Lupus Nephritis Reshapes Pregnancy Risk Save

At #ACR26, investigators presented an abstract using Epic Cosmos to examine pregnancy outcomes in women with SLE across health systems nationwide. Beyond confirming that SLE pregnancies remain higher risk, the study highlights two important opportunities: the ability to use large-scale, real-world data to study relatively uncommon outcomes in lupus, and the importance of lupus nephritis history when counseling patients and planning pregnancy (abstract 2462).

Using Cosmos, an Epic-based electronic health record database that includes more than 300 million patients, investigators identified more than 11 million pregnancies in the general population and nearly 4,000 among women with SLE. This scale offers an important opportunity to study relatively uncommon but clinically consequential outcomes in diseases such as lupus, where even large single-center cohorts may have limited numbers.

The findings reinforce what many rheumatologists already discuss with patients: pregnancy in SLE carries increased risk. Women with SLE had substantially higher rates of preterm birth and preeclampsia, and their infants were more likely to be small for gestational age.

What stood out to me, however, was the magnitude of risk associated with lupus nephritis (LN). Among women with SLE, those with LN had approximately twice the prevalence of early preterm birth and preeclampsia compared with women without nephritis. Neonatal demise, while uncommon, was also higher.

These data make a history of LN an especially important part of preconception counseling. For our patients, the message should not be that women with LN cannot have successful pregnancies. Rather, a history of LN identifies a group in whom pregnancy should be particularly intentional: ideally planned when disease is well controlled, medications have been optimized for pregnancy, and rheumatology, maternal-fetal medicine, and nephrology can coordinate care early.

Importantly, these data cannot tell us whether nephritis was active at conception, how well disease was controlled, or how individual medications affected outcomes. ICD-10-based definitions also introduce potential misclassification.

But that may point toward the next opportunity. Cosmos and other large-scale EHR resources could allow us to move beyond simply describing risk toward identifying which potentially modifiable factors before and during pregnancy are associated with better outcomes. The ability to incorporate information from unstructured clinical notes may further add disease activity, medication use, pregnancy planning, and other clinical details that cannot be captured by diagnostic codes alone. Combining this clinical depth with data at scale could help us better identify not only who is at highest risk, but where we may have opportunities to improve outcomes.