Migraine antibodies in early pregnancy tied to higher miscarriage risk in a claims study

Women who took anti-CGRP monoclonal antibodies — the newer class of injectable migraine-prevention drugs — early in pregnancy had a higher rate of spontaneous abortion than unexposed women in a large US claims analysis, while the older preventive drug propranolol showed no such signal [s1]. The study, published in JAMA Neurology, is observational and cannot prove the drugs caused the losses, but it is among the first sizeable looks at a question the original trials could not answer, because pregnant women were excluded from them.

Anti-CGRP antibodies — a group that includes erenumab, fremanezumab, galcanezumab and eptinezumab — work by blocking calcitonin gene-related peptide or its receptor, a signalling molecule central to migraine attacks. They are large antibodies with long half-lives, which is part of why their use around conception has been a lingering worry: a drug given before a woman knows she is pregnant can still be circulating well into the first trimester. CGRP also has roles in blood-vessel function and is present in the placenta, giving a biological reason to look closely rather than assume safety.

What the study did

The researchers used the AM-PREGNANT cohort, built from MarketScan commercial insurance claims for 2018 to 2021 [s1]. They identified 7,579 singleton pregnancies among 7,119 individuals with migraine and compared early-pregnancy exposure to anti-CGRP antibodies or to propranolol against no exposure, tracking spontaneous abortion through 20 weeks' gestation [s1]. Across the cohort, 1,862 pregnancies ended in spontaneous abortion [s1].

Using a claims database has real strengths — large numbers, real-world prescribing, no reliance on recall — and real limits. It captures filled prescriptions, not whether a woman actually injected the drug or when; it cannot see over-the-counter remedies, lifestyle factors or the severity of the underlying migraine; and women prescribed a newer antibody may differ systematically from those on an older drug or on nothing. The authors adjusted for measured differences and ran additional bias-focused analyses, but no statistical adjustment removes confounding you cannot measure.

The result

Early-pregnancy exposure to anti-CGRP antibodies was associated with a higher risk of spontaneous abortion, with an adjusted hazard ratio of 1.42 (95% CI 1.12–1.80) [s1]. The association persisted in the authors' bias-adjusted analyses [s1]. Propranolol exposure, by contrast, was not associated with increased risk (adjusted hazard ratio 1.02, 95% CI 0.71–1.47) [s1].

The contrast between the two drugs is part of what makes the antibody signal harder to dismiss as pure confounding by "having migraine and taking something for it," since propranolol-treated pregnancies — also drawn from women with migraine — did not show the effect [s1]. But the confidence interval for the antibodies, running from 1.12 to 1.80, also means the true increase could be modest; the lower bound sits only just above no effect [s1].

It is worth being precise about what a hazard ratio of 1.42 does and does not mean. It describes a relative increase, not an absolute one: most pregnancies in the exposed group did not end in loss. A spontaneous abortion rate that is, say, 20% in unexposed pregnancies would rise to somewhere in the high twenties under a 1.42 hazard, not double. Whether that shift reflects the drug, the migraine it treats, or unmeasured differences between the women who are prescribed it is exactly what a claims analysis cannot settle on its own. The researchers' bias-adjusted analyses are designed to probe how much unmeasured confounding it would take to explain away the result, and the association survived that scrutiny — but surviving a sensitivity analysis is not the same as a randomised trial, which for obvious ethical reasons will never be run in pregnancy for this question [s1].

What it means

For now, this is a safety signal that calls for caution and counselling, not a verdict. The authors frame it exactly that way: the anti-CGRP finding supports further safety evaluation and pre-pregnancy counselling for women of reproductive age [s1]. That aligns with the broader evidence base, in which migraine itself is linked to some adverse pregnancy outcomes and the data on individual preventive drugs remain thin [s2].

Practically, women who use anti-CGRP antibodies and are planning a pregnancy, or who could become pregnant, have reason to discuss timing and alternatives with a neurologist and obstetrician rather than stopping or starting anything on the strength of one claims study. The long half-life of these drugs means decisions may need to be made months ahead of conception. Propranolol, an older and cheaper option with reassuring data here, is one of several alternatives, though it carries its own considerations. What the study does not support is the reassuring assumption that because these antibodies are targeted and generally well tolerated, they are safe to carry into early pregnancy. The honest reading is that the question is now open, and the first large answer points toward caution.