Putative “ovarian awakening” in diminished ovarian reserve: a multicenter analysis demonstrates indistinguishability from regression to the mean

In clinical practice, repeated oocyte retrieval cycles are often necessary for patients with diminished ovarian reserve (DOR). It has been hypothesized that the mechanical stimulation from the initial oocyte pick-up (OPU) procedure might mimic the effect of in vitro activation (IVA), potentially activating dormant follicles and improving outcomes in subsequent cycles. This study aimed to determine whether repeat ovarian stimulation cycles yield superior oocyte and embryological outcomes compared to the initial cycle in patients with DOR. This multicenter retrospective cohort study analyzed data from 2,883 DOR patients who underwent at least two ovarian stimulation cycles within 12 months of their initial cycle between 2016 and 2023. Generalized Linear Mixed Models (GLMMs) were employed as the primary analytical method. To formally evaluate the potential impact of regression to the mean (RTM), a simulation-based permutation test was conducted, fitting a null model with only patient-level random intercepts to simulate datasets without any cycle order effect. GLMM analysis initially indicated that the second and third retrieval cycles yielded significantly more oocytes, 2PN zygotes, transferable embryos, and high-quality embryos compared to the first cycle (all P < 0.05). Mediation analysis suggested that an increase in antral follicle count (AFC) explained over half (51.4%) of the benefit in oocyte yield in the second cycle. However, the formal RTM analysis revealed a consistent numerical trend where the observed “improvement” in oocyte yield was lower than the effect size expected from RTM alone (adjusted effects ranging from − 0.44 to -1.18 oocytes). Crucially, the observed effects were not statistically distinguishable from those generated by the RTM phenomenon (all p-values > 0.49). The apparent improvement in oocyte yield and embryological outcomes in subsequent cycles for DOR patients, is statistically indistinguishable from the effect expected due to regression to the mean. Caution is warranted in interpreting these outcomes as a genuine biological effect, such as “ovarian awakening”.