The Impact of Dietary Factors during Pregnancy on the Development of Islet Autoimmunity and Type 1 Diabetes: A Systematic Literature Review

Aims and hypothesis: The incidence of type 1 diabetes mellitus in children is considerably increasing in western countries. Thus, identification of the environmental determinants involved could ultimately lead to disease prevention. Here, we aimed to systematically review (PROSPERO ID: CRD42022362522) the current evidence of the association between maternal dietary factors during gestation and the risk of developing type 1 diabetes and/or islet autoimmunity (IA) in murine and human offspring. Methods: In accordance with PRISMA guidelines, the present systematic review searched PubMed and Scopus (n = 343) for different combinations of MeSH terms, such as type 1 diabetes, diet, islet autoimmunity, prenatal, nutrient, gluten, gliadin, vitamin, milk, and fibers. Results: We found that the most investigated dietary factors in the present literature were gluten, dietary advanced glycosylated end products (dAGEs), vitamin D, fatty acids, and iron. The results concerning prenatal exposure to a gluten-free environment showed a consistently protective effect on the development of IA. Prenatal exposures to vitamin D and certain fatty acids appeared to protect against the development of IA, whereas in utero iron and fat exposures correlated with increased risks of IA. Conclusion: We conclude that a definite association is not established for most factors investigated as the literature represents a heterogeneous pool of data, although fetal exposures to some maternal dietary components, such as gluten, show consistent associations with increased risks of IA. We suggest that human prospective dietary intervention studies in both cohort and clinical settings are crucial to better evaluate critical and protective prenatal exposures from the maternal diet during pregnancy.


Introduction
Autoimmune diseases result from attacks on the body's tissues by the immune system and comprise at least 80 diverse diseases where the etiology and pathogenesis are not fully elucidated.The incidence of autoimmunity is approximately 3-5% worldwide, and many of the associated diseases, including type 1 diabetes, are increasing in western societies, especially in children below the age of 5 [1].Type 1 diabetes is among the most common childhood autoimmune diseases [2].Considering the geography-dependent rise in incidence as well as the wide variations in disease incidence between neighboring areas, the increase in type 1 diabetes incidence cannot exclusively be explained by genetic changes [3].Further, from a genetic viewpoint, the high-risk HLA haplotypes and genotypes [4] seem to be less prevalent factors for type 1 diabetes development than previously thought [5,6], underscoring a potentially bigger contribution of environmental determinants.Supporting this environmental or lifestyle-dependent impact of type 1 diabetes etiology, migrants tend to obtain a similar type 1 diabetes risk as the population in their region of relocation [7][8][9].Additionally, Europeans who are closely related genetically, but separated socioeconomically, tend to acquire substantially different type 1 diabetes risks [10].Many different environmental factors could contribute to the development of type 1 diabetes such as infections, toxins, and dietary factors [9,11].The impact of pre-and postnatal environmental factors on disease seems to be crucial because the process leading to IA starts very early in life; seroconversion has a peak incidence or median at around 9-12 months of age in children later diagnosed with type 1 diabetes, illustrating the importance of early environmental determinants [12][13][14].
This systematic review aims to evaluate the existing literature concerning the impact of prenatal, dietary factors as part of the environmental determinants of type 1 diabetes and islet autoimmunity, in murine models and in human case-control and birth cohort studies.We will answer the following questions: (1) Which prenatal dietary factors have already been investigated?(2) Can available data suggest protective advantages or hazards of certain prenatal dietary factors for the development of type 1 diabetes and IA in the offspring?(3) Does the pool of accessible data represent homogenous and conclusive information?

Research Design and Methods
The systematic review is registered in PROSPERO (CRD42022362522), reported in accordance with Preferred Reporting Items for Systematic review and Meta-Analysis Protocols (PRISMA-P), and follows the PRISMA guidelines, as described extensively in Shamseer et al. and Moher et al. [15,16] to systematically review existing data on the research question (Figures 1A and S1).Our search was performed in October 2022 and included all English-published studies issued until September 2022.The systematic strategy searched for the following medical subject headings terms: type I diabetes, type 1 diabetes, prenatal, diet, islet autoimmunity, nutrient, gluten, gliadin, vitamins, milk, fibers, and animal (complete search strings are available in Table S1).Combinations were narrowed via the boolean operator "AND", broadened via "OR" and the use of asterisks allowed our searching strategy to include multiple unknown and variable characters in the search query.The advanced [TITLE-ABS-KEY] and the [ALL FIELDS] functions were used in the cross-disciplinary databases Scopus and PubMed, respectively.The "NOT" boolean was omitted to avoid exclusion of relevant articles.Following filtering of duplicates, relevant articles eligible for full-text review were selected based on the title-and abstract screenings.Potentially relevant articles underwent a full-text evaluation to determine eligibility, and references of both relevant reviews and included articles were reviewed, to possibly identify additional publications that were not disclosed from the database searches (Figure 1A).The present systematic review assesses the following population, intervention, comparison, and outcome (PICO) elements: The present systematic review assesses the following population, intervention, comparison, and outcome (PICO) elements:

•
Poulation: pregnant murine or human females where data about nutrient factor exposures during prenatal life and the risk of developing type 1 diabetes and/or islet autoimmunity in the offspring are available.

•
Intervention: nutrient interventions and factors that may affect the incidences or occurrences of type 1 diabetes and islet autoimmunity.

•
Comparison: only studies comparing a proper, nonexposed control group with an exposed group were considered eligible.

•
Outcome: changes in type 1 diabetes or IA incidence in offspring of exposed mothers.
In addition to the PICO elements, we anticipate that the identification of prenatal dietary factors that influence the pathogenesis of type 1 diabetes could be used in the prevention of the disease.

Study Design
Reviews and meta-analyses from the search were screened for references, but not included in the final report.Regarding murine studies, only dietary intervention studies were included, whereas both prospective and retrospective population-based human birth cohort and case-control studies were included.

Eligibility Criteria
For murine studies, inclusion criteria were intervention studies, where a defined maternal dietary factor was the exposure, and the readout was either IA or diabetes incidence or time-to-onset assessments in the offspring.Studies where the dietary intervention lasted for only the in utero period and during gestation until weaning were included.As with murine studies, human studies eligible for inclusion had one or more defined maternal dietary factors as prenatal exposure and examined either IA, type 1 diabetes, or both as endpoints.For animal studies examining islet autoimmunity, considered parameters were degree/score/index of insulitis (infiltration of immune cells in the pancreatic islets).Where information of the amount of apoptotic cells, inflammatory cell signaling, inflammatory marker RNA levels, and cytokine protein levels were included in the reviewed articles, this information was also extracted and included in the extensive data tables.Both studies prospectively and retrospectively assessing exposure and outcome were eligible for inclusion.Studies that did not satisfy the stated inclusion criteria were excluded and exclusion reasons were reported (Figures 1A and S1).

Study Selection
An independent reviewer (V.B. I. J.) selected the studies using a two-step screening approach to assess eligibility as described above and in accordance with PRISMA guidelines.The reference lists of selected articles were further searched for relevant articles that were subsequently subjected to the two-step screening method described (Figure 1A).Reviews relevant to the research question allowed relevant publications to be subjected to the twostep screening method following the defined eligibility criteria.J. C. A. and K. J. checked the extracted data.

Data Collection Process
Briefly, data collection concerned study design and outcome (Tables 1-3).

Murine Dietary Intervention Studies
All murine studies examined both diabetes and IA as endpoints, except Essien et al. [25] reporting only the impact of maternal nitrosamine-rich charred meat consumption during gestation and lactation on autoimmune diabetes risks in the litters, and McCall et al. [33] reporting on in utero exposure to cadmium and the risk of autoimmune diabetes in the litters (Table 1).The most studied nutrient in the murine dietary intervention studies was gluten [27,29,31], and dietary advanced glycosylated end products (dAGEs) was the second most studied nutrient factor [22,30].

Human Population-Based Studies
Of the 23 eligible human studies, 18 evaluated the association between prenatal, dietary factors and the incidence of type 1 diabetes in the offspring [34][35][36][37][38][39][40][41]45,46,[48][49][50][51][52][54][55][56], of which 6 studies additionally reported IA-related readouts [45,46,48,49,55,56].The first study to improve our understanding of a possible link between prenatal dietary factors and the occurrence of IA was Fronczak et al., [42] and since then, 4 other human population-based birth cohort studies have been focusing entirely on this link [43,44,47,53] (Tables 2 and 3).Maternal intake of saturated palmitic acid was weakly associated with a decreased risk of type 1 diabetes (HR = 0.82, 95% CI: 0.67-0.99,p = 0.039).Considering clinical type 1 diabetes as an endpoint, onset time and consumption of the following fatty acids were significantly associated: palmitoleic acid isomers 16:1n-7 (p = 0.019) and 16:1n-9 (p = 0.014), EPA (p = 0.037), dihomo-γ-linolenic acid (p = 0.013).High cheese, low-fat margarines consumption associated with a decreased risk of type 1 diabetes (HR = 0.52 and 0.67, 95% CI: 0.31-0.87 and 0.49-0.92respectively) and high consumption of sour milk, fat from fresh milk and protein from sour milk with increased IA or type 1 diabetes (HR = 1.14, 1. 43  The incidence rate of type 1 diabetes among children exposed to iron supplementation in utero was higher after adjustment (HR = 1.33, 95% CI: 1.06-1.67)and this was still the case when mothers only supplementing with either iron-only or iron-other supplements were analyzed alone Prenatal iron exposure led to a higher risk of type Comparing offspring of mothers with highest gluten intake with those of mothers with lowest gluten intake, offspring had double the risk of type 1 diabetes development later in life (HR = 2.00, 95% CI: 1.02-4.00).Offspring type 1 diabetes risk was positively associated with prenatal exposure (p trend = 0.016).This association persisted after expressing it as energy adjusted residuals (p trend = 0.028) and after describing gluten intake as groups of 20% (p trend = 0.035) In utero gluten exposure was strongly associated with the risk of type 1 diabetes later in life The global likelihood ratio test showed no linear association between increases in gluten intake and type 1 diabetes childhood risk (p = 0.11).For each 10 g/day-increase of gluten intake, adjusted HR was 1.02 (95% CI: 0.73-1.43,p = 0.91), and for each standard deviation increase of gluten intake, adjusted HR was 1.01 (95% CI: 0.85-1.20,p = 0.91) Maternal gluten intake in pregnancy was not associated with offspring type 1 diabetes risk

Vitamin D
Low vitamin D levels have been linked to the development of autoimmune diseases [57], possibly due to its impact on both the innate and adaptive immune systems [58,59].Data from both animal and human studies showed conflicting observations in terms of prenatal vitamin D exposure and the type 1 diabetes risk during early life.In a study performed by Hawa et al. [23], maternal administration of olive oil containing vitamin D showed no significant effect on type 1 diabetes incidence, onset time, and IA when comparing NOD mice offspring from control and exposure groups.However, early life administration of vitamin D has also shown protective effects in animal studies [60,61].
In 2003, Fronczak and colleagues reported that maternal vitamin D intake during pregnancy had a protective effect on the child's risk of developing IA by evaluating measures of GAD 65 , IA-2, and insulin autoantibodies in children from the Diabetes Auto Immunity Study in the Young (DAISY) study [42].Interestingly, a similar effect was reported in children from the All Babies in Southeast Sweden (ABIS) study, a large, prospective, population-based cohort study of children born during 1997-1999.Here, maternal vitamin D supplementation during pregnancy protected against IA, when evaluating offspring at 1 year of age [43].However, in a Finnish type 1 Diabetes Prediction and Prevention (DIPP) prospective birth cohort, including children genetically at risk of type 1 diabetes born between 1997-2004, neither in utero exposures to vitamin D from food, supplements, or in combination could support the findings of advanced β-cell autoimmunity from the DAISY and ABIS Children [46].Also, when evaluating the ABIS children cohort 9 years after the first publication, Granfors and colleagues [50] could not support the protective role of vitamin D from 2007 [43], when they analyzed the in utero vitamin D exposure as a continuous and binary variable retrospectively, and when assessing the IA-risk in Finnish children with increased genetic risk for type 1 diabetes.The Environmental Determinants of Diabetes in the Young (TEDDY) study found no association between the intake of Vitamin D during pregnancy and the risk of developing IA in the offspring [53].Although they show inconsistencies in their findings, the obvious strengths of the prospective DAISY, ABIS, DIPP and TEDDY studies are the exactness of data collection concerning confounders and endpoints.However, the accounts for exposures were obtained retrospectively in those studies.
In the early 2000s, a Norwegian research team studied the effect of administrating cod liver oil and other vitamin D-containing supplements during pregnancy.They found that consumption of cod liver oil during pregnancy had a protective effect on the risk of offspring type 1 diabetes [35] whereas other vitamin D-containing supplements, such as maternal multivitamin supplementation, showed no significant effects.Assuming that vitamin D in cod liver oil was responsible for the observed effect, they argued that the bioavailability of vitamin D from cod liver oil might be better than that of vitamin D from multivitamin supplements as a possible explanation for their observations [35].It is, however, still a possibility that the negative association between offspring type 1 diabetes incidence and maternal cod liver oil intake came from other cod liver oil constituents.This publication was a retrospective, human, population-based case-control study design with people from a defined region in Norway as the study population [35].The research group extended their study with cases and controls covering a much broader area of Norway 3 years later [36].Importantly, they saw no significant, protective effects of administering cod liver oil and other vitamin D-containing supplements in this extended study [36].A Danish study, including data from 331,623 children, used vitamin D-fortified margarine to assess a potential effect on the development of type 1 diabetes when exposed in utero, but they also did not identify any association between maternal vitamin D supplement and infant type 1 diabetes risks [39].

Gluten
Proline and glutamine-rich gluten proteins are highly hydrophobic and partially resistant to intestinal degradation, which makes them more immunogenic than other nutritional proteins that are efficiently hydrolyzed into single amino acids [52].Markedly, all animal studies examining a maternal gluten-free (GF) diet as the dietary intervention consistently showed that a prenatal, GF environment protected against autoimmune diabetes and IA.As demonstrated by Antvorskov et al. [29], offspring of NOD mice, fed a GF diet only during pregnancy, had a significantly reduced diabetes incidence (from 8.3% in mice, where the mother was GF specifically during pregnancy, to 62.5% in offspring of mothers, eating a gluten-containing diet during pregnancy).Pups from mothers, fed a GF diet, also showed a significantly lower insulitis score in the same study [29].In mice, that were not GF in utero, this correlated with increased expression of RORγt, a nuclear transcription factor characteristic for Th17 cells, that was significantly changed by diet.These findings were supported by Hansen et al. who found that the intestinal gene expression profile was skewed towards an anti-inflammatory phenotype in offspring of GF diet-fed mice [27].
Although not detected in the present systematic literature review, a recent study also found that the insulitis score in GF mice was significantly reduced compared with mice fed a standard diet, and that the markers for regulatory T cells and T helper 2 cells were upregulated in the pancreas of GF mice [62].
The human DAISY study revealed that maternal consumption of neither glutencontaining foods nor non-gluten cereal grains during pregnancy was associated with the incidence or onset of IA in the children [44].In the DIPP cohort with 3727 children in 2011, Virtanen and colleagues found no association with advanced IA in children from mothers consuming cereal products [48].However, results from a large Danish National Birth Cohort Study [52], comprising 101,042 pregnancies, were published in BMJ in 2018.In this study, maternal gluten intake, based on maternal consumption of gluten-containing foods, was reported from a food frequency questionnaire during pregnancy.The average gluten intake was 13.0 g/day, ranging from less than 7 g/day to more than 20 g/day.The incidence of type 1 diabetes in the participants' children was 0.37% (n = 247) with a mean follow-up period of 15.6 years, and the risk of type 1 diabetes in the children increased proportionally with maternal gluten intake during pregnancy.Women with the highest gluten intake doubled the risk of type 1 diabetes development in their offspring, which illustrates a potential for gluten intake during pregnancy to significantly affect the childrens' risk for type 1 diabetes development, supporting the animal studies.However, the findings were not confirmed in the Norwegian birth cohort study, the MoBa cohort [54].The difference between the two studies could be due to the different gluten content in the wheat, rye, and barley used for manufacturing and consumption in the two different countries [52].

Fatty Acids
Evaluation of NOD mice offspring from 0-50 weeks after birth showed that offspring, born from and lactated by mothers fed a chow diet with high n-6/n-3 fatty acid ratio during the gestation and/or lactation period, developed diabetes faster than offspring from mothers, who were fed a chow diet with low ratio during pregnancy.Interestingly, pups fed a high-fat chow diet after weaning, but born from mothers fed a low-fat chow diet, had a significantly lower diabetes incidence compared to the group of pups fed a low-fat chow diet, but born from mothers fed a high-fat chow diet during gestation and/or lactation [26].Additionally, a maternal high-fat diet accelerated autoimmune diabetes development in offspring from NOD mice [28].Similarly, among 4887 children in the DIPP study, maternal intake of certain fatty acids, including EPA, was associated with a decreased risk of type 1 diabetes [49].However, in the DAISY study, no significant association was found between a combined intake of EPA and DHA and the risk of IA in the children [42], and in the TEDDY cohort, maternal consumption of n-3 fatty acid supplementation during pregnancy was not associated with the risk of IA in the offspring [53].

Iron Supplementation
A case-cohort study, using the Danish National Birth Cohort comprising approximately 100,000 Danish pregnant women, and CPR linkage to the Danish Childhood Diabetes Register, found no association between maternal consumption of iron during pregnancy and type 1 diabetes in the offspring [40].Iron overload can, however, lead to the formation of reactive oxygen species (ROS) in pancreatic islet cells which possess a classical iron metabolism and are sensitive to ROSs [63][64][65].In contrast, in the Norwegian MoBa study, it was found that prenatal exposure to iron could be a risk factor for type 1 diabetes [51].Importantly, maternal, but not fetal, human homeostatic iron regulator protein (HFE) genotypes, causing specific iron storage levels, were associated with offspring type 1 diabetes [51], potentially underlining the maternal dependency.Further, a very recent study focusing on the DIPP children found no association between IA or type 1 diabetes risks in children, exposed to iron in utero by maternal supplementation, when compared to nonexposed children [56].

Conclusions
In utero dietary deficiencies or overloads might alone, or in combination with other environmental factors, play a role in the process leading to the development of type 1 diabetes and IA in children.A potential mechanism of how diet during pregnancy could influence the development of IA and type 1 diabetes in the offspring could be the promotion of either an inflammatory or anti-inflammatory environment [29,31,62,66,67] affecting the child's developing immune system.Furthermore, maternal nutrition during pregnancy can potentially induce epigenetic changes in the child [68] or influence the composition of the resident microbiota that is transferred to the child during a vaginal birth.
This systematic review followed the PRISMA guidelines and evaluated the present literature focusing on in utero dietary exposures and the impact on type 1 diabetes and IA.Maternal consumption of nutritional products, such as gluten, iron, and high fat diet during pregnancy was associated with increased risks of type 1 diabetes and/or IA in the offspring, whereas in utero vitamin D and specific fatty acid exposures protected against these endpoints in some studies.However, these results represent a rather heterogeneous pool of data, and inconsistencies are present, illustrating the importance of research in this field.No prenatal, dietary factor has unequivocally been identified that modulate the frequency of T1D in the child.Thus, it is at present difficult to recommend or discourage specific components from the maternal diet to minimize the disease development in a child with high genetic risk for T1D.Our findings should initiate novel human, prospective, dietary intervention studies in both cohort and clinical settings, which are needed to better understand critical and protective prenatal exposures from the maternal diet during pregnancy.Conclusions from such studies could be an important way to prevent the development of type 1 diabetes in children at genetic risk of the disease.
A contributing factor that may be responsible for the lack of consistency in the current literature is that we likely do not know all the risk factors of IA and type 1 diabetes at present.This would explain the heterogenous results and it suggests basic research within this field to further elucidate the environmental and genetic determinants underlying type 1 diabetes.Importantly, committees, guiding and advising pregnant women concerning prenatal dietary exposure, need to carefully consider the findings in the literature.Furthermore, there is a lack of a concluding studies and review/meta-analysis investigating the effect of non-medical interventions (for example, the role of psychical activity, diet, probiotics, sugarcontaining beverages, psychosocial support) that could be used in IA positive children to prevent progression to type 1 diabetes.

Figure 1 .
Figure 1.PRISMA flow chart and summary of data extracted.(A) Flow chart of the systematic literature review and study selection process by following the PRISMA guidelines.Exclusion reasons summarized to the right.Find official PRISMA flow chart in Figure S1.(B) Pie chart of the included studies from (A) showing the proportion of the investigated dietary factors during pregnancy.Dietary factors investigated in the exploded category named "other" are shown in (C) in a heatmap format, where each dietary factor is scored according to how many times it has been investigated in the included articles from the "other" category.High-fat diet (HFD).

Figure 1 .
Figure 1.PRISMA flow chart and summary of data extracted.(A) Flow chart of the systematic literature review and study selection process by following the PRISMA guidelines.Exclusion reasons summarized to the right.Find official PRISMA flow chart in Figure S1.(B) Pie chart of the included studies from (A) showing the proportion of the investigated dietary factors during pregnancy.Dietary factors investigated in the exploded category named "other" are shown in (C) in a heatmap format, where each dietary factor is scored according to how many times it has been investigated in the included articles from the "other" category.High-fat diet (HFD).

Table 1 .
The impact of in utero dietary factor exposures to offspring in murine studies.
a Founder generation.b Generation one.c Generation two.d Advanced Glycosylated End-products.e Chow diets containing n-6/n-3 fatty acid ratios of 14.5 and 3.0, respectively.f Insulin autoantibody.g L-chow fed during gestation and lactation, but H-chow fed after weaning.h H-chow fed during gestation, lactation, and after weaning.i Gluten free diet.j 26.2% protein, 26.3% carbohydrates, 34.9% fat with 60% energy from fat. k Retinoic acid receptor-related orphan receptor γ t (a Th17 characteristic nuclear transcription factor).l Transglutaminase.m Interferon γ. n Isoflavone genistein.o Post natal day.

Table 2 .
The impact of in utero dietary factor exposures to offspring in human population-based case-control studies.
a Food frequency questionaries.

Table 3 .
The impact of in utero dietary factor exposures to offspring in human population-based birth cohort studies.

Table 3 .
Cont.The Diabetes Autoimmunity Study in the Young.b All Babies in Southeast Sweden.c The Type 1 Diabetes Predicition and Prevention Study.d The Norwegian Mother, Father and Child Cohort Study.e The Environmental Determinants of Diabetes in the Young. a