Likelihood ratios for the prediction of preterm delivery with biomarkers

Abstract

Objective: To conduct a literature search for selected biomarkers on preterm delivery and estimate their likelihood ratios (LR).

Design: Structured review.

Population: Low and high-risk populations and women with symptoms of preterm delivery. METHODS. Publications were identified in PubMed.

Main outcome measures: LR on selected biomarkers for preterm delivery.

Results: In asymptomatic women with low risk of preterm delivery, the following biomarkers gave major shifts in probability (LR above 5): twins (LR+ 10), Ureaplasma urealyticum in amniotic fluid (LR+ of 10), cervical length <25mm (LR+ 6), salival estriol (LR+ 5) and various combined tests. In asymptomatic women with high risk of preterm delivery, short cervical length (LR+ 11, LR- 0.7), high serum tumor necrosis factor-alpha (LR+ 10, LR- 0.6) gave major shifts in probability. In women with symptoms of preterm delivery, major shifts in probability can be obtained from the following amniotic fluid biomarkers: high matrix metalloproteinase-8 (LR+ 23, LR- 0.6), Ureaplasma urealyticum (LR+ 19, LR- 0.8), high interleukin (IL)-6 (LR+ 9, LR- 0.2), IL-8 (LR+10, LR- 0.2) and tumor necrosis factor-alpha (LR+ 8, LR- 0.4). In serum IL-6 (LR+ 12, LR- 0.2), Cluster of Differentiation 163 (LR+9, LR-0.8) and various combined tests. Vaginal fetal fibronectin (LR+ 3 and LR- 0.5) and short cervical length (LR+ 2, LR- 0.3) gave LRs of some importance (LR below 5).

Conclusion: Several biomarkers have been identified for assessment of risk of preterm delivery. Their clinical relevance depends on the efficacy of the interventions which can be offered to these patients.