Discussion
Findings of This Study
To our knowledge, this study was the first to examine maternal iron intake in relation to ASD. We found that mothers of children with ASD had significantly lower iron intakes during the index period than mothers of children with TD. The highest quintile of maternal iron intake during the index period was associated with an approximately halved risk of ASD compared with the lowest quintile. The association between higher maternal iron intake and reduced ASD risk was strongest during breastfeeding, after adjustment for folic acid intake. Further, ASD risk associated with low maternal iron intake was much greater when the mother also was older at the time of the child's birth and/or had metabolic conditions during pregnancy.
Average iron intake from cereals and supplements throughout pregnancy for both groups in this study was above the Tolerable Upper Intake Limit for iron but was in line with national estimates of intake from supplements for pregnant women (48 mg/day) and women of childbearing age (18 mg/day). Notably, the lowest iron quintile (<30 mg/day) for the index period included the amounts recommended for pregnant (27 mg/day), nonpregnant (18 mg/day), and lactating (9 mg/day) women. If the results of this study are replicated, it would suggest that the current Recommended Daily Allowance for pregnancy and/or lactation could be set too low to prevent adverse outcomes in the child. A randomized, double-blind trial demonstrated that during pregnancy, 40 mg/day of supplemental iron was needed to prevent iron deficiency in the majority of women and that up to 80 mg/day was well-tolerated. On the other hand, even though the amount of iron absorbed is regulated by the gut to correspond to needs and avoid accumulation of excess iron in healthy individuals, very high iron intake in supplemental or fortified forms is known to be toxic and has been associated with negative health impacts, including adverse neurodevelopmental outcomes, and could be especially harmful for certain genetically susceptible individuals. The stronger association for higher maternal iron intake while breastfeeding is intriguing given the fact that maternal iron intake has been shown to not affect iron concentrations in breast milk; however, the very low concentrations of iron in breast milk (relative to those in serum) present measurement problems that could have obscured a relationship, and concentrations in the early postnatal period (before 9 months) have not been examined using modern iron indicators. Alternatively, maternal iron intake could influence breast milk production or composition in other ways.
This study examined maternal iron intake, not maternal or child iron status. Maternal circulation constitutes the only source of iron for the developing fetus, and maternal iron intake can influence both the mother's iron status and her child's status during brain development. Our findings may reflect compensation through iron supplementation for poorer iron status resulting from inadequate intake, inefficient uptake or metabolism, or increased needs for iron, producing a functional iron deficiency. It has been demonstrated, primarily in animal studies, that reduced iron supply at several stages of development generates enduring changes in dopamine neurotransmission that outlast the iron-deficient periods. Long-term effects are observed in adulthood, long after iron repletion, in hippocampal structure and function, monoamine metabolism, and myelination, indicating that early developmental periods are critical and that prevention of iron deficiency might be key for protecting against adverse neurodevelopmental outcomes. Iron deficiency can also impair the function of several enzymes that are directly involved in antioxidant and nucleic acid metabolism, which could affect genomic stability during periods of DNA synthesis and cell proliferation during development. In addition, mothers of children with ASD have been shown to have elevated markers of inflammation, and prenatal inflammation is an independent risk factor for ASD. Prenatal inflammation can produce a cytokine-mediated reduction of circulating nonheme iron, or hypoferremia, that can disrupt fetal brain development and lead to persistent structural and functional brain defects. Maternal iron supplementation has been shown to prevent effects of inflammation-induced hypoferremia.
Metabolic conditions like diabetes and obesity lead to iron deficiency, and as their prevalence continues to rise dramatically, suboptimal iron status during pregnancy can be expected to increase as well. Notably, metabolic conditions are independently associated with a 1.7-fold increased risk of ASD and nearly a 2-fold risk of developmental delays. Our study shows that the combination of maternal metabolic conditions and low supplemental iron intake is associated with a nearly 5-fold increased risk of ASD, and that the ASD risk associated with maternal metabolic conditions was nearly null for persons with the highest supplemental iron intake. This interaction effect, if replicated, implies that maternal supplemental iron is associated with prevention of ASD in children of mothers with metabolic conditions during pregnancy.
The significant interaction between older maternal age and lower supplemental iron intake seems biologically plausible given changes in iron metabolism and storage with age, especially in women. If this finding is replicated, more work would be needed to delineate the mechanistic pathways behind this interaction.
Limitations and Strengths of This Study
The retrospective reporting of vitamin and supplement information after the child's developmental status was known, whereby mothers were asked to recall a period several years before the interview, raises the issues of recall accuracy and bias in this study. A scenario in which recall bias explained part of the association between maternal iron intake and ASD would involve case mothers underreporting or control mothers overreporting their intake of supplements containing iron. Notably, the association between iron supplementation and reduced ASD risk was stronger when women recalled the information for a more recent pregnancy versus a less recent pregnancy, which argues against recall bias in this direction. However, we cannot rule out some role for differential recall across case status.
In addition, during these study years data were not collected on other dietary sources of iron, so information was not available with which to completely assess dietary iron intake. However, fortified cereals, which were included in this study, are the largest source of total dietary iron consumed in the United States. In addition, the amount of iron in supplements tends to outweigh the amount of iron found in the diet. Finally, iron supplementation is probably more amenable to prevention strategies than diet is.
We did not collect information on why the mothers took iron supplements, which are often taken after physician recommendations when iron deficiency anemia is detected during pregnancy. However, case mothers reported lower intake of iron from sources other than iron-specific supplements (breakfast cereal) that would not have been as likely to increase in response to an anemia diagnosis. In addition, differences in iron intake between groups were observed not only in late pregnancy, when anemia is more likely to be diagnosed, but throughout the index period. This provides evidence that the association was not entirely due to confounding by indication.
Supplemental intakes of folic acid and iron tend to be correlated, and thus it is difficult to examine their independent contributions. Odds ratios for quintiles of iron intake during the months before pregnancy and during early pregnancy were attenuated after adjustment for folic acid intake, as expected given that periconceptional folic acid is associated with reduced ASD risk. However, the association with maternal iron intake during the index period remained strong after adjustment for and stratification across folic acid intake. In addition, the association with iron-specific supplements, which was not correlated with folic acid, was consistently below the null and significant during breastfeeding. Differences in associations by year were probably artifactual, given that the associations were not influenced by year in a consistent pattern. Still, other unmeasured confounding factors associated with taking supplements could have played a role in our findings.
The efficiency of iron uptake and metabolism differs vastly between individuals on the basis of genetic differences. It is likely that genetically determined metabolic efficiency could modify the need for and the effects of iron supplementation. These genetic differences were not considered here, and they deserve further evaluation.
Strengths of this study include detailed information systematically collected on numerous potentially confounding variables, clinical confirmation of all ASD diagnoses, and confirmation of typical social and cognitive development for the population-based controls. Additionally, the study's large sample size allowed for stratification of results by case subgroup and by maternal and child factors likely to modify the association.
Conclusions
This study provides initial evidence for an association between increased maternal supplemental iron intake and reduced risk of ASD. Researchers should attempt to replicate this association in additional studies and to further delineate who is metabolically susceptible, clarify what dose of iron during pregnancy and breastfeeding is ideal for neurodevelopment, and identify and refine strategies for prevention of ASD through supplemental iron intake.