Fresh fish findings?

See corresponding article on page 1295.

Fish and other seafood are healthful foods that are rich in protein, low in saturated fat, and the primary dietary source for long chain n–3 (omega-3) PUFAs, essential nutrients important for optimal fetal brain development. However, fish may be contaminated with environmental pollutants, most notably methylmercury. These nutrients and toxicants may act in contrary directions to influence the same neurodevelopmental processes. The question thus continues to bubble up regarding whether maternal prenatal fish consumption on balance confers harm or benefit for offspring neurodevelopment.

In this issue of The American Journal of Clinical Nutrition, Hamazaki et al. (1) report associations of maternal prenatal fish consumption with offspring developmental outcomes at 6 and 12 mo in ∼80,000 mother–child pairs in the Japan Environment and Children's Study. They evaluated diet during pregnancy, including fish intake, with an FFQ, and parent-reported psychomotor development with the Ages and Stages Questionnaire (ASQ-3). Results indicated that women in the highest quintile of fish consumption had infants with lower odds of a low score (≥2 SDs below normal) on 1 of 5 subscales of the ASQ-3 at 6 mo of age, and 2 of 5 subscales at 12 mo. The authors also estimated n–3 and n–6 PUFA intake from the same FFQ, and found that both higher n–3 and n–6 PUFA intakes were associated with lower risk of lower developmental scores, but that a higher n–6:n–3 ratio was associated with an increased risk of delay in problem-solving skills. None of the results suggested harm from higher fish consumption.

These results nicely confirm what we already know. Whereas early investigations on this topic mainly focused on studies of only methylmercury or fatty acids, over the past decade studies have increasingly included investigation of fish intake as the independent variable of interest (2). These observational studies of prenatal fish intake have either shown beneficial associations or been null. In a 2019 systematic review of 29 articles published during 2001–2019 including a total of 106,237 mother–offspring pairs from 24 distinct cohorts, Hibbeln et al. (2) concluded that seafood consumption confers benefits to offspring neurocognitive development starting with the lowest amounts of seafood consumed in pregnancy (∼4 oz or ∼120 g/wk) compared with no seafood intake. No adverse effects on neurocognitive outcomes were reported from maternal seafood consumption despite high consumption of 30 to >100 oz/wk in some studies. However, almost all of these studies were conducted in the United States and Europe, and the median sample size was 498 mother–child pairs. The Hamazaki et al. article is thus confirmatory, while adding to the literature because of its very large sample size and also because it augments the number of non-European children.

The reason that this topic remains on the front burner of active investigations is primarily because these consistent findings from observational studies contrast with those from randomized controlled trials (RCTs) of n–3 PUFAs. A 2018 meta-analysis of RCTs that compared n–3 PUFA interventions (supplements or food) with placebo or no n–3 PUFAs found few differences in cognition, IQ, vision, other neurodevelopment and growth outcomes, language, or behavior, but recognized that evidence was mostly low quality to very low quality (3). There are several possible reasons for this disagreement. Firstly, a limitation of all observational studies is the likelihood of residual and unmeasured confounding, and this issue is certainly present in studies of fish intake. In the Japan Environment and Children's Study, predictors of fish consumption were generally those associated with higher socioeconomic status, and remarkably similar to what has been observed in US and European populations (4, 5): women with higher fish consumption were slightly older and more likely to be multiparous, have a higher education level and annual household income, be a current drinker, be a nonsmoker, and be more physically active. Moreover, because they used a parent-reported rather than researcher-measured measure of development, it is highly likely that these higher-socio-economic status (SES) parents who consumed more fish may have also over-reported their children's accomplishments. Thus, although this study was conducted in a different part of the world and includes a study population representative of Japan, the combination of confounding and social desirability bias probably does not substantially overcome limitations of the existing body of observational data.

A second reason for the discrepant findings between observational studies of habitual fish consumption and RCTs might be that the RCTs generally began after the end of the first trimester, and thus may have missed a sensitive window for neurodevelopment. Even if a dietary questionnaire specifically queries consumption in mid- or late pregnancy, the same women were probably higher fish consumers from preconception throughout pregnancy; and if anything fish consumption is likely to be lower during pregnancy compared with before because of nausea, avoidance of uncooked seafood such as sushi, and guidance about limiting exposure to mercury during pregnancy (6). It might be that ongoing exposure to n–3 PUFAs before and during early pregnancy is key to achieving the benefits.

Thirdly, it is possible that neurocognitive benefits from maternal prenatal seafood consumption accrue from more than just the n–3 PUFAs. Fish are rich in iodine, selenium, vitamin D, and other beneficial nutrients. Just like fruits and vegetables appear to provide health benefits that antioxidant supplements do not (7), it may be that seafood is more than a delivery system for n–3 PUFAs. RCTs providing or recommending fish as a whole food are possible and have shown that higher fish intake need not confer higher methylmercury intake (8, 9), but to date none have evaluated effects on offspring neurocognitive outcomes. However, three RCTs conducted in school-aged children have shown benefits for several neurocognitive outcomes (2). Such longer-term trials during pregnancy, if performed, would overcome the issue of confounding, but likely would not be able to overcome the challenge of timing.

At this time, more observational studies demonstrating benefits or lack of harm from prenatal fish consumption will likely not contribute fresh knowledge to this topic. Rather, the most helpful future work on this topic will need to include diverse populations, ideally with different confounding structures. Our knowledge will also benefit from a deeper dive into understanding which specific types of seafood, e.g., based on methylmercury content, are likely to be most beneficial.

Notes

Dr. Oken is funded by the National Institutes of Health, award UH3OD023286.