Missing Research on Preventing Birth Defects
Higher-dose vitamins could make more healthy babies -- but would they also kill some mothers in the process?
What do neural tube defects (NTD) including spina bifida and anencephaly, facial clefts, Down’s syndrome, and neurodevelopmental disorders like autism and ADHD have in common?
Research suggests they may all be linked to maternal folate deficiency before and during pregnancy, including problems with folate metabolism. And, when it comes to preventing some birth defects, folate supplementation, usually with folic acid (synthetic folate), is a well-established intervention.
It’s one of the great success stories of late 20th century public health: By adding folic acid to flour and other “daily bread” foodstuffs, more than 80 countries reduced the prevalence of a number of birth defects, including potentially lethal ones. Costa Rica saw the largest NTD reduction after putting folic acid not just in wheat flower, but also maize flour, cow’s milk, and rice. Data from Chile show there can be large variations in how much folic acid actually gets into flour — with some levels dangerously low for NTD prevention, and others dangerously high for other risks…
Folic Acid Supplementation and Cancer
Those other risks have to do with cancer. In a randomized trial intended to assess whether folic acid could prevent colorectal adenomas in 1021 people with a recent history of them without invasive large intestine carcinoma, Cole et al discovered instead that “Folic acid was associated with higher risks of having 3 or more adenomas and of noncolorectal cancers” (JAMA 2007).
In a broader view, this result was not sui generis. The same thing seems to have happened with beta carotene and high-dose vitamin A when researchers conduced a randomized trial to assess whether they could reduce lung cancer in smokers, former smokers, and workers who had been exposed to asbestos (Omenn et al, “Effects of a combination of beta carotene and vitamin A on lung cancer and cardiovascular disease,” NEJM 1996 May 2;334(18):1150-5). In this case, they stopped the trial early because it looked like they were accidentally killing people: The treatment group’s risks of lung cancer, all-cause death, lung cancer death, and cardiovascular disease death were higher (95% CIs 1.04-1.57; 1.03-1.33; 1.07-2; and .99-1.61).
So vitamins are not always good for you. Folic acid supplementation specifically may be dangerous if you’ve had or are at risk for colon cancer. Unless it’s actually protective, after all…
The evidence on folate and colorectal cancer risk is mixed, ambiguous, and some evidence suggests possible protective effects, too. Reviewing that evidence, Kim suggests:
Folate deficiency has an inhibitory effect whereas folate supplementation has a promoting effect on the progression of established colorectal neoplasms. In contrast, folate deficiency in normal colorectal mucosa appears to predispose it to neoplastic transformation, and modest levels of folic acid supplementation suppress, whereas supraphysiologic supplemental doses enhance, the development of cancer in normal colorectal mucosa (“Folate and colorectal cancer: an evidence-based critical review,” Young-In Kim, Mol Nutr Food Res, 2007 Mar;51(3):267-92).
But it’s not just colorectal cancer, and it’s not just one substance with one effect that we’re talking about here. Figueiredo et al reported random assignment to folic acid supplementation increased prostate cancer risk (see Figure 2, “probability of remaining free of prostate cancer, age-adjusted HR 95% CI 1.23-5.65,” “Folic acid and risk of prostate cancer: results from a randomized clinical trial,” J Natl Cancer Inst, 2009 Mar 18;101(6):432-5).
“In contrast,” they noted, “baseline dietary folate intake and plasma folate in nonmultivitamin users were inversely associated with risk of prostate cancer, although these associations did not attain statistical significance in adjusted analyses.” In other words, synthetic folic acid might increase some cancer risks, while natural dietary folate intake might decrease them. This would be consistent with some evidence on similarly contrasting effects of vitamin A supplementation (particularly high-dose and in at-risk subgroups) versus dietary beta carotene intake.
So men seem to incur some particular cancer risks from folic acid supplementation. What about women? This question is especially important, since women are universally advised to take folic acid if they are or may become pregnant.
Moreover, colorectal tissue shares some potentially salient characteristics with breast tissue. Like colorectal mucosa, breast tissue during pregnancy is rapidly proliferative. This matters because, if rapidly proliferative pregnant breast tissue responds to high-dose folic acid supplementation the same way as rapidly proliferative colorectal tissue, then one might expect to see increased breast cancer deaths in women who took high-dose folic acid in pregnancy.
Folic acid and breast cancer
That is what evidence from a 1960s randomized trial with 2928 Aberdeen women seemed to show: Charles et al found a daily dose of 5 mg was associated with substantial possible increases in all-cause mortality (95% CI 0.84-1.71), cardiovascular mortality (95% CI 0.36-2.12), all-cancer mortality (95% CI 0.90-2.27), and especially breast cancer mortality (95% CI 0.83-4.47), as compared with the daily dose of 200 mcg and the placebo (“Taking folate in pregnancy and risk of maternal breast cancer,” BMJ 2004;329:1375; PDF). They noted, though, that the confidence intervals were wide these findings could be due to chance.
Follow-up analyses on the same evidence at a later date spun its findings as being due to chance and disconfirming the possible risks of folic acid supplementation in pregnancy. But they actually also found similarly substantial possible risks, especially from high-dose supplementation and especially for breast cancer mortality. Taylor et al misused statistical significance testing, repeatedly saying they found “no associations” when they actually found evidence of substantial possible associations (“Folic acid in pregnancy and mortality from cancer and cardiovascular disease: further follow-up of the Aberdeen folic acid supplementation trial,” J Epidemiol Community Health, 2015 Aug;69(8):789-94). Table 3 shows this evidence again has wide confidence intervals, such that the findings could be due to chance. But they could also reflect substantial risks (e.g., a 95% CI 0.56-2.29 for high-dose folic acid and breast cancer mortality risk).
Koenig et al noted two prospective cohort studies linked high folic acid and breast cancer risk (the Prostate, Lung, Colorectal, and Ovarian (PLCO) Cancer Screening Trial and the Swedish Mammography Cohort). They also noted, however, Misotti et al observed other prospective studies found no association. Then, Koenig et al reported results from a nested case-control study finding substantial possible risks of invasive breast cancer associated with serum unmetabolized folic acid (Table 3) (“Circulating unmetabolized folic acid and 5-methyltetrahydrofolate and risk of breast cancer: A nested case-control study,” Eur J Clin Nutr. 2020 Sep; 74(9): 1306–1315). Of note, they looked at cases diagnosed before mandatory U.S. grain fortification with folic acid.
Conversely, not all evidence on folic acid and breast cancer risks points in this direction. In line with the mixed nature of evidence on colorectal cancer risk, other breast cancer research suggests possible protective effects:
In a Canadian case-control study on never or ever supplement use in BRCA1 and BRCA2 mutation carriers, Kim et al found decreased breast cancer risk associated with folic acid supplementation (95% CI 0.25-0.79) (Breast Cancer Res Treat, 2019 Apr;174(3):741-748). (This seems to be a subgroup analysis, because its reported p-value is .006, and Kim et al then add “This was significant for BRCA1 mutation carriers only.”) A moderate dose may have been more protective than a higher dose, and moderate B12 supplement use was also associated with decreased breast cancer risk (95% CI 0.24-0.96), again especially for BRCA1 carriers.
Wang et al found “Folic acid prevents the progesterone-promoted proliferation and migration in breast cancer cell lines” (Eur J Nutr, 2020 Sep;59(6):2333-2344).
Sie et al found “Folic acid supplementation provided in utero and during lactation reduces the number of terminal end buds of the developing mammary glands in the offspring” (Cancer Lett, 2009 Jul 18;280(1):72-7). They noted terminal end buds reliably predict mammary tumor risk at adulthood in rodents, and so folic acid supplementation during pregnancy may lower breast cancer risk in offspring— in rats.
Overall
Women who are or want to become pregnant don’t take folic acid on the chance that it might reduce their potential future daughters’ breast cancer risks. They do it because it’s recommended, at least at a low dose (400 mcg/day), for preventing birth defects. Moreover, supplementing with higher-dose folic acid than is generally recommended “has been consistently related to lower frequency of infertility, lower risk of pregnancy loss, and greater success in infertility treatment.”
This complex picture of folic acid supplementation risks and benefits puts public health authorities in the unenviable position of quietly trading an estimated probable increase in some cancers for an estimated probable decrease in pregnancy losses, elective abortions of non-viable pregnancies, neonatal deaths, and developmental disabilities. They balance these conflicting interests by fortifying foods like flour at a lower level than is probably optimal for a lot of reproductive-aged women, and a higher level than would probably be best for a lot of older people… A balance that observers like Mason et al argued has resulted in small increases in colorectal cancers in places that implemented flour fortification programs, like the U.S. and Canada.
Synthesizing this and other evidence, Vidailhet et al observed folic acid fortification also correlated with decreased stroke and coronary accidents in elderly people, on one hand — but impaired cognitive functions with B12 deficiency, on the other. They concluded, on the whole, that France was wise to forego fortifying foodstuffs with folic acid to prevent NTD, because the trade-offs were unclear — and the evidence suggestive it could do harm as well as good. In contrast, Morris et al call Europe’s failure to fortify flour with folic acid to prevent NTDs “A Public Health Failure.” Whatever you think about that, there are around 195 countries in the world today, and less than half of them mandate this fortification. It’s a divided world.
Folic Acid and At-Risk Moms
Zooming back in on women who want to have healthy babies, but who are at-risk for a variety of pregnancy problems — are they in a double bind? Does taking high-dose folic acid to protect and promote normal fetal development come at the price of heightened cancer risks for these subgroups? And if so, what exactly are the trade-offs?
We don’t know. On one hand, maybe before public health and medical authorities go about promoting heavier folate supplementation in a broader range of subgroups, we should find out. On the other hand, there are a lot of subgroups who research already suggests might benefit from more preconception and antenatal folate:
Women taking the anti-epilepsy drug valproate (more on this later).
Women taking other potentially teratogenic medications and/or folate antagonists.
Women who drink alcohol before and during pregnancy (see, e.g., some evidence in animals and humans).
Women who smoke.
Women who were themselves adversely impacted by valproate exposure in utero; and, by extension, maybe people so affected by other teratogens.
Neurodivergent women, Hispanics, Celts, Sikhs, Northern Chinese, and others at heightened risk of common MTHFR polymorphisms that can lead to higher homocysteine and accordingly lower serum folate levels.
Older women and others at heightened risk of Down’s syndrome pregnancies (e.g., women have who had one before or are at risk of early menopause).
Women at heightened risk of NTD pregnancies, because they had one or more before. In this context alone, researchers note that “Uncertainty as to which high risk groups benefit from high dose folic acid use has led to variations in guidelines internationally.” This uncertainty pervades all these subgroups with regards to whether there is a benefit, what it is, what form of folate is best, what dose is best, and whether other vitamins (e.g., B12) are also necessary (more on this later).
Women at risk of fever or overheating; and, by extension, maybe women vulnerable to adverse health effects from global warming. (Here is some evidence on periconception maternal fever, risk of NTDs and congenital heart diseases, and folic acid.) This one is big: We need to know if global warming is likely to increase potentially fatal birth defects, and higher-dose folic acid supplementation can prevent this.
Maybe men belong in some or all these subgroups, too. More research is needed, with paternal preconception health woefully understudied. Previously, I looked at risks of antidepressants in pregnancy, and noted evidence suggesting men don’t get a free pass when it comes to substantial possible neurodevelopmental risks of preconception exposure.
Many Unknowns
There’s a lot we don’t know: Does supplementing only women before conception with extra folate improve pregnancy and child health outcomes — or would supplementing men more also help? On what terms do we trade better pregnancy and child health outcomes in some for increased cancers in others? To what extent do these groups overlap, such that it’s not just governments mandating flour fortification, but also clinicians recommending additional preconception/antenatal supplementation and women taking prenatal vitamins who need to think about this balance?
We’re also not sure what form of folate is best. Folic acid (synthetic folate) is the standard recommendation, but some evidence suggests it may be associated with heightened cancer risks — while dietary folate may be inversely associated with those risks. Taking the more bioavailable form, methylfolate, prenatally probably makes sense for people at risk of deficiency, including those on folate antagonists and with MTHFR polymorphisms.
We’re not sure what dose is minimally effective for NTD prevention. But the effect is dose-responsive, leading some to recommend women planning a pregnancy should take 5 mg/day. That’s folic acid. We don’t know if methylfolate has a lower effective dose due to its higher bioavailability; and, if so, what that dose is.
Some research suggests B12 also decreases NTD. It’s not uncommon to see folic acid packaged with B12 as a supplement, because otherwise folic acid supplementation can mask B12 deficiency, which can then progress to neurological damage. B12 also comes in the more bioavailable methylcobalamin variety. We would like to know more about how these forms compare, effective prenatal dosing, and interactions of folate with B12.
We don’t know if the more bioavailable methylfolate and methylcobalamin forms are less risky than other (standard) forms in terms of cancer. Some say methylfolate may be. Others note methyl-donors including methylfolate and methylcobalamin interfere with tumor proliferation and may support chemotherapy. But we still don’t know whether methyl-donor folate and B12 supplementation increases, decreases, or has no effect on cancer risks.
We also don’t know if methyl-donor supplementation leads to net beneficial or promiscuous, likely deleterious epimutations. Specifically, we don’t know if such supplementation regimes “are preventative or exacerbate the risk for Autism Spectrum Disorders (ASD) in children.”
We’re don’t know if supplemental folate is necessary at all, or if it could be replaced by a high-folate dietary educational and supportive intervention modeled on SMILES (Jacka et al’s randomized trial on a Mediterranean diet for depression). If there’s a possible cancer risk of vitamin supplementation, versus a likely cancer benefit of eating a healthy diet, it would be useful to know if eating a folate-rich diet alone is enough to prevent NTD and otherwise improve pregnancy/child health outcomes while likely lowering instead of potentially raising future cancer risks. But it would probably be considered unethical to run a trial with no preconception/antenatal folate supplementation, to find out.
Target Trial Necklace
Bracketing population-level effects of mandatory fortification, one overarching question here is how we can best balance fetal and parental interests by improving pregnancy and child health outcomes without increasing parental cancers. So an idealized set of trials might include studies with the identified at-risk subgroups, primary short-term endpoints of pregnancy outcomes, and medium-term endpoints of child health outcomes. Then, those studies could be aggregated later to analyze primary long-term endpoints of parental all-cause, cardiovascular, cancer, and maternal breast cancer death.
Since there’s no consensus on it currently, the trials could randomize men to preconception supplement and control groups.
Some guidelines already recommend high-dose folate supplementation for at-risk groups. One way to approach open empirical questions about benefits and risks of this intervention is to compare different high doses, e.g., 4 mg/day versus 5 mg/day. Another approach would be comparing high-dose folic acid versus high-dose methylfolate. Another possibility would be to compare a moderate dose like 800 mcg/day versus a high dose like 4 mg, both of methylfolate, with and without the dietary intervention.
It’s not obvious how to design this series of trials — just as it’s usually not obvious at which doses to set which medications with a particular action to mitigate a particular risk. Science is full of best guesses and judgment calls, whether implemented through randomized experimentation or not. But it is obvious that a lot of people have an interest in doing more and better science on this. Including people who seem to be on diametrically opposed sides of some related issues…
Two Voices on Valproate
I got to thinking about all this after reading two very different, recent takes on prescribing regimes governing epilepsy drug sodium valproate. Valproate is linked with considerably heightened risks of birth defects (10% of exposed babies) and developmental disorders (40% of exposed). These risks include “a 20-fold increase in neural tube defects, cleft lip and palate, cardiovascular abnormalities, genitourinary defects, developmental delay, endocrinological disorders, limb defects, and autism.” Malformation risks are dose-dependent.
One recent voice on valproate favors extending the UK’s more restrictive regime to France and the EU more broadly. The other criticizes this regime as too strict.
Conservative regime advocacy
Marine Martin, a patient advocate, was interviewed for the spring HealthSense newsletter. HealthSense (formerly HealthWatch) is a UK charity that fights myth-based medicine. Martin founded and presides over APESAC, a French association supporting families affected by maternal valproate exposure during pregnancy — like her own. She also serves as a patient expert at the French Medicines Agency and the European Medicines Agency.
Martin took valproate for epilepsy not knowing that it posed serious risks to her children, both of whom have problems that could result from the exposure. Taking her case public helped bring about changes in France that protect others from the same harm. Those changes include requirements that only specialists can prescribe valproate to reproductive-aged women, they can only do so with women’s signed informed consent, and teratogenic drug packaging must feature pictograms indicating contraindication in pregnancy.
Still, Martin notes that France lags the UK in preventing valproate harms, with more than 220 valproate exposure-affected children born in France in 2021, versus 17 between October 2021 and March 2022 (six months) in the UK. She notes the UK’s prescribing regime is more restrictive, requiring two specialists sign off on valproate prescriptions for people under age 55, and hopes the French and European medical regulatory agencies will adopt that requirement.
But Martin worries some of her work amounts to “a tick-box exercise,” with bureaucrats excluding patient advocacy stakeholders like her and her group from important decision-making processes. And she notes additional research raises concerns about possible risks of paternal preconception valproate exposure, as well as possible transgenerational risks (in research Martin herself spearheaded). Her view contrasts with that of another recent voice on valproate…
Liberal regime advocacy
Rachel Arkell, a PhD researcher with the University of Kent’s Centre for Parenting Culture Studies and the British Pregnancy Advisory Service, argues the UK’s valproate regime, the 2018 Pregnancy Prevention Programme, is too restrictive, prioritizing harm prevention from valproate-exposed pregnancies over women’s right to choose the most effective seizure medication. Under the PPP, valproate can no longer be the first-line treatment for reproductive-aged girls or women. Two specialists must confirm alternatives aren’t tolerated or effective for these patients to be initiated on valproate. These patients then must use long-acting reversible contraceptives (e.g., IUDs or hormonal contraceptive implants).
Arkell interviewed 20 consultants responsible for prescribing valproate. They expressed concerns that formalization may degrade informed consent, warnings about the risks of stopping valproate (e.g., Sudden Unexpected Death in Epilepsy) should be better communicated, alternatives to valproate could pose similar risks to fetuses in exchange for inferior seizure control, maybe society shouldn’t force women with diminished capacity to use birth control, and perhaps informed women should have the right to conceive on valproate. (Note, however, that it appears they do: UK women can continue valproate to treat epilepsy during pregnancy when no other effective treatment is available.)
Some of Arkell’s concerns, and those raised by the consultants she interviewed, echo those raised previously by Alastair Macfarlane and Trisha Greenhalgh, particularly about whether a shared decision-making approach is appropriate for women’s contraceptive choices (“Sodium valproate in pregnancy: what are the risks and should we use a shared decision-making approach?” BMC Pregnancy and Childbirth, Vol. 18, No. 200, 2018). This is not a question that can be answered with science; it’s a value judgment. Society has interests in restricting women’s choices to protect unborn children; societies differ on the specifics of those restrictions (e.g., different norms about pregnant women drinking alcohol, different abortion restrictions). The question is when, why, and how (society agrees) a woman’s right to choose supersedes those interests (or doesn’t).
Contextualizing science in society
Situating Martin and Arkell’s contrasting voices in their social and political contexts also raises questions about how network interests shape science and scholarship — and how we then adjudicate between competing claims associated with different interests. Martin’s personal experiences as a patient and mother drive her advocacy and research. She advocates for the more restrictive regime Arkell criticizes.
Arkell’s critique could be understood as one of a new paternalism (she calls the old norm of physicians prescribing valproate without telling women the risks “the pinnacle of medical paternalism”). In her narrative, the UK now tells women what to do because it knows best. But Arkell interviews consultants to critique the regime. It is ironic to use authorities instead of affected patient voices to construct what is essentially a critique that the regime is paternalistic.
Is it also dangerous epistemologically that we don’t hear affected women’s voices instead or also in Arkell’s research? Why wasn’t a snowball survey conducted of affected patients in the interviewed consultants’ practices to ask what they themselves want, need, and think about the current regime? Without that sort of access, this type of scholarship risks elevating expert voices on hypothetical problems with the UK’s valproate prescribing regime — over Martin’s observations on the demonstrable inferiority of France’s more liberal regime in terms of preventing harm to children like her own.
Moreover, Arkell’s BPAS (British Pregnancy Advisory Service) affiliation situates her as pro-choice. BPAS’s website describes the group as “an independent healthcare charity which, for more than 55 years, has been advocating and caring for women and couples who decide to end a pregnancy.” Arkell calls BPAS a collaborative partner for this, her PhD research.
BPAS has a financial interest in liberalizing valproate policy; it charges for abortion services. Currently, the UK’s restrictive valproate regime prevents abortions at least in part by preventing pregnancies that would be nonviable due to major malformations including NTD.
More broadly, BPAS has an ideological interest in promoting the narrative of abortion as healthcare that is arguably most valid in the context of the small subset of abortions where pregnancies are nonviable (e.g., some NTD). Similarly, the organization also has an interest in promoting the choice ideology that Arkell highlights in her consultant interviews.
We might expect BPAS’s interests to color Arkell’s scholarship, not because either the organization or the scholar have any bad intent, but because everyone has a perspective that colors observation, analysis, and interpretation. Everyone has bias. Situating science in its social and political contexts is an ongoing exercise in critical reflection that demands we remember our limitations and try to maintain humility about what we know.
Given this context, how do we adjudicate between Martin’s claim to speak for thousands of parents whose families were harmed by valproate (including her own), and Arkell’s claim to speak for an unknown number of women represented by consultants critical of the UK’s valproate regime? The former appears to be more direct, the latter potentially conflicted. To the extent that Martin wants to maximize women’s information and harm prevention, while Arkell wants to maximize women’s choice and (even when their capacity is diminished) freedom from constraint, is it possible that their advocacy aims are not necessarily in conflict, or only minimally conflictual, after all? Might their views reflect differently prioritized values that can yet animate compatible policy worlds?
Shared Interests in Doing More Science
Whatever you think (if anything) about valproate and the ongoing debate about its prescribing regimes, it’s worth noting that there are open empirical questions with possible empirical solutions here — and everyone has a stake in doing more science. As Macfarlane & Greenhalgh noted:
Reasons for the teratogenic effects are not fully understood, but possibly involve epigenetic effects, including the inhibition of histone deactylase with associated changes in gene expression [41], increases in foetal oxidative stress, or the antagonism of folate required for DNA synthesis [42].
This suggests a possible preventive care intervention for at-risk women like those on valproate, featuring additional antioxidants (whether by diet, supplements, or both) and additional folic acid supplementation. Some experts are already sold on the latter. Reynolds & Green suggested:
up to 5 mg of folic acid periconceptually in at-risk women with the caveat that the addition of supplementary vitamin B12 may also be prudent because vitamin B12 deficiency is common in pregnancy in some countries and is an additional risk factor for developmental abnormalities” (“Valproate and folate: Congenital and developmental risks,” Edward H. Reynolds and Ralph Green, Epilepsy Behav, 2020 Jul:108:107068).
This is well-intentioned advice with some empirical support — but arguably insufficient grounding in complex bigger pictures. We need to know more about the possible risks and benefits of different folate supplementation protocols before doubling down on recommendations that women who are or may become pregnant should shovel down folic acid. We need to know how such supplementation affects mortality decades later. Whether different forms or dose levels affect maternal mortality and pregnancy/child health risks differently. Whether diet can mitigate some or all of these risks across the board, or even be sufficient in the absence of supplementation for some groups (e.g., for NTD prevention in women not at heightened risk of NTD pregnancies).
Society has been running mass, uncontrolled experiments on folic acid supplementation for decades, without first trying to answer some of these questions. The intent was good. The evidence was incomplete. It still is.
We can do better. And we can do it ourselves when it comes to interventions like vitamins and diets. As always, I’m fantasizing about seeing these studies on the sort of citizen science platform for people-oriented science that is also well-suited to answering questions about ibuprofen as contraception, competing infant feeding paradigms, nicotine and Covid, and DHEA and fertility, but that doesn’t seem to exist yet.
At the same time, in a context like known valproate risks versus unknown high-dose methylfolate risks, maybe most of the affected people wouldn’t want a randomized trial. Maybe people who want to have kids and are affected by valproate — either by medically necessary current use, or by possible intergenerational effects — would understandably want to take their best shot at having healthy babies. They could still collaborate on research to do that, e.g., following a standardized protocol of high-dose methylfolate in divided doses with long duration, antioxidants (e.g., R alpha lipoic acid and N-Acetyl cysteine are widely used in medicine and considered safe in pregnancy, tho NAC decreases testosterone), and a modified, high-folate Mediterranean diet. Since valproate has established risks, a case series using such a harm prevention protocol might usefully compare expected with actual outcomes.
People should get the chance to help design the research they want and need. Institutional science doesn’t normally give ordinary people that much voice, and structural incentives work against changing bad norms like that.
But, occasionally, affected people like Marine Martin ensure that more diverse people’s voices are heard — in advocacy and science alike. It’s a notable result.