Role of serum biomarkers in the prediction of outcome in women with threatened miscarriage: a systematic review and diagnostic accuracy meta-analysis

Threatened miscarriage affects one in five women and is associated with significant emotional distress. The uncertainty around the prognosis of threatened miscarriage makes it equally challenging to the healthcare professionals. Various biochemical markers have been investigated in the past to predict the outcome of threatened miscarriage; however, the results have been conflicting. Therefore, we have conducted a systematic review and meta-analysis to determine the diagnostic accuracy of biochemical markers in predicting the outcome in women presenting with threatened miscarriage. This is a systematic review and meta-analysis of prospective studies that investigated biochemical markers to determine outcomes for women with threatened miscarriage at 5-23 weeks gestational age. Electronic databases were searched up to June 2015 and study quality assessment was performed using QUADAS-2 (Quality Assessment for Diagnostic Accuracy Studies-2: A Revised Tool) for evaluating the diagnostic accuracy studies. Statistical analysis was performed using the Cochrane systematic review software. A total of 19 studies were included in the qualitative data synthesis of which 15 (including 1263 women) were eligible for the meta-analysis. The review highlights the role of biochemical markers serum progesterone, hCG, pregnancy associated plasma protein A, estradiol and cancer antigen 125 (CA 125) in the prediction of outcome in women with threatened miscarriage. Interestingly, serum CA 125 appears to be the most promising marker (n = 648 women in seven studies), whereas serum progesterone and hCG are less useful once fetal viability is established. The summary receiver operating characteristics for CA 125 showed a sensitivity of 90% (95% confidence interval (CI) 83-94%), specificity of 88% (95% CI 79-93%), positive likelihood ratio of 7.86 (95% CI 4.23-14.60) and negative likelihood ratio of 0.10 (95% CI 0.06-0.20). The inverse of negative likelihood ratio was 9.31 (95% CI 5-17.1) indicating that a negative test is likely to identify those who are likely to continue with the pregnancy. Serum estradiol was the next best marker with a sensitivity of 45% (95% CI 6-90%), a specificity of 87% (95% CI 81-92%), a positive likelihood ratio of 3.72 (95% CI 1.01-13.71) and a negative likelihood ratio of 0.62 (95% CI 0.20-1.84). In women with threatened miscarriage, serum CA 125 has high predictive value in identifying pregnancies that are 'likely to continue', whereas the most commonly used biomarkers of serum hCG and progesterone are not useful in predicting outcome of a pregnancy with a viable fetus. Other markers such as inhibin A and a combination of markers need to be investigated to hopefully improve the prediction of outcome in women with threatened miscarriage.