Skip to main content

Pregnancy Study Points to New Ways to Accurately Predict Births

chart of protein changes before labor
Proteins, hormones and immune factors in maternal blood can help predict labor timing. | Stelzer et al., STM

Doctors have long sought accurate ways to predict the onset of labor in expecting mothers, but the shift from pregnancy to labor isn't fully understood.

Now, a new study of 63 pregnant women has revealed how the immune system, various proteins, and other aspects of human biology change in the weeks before the onset of labor. The findings allowed the scientists to build a predictive model that could estimate the onset of labor in the women within two weeks.

The research — published in the May 5 issue of Science Translational Medicine — could inform efforts to develop blood tests that can reliably estimate the time of delivery. These tests could in turn help doctors identify women at risk of premature or late births, allowing for earlier management and interventions.

The timing of labor can have a big impact on the outcome of the pregnancy and the health of a newborn. Studies have linked preterm births — those that occur 37 weeks or earlier during gestation — to a higher risk of complications and health disorders in infants, ranging from heart conditions to a weaker immune system.

On the other end, postterm births that occur after 42 weeks have a higher risk of stillbirth and infection compared with term deliveries, according to the American Academy of Family Physicians.

Obstetricians can prepare for preterm or postterm births, but they must first be able to predict the time of birth in pregnant women. Accurate predictions would allow doctors to mobilize resources and intervene with treatments, but there aren't any methods that can reliably and consistently predict births in the clinic.

The Problems with Predicting Labor

Current prediction methods use a metric called gestational age, which is a rough estimate of how far along a pregnancy is. Doctors base gestational age on ultrasound measurements of the fetus and the mother's last menstrual period. They then derive the day of labor by comparing the gestational age to a standard pregnancy length of 40 weeks.

But this method often doesn't yield accurate predictions in the real world, as most pregnancies stray from the 40-week norm. Even normal term labor can differ from estimates by as many as five weeks, and preterm or postterm labor can stray even further, according to Brice Gaudillière, an associate professor at the Stanford University School of Medicine and senior author of the new study.

Researchers therefore need new methods to predict the onset of labor. New techniques would likely test for changes in molecules during the transition from pregnancy to birth. However, few studies have deeply explored how human biology changes in the weeks preceding birth.

"Our study was motivated by the aim of building a better predictive tool to predict the time to labor by measuring biological factors in the maternal blood that are temporally linked to labor," Gaudillière said. His team made sure to design the study to yield predictions without depending on assessments of gestational age.

Ina Stelzer, a postdoctoral researcher at Stanford Medicine and lead author of the new study, Gaudillière, and colleagues used lab techniques to comprehensively characterize how proteins, metabolic molecules, and immune cells in the blood shifted before the onset of labor.

The researchers followed 63 pregnant women as they entered the last 100 days before labor, corresponding to the third trimester. The team gathered blood and plasma samples from the women one to three times over the course of the 100-day period and studied more than 7,000 molecules, proteins, and immune cell features.

Importantly, all of the women enrolled in the study went into labor spontaneously. This allowed the researchers to determine the time of labor clinically and symptomatically, rather than with arbitrary estimates of gestational age, Gaudillière said.

The scientists also used machine learning techniques to build a model that integrated the data to predict the time to spontaneous labor in the women.

Testing for Transition to Labor Could Improve Prediction

"The overarching theme we observed was a transition from a phase of pregnancy progression to a pre-labor phase two to four weeks before labor onset," Gaudillière said.

This transition appeared in the form of changes in proteins in the placenta and molecules involved in inflammation and the formation of blood vessels. The women also showed a surge in the formation of steroid hormones and in IL-1R4, a receptor on immune cells that suppresses inflammation.

When asked about how the results could inform clinical practice, Gaudillière and Stelzer speculated that a blood draw measuring these changes could provide useful estimates when combined with their model. They noted their model was accurate at predicting labor within two weeks.

Depending on the blood results, doctors could then tailor pregnancy management to best suit the needs of the patient. For example, they could reassure patients that labor is going as expected, or explore inducing labor in women who are estimated to have postterm births.

In the case of women at high risk of preterm birth, doctors could prepare high-risk delivery suites ahead of time or even intervene with progesterone to delay birth, the researchers said.

Gaudillière and his team caution that they trained their model solely with samples from patients at Stanford. They stress the need for studies with more patients from a wider range of demographic and socioeconomic backgrounds, and are currently collaborating with other groups in the U.S., Europe, and south Asia to expand their database.

The team is also working on a reduced, streamlined version of their model that contains 12 features — compared to the 45 features in the current model — which can all be measured with medical platforms available in clinics.

[Credit for associated image: sergio santos/ Flickr]