First Bologna, then Poseidon: What are we still missing to personalize care for patients undergoing assisted reproductive technology with a poor prognosis?

The management of poor ovarian response (POR) poses a significant challenge to both patients and fertility specialists. When a woman presents for fertility evaluation, the initial assessment includes ovarian reserve testing to predict the response to stimulation, craft the appropriate treatment protocol, and identify the risk of POR. For decades, investigators have worked to formulate an accurate prediction algorithm from available baseline characteristics, an algorithm to stratify, prognosticate, and tailor fertility treatments for patients with low ovarian reserve parameters. In 2011, the European Society for Human Reproduction and Embryology introduced the Bologna criteria, a standardized definition of POR intended to be simple and reproducible (1Ferraretti A.P. La Marca A. Fauser B.C. Tarlatzis B. Nargund G. Gianaroli L. et al.ESHRE consensus on the definition of ‘poor response’ to ovarian stimulation for in vitro fertilization: the Bologna criteria.Hum Reprod. 2011; 26: 1616-1624Crossref PubMed Scopus (1276) Google Scholar). Truly, no good deed goes unpunished. The publication of these criteria stirred a lively debate and criticism. Critics argued that the thresholds used in the Bologna criteria were based on limited scientific evidence and encompassed marked “clinical heterogeneity,” lumping together diverse “subpopulations,” including those with genetic or acquired risk factors for POR, for example, Turner syndrome; FMR1 premutation; endometriosis; and/or history of pelvic infection, ovarian surgery, or chemotherapy. In short, critics noted that the criteria defined a population with “diverse baseline characteristics and unknown clinical prognosis” (2Papathanasiou A. Implementing the ESHRE ‘poor responder’ criteria in research studies: methodological implications.Hum Reprod. 2014; 29: 1835-1838Crossref PubMed Scopus (62) Google Scholar). In response to the perceived shortcomings of the Bologna criteria, the Poseidon Group introduced alternate criteria. Rather than defining “poor response,” these newer criteria set out to define “low prognosis.” Success was defined as “the number of oocytes necessary to obtain at least one euploid embryo for transfer” (3Alviggi C. Andersen C.Y. Buehler K. Conforti A. De Placido G. Esteves S.C. et al.A new more detailed stratification of low responders to ovarian stimulation: from a poor ovarian response to a low prognosis concept.Fertil Steril. 2016; 105: 1452-1453Abstract Full Text Full Text PDF PubMed Scopus (314) Google Scholar). Both the Poseidon and Bologna criteria were aimed toward a consensus definition of a population undergoing assisted reproductive technology (ART) with a poor prognosis. Nonetheless, the lack of consensus persists, and both sets of criteria remain largely unvalidated and underutilized. We applaud Reig et al. (4Reig A. Garcia-Velasco J.A. Seli E. Bologna vs. POSEIDON criteria as predictors of the likelihood of obtaining at least one euploid embryo in poor ovarian response: an analysis of 6,889 cycles.Fertil Steril. 2023; 120: 605-614Abstract Full Text Full Text PDF Scopus (1) Google Scholar) for their study comparing patients with POR as defined by the Poseidon Group and Bologna criteria. They retrospectively analyzed 6,889 cycles from 4,928 patients undergoing ART with preimplantation genetic testing for aneuploidy. The primary outcome was likelihood of obtaining at least 1 euploid blastocyst for transfer, and the exposure was POR designation by either or both criteria. The chances of obtaining a euploid embryo were the lowest among patients classified as having POR based on the Bologna criteria, with only 32% having a blast suitable for transfer. Among patients classified as having POR based on the Poseidon criteria, the lowest likelihood of obtaining at least 1 euploid blastocyst was observed in group IV (aged ≥35 years with an antimüllerian hormone (AMH) level of <1.2 ng/mL or antral follicle count (AFC) of <5) (44%), with increasing likelihood in group III (aged <35 years with an AMH level of <1.2 ng/mL or AFC of <5) (71%), group II (aged ≥35 years with an AMH level of ≥1.2 ng/mL or AFC of ≥5 but unexpected POR [<10 metaphase II oocytes retrieved]) (78%), and finally group I (aged <35 years with an AMH level of ≥1.2 ng/mL or AFC of ≥5 but unexpected POR [<10 metaphase II oocytes retrieved]) (97%). Of note, the likelihood of obtaining a euploid embryo in Poseidon group I was similar to that in patients without POR (97% vs. 92%, respectively), and thus, the investigators concluded that Poseidon group I should not be regarded as having POR. In addition, they found that euploidy rate was associated with age and not ovarian reserve (as expected). Unfortunately, neither the Bologna nor Poseidon criteria have been widely incorporated into clinical practice. We still hope that they will inspire POR research, defining standardized cohorts to test interventions; however, they have limited application for individual patients. For example, a 35-year-old with an AFC of 4 and a 43-year-old with an AFC of 4 would both be classified as Poseidon group IV. Based on the results of this study, patients in this group have a 44% chance of obtaining a euploid embryo; however, clearly, one should not counsel the latter patient that their chance of obtaining a euploid embryo is 44%. Hence, the generalizability of these results is limited when applied to clinical practice. For a personalized approach, predictor models, such as Society for Assisted Reproductive Technology and Centers for Disease Control and Prevention calculators, incorporate much more patient-specific data to provide personalized prediction, which is, thus, more accurate and useful than providing prediction based on the Bologna or Poseidon criteria. Moreover, these calculators report live birth rates, ultimately a more important outcome than the number of eggs retrieved or having a euploid embryo. Because of these limitations, we may see that the expertly developed Bologna and Poseidon criteria are soon replaced by machine-learning algorithms. These may ultimately serve not only the prediction of individual treatment success but study design as well. The recent findings of Reig et al. (5Reig A. Franasiak J. Scott R.T. Seli E. The impact of age beyond ploidy: outcome data from 8175 euploid single embryo transfers.J Assist Reprod Genet. 2020; 37: 595-602Crossref PubMed Scopus (34) Google Scholar) represent an important contribution to the field of POR research, especially their elegant and robust finding that age—not ovarian reserve—predicts euploidy. However, the investigators did not report live birth, a paramount outcome of ART. The same group of investigators previously reported an age-related decline in implantation and live births even when euploid embryos were transferred, underscoring that the odds of having a baby from autologous ART should be the outcome on which prediction algorithms are based. The question about what factors, besides embryo chromosome segregation errors, contribute to age-related fertility decline remains unanswered. Some hypotheses include endometrial aging, including that due to changes in endometrial gene expression; nonchromosomal embryo factors such as metabolism and embryonic epigenome that may deteriorate with advancing maternal age; and lastly, increase in uterine anatomic pathology. Therefore, to develop future interventions aimed at managing patients with POR, many of whom are of advanced age, we need to better understand these factors and their impact on live birth. Finally, it is worth highlighting that the path to obtaining an adequate number of embryos, leading to a euploid embryo and eventual live birth for patients with POR, may entail undergoing multiple stimulation cycles, including canceled ones. This can impose significant physical, emotional, and financial burdens on these patients. Although the findings of the studies by Reig et al. (4Reig A. Garcia-Velasco J.A. Seli E. Bologna vs. POSEIDON criteria as predictors of the likelihood of obtaining at least one euploid embryo in poor ovarian response: an analysis of 6,889 cycles.Fertil Steril. 2023; 120: 605-614Abstract Full Text Full Text PDF Scopus (1) Google Scholar) add significant information pertaining to the probability of obtaining a euploid embryo among patients with POR, the lack of data on canceled cycles, live birth rate, and cumulative live birth rate inevitably leaves readers with unanswered questions and a yearning for further knowledge. Bologna vs. POSEIDON criteria as predictors of the likelihood of obtaining at least one euploid embryo in poor ovarian response: an analysis of 6,889 cyclesFertility and SterilityVol. 120Issue 3PreviewTo study the likelihood of obtaining at least 1 euploid embryo for transfer in poor ovarian response (POR) diagnosed per Bologna and Patient-Oriented Strategies Encompassing IndividualizeD Oocyte Number (POSEIDON) criteria, and compare it between groups and with patients without POR. Full-Text PDF Open Access