As many as 10% of births in the U.S. are premature—meaning they occur before 37 weeks—which can lead to potentially life-threatening effects for the baby and a severe impact on the mother’s mental health. The complication is usually only detected once physical symptoms set in, typically in the third trimester of pregnancy.
San Francisco-based startup Mirvie, however, is attempting to open up an earlier diagnostic window, in hopes of expanding the treatment options for patients predicted to experience preterm birth.
It’s doing so with the development of a machine learning-based platform that scours the messenger RNA signatures of a fetus, placenta and expectant mother for biological changes linked to preterm birth and other pregnancy complications like preeclampsia—and so far, studies show, it seems to be working.
The most recent study of Mirvie’s tech was published in the American Journal of Obstetrics and Gynecology this month. It studied tens of thousands of messenger RNA pieces in the blood samples of more than 240 participants that were collected during their second trimesters, between the 12th and 24th weeks of pregnancy.
The machine learning analysis was able to pinpoint dozens of genetic indicators that were associated with preterm birth. Several of those indicators—including some linked to amino acid metabolism and insulin-like growth factor pathways—were unique to patients who went on to experience extremely premature birth, which occurs before 25 weeks.
From there, the platform could then be used in the other direction, to look for those identified biomarkers and predict a patient’s chances of experiencing preterm or extremely preterm birth. By looking for certain mRNA signatures associated with premature changes to the expectant mother’s cervix, for example, the platform was able to predict three out of four preterm births, the researchers found, with the predictions arriving an average of two months in advance.
In addition to giving patients and their healthcare providers more time to prepare for and begin treating a potentially early birth, according to Michal Elovitz, M.D., chief medical advisor at Mirvie, since the study results shed new light on the underlying biology of preterm birth, the findings could also be used to help develop even more targeted treatments for the condition.
The preterm birth study follows the publication of research at the beginning of the year that demonstrated the platform’s ability to predict preeclampsia before symptoms occur. Preeclampsia typically shows up after the 20-week mark and is characterized by new-onset high blood pressure and damage to the liver, kidneys or other organ system.
In the study, the Mirvie platform analyzed the mRNA in blood samples from more than 1,840 patients around the world. Once again, it was able to predict about 75% of those who would go on to develop preeclampsia, compared to the average of just 20% identified by current predictive models that only account for factors like pregnancy and medical history, race, ethnicity and body mass index.