Presentation Transcript

Evaluation of Embryonic Reproductive Competence
presented by Richard T. Scott, MD, HCLD
 
View Dr. Scott's  presentation 
View the video introduction to Dr. Scott's presentation in Windows Media or  RealMedia
Read the written transcript of Dr. Scott's video introduction
Read Dr. Scott's Curriculum Vitae
access the printable version of this page

If you look at reproductive competence, certain fundamental tenets that we all have when we are trying to take care of our patients really have to be questioned.

Because we know, quite frankly, that two cells on day three probably do not make babies very often and we know that embryos that fail to blastoate by day five do not do quite as well. But across the board, the overall driving impact on implantation rates relates principally to ovarian reserve and the age of the patient. Interestingly enough, the number of cells, this original work by Carol Brenner has been confirmed in much larger populations, cell cleavage rates do not change with age. Say you would like to have six- to eight-celled embryos depending on your culture system across the board and you would like to have those present on day three and you have some in your lab, is that more common with 25-year-olds or 42-year-olds and the answer is, 'it is the same’. While we like fast embryos, certainly reproductive competence is not going to be related very much to cleavage rates, not across the board.

Similarly, high fragmentation rates are bad, but again we look at the fundamental reproductive competence of embryos, specifically that portion which relates to reproductive synesis. This work by Alikani in a very nice paper that she did, showed that, in fact, fragmentation rate declines with age. If you have these low fragmentation embryos and you have one thousand of them and you break them into groups and say, who is the prettiest? The 28-year-olds or the 45-year-olds? The 45-year-olds will cleave as fast and have lower fragmentation rates and by many of the very fundamental scoring systems, you will have higher embryo scores in older patients than in younger patients, obviously not related to their ultimate outcomes. The fundamental things we look at in laboratories to distinguish embryos within a cohort may not necessarily provide us with a lot of insight into the true reproductive potential of those embryos. That results in this phenomenon.

These are the transfer number data from the most recent CDC report. As you can see, 2 and 3 embryo transfers have gained enormous momentum in our country and we are, I do believe as a specialty, we are practicing better medicine and putting fewer back. Having said that, 7 embryo transfers 1% of the time, 6 embryo transfers 3% of the time, 5 embryo transfers 7%. We are still transferring large numbers of embryos. Why? Because we cannot tell which ones are reproductively competent. 

This particular data set, which is based on mature follicles on the day of ACG, 14mm or more, are data from our own centre and show that for every 100 follicles, at the end of the day, if you look at mature oocytes, 2 PNs, cleavage, etc., you are going to get between 7 and 8 babies out of every 100 follicles in the fresh cycle.

We have an extraordinarily inefficient process, even when pregnancy rates are well over 50%. Part of that inefficiency means that we have a hard time, if you look at all the embryos that are 6 to 8 cells here, all the expanded blasts here, still we have a hard time making the leap from here to here in talking about delivery rates.

Ultimately, we would rather have pretty embryos than ugly embryos, but when you have pretty embryos, how do you know which one makes the baby?

We know there are lots of reasons why morphology is not very helpful. This landmark paper from the University of Washington group by David Battaglia shows that if you look at the way spindle dynamics go on, and this has been known and demonstrated in many ways in many studies, but this is just such an elegant study, this is from natural cycles in fertile women who had every possible endocrine parameter you can imagine measured and demonstrated as normal. In this particular study, they showed that in the youngsters, 10 of the 12 eggs had normal spindle architecture at the time of retrieval and the older patients, again these people were not anomalous, these people were not infertile, these people did not have some horrible history; in fact, they were probably ultra-selected for normality, look at the chromosomes the way they assemble on the spindle – completely abnormal. In fact, 3 of 14 were normal, indicating this dramatic change. We have known that that relates to aneuploidy.

These are some data involving many hundreds of embryos from our own laboratory, saying if you look at 9 chromosomes, and obviously you can look at more now, but even just looking fundamentally at 9, there is an enormous trend of aneuploidy but even look at the younger couples out here – the prevalence of aneuploid embryos is very, very high, even from great youth. So those are things that are not reflected by morphology. Those are things that are not reflected by multi-nucleation or anything else, because were the embryos that were actually of such sufficient quality to biopsy and these are not discarded embryos. These are embryos that were available for potential transfer. A very high prevalence of aneuploidy across the board.

You would think that PGD, and perhaps it is, something we can do that might allow us to look at improving our implantation rates. In this very elegant summary from Santiago Munné, whom many of you know has been a leader both in doing these procedures and thinking about them in his field, it is shown that as age goes up, the proportion of normal embryos goes down. That is very similar to that which we found in our experience, as well. If you then use those data to select embryos for transfer compared to just random selection based on morphologic criteria, the improvement on a percentage basis of pregnancy rate that you would expect in the various age groups is demonstrated by the yellow bars over here on the right-hand graph. So in youngsters, you would expect a 50% improvement in implantation rates. In people 40 to 41, it should go up 120%. In people 42 to 44, it should go up almost 160%. Yet, even in perhaps the most optimized and maybe even optimistic, because there are always issues in how you select controls, so I will say the most optimistic possible review of the data, the improvement in implantation rates does not even begin to achieve that which we might have expected based on what we could have gained from selecting non-aneuploid embryos. So the improvements have been very modest compared to the potential.

Some of that may be because biopsy is harmful. Is it or is it not? There is just enormous debate about this. If you look at the individuals undergoing biopsy for single gene defects, you will find that have implantation rates which are comparable to their age-controlled and infertile peers. Again, is the glass half-full or half-empty? You can interpret those data, though some people will interpret those data to say, look, they have the same implantation rates. Clearly, these people are doing just as well, there is no adverse impact at biopsy. Others would say these are couples with no improvement in infertility, you can see readily on their own. Maybe their implantation rates ought to be a little higher. So I think just looking at those implantation data provides a basis for a debate, but not an opportunity for an answer.

If you look at embryo biopsy and how benign it is, I have to tell you that this is an extremely important issue. If it becomes more important as we start learning in the last couple years, and I am sure many of you have heard of these data, embryo polarity may start occurring much earlier than we had originally thought. 

It was traditionally felt that the cells remain completely undifferentiated totipotential at least to the compaction stage, so say in the 6 to 8 cells, even a little later, and that genomic activation in many of the modifications did not really get into full swing perhaps even to the blastocyst stage. We now know there are some data suggesting that polarization may occur by the 4 cell stage, so they do not always retain their total potential and there may be polar cells, there may be some totipotential cells, but there may be some polar cells even by the time we are doing these biopsies. Maybe the impact of biopsy depends on which blastomere you ultimately end up biopsying and, of course, you cannot detect these things morphologically. There may be a luck-of-the-draw aspect. These are important theoretical considerations and I do not think we have the answers. I will come back to how we might answer them in a moment.

If you look at PGD, the other piece in this is to remember the limits of PGD. PGD has a finite error rate and in different papers published by Santiago Munné, the error rate depends somewhat on probes, but certainly increases as you study more and more probes. Florescence in situ hybridization is a subjective technology. For those of you who have a FISH lab or have spent time in someone else’s FISH lab, there is interpretation to those signals and there are some biases there. Of course, the cytogeneticists in our field have this great opportunity to always sort of forgive their errors because if they take two cells from an embryo and get a different result, could that be a laboratory error? Of course not. That is mosaicism. We all should have a such a handy way of explaining our differences in our outcomes. Of course, sometimes it is mosaicism, but how do you distinguish those two things? It is really not possible.

If you look at outcomes in prospective randomized studies, what you are going to find is that benefits are modest. In this particular study we are doing, we are using a very idealized population, 32 to 38 years of age, doing down-regulated stimulations with FSH, what we are able to show, of course these are very idealized patients but still they should have a prevalence of aneuploidy in the 40% to 50% range or higher, the on-going pregnancy rates, with biopsy and then culturing the blast, is not better than that with just culturing the blastocyst alone. We have done approaching 100 cycles now, 100 cycles starts in this group. We need about another 50 to get the study completed. Across the board, our data is very consistent with the data we did when we did the study with Santiago Munné. He has yet to show a definitive benefit to biopsy in a cross-sectional general IVF population.

Is that because we test too few chromosomes? Well, we certainly have a lot of chromosomes here that we do test for: you know, 13, 16, 18, 21, 22, X, Y, etc., but there are still several that we do not test for and their prevalence of aneuploidy when you look for them in oocytes and embryos is just as high. These embryos, when they are abnormal at the plant, at least they plant very rarely, so we tend not to look at them when we are doing PGD, but that does not mean that they are not abnormal and that they cannot potentially impact outcome. Certainly, when we tell patients they are doing PGD, we can say we can reduce the prevalence of abnormalities that can cause them to get pregnant and miscarry or become pregnant and have an anomalies child. We certainly cannot exclude the prevalence of aneuploidy in their embryos.

Why not just do more? This particular image, which was given to me by Bill Kerns and the group at Shady Grove, Bill is an outstanding investigator in this field, shows you why when you get too many signals these images blur and it becomes impossible to read. You can do hybridization after hybridization after hybridization; but at some point you do get degradation of the DNA and you get greater interpretation issues and the signaling making the technique less precise.

Spectral karyotyping, and there are at least three different ways of doing this now, is offered as a way of potentially painting chromosomes and looking at the entire chromosomal complement; however, that does not work for us clinically very well because it requires metaphases to be interpretable and it is also a relatively slow process. We have known since the mid-‘90s that it is possible to take blastomeres, fuse them with immature oocytes from animals, have them progress through metaphase and then be valuable. But, unfortunately, that is not practical and certainly not time-practical in terms of getting an interpretable result that you could have to look at your results and make transfer decisions.

Comparative genomic hybridization, a big hit at this meeting four or five years ago, has not found its way into mainstream practice yet. It is not that CGH is a bad technique, it is an excellent technique, but it is a relatively slow technique requiring great expertise and again there are some interpretation issues. There are groups in Australia who have got this down to where they can reliably do it in as little as two days time, but that is really pushing the envelope and whether CGH in the classic sense of metaphases and spreads and looking at competitive binding will ever find its way into routine clinical practice is, in my opinion, highly dubious.

What about single genes? Again, we have issues in our clinical practices that we have to think about because there are significant error rates even in single-gene defects, issues such as allele drop-out and other ways we find imprecisions when doing amplifications create problems and I think we mentioned this last year but, across the board, if you look at the percentage of unaffected cells correctly diagnosed for recessive disorders, it is about two-thirds if you biopsy one cell, it is about 85% if you biopsy two cells. For dominant, it is lower. While the error rate if you say something is normal, it is very likely to be normal. The precision for that is very high – you are very unlikely to have an abnormal offspring based on that; it is not zero, but it is low. On the other hand, if you say something is abnormal or you do not get a result, you really do not know if it is abnormal or not. The imprecision is very high. Again, a single cell for a dominant disorder may approach 50-50. So these techniques also need to mature beyond those which are available contemporary with just typical multiplex PCR amplifications. The future of this has to also change as we move forward to start thinking about chips and other things.

How are we going to sort this mess out? Well, one way it to think about finger-printing embryos. In this particular study which is, in my mind, a landmark study, they looked at 21 blastomeres from 9 embryos. They were able to get amplification successfully 16 times, but that efficiency is improving. This is something we are working on in our lab, as well, and I can tell you that that is definitely improving. It was informative that 13 of those 16 had about a 10% drop-out rate where they were unable to get signals, but then they were able to marry back those embryos to say which ones came from where.

This represents an extraordinary opportunity in our field because, for the first time, we are going to be able to look at 2 embryos from one cycle and one transfer onto one endometrium and so tell which embryo made the baby. It may be possible have different culture conditions in the lab and put one egg here, one egg there, make embryos, put them both back, look at which one implanted and say, okay, because it is completely paired analysis, completely internally controlled, this culture system is better or lesser in this system. This particular approach to almost anything, this morphologic parameter of looking at embryos, these proteins given off by an embryo, this message which we can detect from biopsy of a blastomere or cytoplasm, all of these things may ultimately predict outcome and we will be able to control for transfer technique and endometrium culture conditions cycle to cycle variability. It is going to give us statistical power that we have never approached having in our field and I do believe that this is going to be a huge part of our future.

It is going to take us far beyond the basic things we have looked at today, like pyruvate and lactate and different leucine and alanine because none of these has been adequately spored and I think in extremely well-done studies has allowed us to use them to correlate them with also with implantation rates such that we could change our transfer order.

I think there are other things that will allow us to start understanding meiotic spindle errors. This particular slide is from Pat Hind and actually not in a human model, in a small animal model, and shows there are all kinds of spindle errors and are there going to be ways to image those non-invasively in the future? I think the answer is maybe.

Are we going to have the ability to start looking at some of the cell cycle checkpoint regulatory proteins? This data received a lot of attention this year, there was an excellent publication looking at BUB, showing that BUB declines with age in animals, of course.

That work was originally done by Carol Brenner and demonstrated several years ago that in this particular regression line, which is the mature eggs, showing us that eggs from women are evaluated as they go through reproductive life, that the message for these cell-cycle regulatory proteins goes down for MAD and here again for BUB.

Our potential opportunity to assess those proteins, assess the message for them, represents the kind of thing we need to be thinking about and this could easily be done as a component of this finger-printing process as we become more sophisticated.

That, in turn, can get married with some outstanding chip technologies which are coming along. This particular graph which I borrowed from Michael Alper and the investigative group at Harvard and Boston IVF shows that they are characterizing patterns in genes which are appropriately up- and down-regulated at different points in the developmental process. As we start understanding which appropriate messages are up- or down-regulated at a given point in the process, we may be able to test for those and against assess reproductive competence. There are now chips that can look genomically at as many as 120,000 sites. There are chips that can look for messages in the thousands and thousands. So the ability to potentially look with good amplifications at different expression may really change what we do in the future. The key to that, of course, is going to be the amplification.

Across the board, we have single cells that contain about approximately 6pg of DNA and we are going to have to magnify that in the vicinity of one million to be able to look at things reliably. There are some technologies where perhaps we only have to amplify it one thousand times, but that is still an enormous burden for precision.

Moving along very quickly, we have to learn how to deal with telomeres and most of you are familiar with David Keith. We do not have a good way of assessing that clinically.

I just want to talk in my last two or three minutes before we move on to the question and answer session here.

We are going to have to do all this in the context of condensing our patients and our government and the media and the ethicists and our academic colleagues from different areas that we are not doing anything that might be changing the normality of the embryos. That begs the question: how do you demonstrate what is normal? Anatomic development, intelligence, behaviour, achievement, reproductive potential in the future. Obviously, those are important, but it means every study is going to take 30 or 40 years to do. That is probably not going to work out terribly well.

I want to show you something – you have probably already seen it in the information provided to you this morning – but this is sort of a trick question that I used at a talk I gave which a lawyer who has been on the president’s panel and is a thinker about what we do and has been involved in planning it. I presented this. I said, what if there were a theoretical syndrome? It had some pretty severe things that went with it: mental retardation, severe cardiac anomalies, respiratory problems, increased risk of leukemia. Patients in the general population develop this syndrome around 1 in 800 times on average in this country. But after you do IVF in this particular group of IVF patients, the risk is 1 in 82, a relative risk of 9.76. This person absolutely says, you should never do that. How could you possibly ethically ever consider doing that? This person truly became apoplectic. Think about this. If you look at the prevalence of Down Syndrome across age, it certainly rises.

All that is, is the risk of Down Syndrome in women 40 to 42 years of age.

If you are doing IVF in that group, and who is not because that is certainly one of our principal groups, then you are exposing patients to that relative risk. What we have and the people who evaluate what we do, is a tendency to attribute anything that could possibly show up in the outcome of the children who are born to our patients back to our core technology. They are not always cognizant of the fact that there are intrinsic differences in our populations which predispose our patients to different outcomes. This is a classic example of how people who do not understand the science do not necessarily look at the data in as rigorous a way to come up with a real answer. On the other hand, have we done enough yet? We certainly have problems with imprinting disorders and the prevalence of those appears to be higher.

I heard an excellent lecture at ESHRE this year where the speaker got up and happily quoted that there were more review articles on imprinting disorders in IVF than the total number of abnormal children from all studies published combined. This tells you what people think about us and there are lots of hungry young geneticists who are going to make their careers by talking about the evil things that we do. Across the board, we do have to be responsible.

There are issues with models like monozygotic twinning, low birth weight, some structural birth defects and potential imprinting disorders, although while relative risks may be 2 to 5, the prevalence is still probably less than 1 in 3,000. Across the board, we are going to have to stand up and start being more comprehensive and I think more committed to following the outcomes in our babies but with appropriately selected control groups.

One such study is going to be the footprint study, which is being done by one of the patient advocacy groups, the American Fertility Association, in conjunction with the Rand Corp. We will be starting pilots within the year, but we will be reaching out to virtually everyone in this room to help us encourage patients to participate in this study. We are going to be looking for IVF patients, as well as non-IVF patients, to hopefully allow us to follow outcomes with appropriate control groups and learn things about the lives of the children born through our care.