Comparison of beta‑human chorionic gonadotropin‑based prognostic models on the clinical outcomes of gestational trophoblastic disease patients in the Philippines / Philippine Journal of Obstetrics and Gynecology

ABSTRACT

Objective@#Despite the widespread use and measurement of beta‑human chorionic gonadotropin (β‑HCG) among hydatidiform mole (HM) patients, models derived from this biomarker to predict the remission or postmolar gestational trophoblastic neoplasia (GTN) rarely perform well. The study aimed to generate cutoff points for postevacuation β‑HCG levels and evaluate their performance among women with complete molar pregnancies@*Methods@#A retrospective cohort study composed of women with complete HM underwent bivariate procedures comparing characteristics between the comparison groups. Cut points using Liu’s and Youden’s indices were estimated, and their performance was evaluated using receiver operating characteristic curve analysis. Cox regression to compare time‑to‑progression across these proposed β‑HCG cutoffs was also performed.@*Results@#The incidence of postmolar GTN among the 155 women in the study was 15.5% (95% confidence interval 10.2%–22.2%). Postevacuation HCG levels had a better prediction of disease status than preevacuation and HCG ratio models (χ2 163.07, P < 0.01). A cutoff at 508 mIU/mL the 3rd‑week postevacuation (area under the curve [AUC] 0.89, sensitivity 87.5%, specificity 90.1%) was comparable with the 185 mIU/mL cutoff at the 5th‑week postevacuation (AUC 0.89, sensitivity 91.7%, specificity 87%). The hazards ratio of postmolar GTN was 29.74 (8.53–103.71) and 39.89 (8.82–180.38) for the 3rd and 5th weeks HCG after evacuation adjusting for clinically relevant variables@*Conclusion@#The first 3rd‑ and 5th‑week postevacuation levels of β‑HCG demonstrated potential in predicting postmolar GTN. However, further refinement and adjustment for clinically relevant risk factors are still needed.