Patterns of perinatal exposure to PUFAs and child neurodevelopment: evidence from Mendelian randomization using FADS cluster variants

Abstract

Background: The potential causal effects of perinatal exposure to polyunsaturated fatty acids (PUFAs) on child neurodevelopment remains controversial.

Objective: To infer causation, we assessed the association of perinatal PUFA patterns and child neurodevelopment by using conventional regression analyses and 1-sample Mendelian randomization (MR).

Methods: Among 1096 mother-child pairs from the French Etude des Déterminants Pré- et Postnatals du Développement de la Santé de L'enfant cohort, patterns of perinatal exposure to PUFAs were previously identified combining PUFA levels from maternal and cord erythrocytes, and colostrum. Child verbal, performance, and full-scale intelligence quotients (IQs) were assessed at ages 5-6 y. Among maternal fatty acid desaturase (FADS) variants genotyped, 2 candidates, rs174546 (FADS1) and rs174634 (FADS3), were selected, as instrumental variables, for the MR analysis. The association of PUFA patterns with child IQ was examined by conventional multivariable linear regression and 2-stage least-squares MR regression.

Results: In the conventional approach, the first pattern "high omega-3 long-chain PUFAs (LC-PUFAs), low omega-6 LC-PUFAs" was positively associated with verbal IQ [β (95% confidence interval) = 1.24 (0.27, 2.21) points per 1 standard deviation (SD) increase in pattern] and full-scale IQ [1.11 (0.18, 2.05)]. This pattern was independent of FADS variants, rendering MR analysis inapplicable. The third pattern, "colostrum LC-PUFAs," was positively associated with verbal [1.11 (0.19, 2.02)], performance [1.01 (0.09, 1.93)], and full-scale IQ [1.13 (0.25, 2.01)]. The MR approach, based on genetic instruments strongly associated with the third pattern, supported the beneficial effect on performance IQ [2.93 (0.05, 5.81) points per 1 SD increase in genetically predicted pattern]. The MR also suggested a deleterious effect of the fourth pattern "linoleic acid (LA) and dihomo-gamma-linolenic acid (DGLA)" on performance IQ [-1.66 (-3.22, -0.09)].

Conclusions: These findings supported the potential beneficial effects of perinatal exposure to LC-PUFAs on child neurodevelopment while highlighting possible adverse effects associated with exposure to LA and DGLA.

Keywords: EDEN mother–child cohort; FADS variants; neurodevelopment; patterns of perinatal exposure to PUFAs; polyunsaturated fatty acids.