摘要

Objective: The aim of this study was to comparatively study a novel model and existing models of predicting postmolar gestational trophoblastic neoplasia (GTN). @@@ Methods: Two hundred twenty-two patients with complete hydatidiform moles were enrolled retrospectively. A natural regression was noted in 195 patients (spontaneous regression group), whereas the remaining 27 patients entered postmolar GTN (postmolar GTN group). The upper limits of the 95% confidence interval of human chorionic gonadotropin (hCG) values and hCG regression rates were calculated aggregately from the spontaneous regression group. The 4 prediction models (weekly hCG regression curve and weekly hCG regression rate curve reported by previous studies; daily hCG regression curve and daily hCG regression rate curve pioneered by us) were then plotted. The individual hCG curve of the postmolar GTN group was plotted and compared with the prediction models, respectively. The individual hCG curve superimposing the prediction curve was considered showing an elevated risk of GTN. @@@ Results: All patients with postmolar GTN were preidentified by daily hCG regression rate curve. The other 3 prediction models had a considerable rate of failure in identification. Mean diagnosis time of daily hCG regression rate curve was significantly lower (P = 0.008), with an average of 15.3 days gained compared with International Federation of Gynecology and Obstetrics criteria. Cochran Q test showed that daily hCG regression rate curve produced a significantly better performance in predicting postmolar GTN than weekly hCG regression curve (P = 0.01). @@@ Conclusions: Our data showed that daily hCG regression rate curve gives a better prediction of postmolar GTN and might potentially enhance the monitoring of patients with molar pregnancy, especially those who could not adhere to International Federation of Gynecology and Obstetrics protocols. However, this preliminary research should not change current clinical practice until further validation is carried out.