Prediction tool helps spot preterm birth risk in ICP twin pregnancies
Model could avoid unnecessary intervention in low-risk cases
Written by |
A new prediction tool may help assess the risk of preterm birth (before 37 weeks of gestation) in women affected by intrahepatic cholestasis of pregnancy (ICP) during twin pregnancies, a study in China showed.
The tool was particularly effective at identifying pregnant women with a lower risk of preterm birth, offering a potentially “user-friendly approach to identify low-risk patients and reduce unnecessary intervention in twin pregnancies with ICP,” the researchers wrote.
The model combines dynamic parameters related to blood levels of bile acids (a marker of liver damage), history of ursodeoxycholic acid (UDCA) treatment, and therapeutic response via changes in bile acid levels. UDCA, marketed in the U.S. under the brand names Actigall and Urso, is the first-line treatment for ICP.
“These variables serve as dynamic surrogates for disease control, complementing static [initial] indicators and making the model more clinically actionable,” the researchers wrote.
The study, “A nomogram integrating dynamic total bile acid monitoring, medication history, and curative effect for preterm birth prediction in twin pregnancies with intrahepatic cholestasis of pregnancy,” was published in BMC Pregnancy and Childbirth.
ICP resolves after delivery, but still poses risks
ICP is a form of cholestasis, a condition marked by slowed or stalled flow of the digestive fluid bile from the liver to the intestines. The most common pregnancy-related liver condition, ICP typically appears late in the second trimester or early in the third trimester of pregnancy.
A key feature of ICP is the abnormal elevation of bile acids, bile’s main components, in the mother’s blood, which can cause cholestasis symptoms such as itching, and readily reach the fetus and exert toxic effects.
Although ICP usually resolves after delivery, it poses potential risks for both the mother and the baby, including preterm birth. This is particularly true in the case of twins.
However, there is limited evidence on the clinical management of the disease in twin pregnancies and whether medication and changes in total bile acid levels are associated with the risk of preterm birth.
The researchers set out to develop a nomogram, or a graphical prediction tool integrating multiple risk factors, to predict preterm birth in twin pregnancies complicated by ICP.
The team retrospectively analyzed data from 258 twin pregnant women with ICP seen at a single Chinese hospital. Most cases (83.3%) resulted in preterm birth.
Pregnant women in the preterm group were diagnosed with ICP earlier during pregnancy and had significantly higher blood bile acid levels. They were also significantly more likely to have moderate to high (36.3% vs. 16.3%) or severe bile acid elevations (2.8% vs. 0%) than those whose babies weren’t born preterm.
The preterm group had significantly higher rates of UDCA use (97.2% vs. 79.1%), alone or in combination with other medications to promote bile flow and protect the liver. However, the rate of curative effect (bile acid response after treatment) was significantly lower in the preterm group (52.1% vs. 79.1%).
Preterm babies were born significantly earlier (35 weeks of gestation vs. 37 weeks) and with a significantly lower birth weight than non-preterm babies. The preterm group was significantly more likely to experience premature break of the amniotic sac surrounding the fetus (17.2% vs. 4.7%) and fetal distress (9.3% vs. 0%). In contrast, the rate of births by cesarean section was significantly lower in the preterm group (94.4% vs. 100%).
These data were used to train and validate the predictive model, which combines two conventional variables — gestational age at ICP diagnosis and blood levels of aspartate aminotransferase, a liver damage marker — with three innovative factors: dynamic bile acid parameters, including gestational age at bile acid peak and bile acid levels just before delivery; UDCA use; and curative effect.
The model demonstrated a robust predictive performance, discriminating preterm cases from non-preterm cases with an accuracy of 81.2% in the training data set and 74% in the validation set.
The model showed a specificity of 93.8%, meaning it correctly ruled out most women without preterm birth, and a negative predictive value of 85.7%, reflecting high efficacy at correctly ruling out non-preterm cases. These values emphasized “the high reliability of “non-preterm” predictions,” the team wrote.
However, the tool’s sensitivity, or ability to correctly identify preterm cases, was 16.7%, and its positive predictive value was 33.3%, indicating a low chance that a preterm prediction corresponds to a real preterm case. Because the model could not predict most cases of preterm birth, the team emphasized that it may be better suited to detecting low-risk patients.
The new model “is only valuable for excluding low-risk patients in this specific population and reducing overintervention, but it is not yet suitable for guiding clinical decision-making (e.g., delivery timing) across diverse settings,” the researchers wrote.
