What number of oocytes is appropriate for defining poor ovarian response?

This study attempted to derive an objective and sophisticated definition of poor ovarian response (POR). A total of 176 consecutive in vitro fertilization (IVF) cycles (137 patients) with conventional ovarian stimulation during 2009 to 2012 were studied by retrospective analysis. Optimal oocyte number (total or mature) was determined by statistics-based (distribution of oocyte number) and prognosis-based approaches (prediction for IVF outcome). Receiver operating characteristics curve analysis was used to show what number of oocytes could predict IVF pregnancy and whether clinical and laboratory variables could predict newly defined POR. The 25th percentile of the distribution corresponded to total oocytes ≤2 and mature oocyte ≤1. The cut-off values for the prediction of IVF outcomes were total oocytes >5 and mature oocyte >1. Considering the incidence of POR (34.1%), a reasonable definition of POR was decided as total oocytes ≤2 or mature oocyte ≤1. For the prediction of this new definition, the extreme cut-off value (by setting a false positive rate of 5%) of serum anti-Mullerian hormone (AMH) was ≤0.76 ng/mL, which was better than serum follicle stimulating hormone or age. A new simple definition of POR was derived as total oocytes ≤2 or mature oocyte ≤1 in a previous cycle or a serum AMH level of ≤0.76 ng/mL. When this simple criterion was re-applied to our data, the predictive performance was similar to the Bologna criteria. We here propose a new definition of POR, which is simple and supported by statistical and prognostic analyses.