Objective: Many serum markers have been proposed to aid in the identification of an ectopic pregnancy, but few have been validated. Most studies have been limited by sample size and design. The goal of this study was to assess putative markers to identify which can be optimally combined.
Methods: We conducted a case-control study using sera from 100 patients with ectopic pregnancy and 100 patients with intrauterine pregnancy who presented to three urban academic centers between September 2000 and April 2009 with first-trimester pain or bleeding. Samples were analyzed for 12 promising biomarkers. Classification tree analysis was used to examine markers simultaneously with the goal of optimizing the accuracy of ectopic pregnancy diagnosis, and validation was performed using bootstrapping.
Results: Six of the 12 markers were differentially expressed between those with ectopic pregnancy and intrauterine pregnancy (P<.001) with fair diagnostic properties (area under the curve greater than 0.6) when examined individually (inhibin A, progesterone, activin A, vascular endothelial growth factor [VEGF], pregnancy-specific β-1-glycoprotein, and pregnancy-associated plasma protein-A). Six additional markers were found to have limited value. Using a two-step diagnostic algorithm with four markers (progesterone, VEGF, inhibin A, activin A), we diagnosed 42% of the sample with perfect specificity and 98% (93-100%) sensitivity. Overall, a single ectopic pregnancy was misclassified, achieving 99% (96-100%) accuracy.
Conclusion: Evaluating a large number of biomarkers simultaneously demonstrates that most of the putative markers of ectopic pregnancy are not useful. However, a select few can distinguish ectopic pregnancy from intrauterine pregnancy with superior accuracy as part of a multiple marker test.
Clinical trial registration: : ClinicalTrials.gov, www.clinicaltrials.gov, NCT00194168.