Model construction and application for predicting pre-eclampsia by Sonoclot coagulation analyzer

Abstract

Maternal age has significantly increased among Chinese women, thereby posing risk of pregnancy-related complications. Preeclampsia is a leading cause of maternal and perinatal morbidity and mortality, and coagulation analysis in conjunction with clinical signs and symptoms are generally used for its diagnosis with limited efficacy. Sonoclot coagulation analyzer is effective in assessing coagulation function used during cerebral surgery and cardiovascular surgery. However, its use has not been explored in preeclampsia. Here, we investigated the potential use of Sonoclot in diagnosing preeclampsia in obstetrics cases. Subjects meeting the screening criteria were divided either into a test group or a control group, according to whether they were preeclamptic or not. We recorded the Sonoclot-derived coagulation and the routine coagulation parameters including platelet function (PF), activated clotting time (ACT) and clot rate (CR), prothrombin time (PT), activated partial thromboplastin time (APTT), thrombin time (TT), fibrinogen (FIB), and platelet count. Regression analysis was done on the relevant parameters to assess the feasibility of Sonoclot analyzer in preeclampsia diagnosis. In parallel, changes in preeclampsia lncRNAs was also evaluated. Significant differences were recorded in PT and ACT between the two groups. In the monovariant logistic regression, PT and ACT appeared to be reliable predictor variables. In the multinomial logistic regression, a total of five regression steps were performed with decreasing AIC values. The K-fold cross validation resulted in an accuracy rate (ACC) of 77.5%, a false positive rate of 16.4%, and a false negative rate of 33.2%. lncRNAs ANRIL and HOXD-AS1 were found deregulated. Our findings indicate that Sonoclot may be useful for diagnosis of preeclampsia in obstetrics.