Choline Intake During Pregnancy and Child Cognition at Age 7
Choline Intake During Pregnancy and Child Cognition at Age 7
In this prospective cohort study, higher maternal second-trimester choline intake was associated with modestly higher child memory score at age 7 years as measured by the WRAML2 Design and Picture Memory subtests. First-trimester choline intake was also positively, albeit more weakly, associated with age 7 WRAML2 score. There was a suggestive positive association between second-trimester choline intake and nonverbal KBIT-2 score as well. However, intakes of vitamin B12, betaine, and folate were not associated with scores on any of the cognitive tests.
Our finding that maternal gestational choline intake was positively associated with child visual memory is consistent with robust data from animal models. For example, offspring of rat dams supplemented with choline in mid- to late gestation (days 11–17 or 18) performed better on the Morris water maze and 12-arm radial maze tests. Choline intake may alter brain function through formation of acetylcholine affecting cholinergic transmission or changes in DNA methylation leading to differentiation and apototosis of neurons. In mice, choline supplementation of pregnant dams from days 12–17 of gestation caused differences in gene-specific methylation and protein expression in the fetal hippocampus, the region of the brain associated with memory. These changes in the brain lasted through at least age 24 months and were associated with increased cell proliferation and decreased apoptosis in fetal hippocampi. Moreover, these cellular results were consistent in rats, in vivo and in vitro, and, interestingly, in cultured human neuroblastoma cells.
We found a mean adjusted difference of 1.4 points in WRAML2 score between the highest and lowest quartiles of second-trimester choline intake. Unlike the usual IQ test with a mean score of 100 points, the mean for this test is about 17 points. The adjusted difference we found is approximately one-third of the standard deviation in this study population (4.4 points). Yasik et al. found a similar order of magnitude (2.0 points) of difference in WRAML2 Design and Picture Memory scores between children and adolescents with posttraumatic stress disorder and those without the disorder. Previous studies have found that better working memory is associated with superior scholastic skills, including arithmetic, reading, and writing, and general academic achievement in school-aged children. Therefore, while the difference in memory in our study based on choline intake was modest and did not translate into overall IQ, it may be relevant in terms of the academic potential of the participants.
We found a stronger association of child memory with second-trimester choline intake than with first-trimester intake. This finding may suggest a stronger effect of choline on brain formation in midgestation than early in pregnancy. Most developmental animal studies focus on choline supplementation in mid- to late gestation, when cholinergic neurons in the forebrain undergo final mitotic division. However, previous animal studies did not directly compare effects of choline in early gestation and midgestation on memory. It is possible that the differences across gestational age are due to chance.
Our finding of a positive association between gestational choline intake and child memory is novel in humans. Signore et al. found that serum choline in umbilical cord blood was not associated with child IQ score. However, the participants in that study came from a disadvantaged inner-city population, in which other factors may have influenced cognition more dramatically than diet. In contrast, the participants in our study were predominantly well-educated and of relatively higher socioeconomic status, so it is possible that we were better able to study the subtle association between diet and cognition in this population. Another difference is that Signore et al. examined serum levels, which may be influenced by other internal factors in the body, whereas we evaluated dietary intake. It is possible that the serum choline results reflected recent choline intake rather than long-term intake, although, as Signore et al. noted, women tended to have consistently high or low choline intakes across multiple time points during pregnancy. In addition, Signore et al. did not adjust for other potential confounders, such as maternal intake of fish or other methyl donors during pregnancy, parity, or paternal education, although in our study population there was little evidence of confounding by these factors. Finally, the full IQ test does not specifically measure visuospatial memory, which was the domain affected by choline intake in animal models. In a previous Project Viva analysis, we did not find meaningful associations between maternal choline intake and PPVT-III or WRAVMA scores at age 3 years, but these tests do not specifically assess visuospatial memory. Signore et al. isolated memory components of the IQ test and still did not find an association between choline and these outcomes (e.g., a 1-unit increase in cord serum choline z score was associated with a mean difference of 0.18 points on the block design subtest; P = 0.22), but in our study we were able to use the WRAML2 Design and Picture Memory subtests, which are probably more representative of the domains affected by choline in animal models. This distinction may explain why we found an association in our study despite the previous null findings.
The association between second-trimester choline intake and WRAML2 score was stronger in males, although the P value for interaction was well above 0.05. Men may be more susceptible to choline deficiency than premenopausal women, because estrogen promotes de novo choline production through the phosphatidylethanolamine N-methyltransferase pathway. Because of differences in choline metabolism, the adequate intake levels set for choline in the United States by the Food and Nutrition Board of the Institute of Medicine are higher for men than for women, although they are identical for male and female children. Few animal studies of gestational choline and offspring cognition have considered effect modification by sex, but Williams et al. found a greater effect of in utero choline supplementation on cholinergic neural cell size and memory tests in male rats. Since power was limited for stratified analysis in our study and these findings were post hoc, we suggest that future studies consider differences by sex to confirm or refute our results.
Our findings that vitamin B12 and folate were not associated with child cognition differed from some previous studies showing positive associations between gestational intake of these nutrients and child cognition. These studies used different cognitive tests, which could explain the differences in our results. We previously found that first-trimester maternal folate intake was associated with a modestly higher PPVT-III score (1.3 points for each 600-µg/day increment (95% CI: 0.1, 3.1)) at age 3 years, which may be explained by differences in cognitive domains tested at ages 3 and 7 years. The differences in our findings at age 7 years from other studies may also be due to the fact that our participants, unlike participants in other studies, were generally folate-replete and had adequate vitamin B12 intake. For example, in our cohort, only 4 women (0.5%) did not meet the Recommended Daily Allowance for vitamin B12 of 2.6 µg/day in the first trimester of pregnancy, while in the Pune study, nearly half of the women had low plasma B12 levels. It is plausible that folate, vitamin B12, and other methyl donors are also important for brain development and cognition but that beyond the high level of these nutrients in our study due to fortification and supplementation, there is no additional benefit for child cognition.
Our analysis had several potential limitations. First, dietary intake is always difficult to measure in epidemiologic studies, since intake of nutrients is continuous and constantly changing. The study visits did not occur at the precise end of the first or second trimester for every woman, which may have added some measurement error in the timing of intake. In addition, conducting the analysis by quartile of nutrient intake introduces imprecision in the exposure, reduces power, and could bias multivariable-adjusted results. However, we prospectively collected dietary data using a modified FFQ calibrated for use during pregnancy that was similar to FFQs shown in previous studies to validly measure choline and other methyl donor nutrients. After adjustment for total energy intake, the FFQ should accurately rank individual intakes of nutrients, and the quartiles should discriminate between women with very high and very low intakes. Second, there was potential for error in measurement of the cognitive domains, but the tests were administered by trained research assistants, and any error in the dependent variable would probably have caused reduced precision in our confidence intervals, making our findings conservative. Third, we only had data on dietary intake from the first and second trimesters of pregnancy, not on third-trimester, postnatal, or child diet. We may have missed associations if intake during the third trimester is important in cognitive development in the offspring. Fourth, there was the potential for unmeasured confounding, particularly from maternal and paternal memory in the WRAML2 analysis. However, we measured many environmental, sociodemographic, and biological covariates, especially parental education, maternal KBIT-2 score, and HOME score, that did not substantially attenuate the associations between choline and child memory.
Loss to follow-up is another concern, given that more than half of the original cohort was not included in this analysis and that those lost to follow-up were more likely to be racial/ethnic minorities and of lower socioeconomic status. However, choline intake did not differ according to follow-up, results were similar when we limited our analysis to underrepresented racial and socioeconomic groups, and associations were robust despite adjustment for confounding. Finally, our finding of a positive association between gestational choline intake and age 7 WRAML2 score may have been due to chance, since we examined many associations. However, these findings are consistent with the robust animal literature showing a causal relationship between gestational choline intake and offspring memory and performance. Other strengths of this study include its prospective design, detailed dietary information, and large sample size.
In conclusion, higher maternal gestational intake of choline, but not intake of other methyl donors, was associated with modestly better child memory at age 7 years as measured by WRAML2 score. We also found a suggestive positive association of maternal second-trimester choline intake with KBIT-2 nonverbal score. In the future, investigators should examine this association in other studies and assess benefits and risks of choline intake in multiple populations before we can recommend choline supplementation in pregnancy.
Discussion
In this prospective cohort study, higher maternal second-trimester choline intake was associated with modestly higher child memory score at age 7 years as measured by the WRAML2 Design and Picture Memory subtests. First-trimester choline intake was also positively, albeit more weakly, associated with age 7 WRAML2 score. There was a suggestive positive association between second-trimester choline intake and nonverbal KBIT-2 score as well. However, intakes of vitamin B12, betaine, and folate were not associated with scores on any of the cognitive tests.
Our finding that maternal gestational choline intake was positively associated with child visual memory is consistent with robust data from animal models. For example, offspring of rat dams supplemented with choline in mid- to late gestation (days 11–17 or 18) performed better on the Morris water maze and 12-arm radial maze tests. Choline intake may alter brain function through formation of acetylcholine affecting cholinergic transmission or changes in DNA methylation leading to differentiation and apototosis of neurons. In mice, choline supplementation of pregnant dams from days 12–17 of gestation caused differences in gene-specific methylation and protein expression in the fetal hippocampus, the region of the brain associated with memory. These changes in the brain lasted through at least age 24 months and were associated with increased cell proliferation and decreased apoptosis in fetal hippocampi. Moreover, these cellular results were consistent in rats, in vivo and in vitro, and, interestingly, in cultured human neuroblastoma cells.
We found a mean adjusted difference of 1.4 points in WRAML2 score between the highest and lowest quartiles of second-trimester choline intake. Unlike the usual IQ test with a mean score of 100 points, the mean for this test is about 17 points. The adjusted difference we found is approximately one-third of the standard deviation in this study population (4.4 points). Yasik et al. found a similar order of magnitude (2.0 points) of difference in WRAML2 Design and Picture Memory scores between children and adolescents with posttraumatic stress disorder and those without the disorder. Previous studies have found that better working memory is associated with superior scholastic skills, including arithmetic, reading, and writing, and general academic achievement in school-aged children. Therefore, while the difference in memory in our study based on choline intake was modest and did not translate into overall IQ, it may be relevant in terms of the academic potential of the participants.
We found a stronger association of child memory with second-trimester choline intake than with first-trimester intake. This finding may suggest a stronger effect of choline on brain formation in midgestation than early in pregnancy. Most developmental animal studies focus on choline supplementation in mid- to late gestation, when cholinergic neurons in the forebrain undergo final mitotic division. However, previous animal studies did not directly compare effects of choline in early gestation and midgestation on memory. It is possible that the differences across gestational age are due to chance.
Our finding of a positive association between gestational choline intake and child memory is novel in humans. Signore et al. found that serum choline in umbilical cord blood was not associated with child IQ score. However, the participants in that study came from a disadvantaged inner-city population, in which other factors may have influenced cognition more dramatically than diet. In contrast, the participants in our study were predominantly well-educated and of relatively higher socioeconomic status, so it is possible that we were better able to study the subtle association between diet and cognition in this population. Another difference is that Signore et al. examined serum levels, which may be influenced by other internal factors in the body, whereas we evaluated dietary intake. It is possible that the serum choline results reflected recent choline intake rather than long-term intake, although, as Signore et al. noted, women tended to have consistently high or low choline intakes across multiple time points during pregnancy. In addition, Signore et al. did not adjust for other potential confounders, such as maternal intake of fish or other methyl donors during pregnancy, parity, or paternal education, although in our study population there was little evidence of confounding by these factors. Finally, the full IQ test does not specifically measure visuospatial memory, which was the domain affected by choline intake in animal models. In a previous Project Viva analysis, we did not find meaningful associations between maternal choline intake and PPVT-III or WRAVMA scores at age 3 years, but these tests do not specifically assess visuospatial memory. Signore et al. isolated memory components of the IQ test and still did not find an association between choline and these outcomes (e.g., a 1-unit increase in cord serum choline z score was associated with a mean difference of 0.18 points on the block design subtest; P = 0.22), but in our study we were able to use the WRAML2 Design and Picture Memory subtests, which are probably more representative of the domains affected by choline in animal models. This distinction may explain why we found an association in our study despite the previous null findings.
The association between second-trimester choline intake and WRAML2 score was stronger in males, although the P value for interaction was well above 0.05. Men may be more susceptible to choline deficiency than premenopausal women, because estrogen promotes de novo choline production through the phosphatidylethanolamine N-methyltransferase pathway. Because of differences in choline metabolism, the adequate intake levels set for choline in the United States by the Food and Nutrition Board of the Institute of Medicine are higher for men than for women, although they are identical for male and female children. Few animal studies of gestational choline and offspring cognition have considered effect modification by sex, but Williams et al. found a greater effect of in utero choline supplementation on cholinergic neural cell size and memory tests in male rats. Since power was limited for stratified analysis in our study and these findings were post hoc, we suggest that future studies consider differences by sex to confirm or refute our results.
Our findings that vitamin B12 and folate were not associated with child cognition differed from some previous studies showing positive associations between gestational intake of these nutrients and child cognition. These studies used different cognitive tests, which could explain the differences in our results. We previously found that first-trimester maternal folate intake was associated with a modestly higher PPVT-III score (1.3 points for each 600-µg/day increment (95% CI: 0.1, 3.1)) at age 3 years, which may be explained by differences in cognitive domains tested at ages 3 and 7 years. The differences in our findings at age 7 years from other studies may also be due to the fact that our participants, unlike participants in other studies, were generally folate-replete and had adequate vitamin B12 intake. For example, in our cohort, only 4 women (0.5%) did not meet the Recommended Daily Allowance for vitamin B12 of 2.6 µg/day in the first trimester of pregnancy, while in the Pune study, nearly half of the women had low plasma B12 levels. It is plausible that folate, vitamin B12, and other methyl donors are also important for brain development and cognition but that beyond the high level of these nutrients in our study due to fortification and supplementation, there is no additional benefit for child cognition.
Our analysis had several potential limitations. First, dietary intake is always difficult to measure in epidemiologic studies, since intake of nutrients is continuous and constantly changing. The study visits did not occur at the precise end of the first or second trimester for every woman, which may have added some measurement error in the timing of intake. In addition, conducting the analysis by quartile of nutrient intake introduces imprecision in the exposure, reduces power, and could bias multivariable-adjusted results. However, we prospectively collected dietary data using a modified FFQ calibrated for use during pregnancy that was similar to FFQs shown in previous studies to validly measure choline and other methyl donor nutrients. After adjustment for total energy intake, the FFQ should accurately rank individual intakes of nutrients, and the quartiles should discriminate between women with very high and very low intakes. Second, there was potential for error in measurement of the cognitive domains, but the tests were administered by trained research assistants, and any error in the dependent variable would probably have caused reduced precision in our confidence intervals, making our findings conservative. Third, we only had data on dietary intake from the first and second trimesters of pregnancy, not on third-trimester, postnatal, or child diet. We may have missed associations if intake during the third trimester is important in cognitive development in the offspring. Fourth, there was the potential for unmeasured confounding, particularly from maternal and paternal memory in the WRAML2 analysis. However, we measured many environmental, sociodemographic, and biological covariates, especially parental education, maternal KBIT-2 score, and HOME score, that did not substantially attenuate the associations between choline and child memory.
Loss to follow-up is another concern, given that more than half of the original cohort was not included in this analysis and that those lost to follow-up were more likely to be racial/ethnic minorities and of lower socioeconomic status. However, choline intake did not differ according to follow-up, results were similar when we limited our analysis to underrepresented racial and socioeconomic groups, and associations were robust despite adjustment for confounding. Finally, our finding of a positive association between gestational choline intake and age 7 WRAML2 score may have been due to chance, since we examined many associations. However, these findings are consistent with the robust animal literature showing a causal relationship between gestational choline intake and offspring memory and performance. Other strengths of this study include its prospective design, detailed dietary information, and large sample size.
In conclusion, higher maternal gestational intake of choline, but not intake of other methyl donors, was associated with modestly better child memory at age 7 years as measured by WRAML2 score. We also found a suggestive positive association of maternal second-trimester choline intake with KBIT-2 nonverbal score. In the future, investigators should examine this association in other studies and assess benefits and risks of choline intake in multiple populations before we can recommend choline supplementation in pregnancy.
Source...