Ngs Ion Torrent Sequencing Boosts Ivf Pregnancy Rates

Well, thanks very much for that kind introduction, and for the opportunity to be here today - a very fascinating meeting with so many exciting innovations to consider. I'm just going to be talking, really, about the very application end of some of the amazing technology we've been hearing about in the last couple of days. I'm sure many of you in the audience know far more about the intricacies of next generation sequencing than I do, personally. But I'll really take you down the route that we've taken this, in terms of clinical application. So I guess first, to frame why we're interested in next generation sequencing. I've been working mostly, as you've heard, in the area of infertility for the last couple of decades. And I think the whole of infertility, really, has been revolutionized, I guess in the last 30 years, with the innovation of in vitro fertilization. Now, it's been an incredible intervention. It's completely changed the way that infertile couples can be treated, and brought hope to millions of families around the world. But despite that, despite the incredible success of it, we have to acknowledge that it's still relatively inefficient. Worldwide now, there's meant to be more than five million babies born. And in fact, in most industrialized countries, you'll find between 1 and 5% of all births come down the IVF route now. So in a typical class at a school, probably one of the two of the kids are going to be IVF children. So very successful on the surface of it. But as I said, somewhat inefficient. So what we're really hoping to do is try and address that inefficiency, to try and make it so that more IVF cycles are successful. As it stands at the moment, worldwide, only about a third of IVF treatments actually result in a baby. And although if you go on the websites of different IVF clinics, they'll be trumpeting success rates that sound an awful lot higher than this, the reality is, if you take all patients together, including those that have a poorer prognosis, typically you only see success in about a third of cycles. Now, it's likely that part of the reason for this inefficiency is our inability to tell which of the embryos produced is the one with the best chance of making a baby. In a typical IVF cycle, several eggs are produced, they're fertilized, creating several embryos. And then it's a matter of choosing which one you're going to transfer to the patient - ideally, just transferring one. Now, at the moment, this is done, based on morphology. So in every IVF clinic around the world, at least until recently, the primary way that the embryo has been chosen for transfer has been based on its appearance down the microscope - the idea being that if it looks textbook perfect, that must be a more viable embryo. Now, while there is some truth in that, it's universally acknowledged within the field of embryology, that this is really only a very rough guide. Very often, an IVF clinic will transfer an embryo that actually, they don't have much hope for; it looks pretty awful; and yet, it will make a baby. And on the very same day, they'll transfer one of those perfect looking embryos, and get no baby at all. So it's a guide, but only a rough one. And of course, it's quite subjective. Even the same embryologist on different days, may score the same embryo slightly differently. So because of this, very likely, about 85% of all embryos transferred during IVF cycles don't implant. So you can imagine, this negatively impacts the success rates of IVF dramatically. Now, how can you get around this problem? Well, hopefully you can develop better ways of identifying the most viable embryo. In the past, the way people have got around it has been simply to play with the statistics and transfer more embryos. If you don't know which one will make the baby, transfer more of them, and then just by luck, hopefully one of them will create a child. Now, this does work to some extent - it can improve pregnancy rates. But the negative side of it is that you end up with a lot of multiple pregnancies. Now, the idea of things like twins might not seem so bad. It seems quite nice, really. But the clinical reality is that those pregnancies are at a much increased risk of a whole variety of problems for both the mother and the babies - things like extreme prematurity, maternal hemorrhage, cerebral palsy - all much elevated in twin pregnancies, and even more dramatically, in triplets. So although the majority of twin pregnancies go off without a hitch, ideally, it would be better to achieve the same pregnancy rates with just one embryo. So the question has been, if we want to move towards elective single embryo transfer - in other words, just transferring one embryo in the cycle - we're going to need better ways of identifying the most viable embryo. And the question has been, could we do this using genetic methods. Could they provide us with a more definitive, less subjective measure of embryo competence? There are some pretty good reasons for thinking that they might. It's well known that chromosome abnormalities are extremely common in the oocytes, and as a result of that, the embryos, of our species. In fact, it's something like tenfold more common in humans than any other mammalian species that we have good data for. So because of this, we're likely to see a lot of embryos which are not viable. Of course, the aneuploidy issue, the chromosome abnormality issue, increases dramatically with advancing age. And you can see from this rather scary looking graph, two things, really. The first is that yes, chromosome abnormality rates, and this is actually showing in human eggs, do increase a lot as a woman gets older. In fact, for a woman over 40, it's typical for more than three-quarters of the eggs to be chromosomally abnormal. But the other thing you can see, is actually they're reasonably common even in younger women. So for a woman in her early 30's, you're probably already talking about a quarter of the eggs being chromosomally abnormal. Now, of course, most chromosome abnormalities, most aneuploidies are lethal. We see this, of course, all the time in miscarriages. About 70% of first trimester miscarriages are chromosomally abnormal. So we know that these kinds of problems are usually lethal to the embryo. So it's not surprising that as the chromosome abnormality rate goes up, the implantation rate of the embryos goes down - implantation rate being the probability of an embryo that was transferred, actually making a clinical pregnancy. You can see that these two lines here are almost a mirror image of each other. And we think that's no coincidence. We think that's because they are causally related. So this has led to the proposal that maybe we should actually test embryos before we decide which ones to transfer, and only transfer those to the uterus which are chromosomally normal. It's a concept that's been called various things, but it's probably best known as pre-implantation genetic screening. So the idea is, do an IVF cycle. You begin with a group of embryos. Maybe half or more could be chromosomally abnormal. And rather than transferring one of these, not really being sure whether or not it has the correct number of chromosomes, do a test that allows you to fill in that missing piece of information. And be sure that you're transferring a chromosomally normal embryo. You're not going to guarantee a pregnancy, but you're probably eliminating the single biggest factor in embryo failure. So there's different ways we can take the cells from the embryos for this. You have to appreciate, of course, that we're dealing with a minute amount of tissue when we do these tests. The traditional time to do this sort of analysis was three days after fertilization of the egg, by which point the embryo usually consists of only about six to 10 cells. Now, you can only really safely take one cell for analysis. In fact, there's some question whether even that is entirely safe. I'll show you a very ancient video here, where you can see the embryo with its individual cells. This is just a glass pipette, very magnified, which is applying a gentle suction to hold the embryo in place. You may be able to just about make how the embryo has a membrane around the edge. That's called the zona pellucida. And obviously, to get a cell out for analysis, we're going to have to breach that. Now, these days that's done with a laser. But in this very old video, it's being done with a weak acid. Let's see if I can set that going for you. There, you see the zona pellucida being breached, and then another pipette comes in to tease out one of those cells. So any technology that we're going to develop in order to test these embryos is going to have to be exquisitely sensitive. It's going to have to be able to work at the level of a single cell. And obviously, that comes with a whole bunch of challenges all of its own. And there goes the cell. That cell would be taken away; it would be washed, and then placed into a micro centrifuge tube, ready for the genetic test. This is a more recent version of this kind of approach. This is an embryo five days after fertilization. It's what's called the blastocyst stage. And at that stage, the embryo will spontaneously start to emerge out of the zona pellucida surrounding it. And you can see that's what's happening here. Half the embryo is still inside, and the other half is sort of herniating out of the hole. And if we looked to this embryo again a couple of hours later, the whole embryo would have come out. They actually call this process hatching, like hatching out of an egg shell. So again, these can be sampled. Here's the pipette. It's sucking in a few of these cells from around the edge of the embryo. The embryo has formed a fluid filled cavity at this point. And these cells that have been taken, these trophectoderm cells, are what would ultimately become the extra embryonic tissues - things like the placenta. So you can kind of think of this as like a very early prenatal test. So here, the cells are being tugged. You see the occasional jump of the embryo there. That's because it's also being lasered in this sort of area here. So there's some pulling, and eventually, a piece of tissue is removed. There's the embryo, looking a little bit sorry for itself, and there's the cells that are removed. Now, although that looks a little bit brutal, these embryos are incredibly robust, and they tolerate this kind of insult really almost as if nothing had happened to them. That's probably the main difference compared to the analysis I showed you first, on the single cell taken on day three - those embryos are an awful lot more fragile. So in theory, this can be used in different ways. We could look at inherited conditions. And that's usually referred to as pre-implantation genetic diagnosis, or PGD - looking at the pre-implantation embryo, and trying to diagnose an inherited condition. It could be cystic fibrosis, muscular dystrophy. It could be a chromosome rearrangement, like a translocation, causing a lot of unbalanced gametes. The patients who have this are really looking to avoid transmission of this familial problem. The other thing we can do, which is more for IVF patients in general, is pre-implantation genetic screening, as we've already mentioned. Those patients are really hoping to see three major advantages. You're not going to transferring any chromosomal abnormal embryos. So in theory, you should not see any aneuploidy syndromes - Downs, Edwards, Pataus, Klinefelters, Turners. All of those should go away. You should get a reduction in miscarriage rates, as well, because so many miscarriages are caused by chromosome abnormalities. Not transferring those abnormal embryos should give you a significant reduction in miscarriage rate. And finally, the main point, which motivates most of the IVF patients to consider this, is the possibility of an increased pregnancy rate, a more successful IVF treatment. Just *** past that. It's been quite a controversial area. Although the theory sounds really solid, it really is true that we produce a lot of chromosomal abnormal embryos. And it really is true that those abnormalities are usually lethal. And yet, when people got around to doing randomized trials of this kind of technology a few years ago, it was found that really it didn't make any improvement to the IVF success rates, which is kind of surprising. It now seems that those initials failures were down to technological problems. All of the analyses were done on the embryo three days after fertilization. Maybe looking five days later is perhaps better. And the tests available used fluorescent in situ hybridization, fish, and could only look at about half of the chromosomes in the embryo. So often, an embryo would have been called normal, and actually was abnormal for a chromosome that had never been tested. Now we have a whole variety of methods that look at the entire chromosome complement. And there have been multiple studies now, showing - including in randomized clinical trials - an improvement in IVF outcome. So here's just a few of them. All of these ones, these bottom four are all randomized trials, they're all really from the last 18 months to two years. They used a variety of different methods - array CGH, single nucleotide polymorphism, micro arrays, quantitative PCR - but all essentially delivering similar information - copy number of all the chromosomes, transferring only those embryos that are found to be chromosomally normal. The thing that strikes me is despite the fact that these were going in different clinics, using different genetic methods, and somewhat different patient groups, they all really came out with quite a similar improvement, about a one-third increase in the likelihood of an embryo implanting, if it had been shown to be chromosomally normal. And for me, I find that very reassuring. It's always a little bit suspicious, isn't it, when you get one clinic shouting very loudly, saying, 'We're doing this fantastic thing. It's great.' And no one else is. But this is from different groups, different technologies, different patients, and yet similar kinds of results. So as I've mentioned, there are some challenges to this kind of analysis. We've only got maybe as little as one cell for testing. It can be quite difficult to combine testing of genes and chromosomes. We sometimes get a couple who come for PGD, for a single gene disorder. You find a nice, unaffected embryo, you transfer it, and then you get the really sad news that a pregnancy was established but then miscarried due to a chromosome abnormality. So obviously, it would be nice to be able to combine the two things. At the moment, what we do look at is still relatively restricted. We look at single genes, or we look at chromosomes. But of course, these are just two out of many different aspects of embryo biology that could be relevant to their ability to form a successful pregnancy. And then also, we face issues of cost. We usually think of a genetic test as the patient comes in, you take their blood, and you give them the test. But in the case of PGS, one patient can produce many embryos. In fact, a patient could produce as many as 20 embryos. That's 20 tests. So you have to multiply the costs of whatever you're doing by potentially a lot of different samples. So that has limited the access of some patients to this beneficial kind of technology. And then the final point is we need it quickly. Increasingly, people are starting to freeze embryos after they've biopsied them, allowing them quite a lot of time to do the tests. But that certainly hasn't been the tradition in this area. Mostly, worldwide, embryos are still biopsied on Day Three; cell is taken away for testing. The embryo goes back in the incubator, and he carries on developing. Now, you've only got a very limited window of time before that embryo has to be transferred to the uterus. There's only a limited window for implantation to occur. Either the embryo can go past that, or - of course, the woman's cycle is also still progressing. She can go past the time when she's actually receptive for the embryo to implant. So typically, we need results within about 24 hours. So quite a few challenges. So really the aims that we had in terms of the Next Gen sequencing, which have developed methods that allowed us to have a low cost screening for aneuploidy, ideally giving some potential for combining a single gene analysis if needed, and to look at other aspects of the embryo's biology. In particular, we focused on the mitochondrial issues with those embryos. The strategy was to take single cells, to lyse them in the tube, to release their DNA, do a whole genome amplification, again, still in the same tube so that we wouldn't lose any material, and then go on to use a Next Generation sequencing technique. Now, obviously we needed something that was going to be relatively straightforward to use, because ultimately we wanted to be able to move this into the clinic. So we needed something fairly simple; robust, certainly; accurate, of course; but also not too expensive. We also needed it to be scalable. There's no point having a fantastic result, fantastic method, if we needed, you know, 10 pieces of equipment to run each of the embryos individually. So we needed it to be scalable. And of course, as I've mentioned, we needed it to be fast, ideally giving us results within 24 hours. And we think really, the Ion Torrent™ hit all of these requirements very nicely. Of course, like most Next Gen approaches, it was using natural nucleotides and polymerases, so the costs were not prohibitive. We can get, certainly, enough sequence information to allow us to barcode and multiplex multiple samples together. And of course, it's the fastest option in terms of sequencing. So we were very happy to do some work on this initially. And we started out looking at single cells from cell lines with known chromosomal abnormalities. Obviously, we wanted to look at something where we would know what the outcome should be, at least in theory. We were initially using a 200 base pair chemistry, getting about two and a half million reads, about a gigabase of sequence data. We have also run products on the (Ion) Proton™, and of course, you can get about 20 times that, using that approach. Now, we weren't interested in ever looking at the whole genome. In fact to be honest with you, that's the last thing I want, because think what you might find, and what are you going to do in terms of interpretation. We didn't want to have any kind of findings of unknown significance. So we deliberately kept this at a very low level of coverage. In fact, we were getting, I think, less than half a percent coverage of the genome. It sounds kind of pathetic when you hear about some of the things that we've heard about in the last couple of days. But for us, that's exactly what we wanted. Typically, we're talking about - maybe about 100,000 reads per sample. When we did a 32-plex, we were getting about - just under 80,000 reads per sample. And in terms of reads of the chromosome, because ultimately we're interested in aneuploidy, youíre talking about - it's probably only about 5,000 reads per chromosome. So it's not a lot. But it's enough to call chromosome abnormalities. And that's the key thing. So as I said, we looked at some single cells, and these are the kind of results that we would typically get from a normal female cell. And you can see that these are actually the percentage of the total reads that are attributable to each chromosome. Now, you might expect, of course, Chromosome 1 to give you more reads than any of the other chromosomes, just because of its great size. But you can see, there isn't a perfect size relationship with number of reads. And of course, this is due to biases introduced by the whole genome amplification. It likes to amplify certain sequences better than others. And I guess that's no surprise - such as Chromosome 19, very GC rich, a lot of amplification there, a lot of reads. The important thing, though, is that this pattern was very reproducible. So in multiple normal samples, we always saw the same pattern. So that meant that we could compare a sort of normal reference against what we would get from an embryo, for example, and we might see something like this, where most of the chromosomes look very similar to our normal control, but as you can see there, in this case Chromosome 22, with far more reads than we would normally expect to see. And indeed, this is from a trisomy 22 sample. So this actually worked very nicely, very robust. I mentioned we looked at cells from cell lines with known aneuploidies. We also looked at a bunch of embryos that had previously been diagnosed, using Microarray CGH, which worldwide is the most common way of looking at chromosomes and embryos at this moment. We essentially calculated the amount of reads from each chromosome, as I've shown you. We got a result from every single sample that we tried, and they were 100% concordant within this set. Some of the embryos had more than one chromosome abnormality. In fact, there were 58 different aneuploidies in there, and all of those were successfully detected. Now, this was done in a blinded experiment. The samples had been coded. They were decoded by a third party who was not involved in the experiment, and so it gave us good confidence in the diagnostic accuracy. Here's just an example of one result. This is an Array CGH result, using the common Blue Note Microarray. And you can see essentially we're looking at a ratio of red to green fluorescence, comparing a normal sample in red to our embryo sample in green. And you can see all of the clones on the microarray have been lined up in the order they appear on the chromosomes so starting at the top of Chromosome 1, all the way down to the bottom of that chromosome, and then on to Chromosome 2 and so on. You can see that almost all of the chromosomes have a normal amount of material here, an equal ratio of red and green, but quite clearly, we have an excess there of Chromosome 18. This is a trisomy 18, and we can also see that the sample was female, by the additional X chromosome material and no Y chromosome material. Same embryo, but run using the sequencing data from the PGM. And again, you can see really very much the same story - an excess of material from Chromosome Number 18, and the difference on the sex chromosomes. So once we'd validated this we went on and performed two clinical cycles. This was done in, very much in a research context, with the patients counseled as such. They were done for infertile patients who were going through IVF, who were considered to be at elevated risk of chromosome abnormalities, purely on the basis of maternal age - so increased risk of Down's syndrome, miscarriage, and abnormal embryos. Seven embryos were biopsied at the blastocyst stage, so that's like the second video that I showed you. We did Next Generation sequencing, just to look at chromosome copy number. The results from all seven - single embryo transfer in both cases, giving rise to two healthy babies that were delivered earlier this year, one at New York University's IVF Clinic, and one at the Main Line Fertility Clinic in Pennsylvania. There have also been other births reported from these kinds of techniques, such as from BGI. So this does seem to be a relevant technology to use in this area. And there's the consequence of one of those two cycles. So in terms of costing and speed, we have to of course consider this very carefully. This was the initial work flow that we were using when we first started doing these cases. And the total time required was actually under 13 hours. So it fits very nicely within our 24 hour window that we have to try and satisfy. The other thing, very importantly, is that with multiplexing, the actual cost was coming down to about $70.00 per sample. And I think, you know, that's just the beginning. I'm sure it's possible to get even lower than that. That's already, though, about 30 to 50% cheaper than current methods using microarray technology. So again, very promising. We've been very lucky, actually, to have access to the Ion Avalanche protocol, as well, and we're extremely excited by that. Of course, that gives us even faster data, and I think realistically, we're getting data out in about eight hours now, using that approach. So that really means that you could potentially use these technologies and not have to contemplate freezing the embryos if you don't want to. So that's, I think, enormously exciting. Here's one of those Avalanche results, showing an embryo with a monosomy six. And in this case, I haven't crunched the numbers myself, but I've let Ion Reporter do the hard work for us. And you can see that it calls it very clearly, just there. Of course, I think this is just the beginning in terms of what we can do, in terms of cost and speed. As you're aware, the Ion Chef is not far off now, being able to get put into our labs, and that should streamline things even more. So that's going to be great, I think. Okay, just to talk about a few other results, if we leave chromosomes for a moment. As I mentioned, it would be nice to be able to look at single genes, as well. It's a little bit of a niche area, I guess. There's a huge number of IVF patients who can benefit from chromosome screening. Perhaps the number of families that would actually have testing of individual genes is somewhat smaller, but still an important group, nonetheless. Many of these patients that we have are actually fertile, and they choose to go through IVF, with all the stresses and uncertainties of that procedure, purely because they do not want to contemplate a pregnancy termination. So for them, it can be an extremely important reproductive option. Also of course, you do get those patients who are kind of doubly unlucky - they're unlucky enough to be infertile and need IVF, and also have a single gene disorder running through their family. Of course, it doesn't make sense for them to struggle sometimes years with IVF to attain a pregnancy, only to then find that it's affected, and have to think about whether theyíre going to terminate it or not. It makes much more sense to test the embryos for those kinds of patients. This is just a very brief proof of principle that we did, just to show that you could if you wanted to detect mutations in individual genes. Essentially what we did was take the single cell, do a whole genome amplification, split that product, make two - make a library from the MDA. With the other product, we would amplify our gene of interest, make a library from that, and combine the two libraries and then sequence them together. The whole genome amplification gives us this low coverage of the genome, and you can see the occasional bits of sequence just there from the gray bars - very low coverage, but that's what we expect - enough to give us chromosome copy number information, however. And there you can see the area that we've enriched, just through a simple PCR strategy with multiple sequences. This is actually detecting the Delta F508 three base pair deletion. When you do the chromosome analysis using the Next Gen sequencing approach, of course you get sequence data on any of the DNA that's in there. So as well as our light covering across the genome as a whole, we also of course, get information on the mitochondrial genome. That gets sequenced without us even really trying. And here's a typical result. This is the mitochondrial genome, right the way across. And you can see that we got, you know, reasonable depth of coverage there, going across. This is a single cell that we looked at from a cell line with a known mitochondrial DNA deletion. And you can see that it's picking it up quite nicely. You've got a real distinct break in the sequence here where the deletion begins, and then it starts off again right there. Of course, you do get some underlying coverage because it's very unusual, well, if not impossible, to find cells that have every single mitochondria affected by this kind of problem. There's usually a degree of heteroplasmy, a mixture of normal mitochondria and those that are carrying the mutation. So this actually allowed us quite clearly to define the limits of the deletion, and also actually get some kind of estimate of what proportion of mitochondria might be affected. I think really if you were using this diagnostically, you'd want to do it at considerably more depth than we did it here, but at least it shows in principle that this should be possible. Now, mitochondrial DNA disorders have been something that have been notoriously difficult to diagnose in human embryos, and very difficult to get real good, quantitative data, allowing you to do PGD for them. So I think this is going to be extremely useful for those kinds of patient. We've also generated some quite interesting - I think interesting, anyway - research data from the embryos that we've looked at. This is data I'll show you - on 47 blastocysts, so 47 five day old embryos, where they had been biopsied and analyzed using the (Ion) PGM™. We were primarily looking for aneuploidy detection. But we also looked back retrospectively at the amount of mitochondrial DNA that was in there. So we looked, really took a simple ratio of how much - how many mitochondrial DNA reads were they, compared to nuclear DNA reads. And this is what we saw when we plotted them out and broke them into chromosomally normal and aneuploid embryos. And I think you can see quite clearly, although there's quite a lot of overlap between the two groups, aneuploid embryos quite often have an unusual amount of mitochondrial DNA, more than expected. And that is statistically significant. We also broke down that data by age. And so this is showing you female age, along the bottom; chromosomally normal embryos in blue; and chromosomally abnormal embryos in red. And essentially what we see is that there's an increase in mitochondrial DNA with age, but also with chromosome abnormalities. As you see, the chromosomally abnormal embryos for a given age group almost always have higher amounts of mitochondrial DNA. Now, these are just average values, and I do stress, there's a lot of overlap between individual age groups. But certainly, there's a strong tendency there. So this may be telling us something about the origins of aneuploidy in human eggs and embryos, the fact that we do have these elevated amounts of mitochondrial DNA in those cases. It may tell us something about the reproductive aging process in humans. So that's all very useful stuff, scientifically. Of course, it would be nice if we could also apply this kind of information clinically. And we have looked at this, and this was actually reported at the American Society for Reproductive Medicine meeting, which was actually here in Boston, actually in the Conference Center next door, just last week. So what we did was we looked at chromosomally normal embryos that had been transferred, and yet didn't implant - because although I mentioned that chromosome abnormalities are the single biggest cause of embryo implantation failure, even when you transfer a chromosomally normal, morphologically perfect embryo, you still can't guarantee a pregnancy. So we looked at some of these embryos, and we looked at the mitochondrial DNA levels, and this is what we saw. So these are embryos that produced a pregnancy. These are embryos that did not. And again, you don't need any statistics, I think, just to see by eye that these are clearly different. There seems to be a threshold of mitochondrial DNA content above which an embryo just simply won't implant. So this is showing you the kind of data that Next Generation sequencing approaches can give you, above and beyond what we've been able to attain with the tools we had available to us until just recently. This is data that you would obtain without even trying, if you were just using this kind of approach for chromosomes screening. So very interesting. And of course, that difference also is statistically significant. So just to wind up and conclude - I think the Ion Torrent - should be PGM™, I guess - provides a very powerful method of single cell analysis. We've been using it in the context of reproductive medicine, embryo screening. But of course, you know, the sky's the limit, really. You can use your imaginations, and there are so many other approaches using small quantities of DNA, where this could be applied. We look at all 24 chromosomes, and there's some good evidence now that this is going to lead to improvements in IVF success rates. The evidence at the moment isn't from Next Gen sequencing approaches; it's from alternative approaches that look at all the copy number of the chromosomes. But really, there's no reason for thinking that it's going to be any different if we look at the chromosomes this way. This should give us lower risks of miscarriage and Down's syndrome, and more success in IVF. The results are available in under 24 hours, and in fact, could be in less than a third of that time if needed, and that allows a fresh embryo transfer. There is some debate in the field about whether a fresh embryo transfer is really something superior to a frozen embryo transfer. But certainly many, many labs, particularly in Europe and other parts of the world still have a great preference for doing fresh embryo transfers. Certainly, patients prefer it, too. It does give us the possibility of looking at actual DNA sequences, such as gene mutations, at the same time as looking at chromosome copy number. So you can do PGD and PGS together. And of course, we get information of potential biological relevance about other things that are going on in the embryo. And I think the real value of those still remains to be seen, but we're very hopeful that that's going to make a significant difference. So I'll just leave it there, other than to thank again the organizers for giving me the opportunity to be here, and to thank my team at the University of Oxford, the Reprogenetics Lab there, that have worked very *** developing some of these methodologies. We've had considerable help from the teams at Life Technologies, so thanks for all of their input, and also from Kulvinder Kaur and Jenny Taylor at the University of Oxford's Biomedical Research Center. Thanks very much for listening.