Progesterone-induced blocking factor enhances pregnancy loss prediction in threatened miscarriage

Threatened miscarriage is common, and individualized risk stratification remains difficult when progesterone, beta-human chorionic gonadotropin, and ultrasound findings are discordant. This two-center retrospective prediction-model study evaluated whether progesterone-induced blocking factor (PIBF) adds prognostic information for early pregnancy loss before 16 gestational weeks. A total of 396 women were assigned to development (n = 182), internal validation (n = 78), and external validation (n = 136) cohorts. Development-only repeated cross-validation and a one-standard-error parsimony rule selected serum progesterone, fetal-heartbeat visibility, previous miscarriages, and log PIBF. Early pregnancy loss occurred in 45 (24.7%), 19 (24.4%), and 32 (23.5%) women, respectively. Compared with the three-variable conventional model, the PIBF-enhanced model increased the area under the receiver operating characteristic curve (AUC) in development (0.956 vs 0.901; P = 0.004), internal validation (0.942 vs 0.855; P = 0.009), and external validation (0.942 vs 0.813; P < 0.001). External bootstrap AUC was 0.942 (95% CI, 0.903-0.973), and the calibration slope was 0.971 (95% CI, 0.759-1.399). PIBF provided incremental prognostic information. The calculator is not a clinical decision tool without local recalibration. Larger independent external validation, treatment-confounding assessment, and assay standardization are also required before clinical use.