Iron Status Predicts EEG Measures of Brain Function During a Memory Task in 4–6 Month Old Infants and May Reveal Associations Between Early Iron Status and Memory Behavior

Abstract

BackgroundIron deficiency (ID) during late gestation and infancy—the period of peak brain development—is associated with longterm cognitive deficits. Infants of weaning age (4–6 mo) have an elevated risk of low iron stores, yet no clinical definition of ID exists for infants <12 mo. Previous studies are limited by non‐specific measures of cognition.ObjectiveWe hypothesized that decreased neural responses mediate the relationship between low iron status and memory performance.MethodsIn a cross‐sectional study of infants (4–6 mo, n = 23), we assessed iron status with hemoglobin (Hb) and serum ferritin (sFt), as well as brain activity with electroencephalography (EEG) during a memory task and novelty preference (%Nov) during a second memory task. Both tasks measured infant response to novel versus familiar visual stimuli. We used mixed regression models to determine whether the between‐stimulus difference in amplitude of the EEG “slow wave” (ΔLSW), a known correlate of memory, mediates the association between iron status and novelty preference. Relevant covariates were also included.ResultsAs expected, the LSW was more negative for novel vs control stimuli (p = 0.04). For every 1 μV change in ΔLSW, %Nov increased 0.52 ± 0.26 (SE)% (p=0.05). The ΔLSW parameter estimate increased to 0.56 ± 0.24 (p=0.02) when anemia (Hb <110 g/L; β= −4.97 ± 1.64, p<0.01) and age (β= 0.65 ± 0.26, p=0.01) were added to the model. Having low iron stores (sFt <45.0 μg/L) produced a similar result as anemia, except that ΔLSW was no longer significantly associated with %Nov after adding ferritin status to the model.ConclusionsThe infants had a greater neural response to novel versus familiar stimuli, and anemia status predicted the magnitude of ΔLSW. Preliminary analyses suggest that ΔLSW was not a mediator because the relationship between anemia and %Nov was strengthened, not diminished, when ΔLSW was added to the model. ΔLSW could be a negative confounder. The biological theory and data from animal studies provide a strong case for the presence of a mediating relationship between iron status, brain function, and memory performance in early infancy. Our methods should be repeated in a longitudinal follow‐up and ideally a randomized controlled trial, as it may be that the hypothesized mediation is only related to the change in Fe status.Support or Funding InformationFunded by USDA/CUAES (#NYC‐399413) & Cornell Human Ecology Alumni Association; JN supported by NSF GRFP