Around week nineteen of pregnancy, something shifts. The fetal brain, which has until that point been drawing cholesterol from maternal circulation, begins making its own. It’s a critical handover, and the organ doing it is already, by several measures, the most cholesterol-hungry tissue in the human body. Cholesterol isn’t just structural scaffolding here: it’s raw material for the signalling molecules and cell membranes that neurons need to wire up correctly, to grow their long reaching axons, to find one another across the developing cortex. Interfere with that supply, even briefly, and the consequences can ripple outward for years.
Which makes a new study from the University of Nebraska Medical Center rather alarming. Researchers there have found that dozens of widely used prescription medications, ranging from antidepressants and antipsychotics to statins and beta-blockers, share a largely unrecognised side effect: they disrupt the very cholesterol biosynthesis pathway that fetal brains depend on during precisely this window of development.
A Signal Hidden in Six Million Records
The team, led by neonatologist Eric Peeples and geneticist Karoly Mirnics, analysed linked maternal and child health records for 6.14 million births between 2014 and 2023, roughly a third of all U.S. births over that period. What they were looking for wasn’t a single drug class. It was a biochemical signature. The medications they grouped together, which they call sterol biosynthesis-inhibiting medications, or SBIMs, don’t share a primary indication or even a chemical family. What they share is an ability to suppress one or more enzymes along the cholesterol production chain, often as an unintended side effect rather than a therapeutic goal. The fourteen drugs studied included sertraline, fluoxetine, bupropion, aripiprazole, haloperidol, trazodone, buspirone, metoprolol, propranolol, nebivolol, and four statins: atorvastatin, simvastatin, rosuvastatin, and pravastatin. Together these medicines account for more than 400 million prescriptions a year in the United States, with nine of them ranking among the 25 most commonly prescribed drugs in the country.
The finding, published in Molecular Psychiatry, is stark. Children whose mothers were prescribed at least one SBIM during pregnancy were 1.47 times more likely to be diagnosed with autism spectrum disorder. Each additional SBIM prescribed simultaneously pushed that risk higher, by a factor of 1.33 per drug, reaching a 2.33-fold increase when four or more were co-prescribed. That dose-dependence matters. It’s the kind of pattern that distinguishes a real biological signal from statistical noise.
Why Cholesterol, and Why Now
The cholesterol connection to autism isn’t entirely new. Smith-Lemli-Opitz syndrome, a rare genetic disorder that cripples the final step in cholesterol biosynthesis, is among the strongest known predictors of autism: roughly 75% of those affected meet diagnostic criteria. The pathway implicated in SLOS involves an enzyme called DHCR7, which converts a cholesterol precursor called 7-dehydrocholesterol (7-DHC) into cholesterol itself. When DHCR7 is blocked, whether by mutation or by medication, 7-DHC builds up. And 7-DHC happens to be the most chemically reactive lipid known in this context, generating toxic oxidation products that impair cell growth, differentiation, and, in developing neurons, the formation of circuits.
The UNMC team’s insight was to realise that several of the medications already in common use could be doing something similar, at lower intensity, in pregnant women who happen to be taking them. A 2021 analysis of serum samples from pregnant women found elevated 7-DHC levels in those taking SBIMs, with levels rising further in women taking more than one. In the current study, the magnitude of 7-DHC elevation produced by individual medications correlated strongly (R² = 0.78) with the adjusted hazard ratios for autism. That correlation is difficult to explain away. It suggests the link isn’t primarily about the conditions being treated, or the demographics of who receives these prescriptions; it’s about what the drugs are doing to sterol chemistry in the womb.
“Our findings do not suggest that these medications are unsafe for adults,” said Mirnics, who is dean of UNMC’s Munroe-Meyer Institute. “But they raise important questions about their use during pregnancy, a period when even small biochemical disruptions may have outsized effects on fetal brain development.”
Confounders, Caveats, and a Difficult Signal
The study’s authors are the first to acknowledge the methodological tightrope they’re walking. Women prescribed antipsychotics or antidepressants during pregnancy aren’t a random sample; they have higher rates of psychiatric diagnoses that are themselves associated, through genetic and environmental pathways, with autism risk in their children. The research team ran sensitivity analyses controlling for maternal bipolar disorder, schizophrenia, anxiety, and depression. Risk estimates attenuated somewhat for psychiatric medications, particularly aripiprazole, which fell from a 2.18-fold to a 1.31-fold adjusted hazard ratio. But the signal persisted across drug classes with no psychiatric indication at all. Statins showed elevated autism risk (pravastatin at 1.95-fold, simvastatin at 1.64-fold), largely unchanged by psychiatric sensitivity analyses. Beta-blockers were similar. The common thread, across antidepressants, antipsychotics, anxiolytics, and drugs for metabolic and heart conditions, is the shared, often unlisted side effect on the sterol pathway.
There are real limits here. The study couldn’t track dosage or duration precisely, and the average age of autism diagnosis in the United States is about four years, meaning a substantial portion of the youngest children in the cohort hadn’t yet accumulated enough follow-up time to receive a diagnosis. The true numbers may be higher. The researchers also couldn’t fully account for socioeconomic factors, environmental exposures like pesticides and heavy metals, or the potential role of vitamin D deficiency, which may amplify sterol-pathway vulnerability. And perhaps most vexing: roughly 3% of the population carries a single pathogenic variant in a sterol biosynthesis gene, not enough to cause SLOS on its own, but possibly enough to make the fetal brain more sensitive to drug-induced pathway disruption during gestation.
None of this makes the finding go away. What it does is complicate the story in exactly the way that genuine public health signals tend to.
There’s a timeline problem buried in the data, too. Use of SBIMs during pregnancy tripled between 2014 and 2023 in the cohort, rising from 4.3% to 16.8% of pregnancies. In 2021, the FDA removed pregnancy warnings from statin labelling, apparently reflecting growing clinical comfort rather than any new safety data. If that comfort translates to higher doses and longer durations over the next decade, the population-level exposure could grow considerably. The researchers are careful to frame this as a signal requiring investigation rather than cause for immediate alarm, but the trajectory is not one they’re inclined to ignore.
The proposed mechanism is a cascade. SBIMs suppress key enzymes in the sterol pathway, both in maternal circulation and, later, directly in the fetal brain once it begins its own biosynthesis. That suppression causes a dual problem: cholesterol falls short of what developing neurons need, while toxic 7-DHC-derived oxysterols accumulate. The result, in animal and cell models, is impaired progenitor proliferation, disrupted neurite growth, and interference with circuit formation. Whether that same cascade unfolds at clinical doses, in human pregnancies, across the genetic diversity of six million births, remains the central unanswered question. But the correlation between biochemical disruption and clinical outcome, running from SLOS genetics through SBIM pharmacology to this epidemiological dataset, has a coherence that is difficult to set aside. The full study is available at https://doi.org/10.1038/s41380-026-03610-7.
Frequently Asked Questions
Should pregnant women stop taking antidepressants or statins because of this study?
No, and the researchers are emphatic on this point. Many of the medications flagged in the study are essential, sometimes life-saving treatments, and stopping them abruptly carries its own serious risks. The study is a population-level signal, not a verdict on any individual prescription. Anyone concerned should speak with their doctor rather than make changes unilaterally.
Why would cholesterol-related drugs affect a baby’s brain development?
The fetal brain is the most cholesterol-rich organ in the body, and it begins producing its own cholesterol around weeks 19 to 20 of gestation. Disrupting the enzymes involved in that process, even as an unintended drug side effect, can cause a build-up of toxic precursor molecules alongside a shortfall in cholesterol itself, both of which impair how neurons grow and connect. The same pathway, when disrupted by genetic mutation, causes a condition in which roughly 75% of children meet criteria for autism.
What makes this study different from earlier research on antidepressants and autism?
Previous studies looked at drug classes by their therapeutic use, grouping antidepressants together, statins together, and so on. This team grouped fourteen chemically unrelated medications by a shared biological side effect: interference with the sterol biosynthesis pathway. When the risk pattern emerged consistently across cardiac drugs, psychiatric medications, and metabolic treatments, it pointed toward a mechanism rather than a confounding factor linked to any one condition.
Is the rising autism rate partly explained by this?
Possibly, at least in part. The proportion of pregnant women in the study receiving at least one of these medications almost quadrupled over the ten-year period, from about 4% to nearly 17%. Autism diagnoses have risen over the same decades, though better diagnostic recognition accounts for a significant share of that increase. The researchers stop well short of claiming a causal connection, but the timing and the dose-response relationship make this a hypothesis worth taking seriously.
Does taking more than one of these medications make the risk much worse?
The data suggest it does. The risk increased by a factor of 1.33 for each additional sterol-inhibiting drug prescribed during pregnancy, reaching more than double the baseline risk when four or more were prescribed simultaneously. Lab studies support this: women taking multiple such drugs had measurably higher levels of the toxic sterol precursor 7-dehydrocholesterol in their blood, suggesting an additive or synergistic biochemical effect.
ScienceBlog.com has no paywalls, no sponsored content, and no agenda beyond getting the science right. Every story here is written to inform, not to impress an advertiser or push a point of view.
Good science journalism takes time — reading the papers, checking the claims, finding researchers who can put findings in context. We do that work because we think it matters.
If you find this site useful, consider supporting it with a donation. Even a few dollars a month helps keep the coverage independent and free for everyone.
Thank you for writing a fair and balanced view about our study. We greatly appreciate it.