Corpus Luteum Absence Is an Independent Risk Factor for Early Pregnancy Loss After ART: A Risk Stratification Model with Temporal Validation

Abstract Background To determine whether endometrial preparation protocols influence the risk of early pregnancy loss after assisted reproductive technology (ART) through corpus luteum (CL)-dependent mechanisms, and to assess the feasibility of integrating CL status as an independent predictor in a multivariable risk stratification model. Methods This single-center retrospective cohort study screened 6485 cycles achieving clinical pregnancy after IVF/ICSI or FET between 2012 and 2026. After excluding ongoing pregnancies (448), losses to follow-up (78), and ectopic pregnancies (94), 5865 cycles with definitive outcomes (live birth or spontaneous abortion) were analyzed. Cycles were classified by the presence of an endogenous CL into a CL-present group (fresh embryo transfer and FET natural cycle) and a CL-absent group (FET-HRT and FET-GnRHa + HRT). The 2012–2022 data formed the training set (n = 3582); the 2022–2026 data formed the independent temporal validation set (n = 2283). Results After excluding ectopic pregnancies, loss rates showed a clear gradient consistent with the CL hypothesis. The two CL-present groups had the lowest and comparable rates (fresh embryo transfer 14.9%; FET natural cycle 13.3%; P = 0.689). The two CL-absent FET groups were significantly higher (FET-GnRHa + HRT 18.9%; FET-HRT 19.8%; four-group χ² P = < 0.001). The pooled CL+ group (14.8%) differed from the CL− group (19.4%) by 4.6 percentage points ( P = < 0.001). After multivariable adjustment, CL absence remained an independent risk factor (aOR = 1.128, 95% CI 1.029–1.236, P = 0.010), but the cryopreservation effect could not be fully separated because the FET natural cycle subgroup was small (n = 143). The 7-variable model achieved an AUC of 0.6464 in independent temporal validation, significantly outperforming age alone (AUC = 0.6011, ΔAUC = 0.0453, DeLong P < 0.001), with a risk gradient from 3.2% to 39.1% (12-fold). Conclusions Fresh embryo transfer and FET natural cycle (both with an endogenous CL) had the lowest and comparable loss rates. CL absence was an independent risk factor for pregnancy loss, but the independent effect of cryopreservation could not be fully separated because of the limited FET natural cycle sample. The 7-variable risk stratification model effectively distinguished four risk tiers and significantly outperformed age alone.