Serum biomarkers of spontaneous preterm birth

Abstract

Listen

Translate article

The rationale for the prediction of spontaneous preterm birth (sPTB) is three-fold. First, delineating those factors predictive of PTB, we may obtain a better understanding of the mechanisms and biologic pathways that lead to spontaneous preterm parturition. Second, the use of predictors of sPTB permits the identification of a group of women at the highest risk for whom an intervention may be tested and for whom intervention is most critically needed. The third motivation for the prediction of sPTB is a corollary of the second; by identifying women at low risk of PTB, unnecessary, costly, and sometime hazardous interventions might be avoided. With respect to the source of biomarkers, there are two fundamental compartments of interest: the maternal and the fetal. The maternal compartment may be subdivided into serum, saliva, urine, cervix, and vagina. The fetal compartment may be subdivided into cord blood and amniotic fluid. This brief review will focus solely on the maternal serum compartment. The maternal serum compartment offers several distinct theoretical advantages. Obtaining maternal serum is less invasive than obtaining amniotic fluid or fetal blood. Maternal blood is drawn routinely at several points in time during prenatal care, and thus, a serum biomarker could be incorporated into the usual provision of care. Using maternal serum for biomarkers of PTB also facilitates central processing and analysis. After minimal processing at the point of care, blood specimens may be transported to a central facility for processing and analysis. This model of central processing and analysis has already been successfully incorporated into the aspects of routine prenatal care, such as serum screening for open neural tube defects and serum screening for aneuploidy (1, 2). Serum biomarkers of sPTB may be categorized by using the notion of the syndrome of PTB (3) and the four distinct pathways to PTB forwarded by Lockwood and Kuczynski (4): activation of maternal or fetal hypothalamic–pituitary–adrenal axis (e.g. corticotropin-releasing hormone), upper genital tract infection (e.g. defensins and tumor necrosis factor-α), decidual hemorrhage (e.g. thrombin–anti-thrombin III complex), and pathologic uterine overdistension. To date, no serum biomarkers of pathologic uterine overdistension have been described. Given the known pathophysiologic heterogeneity of sPTB, the clinical utility of any individual biomarker for predicting PTB is limited. To that end, Goldenberg and colleagues sought to evaluate a ‘multiple marker test’ for PTB in The Preterm Prediction Study of the Maternal–Fetal Medicine Units Network of the National Institutes of Child Health and Human Development (5). This was a nested case-control study performed within a cohort of 2929 women with a singleton gestation recruited from the general obstetric population of the 10 participating centers. Women, in this study, underwent serial assessment of serum, cervical, vaginal, ultrasound, and historical markers or risk factors. The serum biomarkers considered in this study have been listed below. The markers associated with sPTB at <32 weeks were α-fetoprotein, alkaline phosphatase, granulocyte-macrophage colony-stimulating factor, and defensins. Only α-fetoprotein and alkaline phosphatase were related to sPTB at <35 weeks. Importantly, the overlap between the biomarkers was small, supporting the notion that there are several heterogeneous pathways to sPTB. In fact, the utilization of the serum markers in concert improved the predictive ability. Any one ‘positive’ serum marker had 81% sensitivity and 78% specificity for sPTB at <32 weeks and 60% sensitivity and 73% specificity for sPTB at <35 weeks. Because this was a nested case-control study, the true positive and negative predictive values are uncertain. No individual serum biomarker of sPTB is useful when used alone in an symptomatic low-risk obstetric population. The notion of a ‘multiple marker’ approach may improve the predictive ability, but no prediction model, to date, provides adequate utility to justify routine clinical use. The future development of an effective prediction model may permit the evaluation of targeted therapies. Future directions for research in this arena will involve the optimization and validation of multiple markers in diverse prospective cohorts. Novel marker development will be facilitated by the use of high-throughput proteomic and metabolomic technologies and the consideration of more complex modeling techniques, such as neural networks and artificial intelligence. Corticotropin-releasing hormone, α-fetoprotein, alkaline phosphatase, β2-macroglobulin, ferritin, ICAM-1, interleukin-6, C-reactive protein, cortisol, lactoferrin, defensins, relaxin, interleukin-10, granulocyte-macrophage colony-stimulating factor, and activin.