Pregnancy Drug Patterns in Myasthenia Gravis Reveal Four Distinct Patient Journeys

Myasthenia gravis is an autoimmune disorder in which antibodies disrupt communication between nerves and muscles, producing weakness that fluctuates with effort and often demands lifelong pharmacological control. For women of childbearing age, that requirement collides with one of medicine’s most delicate balancing acts: keeping the mother’s neuromuscular disease quiet while minimizing any drug exposure to the developing fetus. A new population-based study from Norway and Sweden, published in the Journal of Neurology, has now mapped exactly how that balance plays out in real pregnancies, tracking prescription fills from a full year before conception through six months after delivery. The results offer the most detailed longitudinal picture to date of how women with myasthenia gravis actually use their medications around pregnancy, and they reveal patterns that could reshape how neurologists counsel and monitor this growing group of patients.

The research team, led by Jenny Linnea Victoria Lindroos of the University of Bergen and the Norwegian Institute of Public Health, exploited a unique strength of the Nordic healthcare infrastructure: mandatory national registers that record every birth, every hospital contact, and every prescription dispensed at a pharmacy. By harmonizing and pooling data from the Norwegian medical birth register covering 1999 to 2023 and its Swedish counterpart covering 1994 to 2022, and linking these records individually through each mother’s national identity number, the investigators assembled a cohort of 339 pregnancies among women with confirmed myasthenia gravis, drawn from more than 2.5 million births. The diagnosis itself was verified through multiple routes, including hospital diagnoses coded under the International Classification of Diseases and repeated fills of pyridostigmine, the standard symptomatic treatment, with validation studies in Norway showing a positive predictive value of 99 percent for the register-based diagnosis.

The methodological innovation at the heart of the study was group-based multi-trajectory modeling, an unsupervised statistical clustering technique that can follow several medication classes simultaneously across repeated time windows. Rather than imposing predefined categories such as early versus late pregnancy, the model let the data speak, dividing the timeline into four pre-pregnancy quarters, three trimesters, and two postpartum periods, and estimating how each pregnancy’s use of symptomatic drugs, oral corticosteroids, and non-steroidal immunosuppressants evolved together. The algorithm tested three, four, and five latent groups, ultimately settling on a four-group solution that balanced statistical fit, measured by an entropy of 0.92 and posterior probabilities between 95 and 99.6 percent, against clinical interpretability and privacy constraints that ruled out groups smaller than ten pregnancies.

Among the 339 pregnancies, 201, or roughly 59 percent, involved at least one fill of a myasthenia gravis medication during the year before pregnancy or during pregnancy itself, while 138 women, about 41 percent, filled no myasthenia gravis prescriptions at all across that entire window. The four trajectories that emerged were named for their defining patterns: Intermittent users, the largest group at 20.9 percent of the cohort; Persistent symptomatic-only users at 19.8 percent; Immunosuppressant plus symptomatic users at 10.3 percent; and Immunosuppressant-only users at 8.3 percent. The two largest groups relied exclusively on symptomatic medications such as pyridostigmine but differed sharply in consistency. Persistent symptomatic-only users refilled their prescriptions in 75 to 88 percent of every time period, averaging two fills per period, whereas Intermittent users filled prescriptions in only 21 to 28 percent of pre-pregnancy periods and just 13 percent during the second trimester. The average dispensed pyridostigmine dose told the same story, at 269 milligrams per day for the persistent group versus 159 for the intermittent group.

The immunosuppressant groups displayed more complex dynamics. Immunosuppressant plus symptomatic users carried the heaviest treatment burden, purchasing pyridostigmine at an average of 450 milligrams per day during the first trimester and prednisolone at roughly 16 milligrams per day, double the dose seen in the immunosuppressant-only group. Strikingly, their use of non-steroidal immunosuppressant therapy, chiefly azathioprine, fell steeply after conception, from around 60 percent filling prescriptions before pregnancy to only 20 percent in the third trimester. The authors interpret this drop as a deliberate attempt to avoid fetal drug exposure, though they cannot exclude improvement in disease activity, side effects, or pregnancy-related vomiting as contributing causes. In contrast, Immunosuppressant-only users maintained a remarkably stable prednisolone trajectory of about 8 milligrams per day throughout the entire study period, below the 10-milligram threshold generally considered safe in pregnancy, and rarely needed symptomatic medications at all.

Safety signals in the dispensing data were reassuring in several respects. Known teratogenic drugs, specifically methotrexate, mycophenolate mofetil, and cyclophosphamide, were never dispensed later than nine to twelve months before pregnancy, indicating that clinicians and patients successfully cleared these hazardous agents well before conception. Tacrolimus was never refilled during pregnancy, and prednisolone was the only oral corticosteroid used throughout. Biologics such as rituximab appeared only in pre-pregnancy and postpartum windows, and plasma exchange was confined to the pre-pregnancy period. Azathioprine, considered acceptably safe based on observational and animal data, was the dominant non-steroidal immunosuppressant, and its use actually rose during the final nine months before conception among the heaviest treatment group, a pattern the authors read as evidence of preconception treatment optimization.

Hospital contact patterns added a crucial clinical dimension to the medication trajectories. Overall, myasthenia gravis admissions were infrequent, but they clustered conspicuously in one group: 23 percent of Immunosuppressant plus symptomatic users were admitted for myasthenia gravis during pregnancy, and 26 percent had at least one such admission before pregnancy, with 71 percent requiring myasthenia gravis outpatient visits. This same group also showed the highest body mass index, more diabetes and gestational hypertension, and, tellingly, the lowest use of preconception folic acid supplementation, with fewer than five women taking it. Combined with the observation that 23 percent of these women had no specialist outpatient contact at all in the year before pregnancy, the authors argue that many of these pregnancies were likely unplanned and not preceded by neurological counseling, a gap with tangible consequences given that this group also frequently stopped immunosuppression during gestation.

By contrast, the Non-users, who made up 41 percent of the cohort, almost never required myasthenia gravis admissions during pregnancy or postpartum and rarely initiated treatment after delivery, consistent with stable remission or minimal disease manifestation. The Immunosuppressant-only users presented a paradox: minimal myasthenia gravis hospital contacts but frequent non-myasthenia specialist care, likely reflecting the known associations of corticosteroid exposure with hypertension, hyperglycemia, infections, preeclampsia, and preterm birth, alongside a high rate of autoimmune comorbidity at 39 percent and assisted reproductive technology conception in 21 percent of the group. Notably, the trajectory groups aligned closely with the Myasthenia Gravis Foundation of America Post-intervention Status classification, suggesting that dispensing records alone can proxy disease severity classes without requiring clinical examination data.

The study’s limitations deserve honest framing. Infusion therapies such as rituximab, administered in hospitals rather than pharmacies, were invisible to the prescription registers in Sweden and in Norway before 2022, and the drug’s long-lasting immunosuppressive effect may partly explain the large Persistent symptomatic-only group, particularly its Swedish predominance. Spontaneous and induced abortions could not be included because birth registers incompletely capture them, meaning the findings generalize best to pregnancies progressing to birth, and severe cases may be underrepresented. The authors also lacked antibody status, thymectomy history, and formal severity scores. Even so, the cohort’s population basis, prospective data capture, and cross-national harmonization make it an unusually robust evidence base.

The practical message for clinicians is unambiguous. Most women with myasthenia gravis enter pregnancy with no, low, or stable medication use and remain on predictable trajectories, but the minority who combine immunosuppressants with symptomatic therapy face the highest risk of disease exacerbation, hospitalization, and abrupt treatment discontinuation precisely when stability matters most. The authors conclude that identifying these trajectory subgroups, which likely represent distinct disease courses in relation to pregnancy, could enable individualized care pathways, and that preemptive discussions about reproductive plans should be routine when treating women of childbearing age with myasthenia gravis. For a disease whose course in pregnancy has long been described as unpredictable, this register-based cartography of real-world drug use turns a page: the unpredictability, it turns out, has structure, and that structure can now be named, measured, and planned around.

Subject of Research: Medication use trajectories during pregnancy in women with myasthenia gravis

Article Title: Medication use trajectories in myasthenia gravis pregnancies: a population-based drug utilization study

Article References: Lindroos, J. L. V., Gilhus, N. E., Brauner, S., Bjørk, M.-H., Midelfart-Hoff, J., Cesta, C. E., Furu, K., & Cohen, J. M. (2026). Medication use trajectories in myasthenia gravis pregnancies: a population-based drug utilization study. Journal of Neurology, 273(10), Article 572. https://doi.org/10.1007/s00415-026-14115-2

Image Credits: AI Generated

DOI: 10.1007/s00415-026-14115-2

Keywords: myasthenia gravis, pregnancy, pharmacoepidemiology, pyridostigmine, prednisolone, azathioprine, immunosuppression, trajectory modeling, Nordic registers, autoimmune disease, neuromuscular disorders, maternal health