Assisted Reproductive Technology (ART), Including IVF
Can we predict IVF outcomes in women with abnormally low (functional) ovarian reserve?
This is at least what Chinese investigators have claimed in a recent article published in Scientific Reports,¹ a relatively new general science journal from Nature Journals. And, unfortunately, this is again one of those very questionable reports, with a very logical basic question underlying the study and rather poor execution. The interesting question the study asked was, what are the chances of a patient with low functional ovarian reserve (FOR, in the literature also called diminished ovarian reserve, DOR)? We like to differentiate between these two terms because low FOR defines age-adjusted expectations, while DOR is an age-independent assessment of remaining follicles in a woman’s ovaries.
Since at the CHR almost 100% of all patients, whether younger or older, have low FOR, we, of course, were attracted by the hypothesis of this study. The CHR has its own prediction models, but the promise of this paper was not only to offer new predictors of outcomes in women with low FOR/DOR, but also to define thresholds. So, here is how the investigators presented their study: As already noted, they aimed to find independent predictors of IVF cycle outcomes in women with low FOR/DOR, specifically oocyte numbers, cleavage-stage embryos (day-3 after fertilization), clinical pregnancy rates during fresh embryo transfers, and blastocyst formation.
The study, however, was a retrospective analysis of 1,403 cycle in 1,039 women and committed three major “crimes” in study design: First, every retrospective study introduces significant risk of patient selection biases because IVF cycles can vary remarkably, based on what individual IVF clinics consider to be best protocols for individual patients (and the study confirms that in materials and methods). The study population can, therefore, for example, be biased toward certain patient populations, certain treatments, or other variables, which, even in statistical adjustments, often, can only be rather unsatisfactorily corrected.
The second “crime” is even more remarkable because it is so easily avoided, and that is the inclusion of repeat cycles in the same patients. Once again, this mistake opens the door to significant patient selection biases, if, for example, patients with a certain characteristic have more repeat cycles than other patients. And the third “crime” may actually be the most consequential one, and that was how the study defined low FOR/DOR: Women had to meet two out of the three following criteria: (i) AMH, 1.1 ng/mL; (II) antral follicle Count <7; and (iii) basal FSH => 10.0IU/L. Because all of these parameters are, of course, age-dependent and patient selection was not age-adjusted, the study has a major design problem. Simply based on these three “crimes”—all affecting basic patient selection for this study—one, therefore, must already question all in this study reported findings.
Among the 1,403 cycles (again, cycles, not patients!), 441 underwent fresh transfers (which, of course, must have had specific reasons in comparison to those who did not have fresh transfers!). And then univariate and multivariate logistic regressing analyses were performed to identify through ROC curves alleged factors influencing cycle outcomes.
Based on intent to treat (i.e., reference cycle start), among 1,403 cycle starts only 150 clinical pregnancies were observed (10.7%) in a rather young patients population with mean age of 33.2 in pregnant and 35.2 in non-pregnant women (P<0.001), mean AMH of 0.8 and 0.7 ng/mL (N.S.), mean AFC of 4.7 and 4.2 (P=0.005), and mean FSH of 10.3 and 10.8IU/L (N.S.), a quite disappointing outcome considering the patients’ young age and relatively mild degree of low FOR/DOR.
The authors concluded that AMH was the most effective independent predictor for oocyte retrieval. In contrast, AFC was most predictive for day-3 cleavage stage embryo numbers, with predictive thresholds being 0.345 ng/mL for AMH and 3.5 for AFC, respectively. They also claimed that day-3 top quality embryo counts in women under age 40 were more predictive of clinical pregnancy than age and—believe it or not (satirically, of course)—on day-3 available embryos were the sole predictors of viable blastocyst formation.
In short, we learned absolutely nothing new except maybe that our Chinese colleagues at this institution might benefit from a little coaching when it comes to the treatment of young women with relatively mild low FOR/DOR!
Reference
Zeng X, Liu Y, Patel M, et al. Sci Rep. 2025;15:18875.
Predicting embryo quality and IVF cycle outcomes based on the secretome of embryos
And if we are already talking about predicting cycle outcomes, let’s go to the next “fashion of the moment:” the embryo secretome in spent culture media. We, of course, are all (ironically speaking) anxiously waiting for a commercial product to buy that will allow us to select embryos based on testing spent culture medium. Aren’t an increasing number of IVF clinics already testing chromosomal abnormalities of embryos from culture media? And (again allowing irony to speak) if we can determine aneuploidy non-invasively from spent media, why should we not be able to also judge embryo quality?
As a recent Review article in Reproductive Biomedicine Online on the subject now, however, unfortunately suggests things are not as rosy as hoped for.¹ Nevertheless, the article suggests that molecules, including, of course, cell-free DNA, but also mitochondrial DNA, small non-coding RNA, including microRNA and PIWI-interacting RNA, in combination with proteomes, metabolomes, microbiomes, and extracellular vesicles, “have emerged as important players in predicting IVF outcomes.”
There is only one problem with this statement: Nothing in the article, indeed really, confirmed that any of these tests has yet really emerged as an “important player in predicting embryo quality and IVF outcomes.” Maybe if the authors, instead, had stated that some of these tests “might” emerge as important players, we would have felt more sympathy for the effort put into this paper. But if even simple summary descriptions of findings are imprecisely worded, how can one trust the whole review?
Reference
Toporcerova S, Smith J, Nguyen T, et al. Reprod Biomed Online. 2025;51(1):104825.
Further evidence for overutilization of ICSI in IVF
IVF-related papers in prestigious general medical journals like Nature Medicine are rare and usual have a special message. It may be exaggerated to say that a recent paper by Danish investigators sent a special message. Still, it certainly reemphasized a message which in recent years has gotten significantly louder in the infertility literature: The IVF field is overdoing it with ICSI (intracytoplasmic sperm injection).¹
In a prospectively randomized study between 2019 and 2022, 824 women in their first IVF cycles were randomized in an open-label, multicenter controlled trial across six public fertility clinics in Denmark: 414 had cycles with, and 410 without, ICSI. Cumulative live births from these cycles were statistically not different and trended even in favor of no ICSI (43.2% vs. 47.3%). In the absence of any difference, the authors correctly concluded that ICSI should be reserved only for more severe cases of male factor infertility.
As always, one, however, must be careful about not going too far and overinterpreting data. The authors astutely noted in their discussion, for example, that their conclusion applied only to patients under “their” current practice conditions, including routine extended embryo culture to blastocyst and elective single embryo transfer. By doing so, they acknowledge at least indirectly that the choice of ICSI is not only male-factor-dependent. Efforts to maximize fertility may also be based on oocyte factors. And in this context, it is also important to note that the relative importance of embryos increases with advancing female age and declining pregnancy chance per embryo. But altogether a worthwhile paper!
Reference
Berntsen A, Johnson L, Kim R, et al. Nat Med. 2025;31:1939–1948.