Scientist Stories: Matthew Hurles, Causes and Consequences of New Mutations in Humans

Welcome to this, the latest in the Sanger seminar series. My name is Matthew Hurles. I'm a human geneticist at the Sanger Institute. And I'm delighted today to be talking to you about the causes and consequences of new mutations in humans. So I'm going to start by taking you back to the start of your life, the start of everyone's life, when we were all just a single-celled zygote, a fertilized egg. And for the first few days, our lives were pretty similar. So both men and women, as they developed and cells divided, they— and at the end of 10 replications or so, a particular key cell type called the primordial germ cell was— is developed. And that's key because that cell, those cells contain the DNA that gets passed on to the next generation. So when we're thinking about mutations that occur from one generation to the next, we really need to be thinking about the journey that DNA takes through different cell types as it goes from one generation to the next. And so the first 10 replications, whether you're a man or a woman, your DNA has gone— been copied about 10 times, and it's ended up in a primordial germ cell. And then in later early development. Now the paths start to diverge between men and women. So for women to form the eggs that go on then to be fertilized for the next generation, there's another 20 times that the DNA is copied going from one cell generation to the next until the immature oocytes are formed at birth. And those cells stay immature in stasis until they are then mature during the menstrual cycle. So every egg that a woman produces, the DNA has gone through about 30 generations of cells to get from the zygote to the egg, whereas on the male side it's somewhat different. So on the male side, until puberty, it's relatively similar. There's a relatively small number of cell divisions that go to form a key cell type called a spermatogonial stem cell, and from puberty onwards these are the cells that continually divide and differentiate to form sperm, and those sperm are produced throughout a man's life. And that— what that means is that the sperm that are produced by an older man have typically gone through more cell generations, have been copied more times to get into that sperm than the sperm of a younger man. And it also means overall that the DNA that comes from a father has gone through many more cell generations than the DNA that comes from the mother. So once you start thinking about this journey, it leads to a couple of, of obvious kind of hypotheses. The one is that if the process of copying DNA is in any way causing some of the mutations that, that can occur, then it may— we may expect to see that there are more mutations in children of older men than there are in children of younger men. And we may also expect to see that there are more mutations coming from the father in the sperm than there are from the mother in the egg because of the additional numbers of cell generations that have arisen in the male line. So I'll come back to this journey over time because it'll, it'll be a key part to how we think about mutations, how they accumulate. But we have to bear in mind that DNA, as it gets passed on, is going through this kind of complicated cellular genealogy. We also have to bear in mind that there's not just one type of mutation that affects our DNA, there's actually many different types, from the tiny change of a single letter of DNA shown on the left-hand side, to on the right-hand side, the addition of an entire copy of a chromosome, for example, in Down syndrome where there's an extra copy of chromosome 21. And essentially there's a continuous distribution in between these two, so going from the gain or losses of single or small numbers of letters of DNA, bases of DNA, to the gains and losses of larger chunks of DNA, all the way up to the size of a chromosome. But if one compares the DNA of a child to the DNA of their parents and looks at the new mutations, most of those new mutations are single letter changes, single base changes. So 90% of them, say, and only a smaller fraction of those are these larger changes. And even those, most of those are small. So in thinking about DNA as it passes from one generation to the next, we probably need to think about different types of mutation processes. And so how do we go about measuring these mutations? So this slide shows 3 general strategies that people who do this kind of research have taken. The kind of the central, the middle one of these 3 is perhaps the most obvious one, is you just look at the DNA of children and you compare it to their parents. Now, until the advent of very high-throughput sequencing, that was a thing that was much easier to say than it was to do. And so other strategies have also been used. And on the right-hand side, you can see that we know that mutations accumulate over evolutionary time as well. And if one knows how long ago did two species split, here shown as humans and mice, and one can count the number of mutations that have occurred since that point in time, one can start to estimate how rapidly mutations have occurred. And a third strategy, shown on the left-hand side, is if we are— have very sensitive ways of measuring mutations in single cells or small groups of cells, we can potentially measure mutations in a collection of gametes, which is obviously most easily done by looking at a collection of sperm and identifying mutant sperm compared to the rest of the sperm. So all three of these different strategies have been used in recent years to try and look at the different factors that influence mutation rates for these different mutation processes that I mentioned. And I'll give you one kind of example that relates directly to how we've investigated looking at the effect of sex on mutation rates. So this is using this kind of evolutionary approach So if one looks at the amount of single base changes that have accumulated between humans and chimpanzees in the human genome over time, one can see that broadly speaking, that genome splits into 3 groups. There's the X chromosome, which has between humans and chimpanzees about 1 in 100 bases are different. There's the, the rest of the chromosomes apart from the Y chromosome, where it's about 1 in 77 bases are different, and then on the Y chromosome about 1 in 53 bases are different. And so what that seems to suggest is the mutation rate on the Y chromosome is significantly higher than it is on the autosomes, which is significantly higher than it is on the X chromosome. And the key point to understanding this is knowing that the Y chromosome, because it's passed only from fathers to sons, is always going through that male lineage as it goes from one generation to the next. The autosomes are spending half their time in the males and females being passed on, sometimes from an egg, sometimes from a sperm, whereas the X chromosome, two-thirds of the time it's being passed on from a mother and then go through the mother's line. So this is good evidence that for this type of mutation, these single base changes, that sex really does influence the mutation rate, and the mutation rate is higher in men than it is in women. But that's not to say that these kinds of effects of sex are the same for these different types of mutation process. And here I'm contrasting two different mutation processes. On the left-hand side, we're looking at those single base changes that I mentioned, and I already told you that those comparisons between humans and chimpanzees indicated the mutation rate was significantly higher on— through the male lines than the female line. And now that we can do the experiment of sequencing all the entire genome of children and comparing to their parents, we can see that that holds true. So about 75 to 80% of the time If we see a new mutation in a child, it's come from the father, and only about 20 to 25% of the time has it come from the mother. We can also see if we look at individual families, and these three kind of lines here show three different families, each of whom had four children at different times. And we can see over time the children that were born later when the father was older have more mutations than the children that were born younger. So this goes to support that original hypothesis that I mentioned, that there may be an effect of the age of a father on the number of mutations their children have. And so we see a clear effect of sex, we see a clear effect of age for this particular type of mutation process. But if you contrast that to on the right-hand side, uh, with a particular type of mutation which creates a deletion of a large chunk of DNA on one particular chromosome, We can see that when we investigate that further, that about— it's actually 50/50 when it— if it occurs on in the sperm or the egg through the male line or through the female line. And here in this case, we're actually using an assay that looks at sperm, and so we're measuring mutation using that third strategy that I mentioned where we can look at sperm and compare it to other sperm. And when we do that for sperm donors of very different ages, as shown in the bottom plot, on the right-hand side, we can see there's effectively no relationship with age. So for this particular mutation process, there doesn't seem to be an age effect, and there doesn't seem to be a sex effect either. So then we need to try and understand why might that be, and this is where we can go back to looking at the journey that DNA takes as it goes from one generation to the next. And one thing I didn't tell you is that these single base changes, they're where we see big effects of sex and of age, those mutation processes can occur at any one of these cellular generations. So they can occur throughout, whereas the particular type of deletion of a large chunk of DNA that I mentioned can only occur in one very specific cell division known as meiosis, which only occurs once during either the female line or the male line. And because it only occurs once in both lines, that's probably the reason why there isn't a sex difference and there isn't an age effect for this particular type of mutation. So understanding when mutations occur on this journey from one generation to the next helps us to understand what the underlying factors that might influence if you're more or less likely to have that type of mutation. So that's really all I wanted to say on the factors that we investigate with regard to mutation process. I now want to move more on to thinking about the consequences of those mutations. And here we're thinking particularly about the consequences for human disease, and I want to describe to you two studies that we've been part of that have been nationwide collaborative studies which have really used the power of the National Health Service and the expertise of the clinicians that we have in the National Health Service in the UK. So the one on the left is the DDD study, where we've worked with 24 different centers throughout the UK and Ireland who see children with severe developmental disorders, and with the title of the project being Deciphering Developmental Disorders. And those, those 24 centers and over 200 different clinicians have worked with almost 14,000 families, recruited them to this study. They've provided DNA, and then we've sequenced all of the genes in the genome for, for those families. And what we find is that for about a third of those families, when we do that kind of approach, we can actually provide a clear genetic cause for the child's condition. In one of the genes that has already been well associated with developmental disorders. And these diagnoses are fed back to the clinical centres and back to the families. And as shown here, 80% of the time that we can find a diagnosis, that diagnosis is caused by one of these new mutations. So this is a mutation the child has that neither parent has. Now, All of us have new mutations. So as you would have seen on the previous plots, the average number of new mutations of single base changes is about 60 or 70 per individual. And so it's pure chance whether one of those new mutations happens to fall at a letter of DNA that is really critical for development. But despite the fact that there are so few mutations, only 60 or so, given that we have 3 billion letters of DNA we receive from both mother and father, So the process of copying DNA has very high fidelity. It still accounts for a significant fraction of these children, these diagnoses, and underscores why very often, in fact most of the time, in these families the child is the only affected member of the family. Now if you compare that onto the right-hand side, where rather than working with children who maybe have been seen by GP, then a pediatrician, and then referred to a clinical geneticist, as in the DDD study, we're looking prenatally at ultrasound abnormalities that are picked out, picked up during the routine ultrasounds carried out in the UK, typically at 12 weeks and at 20 weeks when a woman is pregnant. And what we found when applying the same kinds of approaches is that a lower proportion of the of the babies we can identify a clear genetic cause of disease. But again, most of the time when we can, it's a new mutation. So these de novo mutations, although they're infrequent and they're things that we all have, they're causing a large fraction of the diagnosable conditions, both postnatally and prenatally. We also did a smaller study where rather than looking at abnormalities picked up by ultrasound, We worked with fetal pathologists who try to understand the causes of death in the very sad circumstances where babies don't survive, and because they are able to look in much greater detail than one can with an ultrasound, they're much better at picking up likely genetic causes of disease, and that's why we see a much higher rate of of diagnosis in fetal postmortems than when there's an ultrasound abnormality. So these kinds of genomic approaches that we're using in these studies, they clearly have great utility for being able to provide diagnoses to families and understanding the cause of their child's condition and giving them opportunities to have counseling and understanding what their future risks might be. So that kind of describes to you a little bit about how we can diagnose conditions that we already knew about in patients where they don't currently have a diagnosis. But we can use the same understanding of mutation to try and identify new causes in the patients that aren't diagnosed. And the simple thing that we do here is by using that understanding of mutation that we've got from the studies that I mentioned at the beginning, we can just predict simply by virtue of the number of families that we've sequenced and how big and how mutable each gene is, how many mutations we might expect to see in each gene just by chance. And then we can look in the patients to try and identify genes like gene 3 here, where there's a much greater number of mutations in patients than we would expect to see by by chance, purely on the basis of how we understand mutations accumulate. Now of course, most of the time we see situations like gene 1 and gene 2 here, where the number of mutations that we see across all these families is entirely consistent with those mutations occurring by chance. It's only in very rare cases that a gene stands out as being unusually highly mutated. And then of course, we need to look to see whether the children that have those mutations share similar clinical problems And thus we can start to define new genetic syndromes. And this gives you just one example of three of those new syndromes that we've discovered in the DDD study to give you a sense of the value of taking this kind of nationwide approach. So in one of these new conditions, we identified three children with essentially the same mutation in a gene called PPP2R1A, As you can see, they came— those 3 children came from 3 different centers around the UK, and it's only by combining the data across all these UK centers, because it may well have been that no one clinician would ever see 2 children with this disorder because they're so rare, but by pooling them together nationally, we can pick up these patterns. And that's actually a general phenomenon that we see for each one of these disorders. That I'm showing you here, which localize onto different chromosomes shown around the outside, we see that children with identical mutations occurred in different parts of the country, and only through this nationwide project can we identify these disorders and show that these children clearly have the same condition. So that gives you a kind of a very national flavor of what we're doing, but the reality, of course, is that rare genetic diseases affect everyone on the planet, every country on the planet. And sometimes to gather enough evidence to be really confident of a new disorder, we're combining it with other clinical centers, other researchers around the world. And just to give you a flavor of that, if one looks at the first 100 papers that have come from the DDD study, the distribution of authors on those papers, while centered on the UK, clearly involves many other countries around the world. And in fact, one of the things that we've done, and led by my colleague Helen Firth, who's a clinician within the NHS, has been to set up a web portal that enables clinicians and researchers around the world who think they may have found the cause of disease in their patient, the genetic cause of disease, but aren't sure, to kind of pull that information with the family's consent providing— it's all anonymized, there's minimal information given— but to find ways of making matches between similar patients with the same condition in different parts of the world. And the kind of global map on the top left just shows you the extent to which this kind of system has enabled thousands of collaborations and thousands of papers and thousands of diagnoses to be made that may not otherwise have been made. And not only can we bring together on this one portal all of these clinicians and their patients, but we can start to integrate the data that they provide on the genetic data, and we can start to identify, link that together with what we know about what those genes do and what the structure of the proteins are. We can link them together with other initiatives around the world through an initiative known as Matchmaker Exchange to link up different rare disease data sharing initiatives around the world, and we can start to build up a picture of what these new disorders, which are often defined when we only have a very small number of patients, what are these new disorders, what characterizes them, what are the kinds of problems that patients have, and what are the kinds of mutations that they have. And by building up these genotypic and phenotypic summaries of disorders, we're providing resources that make it easier for clinicians to recognize these disorders in the future. So one thing we've noted to be doing these kinds of studies is that when we find a gene that's been enriched for new mutations, we often find there's two distinct patterns, one of two distinct patterns that those mutations show. Typically, we observe genetic variants which we think truncate the protein and cause it to have loss of function, or or that alter the protein by changing an amino acid sequence known as a missense variant. And we see quite a few genes that look like these blue genes on the top panel, where the majority of the disease-causing variants are protein-truncating variants, and they're spread throughout the gene itself. Whereas the second class is where actually the majority of variants are in these genes occur in only a very small fraction of the protein coding sequence of the gene, and they're almost all of the amino acid changing type, and so they're very clustered together. And what that suggests to us is firstly, if one wants to try and find new disorders, we have to look for these two distinct patterns in these mutations that are enriched in these families. And the second is that these patterns probably reflect different mechanisms by which these diseases are occurring. And in the top blue genes, this mechanism is probably loss of function, so the functional copy is being destroyed of this gene. And in the bottom panel with the red genes, it's that the function is being altered by specific changes to specific regions of the protein, such the protein is now doing a subtly different role. So we can apply these kinds of approaches to ask the question, not just can we see how enriched individual genes, but how— for damaging mutations in these cohorts, but also how enriched are we across all genes. So how much does new— how much of a role do new mutations in any of the 20,000 genes in the human genome play in these kinds of disorders? And when we do that kind of analysis, we find that roughly speaking half of these patients have a de novo mutation, a new mutation in a protein-coding gene that's being— making a significant contribution to their disease. But only about half of that burden, here shown in the black, is in the genes that we are already robustly associated with disease. So there is another quarter of the cohort who, where we've already discovered the variant that's causing the disease, but we don't yet know concretely that the gene in which that variant has fallen is associated with these diseases. And we think that roughly half of those yet-to-be-discovered diseases are of that loss-of-function type, that blue type, and about half are of that altered-function type, that red type that I showed you before. So this approach gives us a clear direction of travel for where we need to go next to try and diagnose the undiagnosed patients. And to do that, we need to accumulate more data than we currently have. And I showed you the DDD study where we're working with 14,000 families, and about 10,000 of those families we have data from both parents and, both parents and the child, and therefore we can identify these de novo mutations. And we've pooled these data with a large US diagnostic company called GeneDx and perhaps the leading diagnostic center in Europe at Radboud University in the Netherlands. And through doing this, we've managed to generate a dataset on over 30,000 families, which has really helped us to discover some of these new disorders that we couldn't discover just by looking at the UK patients alone. And this is a slide just showing you the 100 most diagnosed conditions in this dataset of 30,000 families. And the key thing to note really is that even the most common diseases that we find here account for less than half a percent of patients, so less than 150 patients. And there's a very, very long tail of diseases. These cohorts of children with developmental disorders are very highly heterogeneous. There are hundreds of thousands of different disorders some of which are very hard to distinguish clinically. And what we can also see here, broadly speaking, is in terms of what these genes do and what their proteins do, there are two broad classes of genes. Now, in developmental disorders that we've worked on, about 80 to 90% of the children have neurodevelopmental problems. So intellectual disabilities are by far the most common of all of the traits that we're looking at. And what we find is that in these most commonly diagnosed disorders, almost all of which are caused by new mutations, that the biggest functional contribution is from genes that are involved in gene regulation and so-called chromatin genes that are often expressed very early on in neurodevelopment and throughout neurodevelopment. Whereas a smaller group is involved in the synapse, which is a particular part of the function of the neurons in the brain, and typically are only expressed later on during neurodevelopment. So just broadly speaking, we can see that some of the contributions to these diseases are occurring— are actually probably having an effect very early on in neurodevelopment and some later on in neurodevelopment. Now taking that data set of 30,000 families, just recently we've published a paper identifying strong evidence for 28 novel disorders. And here I'm again, I'm showing in blue where we've looked for genes that look like they're working by that loss of function mechanism, and in red the novel genes that look like they're operating by that kind of alteration of function mechanism. And you can see that actually, as we predicted, that roughly half and half, loss of function, half altered function. So the glass half full approach is that we've identified 28 novel disorders there. That's led to the diagnosis of hundreds of new patients. But despite the massive scale of this study, by far the biggest of its type in the world, we have only nibbled away at that excess of damaging de novo mutations in protein-coding genes, and there's still this large green block of the excess de novo mutations which remain unaccounted for. So we think that despite the scale of this study, we need to get much bigger data sets and develop much better analytical approaches to try and identify these novel disorders that we know are out there. Now, in terms of What else we can learn from these data, we can use the same kind of data that I showed you of sequencing families to understand a little bit more about the nature of that effect of age on the mutation rate. And what I'm showing you here is that as well as an age effect of the fathers, we can also detect that there is also an age effect of the mothers. In the mothers, it's weaker than in the fathers,, but it's clearly having an effect, showing on the kind of right-hand side here. Now what that suggests is because, as I mentioned at the very beginning, the DNA is always being replicated in the father— every, every couple of weeks the DNA is being copied as it goes in, as it turns over in the spermatogonial stem cell— whereas the DNA after birth is not being copied in the mother. It's only existing in the immature oocyte. So the fact that we see an effect of age in the mothers suggests that it's not just due to the copying of DNA that we see these mutations, it's also due to changes of DNA that happen within a cell that's not being copied. For example, damage to the DNA that's not being repaired correctly. So we can start to use this information on both the proportion of patients that have a de novo mutation as a cause, and the effect of age, to look at what the overall public health burden is of developmental disorders caused by new mutations. And we can produce this kind of model here, whereby what we essentially see is over the population as a whole, there's about an average of about 1 in 300 births will have a developmental disorder caused by one of these de novo mutations, one of these new mutations. And that risk goes down if both parents are young, and it goes up if both parents are old. But it goes down to about 1 in 400 if both parents are aged, say, 20, and it goes up to about 1 in 200 if both parents are in their 40s. So although there's quite a considerable age effect for these kinds of mutations caused by single base changes that occur de novo,. It's much less strong age effect than in for Down syndrome, for example, where many people are aware that the risk of having a Down syndrome increases substantially with maternal age, especially after the age of 35. But it actually increases much more rapidly than the age effect here. And indeed, the age effect here, the father's age has a bigger effect than the mother's age. Now if we extrapolate this to, to the whole world, what that's suggesting is just under half a million babies born per year are probably having a developmental disorder caused by one of these new mutations. Now of course that's spread across many hundreds or thousands of different disorders, but it gives you a sense of the overall global public health burden caused by new mutations. So to end, I just want to reflect back onto this journey that DNA takes and reflect on why this journey and understanding where mutations occur are important in a clinical context. And the reason why it's important is because we need to think about what is the risk of a family who has a child with one of these disorders having another child with the same disorder. So there are— the way in which we can think of this is that, is that here we have a child with a mutation. The mutation is present in the zygote and spreads throughout all of the cells in the body, and that was present in one of the gametes, the sperm or the egg, from the parent. Now, it may be that it's present only in one of the gametes, but it could be present in more of the gametes, and that's where it really depends on how many where does it occur here? So if the mutation occurs down here towards the end of this differentiation, it may well only be present in one or a small number of sperm. If, however, the mutation occurs a little bit earlier in development, it may well be found in many more than just one sperm. And if it occurs very early on in the embryo development, it may be found not just in the sperm itself, but also in the other tissues of the body before the split between the primordial germ cells and the cell types that make up the rest of the body. So what we might be also be wanting to see is to find out is where the germline is what's known as mosaic, or rather the germline— there's more— the mutation is present in more than one gamete, and that of course means there's a higher risk of a subsequent child having the same mutation. Or indeed, if the mutation occurred so early on in development, as I mentioned, it's actually found both in the gametes and in the somatic tissues of the individual. It's also possible some of these new mutations we're detecting weren't in the egg or the sperm at all, but actually occurred very early on in the development of the child, such that it then was passed on to sufficient numbers of cells in the child that it was causing disease. So it becomes important to try and distinguish between these different scenarios to try and give counseling to parents about the risk of having a subsequent child with disease. And because typically we're only looking at the blood or saliva DNA that have come from, from parents and children, we're only really able to distinguish these kind of two different scenarios here. So what we did is we took 1,000 disease-causing de novo mutations and then looked at them very, very deeply and sequenced them very, very deeply. Now we might expect that over this 1,000, the recurrence risk of having a child with the same mutation, the same disease, is about 1%. But as I just said, what we suspect is this average number of 1% probably reflects a small number of parents who had a very early mutation and have a much higher risk of having a subsequent child, and a small number of children who had an early embryonic mutation in the child, and because the mutation was never there in the sperm and the egg, the risk is essentially zero for subsequent children. So when we took this approach and looked about across these 1,000, we found that about 2% of the time the disease had incurred sufficiently early on in the parent that we could see it in the in the saliva of the parent as well, and so they're at much higher risk of recurrence of disease. So it's important to identify these parents. And then for about 4% of the, of the children with a new mutation, the mutation only actually occurred during the embryonic development of the child, so their recurrence risk is essentially zero. And that means for the remainder, on average, the recurrence risk has gone down a bit. The average recurrence risk has gone down a bit. So therefore, this is not just of academic interest to understand how this flow of DNA from one generation to the next, but of clinical interest to understand which mutations, which mutations occur where during this passage of DNA. So where do we need to go next? Well, hopefully I've given you a sense that this understanding of the mutation process has really helped us to identify and understand new severe developmental disorders. But it's clear that many, indeed most of the families we work with, have not yet got a clear genetic cause for their child's condition. So we need to continue to discover these novel disorders. Now, as the kind of sequencing that I've described to you is moved from the kind of research domain where the Sanger Institute operates into healthcare, which is a fantastic thing. But it does mean that to get together the scale of datasets that we really need to bring together to identify these new disorders, we're going to have to combine data from both research studies like the DDD study with the routine clinical testing that's being done throughout the world. And that creates logistic and ethical challenges that we need to kind of embrace and tackle. The second challenge that we have is that when we discover a new developmental disorder-associated gene, we still have a— there's still a challenge that how do we interpret which variants within that gene can actually cause disease, because the majority of variants typically do not cause disease. And that means that there's a risk of misdiagnosis and of missed diagnoses, and we think there are real opportunities for using new genome editing techniques in cells that'll enable us to kind of essentially go, as shown on the right-hand side, from this position that we're typically in with a newly discovered developmental disorder where we know a few pathogenic variants on the bottom, a few benign variants on the top, but what we really need to go is to a comprehensive map of all of the variants in a gene that then gives support to clinicians to interpret these variants they see in patients. As accurately as possible. So we think the construction of these variant effect maps will really transform how the diagnosis of these disorders happens, even after we've discovered the disorders in the first place. And the third real challenge that we have is within these rare disorders, or even just one rare disorder, there's often a huge variation in disease severity, and we don't understand that disease severity. And that creates challenges for providing information for families as to the likely prognosis of their child's condition. But it also potentially presents an opportunity, because if we can identify individuals who are unusually resilient to the mutation that's occurred, and we can understand why that they are resilient to that, then we can potentially identify strategies that will inform on how we might develop therapies for those individuals where the disease is more severe. So this is going to require a lot of working closely with families to really understand how to characterize disease severity, potentially to start to bring together different resources that we don't currently have access to and really involve the families more in our research. So with that kind of future perspective on where we need to go next, It just remains for me to really thank the key members of my group, many of whom have contributed to this in lots of different areas. This is 10 years of work that I've described to you, especially my colleagues on the DDD study with the clinical leads Helen Firth and David Fitzpatrick, the Decipher team, our collaborators in GeneDx and in Nijmegen at Radboud University Medical Center. And the PAGE study that I mentioned on the prenatal ultrasound anomalies. And so with that, I'd be very happy to take any questions or comments that you might have.