Female Age and Sperm DNA Fragmentation Linked to Embryo Euploidy, but Predictive Power Is Weak

A retrospective study of 588 preimplantation genetic testing for aneuploidy (PGT-A) cycles, largely involving patients with recurrent pregnancy loss, has found that only two parental factors—female age and sperm DNA fragmentation index—remained independently associated with whether a blastocyst was euploid after multivariable adjustment. Even so, the authors report that the predictive model built from these variables performed poorly, and they conclude that the measured clinical characteristics lack sufficient value for clinical use.

The research, published in Frontiers in Endocrinology, analyzed 2,357 blastocysts from the 588 PGT-A cycles. Of these embryos, 1,299, or 55.1 percent, were euploid. The team used embryo-level generalized linear mixed models to evaluate six candidate factors: female age, female body mass index, anti-Müllerian hormone (AMH), male age, male body mass index, and the sperm DNA fragmentation index (DFI).

In univariable analyses, three of the six factors showed significant associations with euploidy: female age, male age, and male DFI. However, when the variables were entered together in a multivariable model, only female age (odds ratio 0.87, p = 0.011) and male DFI (odds ratio 0.89, p = 0.006) remained independent predictors. Male age, significant on its own, did not survive adjustment alongside the other variables.

The most striking result concerned model performance. The apparent area under the receiver operating characteristic curve was just 0.571, and after internal validation using a patient-clustered bootstrap procedure with 200 replicates, the corrected AUC was 0.574—barely better than chance, which corresponds to 0.5. The model’s explained variance was similarly minimal, with an apparent marginal R² of 0.018 and a conditional R² of 0.036.

Calibration, the agreement between predicted and observed probabilities, was assessed alongside discrimination. The apparent calibration slope was 0.985, and the bootstrap-corrected calibration slope was 0.913. The authors interpret this combination of weak discrimination and limited generalizability as evidence that the signal carried by these parental characteristics is too faint to support clinical decision-making, despite the statistically significant associations themselves.

The study also explored whether the association between male DFI and euploidy differed across female age groups. These subgroup analyses produced some nominal associations, but none remained statistically significant after correction for the false discovery rate, with all adjusted q values exceeding 0.05. The authors describe these subgroup findings as exploratory and hypothesis-generating rather than confirmatory, and they note that no statistically significant age-modified association between male DFI and embryo euploidy persisted after the correction.

At the cycle level, the chances of obtaining at least one euploid blastocyst were high across the cohort. Among the 588 cycles with complete covariate data, 90.3 percent had at least one euploid blastocyst. When all 613 biopsied cycles were considered, the rate was 89.89 percent, and across all 637 started cycles it was 86.50 percent. These figures underscore that even in a cohort dominated by recurrent pregnancy loss, most cycles yielded at least one chromosomally normal embryo for transfer consideration.

The authors emphasize that their analysis was retrospective and that the models relied solely on parental clinical characteristics rather than embryo-specific information. They call for future predictive models that incorporate embryo-level data and draw on larger samples, suggesting that the modest associations detected here represent only a small fraction of the variation in blastocyst euploidy.

For clinical practice, the message is one of caution rather than dismissal. Female age and sperm DNA fragmentation are genuinely associated with the likelihood that a biopsied blastocyst will be euploid in this cohort, but the strength of those associations is too small to produce a useful individual-level prediction. Patients and clinicians weighing PGT-A decisions should therefore not expect these parental measures alone to forecast embryo chromosomal status with meaningful accuracy.

The study adds to an ongoing debate about which clinical factors, if any, can reliably inform expectations about embryo euploidy in assisted reproduction. By applying formal internal validation and multiple-testing correction—steps often omitted in smaller retrospective analyses—the work provides a measured benchmark: real but weak associations, and a reminder that statistically significant predictors do not necessarily translate into clinically useful models.

Subject of Research: Associations of parental clinical factors with blastocyst euploidy in a predominantly recurrent-pregnancy-loss PGT-A cohort

Article Title: Parental determinants of blastocyst euploidy in a predominantly recurrent-pregnancy-loss PGT-A cohort: limited predictive value of female age, BMI, AMH and sperm DNA fragmentation

Article References: Tan, C., Wang, X., Tong, D., Zou, P., Liu, X., & Yu, Y. (2026). Parental determinants of blastocyst euploidy in a predominantly recurrent-pregnancy-loss PGT-A cohort: limited predictive value of female age, BMI, AMH and sperm DNA fragmentation. Frontiers in Endocrinology, 17, 1909930. https://doi.org/10.3389/fendo.2026.1909930

Image Credits: AI Generated

DOI: 10.3389/fendo.2026.1909930

Keywords: PGT-A, embryo euploidy, female age, sperm DNA fragmentation index, recurrent pregnancy loss, predictive model, generalized linear mixed model, AMH, BMI, retrospective study, blastocyst, assisted reproduction