foods were calculated with the use of the Finnish food composition database (23) by in-house software of the National Public Health Institute. The use of dietary supplements was assessed by a separate question on the FFQ form. The respondent was asked to write down the type, brand name, and manufacturer s name of all the dietary supplements used during the entire pregnancy, the amount of each supplement used per day or per week, and the weeks of pregnancy during which each supplement was used. The nutrient contents of the dietary supplements registered as drugs were obtained from the Finnish pharmacopoeia. Information on other dietary supplements was acquired from the National Food Administration, the manufacturer, or both. Average daily intakes of antioxidant nutrients from dietary supplements during the entire pregnancy were calculated as separate variables by in-house software of the National Public Health Institute. Genetic methods HLA-DQB1 alleles were analyzed as described earlier (19). In brief, a part of the second exon of the HLA-DQB1 gene was amplified with the use of a primer pair with a biotinylated 3' primer. The biotinylated polymerase chain reaction products were then transferred to streptavidin-coated microtitration plates, denatured, and hybridized with sequence-specific probes labeled with lanthanide chelates: europium, terbium, or samarium. Two hybridization mixtures were used, one containing probes hybridizing with DQB1*0602 and DQB1*0603, DQB1*0603 and DQB1*0604, and a consensus sequence, and the other containing probes specific to the DQB1*02, DQB1*0301, and DQB1*0302 alleles. After appropriate incubations and washings, the specific hybridization products were detected with the use of 3-color time-resolved fluorescence of the lanthanide chelates. Immunologic methods The detection of type 1 diabetes associated autoantibodies was described in Virtanen et al (20). In brief, ICAs were quantified by a standard indirect immunofluorescence method on sections of frozen human pancreas from a blood group O donor (24). Serum antibodies against insulin were quantified with a microassay (25, 26) and glutamic acid decarboxylase and islet antigen 2 antibodies with specific radiobinding assays (27, 28). Transplacentally transferred autoantibodies, which were present in cord blood and thereafter disappeared from the infant s serum (29), were excluded from the analyses. Background characteristics Information on maternal age and education was registered with the use of a structured questionnaire completed after the delivery. In preliminary single covariate analyses, vocational

3 460 UUSITALO ET AL education was more closely associated with dietary factors and with the endpoint variable than was basic education, and it was used in the multiple covariate models. In the questionnaire, vocational education was recorded on a 7-category scale corresponding to the Finnish education system. To achieve sufficient sample size in each category, the data were aggregated into 4 categories. Information on the number of earlier deliveries (parity), maternal smoking during pregnancy, and gestational age and weight and height of the newborn was received from the Medical Birth Registries of the Oulu and Tampere University Hospitals. Smoking during pregnancy was registered on a 3-category scale (nonsmoker, quit smoking during the first trimester, smoker). Because the number of quitters was small, they were aggregated with smokers. Ponderal index of the newborn was calculated by the formula (g/m 3 ) 100. Home municipality was coded as urban, semiurban, or rural according to the classification of Statistics Finland (30). Statistical methods Dietary variables The intakes of the antioxidant nutrients from foods were adjusted for energy intake by the residual method (31). In the linear regression model used for energy adjustment, the variables are often transformed with the use of the ln to improve normality and homoscedasticity. However, when these ln-scale residuals are used as explanatory variables in a statistical model used to analyze the associations between dietary intake and an endpoint variable, the results may be difficult to interpret. If ln transformation had been used in the present study, the interpretation of the hazard ratio produced by the survival model would have been the change in risk per a 2.7-fold increase in energy-adjusted intake of the nutrient. To achieve an interpretation of the hazard ratio that is intuitively more appealing, we used base-2 logarithmic (log 2 ) transformation for the dietary variables instead of ln transformation. When a log 2 -transformed food or nutrient is used as an explanatory variable in the survival model, the interpretation of the hazard ratio is the change in risk per a 2-fold increase in energy-adjusted intake of the nutrient. The total intake of each antioxidant was calculated by summing the intakes from foods and supplements. The total intakes were not adjusted for energy. The total intakes of all the antioxidants had distributions that were close to normal after log 2 transformation, and these transformed values were used as explanatory variables in the survival models. Derived categorical variables were formed both from energy-adjusted intakes from foods and from total intakes from foods and supplements with the use of quartile values as cutoffs. Associations of dietary variables with the endpoint variable The endpoint of advanced cell autoimmunity is interval censored. To accommodate this structure, a piecewise exponential survival model was used to analyze the associations of background characteristics and maternal antioxidant intake with the endpoint (20, 32), with constant hazard in the intervals , , , and 3 y. The models were fitted with the use of maximum likelihood in SAS PROC NLMIXED, with standard errors of estimates derived from the observed information matrix. SAS version (SAS Institute, Cary, NC) was used in the analyses. There was no interval censoring in the diagnosis of clinical type 1 diabetes; therefore, the associations of antioxidant nutrients with the risk of clinical disease could be analyzed by Cox regression models with the use of SPSS 15.0 (SPSS, Chicago, IL). Cox regression analysis was also used to check the associations of categorical antioxidant intakes with advanced cell autoimmunity and the associations of antioxidant intake with early endpoints (advanced cell autoimmunity appearing before the age of 2.5 y). In etiologic analyses the effective sample size (ie, the number of children who had both maternal dietary data and endpoint data available) was 3723, including 138 cases with advanced cell autoimmunity, 57 cases with advanced cell autoimmunity appearing before the age of 2.5 y, and 55 cases with clinical type 1 diabetes. Adjustment for potential confounding factors The possible confounding by background characteristics was controlled by adding the background variables as covariates in the statistical models. Two sets of covariates were used, one including diabetes in a first-degree relative and genetic risk group, and the other including also sex, gestational age, maternal age, maternal parity, maternal education, maternal smoking during pregnancy, degree of urbanization of the home municipality, and region of birth (Oulu compared with Tampere area). All the background characteristics were categorical and were coded as binary indicator variables for model-fitting purposes. RESULTS Among the 4297 children, 144 (3.4%) were repeatedly positive for ICA plus 1 other antibody during the median follow-up time of 4.4 y (range: y) since birth. Seventy-four children (1.7%) had progressed to clinical type 1 diabetes at a median age of 4.1 y (range: y). Among those children, 53 were repeatedly positive for ICA plus 1 other autoantibody. Nine of the remaining 21 children had 1 autoantibody in one single sample before or at the time of diagnosis. Four persistently seronegative children had the last blood sample drawn y before the diagnosis of diabetes. The remaining 8 children with diabetes had not provided serum samples for autoantibody follow-up. Therefore, we decided to include children who progressed to clinical type 1 diabetes in the group of children called later in the study as autoantibody-positive children. This resulted in 165 children (3.8% of all children) being positive for the ICA plus 1 other antibody endpoint, ie, having advanced cell autoimmunity, which term will be used throughout the text for the endpoint. Cumulative incidence of advanced cell autoimmunity was highest for children who had a first-degree relative with diabetes and for those with a high-risk genotype (Table 1). Gestational age was inversely associated with the risk of advanced cell autoimmunity. The intakes of the antioxidant nutrients had skewed distributions; therefore, average intakes are presented as medians (Table 2). The main sources of retinol were meat and milk products (Table 2). Organ meats supplied 43% and cheese 13% of the total dietary retinol intake. -carotene came mainly from roots (65%). Juice was the main source of vitamin C, followed by fruit and vegetables (eg, cucumber, sweet pepper, and tomato; 10%) and citrus fruits (10%). Vitamin E was supplied mainly by vegetable oils and cereal products. The most important sources of selenium and zinc were meat, milk, and cereal products. Manganese was provided mainly by cereal products. The mean (maximum) daily supplementary intake was 18.1 g (2141 g) for retinol (1.9% of total intake), 261 g (10 800

4 MATERNAL ANTIOXIDANT INTAKE AND PREDIABETES 461 TABLE 1 Distribution of background characteristics and cumulative incidence of advanced cell autoimmunity among children with a genetic susceptibility to type 1 diabetes born at the University Hospital of Oulu or Tampere, Finland, Children Cases % (n) % (n) Girls (2029) 3.3 (67) Boys 52.8 (2268) 4.3 (98) HLA-DQB1 conferred risk group Moderate risk (DQB1*0302/x 3 ) (3470) 3.3 (114) High risk (DQB1*02/*0302) 19.2 (827) 6.2 (51) 4 Familial diabetes No (4002) 3.5 (140) Yes 6.9 (295) 8.5 (25) 4 Mother s age (y) (831) 4.5 (37) (1435) 3.3 (47) (1270) 4.3 (43) (753) 3.9 (36) Missing information 0.2 (8) 4.3 (2) Previous deliveries No (1835) 3.4 (63) (1372) 4.5 (62) (628) 3.0 (19) (436) 3.7 (16) Missing information 0.6 (26) 19.2 (5) Mother s professional education Academic education (867) 4.3 (37) Upper secondary vocational education 39.5 (1698) 3.2 (54) Vocational school or course 29.4 (1262) 3.9 (49) No professional education 7.1 (303) 5.3 (16) Missing information 3.9 (167) 5.4 (9) Gestational age (wk) 5 First quarter, (1079) 5.0 (54) Second quarter, (1026) 3.1 (32) 4 Third quarter, (1034) 3.7 (38) Fourth quarter, (1113) 2.8 (31) 4 Missing information 1.0 (45) 22.2 (10) Mother s smoking during pregnancy No (3688) 3.8 (141) Yes 10.4 (446) 2.7 (12) Missing information 3.8 (163) 7.4 (12) Delivery hospital Tampere (2417) 3.6 (86) Oulu 43.8 (1880) 4.2 (79) Total 100 (4297) 3.8 (165) 1 Advanced cell autoimmunity is defined as repeated positivity for islet cell antibodies and 1 other diabetes-associated autoantibody, clinical type 1 diabetes, or both. 2 Reference category. 3 x not equal to *02, *0301, * Hazard ratio significantly different from the reference category, P 0.05 (piecewise exponential survival model). 5 The trend across the quarters was significant (P 0.05). Loglikelihood ratio test was used to test whether the model with and without the variable differed. g) for -carotene (5.7% of total intake), 23.3 mg (1395 mg) for vitamin C (10.5% of total intake), 1.1 mg (191 mg) for vitamin E (7.3% of total intake), 6.3 g (95 g) for selenium (7.9% of total intake), 2.0 mg (30 mg) for zinc (10.6% of total intake), and 0.3 mg (18 mg) for manganese (4.8% of total intake). The minimum and median supplementary intake was zero for all the nutrients analyzed. Maternal intake of any antioxidant nutrient in the study, analyzed with the use of a piecewise exponential survival model, was not significantly associated with the risk of advanced cell autoimmunity in the offspring during the median follow-up time of 4.4 y (Table 3). For each nutrient, the hazard ratio, indicating the change in risk per 2-fold increase in the energy-adjusted intake of the nutrient, was nonsignificant and close to one. Hazard ratios were similar for energy-adjusted dietary intakes and for combined dietary and supplemental intakes. The inclusion of a broad set of background variables as covariates in the model, in addition to basic adjustment for genetic risk group and diabetes in the family, did not change the results. Antioxidant intake was not associated with the risk of advanced cell autoimmunity appearing before the age of 2.5 y or with clinical type 1 diabetes. The hazard ratios were close to one, but the CIs tended to be wider than in the main analysis because of the smaller number of cases. Categorization of antioxidants into quarters of intake did not change the results (data not shown). DISCUSSION In this prospective cohort study with a reasonable number of endpoints, the maternal intake of retinol, -carotene, vitamin C, vitamin E, selenium, zinc, or manganese during pregnancy was not associated with the risk of advanced cell autoimmunity in the offspring with HLA-conferred susceptibility to type 1 diabetes. The presence of 2 type 1 diabetes-associated autoantibodies usually reflects a progressive process that only rarely reverts (33). Dietary intake was assessed by a FFQ developed and validated specifically for the purposes of the present study (21). The variation in the intakes of the analyzed nutrients was substantial. The difference between the 25th and 75th percentiles of intake from foods ranged from 46% for selenium to 147% for -carotene. The mean intakes from dietary supplements covered only a small fraction of the total intakes, but the maximum supplementary intakes were large. There are few previous studies on the association of retinol intake and type 1 diabetes. Retinol deficiency reduced diabetes in diabetes-prone BB rats (34). Because cod liver oil is rich in retinol, along with vitamin D and long-chain n 3 fatty acids, it may be hypothesized that retinol could be the main player in the inverse association that was observed between the use of cod liver oil during pregnancy and subsequent risk of type 1 diabetes in the offspring (4). The use of other kinds of vitamin D supplements was not associated with the risk of diabetes. The present findings do not support any association between maternal retinol intake and risk of prediabetes in the child. To our knowledge, -carotene intake has not been analyzed in the context of type 1 diabetes. We found no association between maternal -carotene intake and advanced cell autoimmunity in the offspring. A rat experiment suggested a protective effect of vitamin C against diabetes induced by interferon- (11). Association of vitamin C with the risk of type 1 diabetes was also assessed in 2 epidemiologic analyses. Fewer children with type 1 diabetes than controls reported the use of vitamin C supplements for 1 mo, before the date of diagnosis of diabetes (13), whereas the intake of foods rich in vitamin C was not related to childhood type 1 diabetes in another case-control study (14). The present results

5 462 UUSITALO ET AL TABLE 2 Intake of antioxidant nutrients and their dietary sources among 3730 pregnant women, who delivered an infant with genetic susceptibility to type 1 diabetes at the University Hospital of Oulu or Tampere, Finland, Supply by food groups (%) Retinol -Carotene Vitamin C Vitamin E Selenium Zinc Manganese Meat products Fish and shellfish Egg Milk products Butter Margarine Vegetable oils Cereal products Potato Leaf and fruit vegetables Other vegetables Fruit Berries Juice Other foods Total Median (25th and 75th percentiles) for the nutrients were retinol: 662 g (443, 1128 g); -carotene: 3407 g (2231, 5140 g); vitamin C: 172 mg (116, 250 mg); vitamin E: 13.4 mg (10.9, 16.5 mg); selenium: 77.4 g (63.9, 93.5 g); zinc: 16.2 mg (13.4, 19.6 mg); and manganese: 6.44 mg (4.94, 8.29 mg). 2 Usually eaten fresh (eg, lettuce, cucumber, sweet pepper, tomato). 3 Usually eaten cooked (eg, cabbages, roots and tubers, onions, peas and beans, mushroom). 4 Alcoholic drinks, coffee, tea, soft drinks, industrial and animal fats, snacks, sweets, biscuits. show no association between vitamin C intake during pregnancy and the risk of advanced cell autoimmunity in the child. Animal models of type 1 diabetes have indicated that vitamin E may have protective effects against type 1 diabetes (9, 10, 35, 36). Serum -tocopherol concentrations were inversely associated with type 1 diabetes in a nested case-control study within a 21-y follow-up of adult men (15). In a cohort of initially nondiabetic siblings of children with diabetes, we found an inverse relation of borderline significance between serum -tocopherol concentrations and type 1 diabetes (16). In the present study, no evidence was observed of an association between maternal intake and advanced cell autoimmunity in the offspring. However, because of the poor validity of the FFQ for vitamin E (21), we consider the data insufficient to rule out a possible protective role of maternal vitamin E intake against preclinical type 1 diabetes in the child. In the nested case-control study referred to earlier (15), serum selenium was not associated with type 1 diabetes. Similarly, we found no association between maternal selenium intake and prediabetes in the offspring. Animal experiments have suggested a TABLE 3 Associations of maternal intake of antioxidant nutrients during pregnancy with the risk of advanced cell autoimmunity among 3723 children with a genetic susceptibility to type 1 diabetes born at the University Hospital of Oulu or Tampere, Finland, Intake from foods, adjusted for energy Intake from foods supplements, log 2 -transformed Model 1 2 Model 2 3 Model 1 2 Model 2 3 Retinol 0.94 (0.76, 1.15) 0.97 (0.79, 1.21) 1.05 (0.84, 1.32) (0.81, 1.30) 4 -Carotene 1.06 (0.88, 1.28) 1.00 (0.82, 1.22) 1.07 (0.90, 1.27) 1.02 (0.85, 1.22) Vitamin C 0.96 (0.76, 1.22) 0.92 (0.72, 1.18) 0.97 (0.80, 1.18) 0.96 (0.78, 1.17) Vitamin E 1.13 (0.58, 2.21) 1.10 (0.55, 2.20) 1.04 (0.74, 1.46) 1.06 (0.75, 1.49) Selenium 1.01 (0.47, 2.21) 1.21 (0.54, 2.70) 1.04 (0.72, 1.50) 1.11 (0.76, 1.64) Zinc 0.84 (0.38, 1.89) 1.01 (0.44, 2.34) 0.99 (0.70, 1.39) 1.05 (0.74, 1.48) Manganese 0.99 (0.69, 1.42) 0.97 (0.65, 1.44) 1.02 (0.77, 1.36) 1.03 (0.76, 1.40) 1 Advanced cell autoimmunity is defined as repeated positivity for islet cell antibodies and 1 other diabetes-associated autoantibody, clinical type 1 diabetes, or both. All values are hazard ratio (95% CI) produced by piecewise exponential survival models and indicate the change in risk per a 2-fold increase in intake. 2 Adjusted for genetic risk and familial diabetes. 3 Adjusted for genetic risk, familial diabetes, sex, gestational age, maternal age, maternal parity, maternal education, maternal smoking during pregnancy, degree of urbanization of the home municipality, and region of birth (Oulu versus Tampere area). 4 Because the amount of vitamin A in dietary supplements was expressed as retinol equivalents, the intake of vitamin A in foods is also calculated as retinol equivalents in the combined variable.

6 MATERNAL ANTIOXIDANT INTAKE AND PREDIABETES 463 protective effect of zinc against developing type 1 diabetes (12, 37). A high groundwater concentration of zinc was associated with a reduced risk of type 1 diabetes in a case-control study (17). We found no significant association between maternal zinc intake and advanced cell autoimmunity in the child. To our knowledge, there are no previous reports on the associations of manganese intake with type 1 diabetes. In the present study, we found no evidence of high maternal intake of manganese being protective against prediabetes in the offspring. CIs for the hazards ratios tended to be wider for energyadjusted nutrient intakes from foods than for the combined intakes from foods and supplements not adjusted for energy. Because the CIs for non energy-adjusted intakes from foods and from combined sources were similar, this difference could not result from supplement use (data not shown). The discrepancy was largest for those nutrients that correlated most strongly with energy intake (vitamin E, zinc, and selenium). For these nutrients, the variances of the residuals produced by the energy adjustment method were considerably smaller than the variances of the absolute nutrient intakes. In the survival models with nutrient intakes as explanatory variables, a 1-unit change in residual implied a greater relative change on the range of nutrient intake among the study subjects than a 1-unit change in absolute intake, leading to larger standard errors of the estimate. Thus, the discrepancy in CIs for the hazard ratios may be considered an artifact created by the different scaling of energy-adjusted and absolute variables. Antioxidant status of the fetus and newborn is likely to be influenced by maternal dietary intake. Although serum retinol and zinc concentrations are poor indicators of dietary intake (38), correlations were shown between maternal and cord blood concentrations of retinol (39) and zinc (40). Cord blood concentrations of -carotene (41, 42), vitamin C (43, 44), selenium (45, 46), and manganese (47, 48) have shown positive correlations with maternal values. For vitamin E, positive correlations were observed for lipid-adjusted -tocopherol concentrations and for erythrocyte -tocopherol concentrations in cord and maternal blood (49, 50). However, the maternal antioxidant intake does not seem to protect the child from developing pre type 1 diabetes in early childhood. One could speculate that the lack of association could be due to the short period of time when the infant faces oxidative stress while simultaneously being dependent on the antioxidants received from the mother. During the fetal period, the uterine environment may offer protection against oxidative stress (7). Free radical challenges increase at birth (8), but the newborn s own dietary intake may soon wipe out the effects of maternal intake. Therefore, studies on the associations between antioxidant intake in childhood and the risk of type 1 diabetes are warranted. The author s responsibilities were as follows LU: participated in designing the study, preparing, analyzing, and interpreting the data, and wrote the manuscript; MGK: participated in designing the study and planned the statistical analyses; SMV: participated in designing the study, planning the statistical analyses, and analyzing and interpreting the data;. UU: participated in designing the study and preparing the data; JN: participated in analyzing and interpreting the data; SN: participated in designing the study and preparing the data; CK-K and RV: participated in preparing the data; M-LO: participated in designing the study and interpreting the data; LM: participated in preparing the data; OS and JI: participated in conceiving and designing the study; MK: participated in conceiving and designing the study and interpreting the data. All authors participated in drafting or revising of the manuscript and approved the final version to be published. None of the authors had a personal or financial conflict of interest. REFERENCES 1. Godfrey KM, Barker DJP. Fetal programming and adult health. Public Health Nutr 2001;4: Lindberg B, Ivarsson S-A, Landin-Olsson M, Sundkvist G, Svanberg L, Lernmark Å. Islet autoantibodies in cord blood from children who developed type 1 (insulin-dependent) diabetes mellitus before 15 years of age. Diabetologia 1999;42: Virtanen SM, Jaakkola L, Räsänen L, et al. Nitrate and nitrite intake and the risk for type 1 diabetes in Finnish children. Diabet Med 1994;11: Stene LC, Ulriksen J, Magnus P, Joner G. Use of cod liver oil during pregnancy associated with lower risk of type 1 diabetes in the offspring. Diabetologia 2000;43: Dahlquist GG, Patterson C, Soltesz G. Perinatal risk factors for childhood type 1 diabetes in Europe. The EURODIAB Substudy 2 Study Group. Diabetes Care 1999;22: Chen X, Scholl TO. Oxidative stress: changes in pregnancy and with gestational diabetes mellitus. Curr Diab Rep 2005;5: Takehara Y, Yoshioka T, Sasaki J. Changes in the levels of lipoperoxides and antioxidant factors in human placenta during pregnancy. Acta Med Okayama 1990;44: Robles R, Palomino N, Robles A. Oxidative stress in the neonate. Early Hum Dev 2001;65(suppl):S Asayama K, Kooy NW, Burr IM. Effect of vitamin E deficiency and selenium deficiency on insulin secretory reserve and free radical scavenging systems in islets: decrease of islet manganosuperoxide dismutase. J Lab Clin Med 1986;107: Beales PE, Williams AJ, Albertini MC, Pozzilli P. Vitamin E delays diabetes onset in the non-obese diabetic mouse. Horm Metab Res 1994; 26: al-zuhair H, Mohamed HE. Vitamin C attenuation of the development of type I diabetes mellitus by interferon-alpha. Pharmacol Res 1998;38: Ho E, Quan N, Tsai YH, Lai W, Bray TM. Dietary zinc supplementation inhibits NFkappaB activation and protects against chemically induced diabetes in CD1 mice. Exp Biol Med (Maywood) 2001;226: Glatthaar C, Whittall DE, Welborn TA, et al. Diabetes in Western Australian children: descriptive epidemiology. Med J Aust 1988;148: Dahlquist GG, Blom LG, Persson L-Å, Sandström AI, Wall SGI. Dietary factors and the risk of developing insulin dependent diabetes in childhood. BMJ 1990;300: Knekt P, Reunanen A, Marniemi J, Leino A, Aromaa A. Low vitamin E status is a potential risk factor for insulin-dependent diabetes mellitus. J Intern Med 1999;245: Uusitalo L, Knip M, Kenward MG, et al. Serum alpha-tocopherol concentrations and risk of type 1 diabetes mellitus: a cohort study in siblings of affected children. J Pediatr Endocrinol Metab 2005;18: Haglund B, Ryckenberg K, Selinus O, Dahlquist G. Evidence of a relationship between childhood-onset type 1 diabetes and low groundwater concentration of zinc. Diabetes Care 1996;19: Kupila A, Muona P, Simell T et al. Feasibility of genetic and immunological prediction of type 1 diabetes in a population-based birth cohort. Diabetologia 2001;44: Ilonen J, Reijonen H, Herva E, et al. Rapid HLA-DQB1 genotyping for four alleles in the assessment of diabetes risk in the Finnish population. Diabetes Care 1996;19: Virtanen SM, Kenward MG, Erkkola M, et al. Age at introduction of new foods and advanced beta-cell autoimmunity in young children with HLA-conferred susceptibility to type 1 diabetes. Diabetologia 2006;49: Erkkola M, Karppinen M, Javanainen J, Räsänen L, Knip M, Virtanen SM. Validity and reproducibility of a food frequency questionnaire for pregnant Finnish women. Am J Epidemiol 2001;154: Parr CL, Hjartåker A, Scheel I, Lund E, Laake P, Veierod MB. Comparing methods for handling missing values in food-frequency questionnaires and proposing k nearest neighbours imputation: effects on dietary intake in the Norwegian Women and Cancer study (NOWAC). Public Health Nutr (E-pub ahead of print 2 July 2007).

7 464 UUSITALO ET AL 23. National Public Health Institute. Fineli Finnish Food Composition Database. Version current 10 May Internet: (accessed 27 November 2007). 24. Bottazzo GF, Florin-Christensen A, Doniach D. Islet-cell antibodies in diabetes mellitus with autoimmune polyendocrine deficiencies. Lancet 1974;ii: Ronkainen M, Hämäläinen A-M, Koskela P, Åkerblom HK, Knip M; Finnish Trigr Study Group. Pregnancy induces nonimmunoglobulin insulin-binding activity in both maternal and cord blood serum. Clin Exp Immunol 2001;124: Williams AJK, Bingley PJ, Bonifacio E, Palmer JP, Gale EAM. A novel micro-assay for insulin autoantibodies. J Autoimmun 1997;10: Savola K, Sabbah E, Kulmala P, Vähäsalo P, Ilonen J, Knip M. Autoantibodies associated with type 1 diabetes mellitus persist after diagnosis in children. Diabetologia 1998;41: Savola K, Bonifacio E, Sabbah E, et al. IA-2 antibodies a sensitive marker of IDDM with clinical onset in childhood and adolescence. Diabetologia 1998;41: Hämäläinen AM, Ronkainen MS, Åkerblom HK, Knip M. Postnatal elimination of transplacentally acquired disease-associated antibodies in infants born to families with type 1 diabetes. The Finnish TRIGR Study Group. Trial to Reduce IDDM in the Genetically at Risk. J Clin Endocrinol Metab 2000;85: Statistics Finland. Statistical grouping of municipalities. Version current 27 November. Internet: en.html (accessed 27 November 2007). 31. Willett W. Nutritional epidemiology. 2nd ed. New York, NY: Oxford University Press, 1998; 32. Collett, D. Modelling survival data in medical research. 2nd ed. Boca Raton, FL: Chapman & Hall/CRC, Knip M. Natural course of preclinical type 1 diabetes. Horm Res 2002; 57(suppl): Driscoll HK, Chertow BS, Jelic TM, et al. Vitamin A status affects the development of diabetes and insulitis in BB rats. Metabolism 1996;45: Slonim AE, Surber ML, Page DL, Sharp RA, Burr IM. Modification of chemically induced diabetes in rats by vitamin E. J Clin Invest 1983;71: Behrens WA, Scott FW, Madere R, Trick K, Hanna K. Effect of dietary vitamin E status in the BB rat during development and after the onset of diabetes. Ann Nutr Metab 1986;30: Tobia MH, Zdanowicz MM, Wingertzahn MA, McHeffey-Atkinson B, Slonim AE, Wapnir RA. The role of dietary zinc in modifying the onset and severity of spontaneous diabetes in the BB Wistar rat. Mol Genet Metab 1998;63: Hunter D. Biochemical indicators of dietary intake. In: Willett W, ed. Nutritional epidemiology. New York, NY: Oxford University Press, 1990; 39. Gazala E, Sarov B, Hershkovitz E, et al. Retinol concentration in maternal and cord serum: its relation to birth weight in healthy motherinfant pairs. Early Hum Dev 2003;71: Zapata CL, Melo MR, Donangelo CM. Maternal, placental and cord zinc components in healthy women with different levels of serum zinc. Biol Neonate 1997;72: Karakilcik AZ, Aksakal M, Baydas G, Sozen R, Ayata A, Simsek M. Plasma beta-carotene concentrations in pregnancies, newborn infants and their mothers. J Pak Med Assoc 1996;46: Dimenstein R, Trugo NM, Donangelo CM, Trugo LC, Anastacio AS. Effect of subacute maternal vitamin-a status on placental transfer of retinol and beta-carotene to the human fetus. Biol Neonate 1996;69: Vobecky JS, Vobecky J, Shapcott D, et al. Biochemical indices of nutritional status in maternal, cord, and early neonatal blood. Am J Clin Nutr 1982;36: Dejmek J, Ginter E, Solansky I, et al. Vitamin C, E and A levels in maternal and fetal blood for Czech and Gypsy ethnic groups in the Czech Republic. Int J Vitam Nutr Res 2002;72: Wasowicz W, Wolkanin P, Bednarski M, Gromadzinska J, Sklodowska M, Grzybowska K. Plasma trace element (Se, Zn, Cu) concentrations in maternal and umbilical cord blood in Poland. Relation with birth weight, gestational age, and parity. Biol Trace Elem Res 1993;38: Lee AM, Huel G, Godin J, et al. Inter-individual variation of selenium in maternal plasma, cord plasma and placenta. Sci Total Environ 1995; 159: Tsuchiya H, Mitani K, Kodama K, Nakata T. Placental transfer of heavy metals in normal pregnant Japanese women. Arch Environ Health 1984; 39: Takser L, Lafond J, Bouchard M, St-Amour G, Mergler D. Manganese levels during pregnancy and at birth: relation to environmental factors and smoking in a Southwest Quebec population. Environ Res 2004;95: Jain SK, Wise R, Bocchini JJ. Vitamin E and vitamin E-quinone levels in red blood cells and plasma of newborn infants and their mothers. J Am Coll Nutr 1996;15: Gonzalez-Corbella MJ, Lopez-Sabater MC, Castellote-Bargallo AI, Campoy-Folgoso C, Rivero-Urgell M. Influence of caesarean delivery and maternal factors on fat-soluble vitamins in blood from cord and neonates. Early Hum Dev 1998;53(suppl):S