Fertility and Sterility On Air - TOC: Dec 2024

Take a sneak peek at this month's Fertility and Sterility! Articles discussed this month are:  

01:48 Trustworthiness criteria for meta-analyses of randomized controlled studies: OBGYN journal guidelines

06:24 Playing the long game for reproductive rights

08:04 Limiting the number of fresh donor oocytes inseminated with sperm as a strategy to minimize supernumerary embryos

15:38 Effectiveness of degradable polymer film in the management of severe or moderate intrauterine adhesions (PREG-2): a randomized, double-blind, multicenter, stratified, superiority trial

26:37 Aneuploidy rates and likelihood of obtaining a usable embryo for transfer among in vitro fertilization cycles using preimplantation genetic testing for monogenic disorders and aneuploidy compared with in vitro fertilization cycles using preimplantation genetic testing for aneuploidy alone

38:13 Antimüllerian hormone levels are associated with time to pregnancy in a cohort study of 3,150 women

47:50 Racial disparities in the outcomes of euploid single frozen-thawed embryo transfer cycles – analysis of the Clinical Outcome Reporting System of the Society for Assisted Reproductive Technology 2016–2018 data

57:14 No association between celiac disease and female infertility: evidence from Mendelian randomization analysis

View Fertility and Sterility October 2024, Volume 122, Issue 6:  https://www.fertstert.org/issue/S0015-0282(24)X0013-8

View Fertility and Sterility at https://www.fertstert.org/

Welcome to Fertility and Sterility On Air, the podcast where you can stay current on the latest global research in the field of reproductive medicine. This podcast brings you an overview of this month's journal, in-depth discussion with authors and other special features. F&S On Air is brought to you by Fertility and Sterility family of journals in conjunction with the American Society for Reproductive Medicine, and is hosted by Dr. Kurt Barnhart, Editor-in-Chief, Dr. Eve Feinberg, Editorial Editor, Dr. Micah Hill, Media Editor, and Dr. Pietro Bortoletto, Interactive Associate-in-Chief.

Good morning, everyone, and welcome back to another episode of Fertility and Sterility On Air. I'm Micah Hill, the Media Editor. We're in December 2024, volume 122, number six.

I'm joined today by Kate Devine. Good morning, Kate. Good morning, Micah.

Good to see you, Pietro, all the way from New Orleans. Not as usual Boston accommodations. Good to see you, Pietro.

I'm now doing double duty this week for F&S, recording podcasts this morning and then jetting off to record a live from the AAGL annual meeting episode with some reputable reproductive surgeons you may have heard about, one Rebecca Flick, Chuck Miller, and Zaraq Khan. That episode should be dropping later this month. Sounds great.

Thank you for doing that and for joining us while on the road, Pietro. And last but not least, our Editor-in-Chief, Kurt Barnhart. Good morning, Kurt.

Good morning, everybody. Happy to be here. I look forward to this.

Good to see everyone. We're fresh off of ASRM 2024. We had a wonderful journal club on obesity, and so that should be also dropping this month, as well as two podcast episodes where we interview some of the best of the science that we saw being presented at the meeting.

So look for all of that coming. Kurt, we're going to jump right into an editorial that you wrote that's actually published in a lot of journals. So this doesn't fit into the traditional sort of front matter of the Journal of F&S, but it's an important one.

It's on trustworthiness criteria for meta-analyses of RCTs, OBGYN journaling. And so I'm just going to summarize it real quick, and then I just want to hear your thoughts as our Editor-in-Chief of why you felt like this was important. So essentially, this lays out 21 trustworthiness criteria for a randomized clinical trial, and they're pretty basic things like pre-stating the outcome of your trial and sticking to that so you don't do what's called endpoint switching.

It goes through a lot of other things like effect sizes. Are these studies showing similar findings to other studies, or are they very different or unbelievable? So it goes through all of these criteria, and essentially the consensus of this group of Editor-in-Chiefs like yourself was that the primary analysis done in a meta-analysis should only include the high-quality RCTs that meet these trustworthiness criteria. So Kurt, if you could just give us some background, why did you think this was important? What's the issue here that this group of Editor-in-Chiefs is trying to address? Sure.

It's two really important issues. The first one is, much to my chagrin and perhaps naivete, there is a lot more medical fraud in the literature than I ever imagined. It used to be that medicine was one of those domains where you really trusted what you saw as the truth or at least a sampling of the truth.

It might have been hard to replicate for lots of reasons, but we're finding out now that there are a lot of studies that simply are just not accurate or made up or are just wrong. And that translates down into meta-analyses because it's relatively easy to take all of your information from the literature, put it into a computer program, and spit out a summary estimate and claim that that's the truth. And in fact, it amplifies the falsity or the falsehood of the other trials.

So what we're trying to bring up as the Editors-in-Chief is saying that while we're working on research integrity in many ways, those that are writing meta-analyses need to pay attention to this. And we need to try to give them objective criteria to say that not every study is graded equal and it's on you, the author, to figure that out. You just can't just say because it's published, it must be true.

And therefore, these criteria are ways of saying, you know, use your judgment and look at what you think are really good, high-quality studies. Don't wait for a study to be retracted or have a letter of concern. It surprises me to date the number of meta-analyses we get where there is no screening of the studies and they're giving us a summary measurement and passing this off as the truth.

So this is an editorial saying, at least pay attention to it as we work out some of these criteria. I personally think when you do a meta-analysis, the primary analysis, the one that you report as your primary finding should be the ones you think are the good studies. And then the sensitivity analysis should be all of them, not the other way around.

Because if you do the primary analysis as all of them, you're not getting the truth. And some people never get to the sensitivity analysis or find that as the better analysis. So that's, sorry for giving you a long answer, but that's the impetus for that editorial.

And I'm glad we're able to do it with a lot of other editors-in-chief, not just one journal. Yeah. I think this is a fantastic statement.

I encourage everyone to read it. And just to drive home that point, you were saying, Kurt, you know, we had the updated clinical guidelines from ASRM on subclinical hypothyroidism management. And there's one study that is either fabricated or unbelievable in the results where everyone gets pregnant using levothyroxine replacement.

And the effect size of that study is about 10 times higher than the other randomized trials. There's only three of them, so there's not even many. And of the nine meta-analyses on those three randomized trials, so we've tripled the number of meta-analyses and we have trials, all of them did their primary analysis including that study.

And so their conclusion is it's helpful. Only one did a secondary analysis where they excluded it. But that gets buried.

It's not in the abstract. And like you say, people don't read that. And it was only done in one out of them.

And so I think that drove a lot of these guidelines around the world for millions of women just because of one fabricated study and people not taking the approach of quality that you're suggesting. So it's not just a nerdy exercise. I think there's really practical implications for how these trials are done.

And it's important to say one of my favorite Kurt Barnhart-isms, which is if there are more meta-analyses than there are articles that contribute to the meta-analyses, that might be a problem. I also want to make sure that this is taken as a hard line. I mean, meta-analyses will not be accepted in fertility and serality unless you go through this exercise.

Fantastic. We have some other front matter this month. We have Playing the Long Game for Reproductive Rights from editorial editor Nanette Santoro.

And she's basically talking about the long-term strategies and implications we should have in the face of shifting reproductive sands in our country, in the United States. And this was written before these most recent election results. So I encourage everyone to read that.

There's a couple of other articles in there looking at advocacy and the Supreme Court in reproductive medicine. There's a fertile battle from Dr. Eric Widra on donor incentives revisited. And for those who aren't aware, there are some companies and models out there that are looking at incentivizing the donor pool.

We've reviewed some articles recently that, especially for different minorities, we don't have enough donors in the United States, donor oocytes and donor sperm. So looking at offering, perhaps freezing up donor's eggs, half of them, and giving the other half to the recipients as a way to try to incentivize donor egg donation or donor sperm donation. And they talk about the ethics and the pros and cons, interesting arguments on both sides.

And then finally, the Inkling is also from Eric Widra. He's very busy this month on how do we communicate failure of IVF risk to patients. And they go through some really good data that patients tend to overestimate their success.

And we've all had this as clinicians, where on the back end, when it fails, you're then counseling the patient who's surprised that it didn't work, even though you felt like you gave them realistic expectations and numbers up front. So definitely a lot of good front matter for us to read through and think about in the journal this month. It's really exceptional.

Pietro, you're on a timeline because you've got a podcast today, AGL to record. So we're going to go out of order and skip right to you. So your first article this month is on limiting the number of fresh donor oocytes to try to mitigate supernumerary embryos.

Tell us about this study. Thanks, Micah. And thanks, Kate and Kurt, for letting me go a little sooner in the podcast.

I know these are towards the end of the table of contents. But Micah, you're on a timeline. So you're right.

I have a great article from my old buddies at Weill Cornell. First author Samantha Spring, recently matched REI fellow staying at Cornell, and a friend of the podcast, Dr. Steve Spandorfer, entitled Limiting the Number of Fresh Donor Eggs Inseminated with Sperm as a Strategy to Minimize Supernumerary Embryos. So here's a fun fact.

As of 2021, at least one and a half million embryos are estimated to be in storage. Wow, that's a lot. Also kind of tough to wrap your head around how they estimated that, but kind of gives you a sense of the magnitude of the problem that embryologists have.

And if you've been paying any attention to the news recently in the United States, you know that these cryopreserved embryos have been the subject of much ethical and legal consternation in the United States. The authors of this paper kind of ran with that idea and took an idea from an earlier FNS paper, 2023 by Correa et al., that employed modeling to arrive at an ideal number of eggs to fertilize for IVF for patients 18 to 45, and they took that idea and extended it to the donor egg population, which they argue are probably the most at risk of having supernumerary embryos. They used data from a single center, while Cornell 2013 to 2022 for patients undergoing fresh donor egg cycles.

They defined supernumerary blasts as a number of blasts remaining after their final cycle, which defined final cycle as the cycle that resulted in the patient's first live birth, or if no live birth was achieved, the last cycle performed at that institution. You can argue back and forth if that's truly the last cycle, and there's some limitations to that, but it's a pretty decent working definition. The secondary outcomes for the study included the live birth rate for a given number of donor oocytes retrieved, and the number of usable embryos for a given number of donor oocytes retrieved.

In total, they had 543 donor egg recipients, which were included with a total of 750 embryo transfer cycles analyzed. Two-thirds of these were fresh embryo transfers from fresh donor eggs synchronized with Lupron and estrogen, and then 38% of them were frozen embryo transfers from fresh donor eggs. The mean recipient age was 43, and the mean donor age was 26.

The mean number of eggs retrieved was 12, 10 of which were mature, 8 fertilized, and the mean number of usable blasts was 4. Now, the mean number of supernumerary embryos was 2 for the entire cohort, and you're probably wondering, well, how many live births did you have? 75% of this cohort achieved at least one live birth. Things became helpful and elucidating, I think, kind of got to their main point when they stratified the group into quartiles based on the number of eggs retrieved. And here's kind of the top-line finding and why I think this paper is interesting.

So there was no apparent difference in the live birth rates, whether less than seven eggs, eight to 10 eggs, 11 to 14 eggs, or greater than 15 fresh donor eggs were fertilized, whereas there were significant increases in supernumerary embryos as the number of oocytes exposed to sperm increased. Kind of makes sense. But I think what this paper tells us is that there's probably a sweet spot for a number of eggs to fertilize if you're trying to balance live birth and minimize the number of supernumerary embryos.

Now, I think the reason why this data is interesting is, one, the aforementioned ethical concerns that have come up in the United States regarding gametes and embryos that are crowd-preserved. There's also some very real-world practical implications. There's logistical implications, there's financial burdens on both families and clinics from having supernumerary embryos after people have completed their family building.

But probably the one that I think resonated with me the most was donor oocytes are a limited resource, and identifying the optimal number of oocytes to fertilize would probably better allocate and improve utilization of donor oocytes, potentially, we think. You'd have to kind of put that into practice and see if it plays out. But I was really thrilled to see this paper published in FNS because I think it was not only new but also thought-provoking and got me thinking a little bit about, what is the number of oocytes to fertilize when using the donor population to minimize the number of supernumerary blocks? Micah, Kurt, Kate, what do you think? I was really happy to see this in FNS as well.

At Shady Grove, we have long used split donation cycles as a way to maximize the resource of donor oocytes while also minimizing the cost to recipient couples. And I couldn't agree more with you that that is the most compelling reason to take such a strategy. And I think it could be extrapolated as well to trying to ascertain what's the appropriate size of an egg lot for frozen donor eggs as well, which I think a lot of centers are moving to due to convenience.

But I was happy to see this. I agree with you that I care more about helping my patients achieve their desired family size at a cost that is manageable and to serve as many patients as possible than I do about necessarily avoiding supernumerary blasts. But that is certainly a priority for many people and increasingly so in this political climate.

I think this is interesting in that usually our field has been focused on the individual. What does it take to get this person pregnant in front of me? And if there's excess, so be it. I'd rather maximize my success.

The fact that we're now considering the greater good, not the individual, is intriguing. And I'm wondering if this is how this is going to play out unless there are mandates on the number of oocytes. I'm not sure I'm going to find a lot of patients that are saying, well, I'm willing to take a lower amount because it's okay if I don't get what I want, you know, at least I helped everybody else.

So this is going to be an interesting story to see how it plays out. And the other story is there's also, we talked about another podcast, not all oocytes are created equal and which oocytes you're getting is also going to be a battle, if you will. Are you getting the best six of the group or are you getting the last six of the group? So intriguing.

And I'm glad that the paper was in there to make us think about it. Agreed that I wouldn't think that patients would want to accept fewer eggs, but we have found over time they gladly will for a lower price tag. And so it really just kind of depends on what that individual patient's cost benefit analysis is.

But yes, as I hope we will be able to better assess oocyte quality, certainly that also comes into the mix. So great point. I think there's the niche case is where I think this data is really helpful for those patients that are living in states that there is an ethical or religious or even legal concern about having excess cryopreserved human tissues, but then also in the niche situation where there are religious concerns about having excess cryopreserved human tissues.

I think this data helps those folks sort out, all right, maybe I don't need to fertilize everything and I can probably still achieve that same live birthrate across these different strata of number of eggs retrieved. Great article, great discussion. As it was said, this one sort of flips the way this question has been asked from this perspective, which is great.

Petra, we're sticking with you and you're at AAGL, so it's a perfect chance to talk about how do we prevent intrauterine adhesions? If you've listened to this podcast before, you've heard me say that I love when a reproductive surgery article sneaks in to fertility and sterility because there are so few of them, but I really do somersaults when I see an RCT in reproductive surgery sneak into FNS, so this was outstanding for me. This paper is entitled The Effectiveness of Degradable Polymer Film in the Management of Severe or Moderate Intrauterine Adhesions, also known as the PREG-2 trial, a randomized double-blind multicenter stratified superiority trial. That's a mouthful.

So if we think a little bit about our clinical practice, we all have patients with intrauterine adhesions. They happen. Our first line is typically a simple hysteroscopic lysis of adhesions and hope that we did a good enough job in getting the cavity normalized and we can get them pregnant before anything else happens, but in the literature, particularly for moderate or severe adhesions, recurrence rates after adhesive lysis can be as high as 65% in the most extreme cases, and we've all tried different versions of hormones, IUDs, balloons, serial hysteroscopy, early semen sonogram for cavity distension, and depending where you are in the world, some have actually used barrier gels or even human amniotic membrane grafts with varying effectiveness.

Probably the major issue with most of these therapies, things that stick in the uterus is that they all have short residency times. They just don't linger long enough to prevent adhesions beyond just a couple of days or even a couple of hours. There's a potential risk of infection.

There's a risk of invasiveness, and then there's the annoyance of having to come back into the office to have something potentially removed. Enter the Womed Leaf. The Womed Leaf is a product available not in the United States, in Europe, which is a tri-block adhesion barrier.

It's a degradable polymer film that's designed to be inserted at the end of a hysteroscopic procedure, and once it's inserted into the uterus, it absorbs fluid and expands to fill the cavity to keep the walls away from each other. And then the real cool innovation is that it has hydrolytic enzymes that degrade the gel into water, and it falls out through the cervix after about a week of dwell time. This paper in FNS, like I told you, was a multicenter double-blind RCT.

It was conducted between the years of 2021 and 2023 in France, the Czech Republic, Spain, Belgium, Italy, Switzerland, and with a huge contribution coming from China. The enrolled patients were patients with infertility, recurrent pregnancy loss, menstrual cycle disorders who were found to have modern or severe intrauterine adhesions as classified by the American Fertility Society classification system. And for those who are unfamiliar with that classification system, because it is pretty niche, this is a 12-point scale with scores of 5, 6, 7, and 8 indicating moderate adhesive disease, 9, 10, 11, and 12 indicating more severe adhesive disease.

So in this trial, patients were randomized one-to-one to the WOMED leaf film versus no intervention via block randomization at the completion of the hysteroscopic license of adhesions procedure, and they did allow for hormones to be used depending on what the local practice pattern was. And for those who are unfamiliar with block randomization, block randomization is a technique to reduce the risk of imbalance but also prevent unanticipated confounders. So it's a nice elegant use of block randomization here for a surgical trial.

Patients all had second look hysteroscopies and the evaluators were blinded to the allocation. The second look was typically performed about six weeks later. In total, they had 160 patients that were randomized, and of these patients, nearly two-thirds of them had infertility as an indication for the procedure with D and C being the leading suspected cause of intrauterine adhesions in this cohort.

70% of the participants had moderate adhesive disease with a mean AFS score of 8.2, and the remaining group had severe intrauterine adhesions with a mean score of 9.2. Now what did they find? The reduction in the AFS score was significantly higher in the group that received this film. The mean AFS score of 5.2 versus a mean AFS score of 4.2. The complete absence of intrauterine adhesions at the second look was significantly higher in the group that received the intervention. 41% of the patients had a completely normalized cavity if they received the leaf versus 24% who did not receive any intervention.

There was no association found between the use of hormones, the patient's age, the use of electrosurgery during the time of the procedure for lysis of adhesions. Really the only important association was the number of prior adhesive adhesive lysis procedures which was performed, which I think if we put our thinking cap on is probably just a proxy for severity of disease. So I said this before, any intervention in the reproductive surgery space is super exciting, and I loved seeing this trial being done in FNS, but when I also put my reproductive endocrinologist hat on, there's a couple of unanswered questions that I would have loved to have seen.

One, I would have loved to see what happened afterwards. I'm sure you normalized the cavity, but did you improve menstrual symptoms? Did you improve fertility outcomes? Did you even improve the endometrial thickness as a proxy measure? I think that's where a lot of these trials fall short is that it's really hard to follow these outcomes afterwards. You want to start the trial and finish it as quickly as possible.

The comparator group I think could have also been a little bit stronger, but I understand why they did it. Comparing it to no intervention is challenging, but there's not really a lot of convincing data that any of the existing standard of care interventions are better than no intervention. So I think there's equipoise here, and they did a good job using I think the best comparator group that they could.

But then I'll pose this final question to the group is, so the AFS score was one point lower on average with this intervention. What did it fix? Is that clinically meaningful? Does that really move the needle, or is this just a artificial proximal endpoint that we feel good about, but does the patient who then leaves the study and goes on with their reproductive life, did we make anything better for them? This is a little bit of what I struggle with these surgical trials. Full disclosure, I am a principal investigator on a multicenter randomized trial for a product that is also aiming to instill a gel inside the uterus for adhesion prevention.

So I have my biases and have the similar gripes in our trial, but I always wonder, what did we actually make better? I think that's a great question, Pietro. For me, clinically, I don't know that moving the score, the AFS score by one from three to four moves the needle on adding an intervention that I'm sure is going to, I don't know what it will cost because I don't think it's available yet, but it's not going to be free. So that's not a big needle mover.

But moving the absence of adhesions from 24% to 40%, that's 16 percentage points. That's one, that's a number needed to treat of seven. So every seven patients, one won't require additional surgery to remove adhesions.

That probably is a number that clinically moves the needle for me. So it's a great question. I think one of the answers is it didn't move the needle.

The other is that it kind of did if we could have this product available and if the price point is reasonable. I was going to comment on that same point from a slightly different angle. One of the things that my patients with intrauterine adhesions really struggle with is wrapping their heads around the idea that this is a diagnosis that often requires multiple surgeries.

They really see it as valuable time that's being lost when they have to go back over and again to get their uterine cavity optimized. So if we have, you know, an absolute difference of 16% of patients that don't have to do that, again, if it's cost effective, I think it's a worthwhile intervention. That said, I'm thrilled to hear that there are other products under investigation as well.

And certainly, you know, as a field, we'll do our best to pick the one that gives our patients the greatest benefit at the lowest cost. I think many people who do lysis of adhesions have kind of lived through this last 12 months where the cook catheter, the one that had that nice triangular shape that filled the cornea up nicely is no longer available on the market. So a lot of us kind of reverted back to just eight French pediatric Foley balloon, inert IUDs if you have them, depending where you are in the world, and oftentimes just serial hysteroscopy and kind of doing our best.

But I would say that almost all of those therapies are significantly less expensive than what I imagine a gel or a film will be when it does become commercially available. And you have to balance is that one extra point on the AFS score meaningful for the patient to spend an extra $300, $500, $600, $700 or not. And I think, unfortunately, our patients are willing to do whatever it takes if it makes the uterus a little bit better and speeds up the time of pregnancy.

So probably remains to be seen, but we've got to be careful. Just to geek out a little bit, Pietro, because I know you're super knowledgeable about these products. You mentioned that there's an enzymatic reaction that degrades the film once it's absorbed the fluid and has been in place for about a week.

What do you know about kind of what the enzymatic material is? And are there perhaps things like MMPs or other active constituents of the product that could potentially also help degrade adhesions? Good question. For this particular product, it's just a hydrolytic enzyme. So the gel absorbs water and then the matrix that absorbs the water just needs to get broken down.

For the product that we're working on, it's basically 98 percent polyethylene glycol or 98 percent water, 2 percent polyethylene glycol. So it just kind of degrades with time and falls out and you don't require an enzymatic reaction. One of the concerns that people have had with enzymatic reactions is, does it then recruit other inflammatory cells into the uterus? That could be problematic.

I think that's probably what you're getting at, Kate, but it doesn't appear that these hydrolytic only enzymes cause an issue. The really geeky thing, and I was wondering if you were going to bring this up, is are these gels, films, polymers, also drug delivery devices? We've talked about it for cavity distention, but could you introduce doxycycline into the cavity for treatment of resistant chronic endometritis? PRP, which if you believe the data at all, probably suffers from a short dwell time, is can you improve dwell time with some of these devices? And even people who are interested in using regenerative therapies like bone marrow derived stem cells, could you, instead of putting them in through the uterine artery, could you get them into the cavity and increase dwell time with them? So there's a bunch of other secondary applications that I think we're all very excited about. Whole new take on the first uterine pass.

All right. Thank you, Pietro, and thanks for doing double duty, and we appreciate your ability, as long as you can hang around to give comments on these other articles. Kate, we're jumping backwards to seminal contributions.

Interesting questions. Does aneuploidy and the likelihood of having a usable embryo differ with PGTA with PGTM versus just PGTA alone? Thanks so much, Micah. So this was a paper that was by Martel et al.

out of Cooper Surgical PGT Lab, and the authors evaluated 4,255 cycles that use PGTM and PGTA versus 68,267 cycles using PGTA only. And they compared as their dual primary outcomes, age stratified aneuploidy rate between the two groups, as well as age and inheritance pattern stratified cycle likelihood of obtaining at least one usable blast. So as many of us, I think, would have.

The authors a priori hypothesized that the group that was doing PGTM would have a lower age stratified rate of aneuploidy on account of this idea that probably a lot of them were not infertile. So NGS methodology was used for PGTA from November 2019 to March 2023, and PGTM was done by a linkage using SNPs. And from 2019 to 2022, this was done using microarray, whereas from June 2022 to March 2023, NGS was used for the PGTM.

So the authors of note did consider low mosaic embryos. So those that were noted to have a 20 to 39 percent aneuploid signal, they classified those as euploid and furthermore carrier embryos. So whether those were X-linked carriers or autosomal recessive mutation carriers were also considered to be usable for the purposes of their analysis.

So surprising to the authors and I think probably to most readers as well, they did not find any differences in the rate of aneuploidy in the two groups. So those that were doing PGTM and PGTA versus those that were doing PGTA only. They found that the PGTM group overall was younger with a higher proportion of their patients younger than 35 years old.

They also found that patients using PGTM actually in the younger age groups had fewer blasts to test, which was a surprising finding. So four blasts versus five in the PGTA only group. And that was statistically significant.

Unsurprisingly, though, they found a lower probability of having at least one usable blast in nearly all of the PGTM groups. So this makes sense because we figure first we find out about the PGTA status and then some of these embryos are not going to be usable because they would be expected to be affected by the monogenic disorder. And so the exceptions to this were that autosomal recessive mutations in patients age 38 to 40 and X-length recessive mutations in those age 35 to 37 exhibited no difference in the probability of usable blasts versus those not doing PGTM at the same age.

Furthermore, in all patients greater than 40 years old, they found no difference in the rate of usable blasts if they were doing PGTM. However, it's worth noting that the sample size was quite low in the PGTM groups in the patients who are 41 to 42 years old or older than 42 years old. So furthermore, patients with autosomal dominant mutations had the lowest probability of obtaining a usable blast.

Again, that tracks with what we would have expected. And among all the inheritance patterns assessed in all the groups, having more embryos to test was associated with a higher probability of having at least one usable blast. So again, no surprise there.

Lastly, patients using PGTM did have to do slightly but significantly more cycles to get to a usable blast. So not a surprise in that domain either. So this was a good study that said there were quite a number of limitations.

So the biggest one is definitely that there were no data on the infertility factors either the group using PGTM or the group not using PGTM. And similar to the first paper that, or actually both papers that Pietro reviewed, we don't know the clinical outcomes, including live birth, which is really kind of what we're most interested in being able to counsel our patients about. And so with regards to their most surprising finding, which was that aneuploidy was not lower in the PGTA group when stratified by age, the authors offered two potential explanations.

So the first was that maybe the infertility diagnosis represented in the PGTA-only group were in large part not those that would be expected to be associated with higher aneuploidy. So they don't know at all why these patients are doing IVF. Maybe some of them were doing it for sex selection.

Maybe some of them were doing it for embryo banking. Again, they're hypothesizing that perhaps a large proportion of those patients actually didn't have a higher rate of aneuploidy than the general population. And I think that that's quite possibly contributing.

The second explanation they offered that I'm a little bit more skeptical about, especially with such a heterogeneous group of disorders being tested with PGTM, is that the monogenic mutations may increase aneuploidy. And I would offer a third possible explanation that the authors don't propose, which is that perhaps a significant proportion of the PGTM group did have infertility factors associated with aneuploidy. Only 42% of the patients in the PGTM group were doing PGTM for autosomal dominant conditions.

And as we all know in clinical practice, many of our patients learn that they have risk for a genetic condition in their offspring when they do expanded carrier screening at the time of their infertility workup. So, similar to the fact that some of the patients in the PGTA group may not have had underlying egg factor predisposing to aneuploidy, many of the patients in the PGTM group may have had it. And so, the authors further conclude that patients undergoing PGTM should be counseled to anticipate similar aneuploidy and mosaicism rates to age-matched controls undergoing IVF with PGTA.

So, you know, I think that is a pretty bold conclusion. I find these data to be most helpful in corroborating the intuitive findings of prior studies, which suggest that patients selecting embryos using PGTM in addition to PGTA will need to do more cycles on average than those using PGTA alone. This is very helpful for counseling and managing patient expectations and minimizing dropout.

That said, I am less comfortable counseling patients doing PGTM without any underlying infertility factors that they should expect the same aneuploidy rates as infertile controls based on these data. I think just given the absence of the data on the IVF indications in both groups, I think that we have to still individualize our counseling and acknowledge what's unknown. So, great study, great counseling tool.

As always, not a perfect study. And so, questions for the three of you gentlemen. Do you think that monogenic mutations by and large cause aneuploidy or have the potential to cause aneuploidy as selected by these authors, you know, as a conglomerate group? And also, you know, how do you think you would change counseling of your PGTM patients based on these data, if at all? Those are good questions, Kate.

I would say to answer your first question, probably not. I can't imagine any scenario, but I think we're lumping all monogenic diseases together. Are there some monogenic diseases with point mutations, frameshift mutations that may affect chromosomal instability? I don't know.

But I think it's very unlikely that as a whole, the monogenic disease is going to affect aneuploidy. I think there was two missed opportunities in the study that I understand why they couldn't do. But one is, what actually happens to the implantation potential of these embryos? Are embryos derived from infertile patients versus non-infertile patients going to implant at different rates despite being euploid? I would have loved to have seen that, but I get the limitations of being able to collect that data.

And then there's that other big question, which I know people have kind of puttered around every so often in the pendulum swings, is does ovarian reserve affect rates of aneuploidy? I feel like that would have been also an opportunity for them to look at ovarian reserve markers like AMH and AFC to be able to also continue to add some data to that question is diminished ovarian reserve driving some of this differential outcome? Yeah, that's certainly a super clinically relevant question. And yeah, the main limitation here is just they have no clinical data to speak of whatsoever, except whether they did PGTM and what for and their age. I'm not sure the question that the rate of aneuploidy drives infertility or infertility drives the rate of aneuploidy.

Do you really think that the rate of aneuploidy in a woman's embryos is what's bringing her to a fertility clinic? That's the cause of infertility? I certainly don't think that's the cause of her fertility for all comers. I would certainly think that there are a subset of patients who have experienced sub-fertility due to poor egg quality, likely even poorer egg quality than would have been expected based on her age, and that the ability of IVF to basically access more reproductive potential per unit time is a large part of what makes it effective. But I agree with you if the population is diluted by multiple different kinds of infertility diagnosis, along with non-infertile patients, certainly not all infertility is caused by aneuploidy.

What do you think? My hypothesis would have been that the aneuploidy rate was the same, that I think aneuploidy rate is independent of most causes of infertility. That's just what I believe. I don't know whether I'm right or wrong.

So in a sense, this finding didn't surprise me. I thought it was confirmatory of what I thought. So I'm curious that there really are two different theories out there, and that intrigues me, that what really is the truth? Well, I mean, this paper would, if we can draw conclusions from it, indicate that you're correct.

So it's useful data. Micah, what do you think? My hypothesis and how I counsel patients has always been along the same lines as what Kurt just said, that aneuploidy is purely related to age. It's unrelated to other things that are individual to that patient in front of us.

And with anything that's a roll of the dice by chance, the older you get, the more likely you are to have that bad outcome multiple times in a row. So in that way, it could be associated with infertility. But I agree with Kurt.

I tend to think of it as a separate mechanism linked only to age. And so this study confirmed my biases, I guess, or my internal hypotheses that I had. Gotcha.

That's super interesting. Yeah. The authors hypothesized that they thought it would be different and were surprised by the finding.

But it's interesting to hear that our a priori assumptions are even different there. Kate, that last paper was interesting because it's a lot of data based on pragmatic information that allows us to get to some of these questions. And I have another paper like that, that I think is worth discussing, which is anti-malarian hormone levels are associated with time to pregnancy in a cohort of 3,150 women with the first author, Scott Nelson, and the senior author, Sharon Briggs.

This is also a really important, big question. And we're answering it in a different way. Where the studies are analogous is we're answering it with big data, so to speak, like pragmatic data, not a hypothesized study where you recruit patients for it.

In this particular study, we're looking at a large group of people that enrolled in basically a commercial group in San Francisco, the Modern Fertility Company, which tracks women's fertility with an app. And they happen to also link AMH levels within three months. So let me back up a little bit.

This is another big question. Does AMH predict fertility or just the number of eggs? And as we all say in many of our lectures, including ACOG, that a single AMH level is really not predicting a woman's innate fertility based on age. What it really is better predicting is the number of eggs we might get with ovarian stimulation or IVF.

Now, I paraphrase a little bit, but that's the going rate. And it's been based on lots of studies, including Ann Steiner's very seminal study, where she followed people that she recruited and followed them in a cohort prospectively in time to pregnancy and found out that there was no association with baseline AMH level at a time to pregnancy. Now, since that seminal study, there's been a lot of controversy in a lot of studies, including some meta-analyses that show that maybe there is an association with AMH and fertility, even though it's modest.

And in Dr. Steiner's own data, she said that maybe it's associated with miscarriage, but not time to pregnancy. So it does beg this whole question. And that's why I'm intrigued by this study that said, we have a ton of data available to us because of current practice that we can analyze to look at the question.

And that's what they did in this paper. They basically took women that signed up for this app. I don't know if they're fertile or infertile.

All they supposedly were doing was tracking their time to pregnancy, and they got an AMH, and they basically did a time to pregnancy analysis on this group. They used the same cutoffs. Well, they used different cutoffs, their own cutoffs, like an AMH of one is lower than normal.

And they also used Dr. Steiner's cutoff, which they used 0.7. And they looked at age stratification, and they found, to their surprise, I guess, that they think that AMH does predict time to pregnancy. A modest effect, albeit around odds ratios are 1.1, I'm sorry, 1.17, things like that. But it definitely intrigued me.

So the big controversy of this is not necessarily the finding. It's, is this the right population to look at? And is the finding basically an artifact of the population? Or is the finding true? And that's why I was contrasting it to what you found in that study with PGTA and PGTL. So I can certainly go into more details on the study, but I'm curious on what you guys thought of this idea of using, again, large amounts of data that's available to us, but just isn't as clean as a real good research study where you know everything about the patients going in and you're following them for amount of time.

So one advantage is size versus, I guess, precision. And this one is basically saying, so I think the stats were well done with the data they have. I don't think they made a mistake in the way they're presenting the data.

I think the question is, is this, does this population, the population we want to know the answer from? So I'm curious on what you guys think of that. And as we talk about it, I can certainly give you more details about the finding. Yeah.

I mean, again, a lot of data, not a lot of granular data, and I would define it as kind of a, in some ways, a worried well population. And so, you know, I really struggle with this question and this is one of the ones that I, if somebody, you know, held a gun to my head and said, make a bet, is AMH associated, you know, with time to pregnancy? I don't know. There are compelling studies in both directions.

And no, I don't think that this is necessarily the right population to check in. I think that if they had more clinical data for us to tease out kind of what was actually going on with, with these patients from a population perspective, in terms of some of their other characteristics, that would have been more helpful. But I do think, I think it adds, I mean, we've had studies that show, yes, it predicts in patients older than 35, but not younger than 35, and vice versa.

And I still think that Anne's studies are the best ones out there. As much as this is a lot of data, it's not, it's not the same level of quality in terms of our ability to extrapolate. Yeah.

The question remains, well, the study gives us a lot of information, some of which, which confirms what I thought I believed, and some of which challenges what I thought I believed. I mean, it was interesting just to rattle off some other findings that I think that are true and important is that the majority of pregnancies in this group of women happened in the first four cycles. And that, so it seems to be that the worried well, or people thinking about getting pregnant, those that are really normal, do relatively quickly.

So that was one finding. It was also interesting was that chlamydia was mildly associated with, with a lower chance of getting pregnant, as was the diagnosis of PCOS, even though theoretically these women had all regular cycles. So there are some things in here that ring is true, but I don't know whether to dismiss the finding about AMH.

So let's, let's, for the listeners, let's contrast the study again. So Anne Steiner's study basically found people trying to get pregnant, enrolled them prospectively, followed them very carefully, very, very carefully understood who they were, when they were trying, when their AMH level was, and found no association. The criticism of Dr. Steiner's study was just the size of it.

It's just really hard to do that kind of study and recruit that many people. And some people suggested that even though the findings were close to null, you know, if you find it in a larger group of people, maybe there's a subtle effect there. The criticism of this study is we really don't know who went in the study.

Not everybody followed for a year and there's a lot of noise. And, you know, when people took the, when people followed, who followed the app, you know, is there a bias in who followed the app? Is the person that they're more infertile than the people that didn't follow the app and therefore weren't included in the study? So you could get to this modest increase with lots of differences in the population, but it does, again, which side of the fence are you on? Does it ring true that your hypothesis is that AMH really is subtly associated with fertility, or does it ring true that I can't figure out why AMH should be associated with fertility because all, and one of the explanations the authors gave was AMH might be associated with increased aneuploidy, which again, I'm not sure is true, but it might be associated with health of the granulosa cells around the egg and the lower egg quality. And maybe that is something to do with subtle fertility.

So good papers to think about, but what do you guys think? The other thing I just wanted to flag is that the absolute difference in AMH between the two groups, the pregnant and not pregnant group is really quite small and you might argue not really clinically significant. So, you know, that's another thing with such a finicky assay and the wide confidence intervals, you know, I'm not so sure that, that, you know, the clinical significance is there for us to draw conclusions. Yeah.

And just to be clear, since you don't have the paper in front of you, when you look at the group that got pregnant and the group that didn't get pregnant, the AMH levels are 2.9 versus 2.5. And you might argue that, how do you tell that difference? But the next statistical approach they did was saying, if I stratify people by those that have less than one compared to those that have one to five, or in one case, one to eight, there is about a 10, 11% lower chance of getting pregnant within a year than those that have a normal AMH. And by the way, high was the same as normal. So there wasn't, they didn't find that there was a too high level, which was, I don't know, again, intriguing.

So the question is not, what is your AMH? Can I look at it and say 2.5 is different than 2.8? But if your AMH is 0.7 or is less than 0.1, do you have a lower chance to get pregnant than someone with a normal AMH? Such a good discussion. I loved how you and Kate framed it. A lot to think about there.

And I'll just end this with Dr. Steiner, and you should read this, did write a very good reflection about this paper. And the bottom line is, and I agree with her, that this paper shouldn't change our recommendations, that there are too many questions about this. It's intriguing that we found this finding.

It makes us talk about it, but it doesn't mean we should now tell people that a low AMH is affecting your fertility. I don't think we're there yet. The evidence isn't that strong.

Well, especially because of how great the consequences of telling someone that are. These are the worried well, right? But for them to then start believing that they're the worried unwell is not something that we want to be imposing unnecessarily anyway. Well said, Kate.

Let's stick with you on this SART study, looking at racial disparities and outcomes. And so this was a study by Brioso et al, which was a collaboration of investigators at Yale and NYU. They looked at 79,000 plus PGTA-tested transfer outcomes from 2016 to 2018 from the SARTCORES database, evaluated by voluntarily self-reported maternal race and ethnicity.

And they adjusted for age at the time of transfer, BMI, AMH, endometrial thickness, smoking history, and infertility diagnosis. And specifically, they adjusted for uterine and tubal factor grouped together versus any other diagnosis. Further, they use generalized estimating equations to account for multiple transfers in the same patient.

So their findings were such that on adjusted analysis, Black and Hispanic women were more likely to experience preterm birth and cesarean section. And Black women were more likely to experience miscarriage compared with non-Hispanic whites. This was not the case on the adjusted analysis.

All other races and ethnicities were less likely to experience term birth, both per transfer and per clinical pregnancy relative to whites. However, differences in live birth overall were not statistically significant on adjusted analysis. So just to break that down one more time, Black and Hispanic women had more preterm birth and c-section on the adjusted analysis.

And Black women had more miscarriage than whites on the adjusted, but not on the unadjusted, but not on the adjusted analysis. Of note, uterine and tubal factor diagnoses were twice as prevalent in Black women versus all others. Further, Blacks comprised only 2.7% of study population versus 12.6% in the population at large.

So dramatically underrepresented. And Hispanics accounted for only 3.4% versus 19.1% in the general population. So even more underrepresented.

So the authors offered as an explanation that the large proportion of patients with race or ethnicity unknown might explain this difference. So 36% of the entire population had an unknown or unreported race and ethnicity. And so, you know, the elephant in the room is that the more concerning explanation for why Blacks and Hispanics were so underrepresented in this population revolves around limited access to ART and or to PGT in these populations.

Since again, these were all PGT-tested embryos. So this is a very important study on a very important topic. It also, though, has some important limitations.

So the authors presume that these were euploid transfers. So actually, they're PGTA-tested, but SARTCORS doesn't contain the data for this data set as to what the results of the PGTA actually are. We only know that they were PGTA-tested and were transferred.

And so, some of these embryos that were transferred were likely tested mosaic or unknown or, you know, no result. And so, the other limitation that really hit home for me was that there were 36%. That's a big percent of the overall sample that exhibited unknown race.

And the missingness was substantial for many of the other covariates that they considered. So the authors accounted for this only by considering the entire group of unknown race as a separate group, rather than conducting any kind of sensitivity analysis. And only cycles with complete data for all the covariates were included in the adjusted analysis, which given the high degree of missingness may have significantly impacted their results.

So again, overall, very high-quality study highlighting inequities in access and outcomes from ART associated with self-reported race and ethnicity. As nearly all of these studies unfortunately end, the conclusion is that more studies are needed to further elucidate the underlying mechanisms of these lesser outcomes and strategies to address them are definitely needed. I also would recommend that readers look at the reflections from my friend and colleague, Jen Kawwass and second-year fellow, Daniela Diego from Emory.

They put forth a very balanced assessment of what is known and what remains unknown and what future research is needed regarding potential causative factors for these poor outcomes in minority groups. And they focus a lot on cultural factors and also underlying medical comorbidities, which is not something that this SART dataset could assess. One other thing I just wanted to mention is that a more optimistic way of kind of evaluating these data, other than, you know, just that on the unadjusted analysis, specifically the, or on the adjusted analysis rather, specifically the black population did much worse, would be to kind of further tease out the issue of the diagnosis.

So, as I mentioned, the authors grouped uterine and tubal versus all other diagnoses. And these diagnoses, both uterine and tubal factor were disproportionately represented among black study subjects. And so we would expect that uterine factor would likely decrease implantation and increase miscarriage and preterm birth, whereas tubal factor may even be a positive prognosticator when we're at the point of a euploid embryo transfer.

So I would have liked to see the model absent diagnosis as a covariate and also kind of considering all of the diagnoses separately. I think that might have helped us get a better sense of to what extent these diagnoses, which are more common among the black population, were causative of their worst outcomes. And I'm interested to hear what Curt and Micah think as well.

I think you summarized it really well, Kate. And I think it confirms what we've learned from other studies on similar topics addressing these questions. I agree with you when you're highlighting limitations of SART data.

The missingness of race ethnicity is a huge issue in these data sets and one that we recognize in SART and have tried to address. But you're always trying to balance the burden of the clinics and the entering of data with the data that we get so that we can use for research. And that's a hard thing to balance.

And as you said, we have not tracked currently aneuploid versus euploid versus mosaic. So we only track usable embryos in different clinics to find usable differently and maybe transferring aneuploid or mosaic. So could that have an effect that's going to change going forward? But I think these studies are helpful from a surveillance standpoint to highlight potential disparities so that we can try to drill down on what the underlying cause is so that we can ultimately come up with solutions.

And I think you hit on a lot of those. There may be differences in underlying diagnosis, and some of this is probably an access to care issue, which is an even bigger question. I just want to add the difficulty in the SART database, again, is this is not perfect data.

It's observational data. It's not research quality data. So we really shouldn't be looking at the unadjusted analyses.

That unadjusted is because there is confounding in there. There absolutely is confounding in the SART database. So it's how you control for that that can really make a difference and see how you can get closer to what the actual truth is.

And I wanted to say that for all analyses. The unadjusted analysis is, or the crude analysis, which is called, is crude. And what the really good methodology is, is how you're going to control for that.

And the biggest thing to control for, and you can control for it, is what's called confounding by indication. We treat different people differently because of our clinical acumen, and that's in the SART database. So you've got to find a way to control for things that might be a surrogate for that.

Sorry to get geeky on you here, but this is why I'm very cautious about the findings of the SART database without really sophisticated analysis. Right. And I mean, again, I think that their modeling may have been imperfect, and it's hard to know exactly how to make heads or tails, again, with data with so much missingness in it.

But just to emphasize, yes, there were differences in live birth in the unadjusted analysis, but there were not in the adjusted analysis. So in a way, that's what I mean by these data can be taken perhaps as reassuring. And if we were to look at each diagnosis separately, we may have gotten as controlled in the modeling, I should say, we might have gotten a better sense of the contribution of each of those factors.

I get a lot of papers submitted before they're reviewed and cleaned up, where they're trying to push the unadjusted analysis as the main finding. And that's what I'm trying to make a comment about here, that the unadjusted finding is not probably the correct finding. Yeah.

Kurt, in the interest of time, I'm going to skip mine and go to yours next. Just let me make sure we get it in. Just because I've seen a lot of articles on Mendelian randomization.

I've been asked to review some. It's a challenging topic. Tell us about what these authors were looking at and teach us what Mendelian randomization even means, and what as readers should we be looking out for as we try to interpret these articles? I had a feeling you'd ask me a question like that.

So I think you picked this article more so for an explanation of what Mendelian randomization is rather than the finding. The paper we're talking about is, there's no association between celiac disease and female infertility, evidence from a Mendelian randomization. And the bottom line is that celiac disease doesn't seem to be associated with infertility.

But the question is, why would you do a Mendelian randomization and do you believe it? And let me see if I can try to say this simply, because it is a little bit complicated because the vocabulary is just not something we're used to. So whenever you're looking at an exposure and an outcome, you're trying to find if that association is true. So if you're looking at someone with celiac disease and infertility, you've got to find people with celiac disease and find infertility.

But there is a lot of difficulty in saying, is that association truly causal or could it be through something else, like behavioral? People with celiac disease have different diets. Maybe they find partners later. Maybe their sexual frequency is different.

Maybe it affects other things. So you really don't know the causative nature of that association if you did it the traditional way. So Mendelian randomization tries to get at the causal effect by saying, I'm going to find something that isn't confounded.

And theoretically, you're using what's called an instrument. So a proxy, if you will, a proxy of a celiac disease, but one that's not associated with confounding. And one proxy could be genetic polymorphisms or genetic aspects.

So the first thing is, can I find something that is really only associated with celiac disease that will never change? And the theory is, I'm going to make my exposure not the disease itself, but polymorphisms that I can detect on someone's genetic analysis and say, that's not going to change with behavior or confounding. So I'll use that as my exposure. So the first step in any Mendelian randomization should be, do you have such an instrument, such a proxy? So the first step in this paper was to use GWAS analysis.

And there was a number of GWAS analysis on this topic. And you can even use a meta-analysis of GWAS analysis to basically say, do I have that proxy? Now this is where it gets tough. They say that the explanation for the association with disease is it explains these polymorphisms can explain 45% of the disease.

That sounds low to me, but from a geneticist, that's exceedingly high. So they say therefore they have a very good proxy. So that's the first step.

The next step is, all right, now that I've got my exposure, which is this polymorphism rather than the disease itself, now I can say, what's my outcome? And the outcome is in this case, they used ICD-9 coding from large biobank studies. They used the Finnish biobank study and also the British biobank study. So there's a lot of people.

So you need a large sample size. So now we could argue whether ICD-9 codes of fertility is the best proxy of infertility, but I think that one is pretty good. And it has the positive nature of the study of having large numbers.

They're talking about thousands, tens of thousands of people, not just a couple of hundred. So you've got the power to make this association. So theoretically, now you have a way to say exposure to outcome without confounding.

And that's what Mendelian randomization is. And we can argue whether that's truly randomizing them, but it theoretically is taking away the confounding. The next step is to find out the analysis.

And it's a traditional analysis, basically association to outcome. But I'll go through a little bit of the vocabulary so people use it. So the analysis they use is something called inverse variance weighting.

It basically is like logistic regression. Is there an association between the exposure to outcome? But because they really don't have one polymorphism that is associated with celiac disease, they have a panel, 40 or 50. They have to say each one has an association.

I need to weight that and make a composite number. And they make the composite number by weighting the association that has the least variances the strongest and the one that has the most variances the weakest. So I didn't explain the math.

I just explained the theory. And that's what the inverse variance weighting is. And then now they have to check two or three very important assumptions to make sure that this makes sense.

And one of them is called pleiotropy. So is it possible that your instrument that you thought was not affected by confounding actually does work through different pathways that might actually be confounded with infertility? And they do something called an MR-eager test for that, which is basically saying, is there an independent pathway that I couldn't see? And then the other one they want to test the assumption is, is it possible that it's working through a different pathway of exposure, not what I just thought with celiac disease? And they analyze it with something called an MR presser analysis. Now, again, I'm not going to explain the math, but it's basically trying to say, is it really working in the direct line that I thought, exposure to disease, or are there other pathways because genetic diseases, they're obviously complex and then you can't be sure.

So that's basically what they did. They basically said, I have, my genetic variant is very good because I looked at tens of thousands of people to find the instrument that is associated with celiac disease. It had low linkage disequilibrium.

It wasn't a weak instrument. I mean, there was a strong association. Then they did their analysis that basically say, I controlled for all the variants in my instrument and I weighted them.

Then I made sure there wasn't other reasons that I could get to this association with these statistical tests called MR Eager and MR Presser. And they did some sensitivity analysis. And at the end of the day, they said, we think we find no association any way we looked at.

So this is probably the easiest Mendelian randomization study to understand because everything came out the same way and ended up with no association. So you're more likely to believe it. Not every Mendelian randomization is as easy to understand because you get different answers based on which test you have or how off of the sensitivity analysis or how strong is the association.

So again, in one sentence, the idea is to take the confounding out of the association by finding something that is started at birth and will not be affected by your behavior or your smoking or anything like that. And therefore you're getting closer to a true causal effect. The proof is in the pudding, which is why many of these studies are not accepted and are controversial is that to make sure all those assumptions are true is difficult.

And you get really lost in the weeds about what these numbers mean. And you lose the idea that I'm not really sure that that instrument or that proxy chose is really the right one. And I'm not really sure that it didn't have other factors associated with it that you're not telling me about.

So therefore, how can I believe that the associate you're telling me is really causal? So that's why we put this in so we can start understanding it. It's difficult to interpret and very difficult to review, but this one was clean enough that I thought it was good for our readership and us to discuss. And let's see if I did it.

Did you guys understand that explanation or what questions do you have? That was so good. It was very helpful in terms of being able to understand as a reviewer as well, how to interpret these and what is a reasonable paper to draw conclusions from and accept. And it seems as though the less complex and the fewer the assumptions are, at least the more congruent the findings, the more we can draw from them.

But I mean, it's even the idea of assigning the appropriate weight alone seems like it's extremely fraught. So thank you for explaining it, Kurt. Yeah.

And these basically have to be done with big data and big data gets us all a little bit nervous because we're not used to dealing with all the statistical assumptions that you have to put in when you're looking at big data to see if it's quote unquote clean or not. But if you really want to geek out, you can learn some of what these statistical tests do. But remember, a Mendelian randomization study is like anything else.

It's just trying to find an association between your exposure and your outcome. We're just moving the exposure to something that we think is not affected by confounding. That's a fantastic summary, Kurt.

Obviously, we can't randomize patients to having celiac disease or not. So we'll never have an RCT. So we're always going to be dealing with observational data.

So my understanding as you're explaining it, Mendelian randomization is trying to adjust for that confounding before we even do the analysis. And the paper I'm not going to present because we don't have time, use propensity scoring and inverse variance weighting to do the same thing before. Other papers we reviewed today use logistic regression or GE models after the fact to try to adjust for the confounding.

But all of these are just different approaches to how we adjust for confoundings that might be affecting the exposure and the outcome. Right. So there is a misnomer in Mendelian randomization.

You're really not randomizing anybody. And it really is an association study. And they call it causal because theoretically, you've removed the problem of all observational studies, which is measured or unmeasured confounding.

The question is when we all read it is, is it really a true assumption that we did that? In other words, the assumptions that they made or the instrument that they chose, is it really as good as they say? But isn't that going back to what we all said? All studies, is it really as good as they say? So that's why we review these studies. And that's why we have these podcasts. Yeah, fantastic.

Wonderful discussion today. As always, I learned a lot. I want to give my shout out to who Kate shouted out on the Reflections, Jen Kawwass.

She and I were emailing this morning. And she said she listened to us this morning on her run and felt like that she already hung out with us today while she listened to the September podcast of FNS OnAir. So shout out to Jen Kawwass.

And thank you for the great work you do and to all of our listeners. As always, we're just talking about some of the articles in FNS. Please log in, read them all.

There's some amazing content there. And we look forward to seeing you in January or talking to you in January 2025 in the new year as FNS rolls into 2025. Kate, Kurt, Pietro is not here.

Good seeing you. Good to see you, Mike. Thank you.

This was fun. I hope that this is both entertaining and educational. So keep the comments coming in and we'll do our best to make the podcast even better.

This concludes our episode of Fertility and Sterility OnAir brought to you by Fertility and Sterility in conjunction with the American Society for Reproductive Medicine. This podcast is produced by Dr. Molly Kornfield and Dr. Adriana Wong. This podcast was developed by Fertility and Sterility and the American Society for Reproductive Medicine as an educational resource in service to its members and other practicing clinicians.

While the podcast reflects the views of the authors and the hosts, it is not intended to be the only approved standard of practice or to direct an exclusive course of treatment. The opinions expressed are those of the discussants and do not reflect Fertility and Sterility or the American Society for Reproductive Medicine.