Dr Todd Lencz is a Professor at the Feinstein Institutes for Medical Research in Manhasset, New York. Lencz’s team were recently awarded a $2.9 million grant from the National Institutes of Health’s (NIH) National Human Genome Research Institute (NHGRI) to fund research into the use and ethics of polygenic embryo screening (PES). Reproductive genetics has been used for decades to identify rare, serious genetic disorders like Down syndrome. Now, PES has the potential to evaluate embryos for more common diseases like diabetes and schizophrenia. Lencz and his team will conduct research that will provide much-needed insight into this novel approach.
Please note the transcript has been edited for brevity and clarity.
FLG: Hello, everyone and hello, Todd. Welcome to the latest interview in the ‘A Spotlight On’ series. Today, we’re going to be talking about polygenic embryo screening, which is going to be really, really interesting, because it’s a topic that I’m unfamiliar with. But I know that there’s loads of social and ethical issues that we can delve into. So, before we begin, Todd, if you can just introduce yourself and tell everyone a little bit about what you do as well.
Todd: Yes. Hi. Thanks for having me. It’s a pleasure to be here. My name is Todd Lencz, and I’m a geneticist working at Northwell Health at the Feinstein Institutes for Medical Research and the Zucker School of Medicine at Hofstra Northwell. I study a variety of different genetic topics. But in the context of this study, we are very interested in learning more about how polygenic risk scores are something that might be used in the context of in vitro fertilisation and screening of embryos – I can explain all of the terms that I threw into that sentence as we go along.
FLG: Yeah, I think it’d be good to start off with embryo screening in general. I know it differs worldwide, but what is the current landscape of embryo screening? What is currently offered to people and when are certain things offered to couples?
Todd: Sure. So, in the process of in vitro fertilisation, a number of embryos are created for potential implantation. Depending on the health and age of the parents, there may be varying numbers of embryos available (generally more than one is available), and different forms of genetic screening might be employed to determine which one to implant. There are two main forms of genetic screening that are currently utilised, and these are monogenic disease screening. So, if one or both parents are known to carry specific genetic variations that might cause a genetic rare disease, then these can be screened for, and embryos that carrying that particular variant can be removed from consideration for potential implantation. So that’s a fairly straightforward process. But at the time that it was developed 30 years ago, it was not considered straightforward – it was considered potentially eugenics or ‘playing God’. Another form of screening that is sometimes employed is looking at whether there are any large-scale chromosomal abnormalities within the embryo. That’s a little bit controversial because the accuracy of these methods is not currently well established.
But the process that I’m studying now is a brand-new concept that has only been recently introduced into the clinic – it’s the idea of using polygenic risk scores. These are very different than looking for an individual gene that is potentially related to a genetic condition like Tay Sachs or cystic fibrosis, which are monogenic diseases caused by a single gene that is well known, well identified and relatively easily screened for. By contrast, most common diseases (like mental illnesses e.g., schizophrenia, or common conditions like type two diabetes, or breast cancer, or prostate cancer) are caused by many, many genes throughout the human genome. And the risk for these is very complicated. But more recently, these polygenic risk scores have been developed, which sort of add up or sum up all of these small, tiny influences across many, many genes throughout the genome. And the idea of using these probabilistic risk scores, which add up all of these tiny genetic variations, to give you a probability of whether one embryo might be more or less likely to develop a common disease like schizophrenia, or type two diabetes, is what we’re currently interested in investigating.
FLG: You mentioned polygenic risk score – how is this currently being applied?
Todd: Well, they’re very new. And so, the application in the clinic is quite experimental. At this point, there’s no clear protocol for how they should be applied. And what we’re still trying to really understand is how much of the risk are they able to capture? And how accurate are they in predicting the future? And the fact is that, right now, they’re still only capturing about 5-10%, of all of the variation that we can possibly explain in the future risk of the disease. So, there’s still 90-95% of the variance that’s unexplained. So, these are only really capturing just the tip of the iceberg. And the way that they’ve been investigated so far in adult patients, is to consider adult patients for possible future screening. So, there are studies ongoing that look very promising, for example, in various forms of cancer, like prostate and breast cancer – if you have a high polygenic risk score, maybe you should have more mammograms or a different type of prostate cancer screening protocol or more frequent check-ups, this kind of thing. Rather than saying you are going to definitely develop cancer, which is out of the question and never will be the case with these types of risk scores, not even in the future. That’s the nature of the genetic contribution to these diseases, they’re so complex. And so much of them is not determined by genetics, that it’s never going to be that the polygenic risk score will be able to pick somebody out or pick one embryo out of a dish. Let’s say there are five or seven in an IVF cycle – they’ll never be able to pick out with 100% certainty, or even anywhere close to that level of accuracy.
So, there are a few companies that are right now offering these services to detect relative risk of one embryo versus another in an IVF setting. So, if a potential family has five or eight, or some number of embryos available in the IVF setting, and one could be implanted, the idea is to take a single cell out of the embryo, screen it for all of these minor tiny genetic variations, add up these scores and determine which one has higher risk for disease X, Y or Z. While that’s currently being offered in the clinic by a couple of companies, in general, this is still something that is very poorly understood of how accurate it might be. And like I say, even in the situation of the adult clinic, there’s no standardised protocol or standard of care where this is very well understood or widely accepted.
FLG: So essentially, polygenic embryo screening is applying polygenic risk scores to embryos during IVF. What is the value of this technique now and how has it been made possible with advanced technologies?
Todd: Well, that’s exactly what my research is all about. We’re trying to understand exactly what are the statistical properties of polygenic risk scores, how accurate are they currently and how accurate might they be in a hypothetical future if we had more knowledge of the genome alongside more refined polygenic risk scores. So right now, we have studied two different situations. First, we have studied polygenic embryo screening in the context of what we call ‘quantitative traits’, like height or intelligence, things that can be measured on a scale of higher to lower. And what we find is that these scores are very inaccurate and not highly useful. We published that paper two years ago and what we showed is that, for example, if you had a batch of five or ten embryos (however many you were choosing from), you could run a polygenic risk score, let’s say for height. And you could identify an embryo that might be a couple of centimetres taller, on average, than one that might have just been selected at random. But there is a great amount of variability around that for any individual couple. So, I could tell you: ‘On average, yes, we can select an embryo that’s going to be two centimetres/two and a half centimetre taller, than if we hadn’t selected just at random’. But that’s plus or minus eight centimetres. In other words, there’s a 25% chance that the one that we select and say is likely to be taller, will actually be shorter than the average child. And we were able to test for this in a unique cohort of families that were very large with 10 or more children – and these were all grown children now in adulthood. We were able to imagine, using these 10 children who are now grown adults and have known heights, how much the polygenic risk score actually matches their actual height. And if these were 10 embryos, and we had selected just one, is this the tallest member of the family? Most of the time it wasn’t.
Now, there’s a second version of polygenic embryo screening, which is for diseases. What is your risk of developing a particular disease? Like I say, type two diabetes risk or schizophrenia – it could be anything that you can think of. Any common disease that is not like a genetic disease of birth (like Tay Sachs) is what we call a common complex disease. And at this point, we’ve looked at it and the situation is a little bit more complex. Here, it really depends on many, many variables that haven’t yet been well articulated by the couple of companies that are currently offering these services and that are not well understood by the public, or even the scientific community. This involves something called the selection strategy. How are you going to select when you have many different diseases to choose from? Let’s say this embryo over here has a high-risk score for diabetes, but a low-risk score for major depression. And let’s say another embryo over here has a high-risk score for Alzheimer’s, but an average-risk score for everything else. How do you make that decision? Are you just going to exclude embryos that have very, very high-risk scores? Are you going to look for one that has the lowest risk score? Are you going to try to combine across multiple diseases? These are all issues that we are currently studying right now. And that’s really the essence of my research. So, there’s no clear answer, as of yet.
There is one other thing that we’re really very interested in studying and in communicating, and that is very complex to communicate. Let’s imagine that you can reduce the risk of a given disease by even 40 or 50% and we think it’s possible that relative risk reductions may be more feasible than we originally anticipated with polygenic embryo screening. The risk for schizophrenia, for example, in the general population is 1%. So, if I tell you, there’s a relative risk reduction of 50%, you might say, ‘That’s great, that’s a huge risk reduction’. But if I tell you that there’s an absolute risk reduction, meaning that the absolute chance of developing schizophrenia is reduced by 0.5%. So, you go from a 99% chance of not developing schizophrenia to a 99.5% chance of not developing schizophrenia. Now, that sounds very different. So, we get into issues not only of statistical genetics, but also of scientific communication and understanding of risk. And that’s the other piece of the research that we’re currently doing, which is trying to understand if a doctor says something about risk, how does the patient hear that? What is the difference between what we think we’re communicating in terms of risk reduction, which is a statistical concept, and what the patient hears in the clinic, and how they make a clinical decision with their doctor. The bottom line of all of this is we feel that a lot more research really needs to be done on both ends of the scale – both the statistical side and the communication side, in order to really understand how this could possibly be applied.
FLG: Yeah, definitely, there’s so many points from what you just said that I want to touch upon. One of the first things you mentioned is that there are companies that are offering these services. In what capacity are these services being offered? And how is this currently regulated?
Todd: Yeah, they’re currently offering it privately, either direct to consumer or direct to doctor. And it’s really depends on whether the patient has heard of it, or whether the doctor has heard of it. It’s a brand-new concept. So, I think, in most cases, it’s not something that people are going to run across. But if they do run across it, it’s not regulated, or certainly in the United States, it’s not at all regulated. Regulations vary dramatically by country. And the whole landscape of how this would work is quite different in every country in the world. So, in the United States, for example, there’s absolutely no regulation. And there’s also no government reimbursement. And so, this is really something that a private individual might pay for if they can afford it. And that brings up all sorts of issues of equity and social justice and so forth. But that’s, of course, part of the larger healthcare debate. In terms of other countries, for example, in the UK, the entire process of embryo screening in IVF is highly regulated. And so, in the UK, there’s a government commission, which directly dictates not only which conditions can be tested (only monogenic conditions can be tested), but it actually dictates which specific genetic variations are allowed to be tested. And they’re sort of on a list or a menu that is specifically dictated by this government agency. And so, my sense – and not only am I not a lawyer, but I’m also certainly not a British lawyer – is that this would not be even allowed in Britain unless there was some sort of loophole. Because it’s not monogenic screening, maybe there would be some kind of loophole. Those sort of represent the two ends of the spectrum of how this is regulated. In Germany, there are individual review boards that have to approve individual screenings. It really varies greatly.
FLG: I think that polygenic risk scores themselves are not refined very well. As you said at the start, they’re still only capturing 5 to 10% of the variation. So, how are we able to then apply this to embryos?
Todd: Absolutely. In the realm of clinical applications of polygenic risk scores in adults, there’s currently a working group that just published a paper in Nature and just called for a process in the same way that the governing bodies tried to regulate or create at least standards of care for monogenic screening. There’s a desire to create a process for standardisation of how polygenic risk scores are reported in the research literature. One thing that’s very important that we haven’t even touched on, is that polygenic risk scores are only as good as the research base that they are drawn from. In order to develop a polygenic risk score, we have to understand all of these tiny genetic components across the genome, and how they contribute to risk. Those are based on large scale, genome-wide association studies that get published in the literature, which can vary in quality depending on the size, or particularly of the cohort that is under the study. So, if you have a GWAS that looked at 10,000 patients, it’s going to probably be okay. But if you have a GWAS that has 100,000 patients, it’s going to be much more accurate. But it’s very critical to understand who are those patients, what is their exact diagnosis, and most importantly, what is their ethnicity and population ancestry?
And in most cases, polygenic risk scores have been developed for individuals of European ancestry. And that’s a big issue that’s been hotly discussed in the field. There needs to be much more genetic studies in other ancestries, and particularly ancestries that have been poorly studied in the literature so far, such as individuals of African ancestry. They are going to be largely left out of polygenic risk score clinical applications because they’ve been largely left out of genome-wide association studies up to this time. There are massive efforts now to try to collect these samples for genome-wide association studies to try and play catch up here. But this really needs to be understood, even for adult clinical applications of polygenic risk scores. So, my urgent call to the field, is that governing bodies, research societies and so forth all get together and try to understand and develop standards for reporting before we can even get to identifying standards for utilisation and application of polygenic risk scores in the clinic and in the IVF clinic, in particular.
FLG: Yeah, that was one of the things that I was thinking as well. Because, at the moment, polygenic risk scores can’t really be applied to other populations other than Europeans because of the data. And you’d hate for things to progress for some but not for others.
Todd: And right now that’s the case. It’s very unevenly distributed. And some would suggest unfairly so. And that would be a very big limitation for a lot of people who just certainly, at this point, would be ineligible because the data don’t exist for them.
FLG: What are the other conditions that are being explored in this context? Because, as you mentioned, schizophrenia is such a complex disease, and we still don’t fully understand the genetics of it. So, how is that then going to be applied in embryo screening when we still don’t really understand it?
Todd: I mention schizophrenia frequently because that’s my original and primary area of research. But even diseases that are perhaps a little bit more well understood, like diabetes and so forth – the amount of genetic variation that we can explain is still quite limited. So even if we understand the physiology of disease, what we understand at the genetic level is not much greater. So again, a difference with utilising polygenic risk scores in the adult clinic, is that we can utilise all of the additional clinical information that we have and that we understand. So, let’s say, there’s a 50-year-old male who is non-smoker but is overweight and has cholesterol numbers of X, Y, and Z, and we want to understand his potential risk for a heart attack in the next 10 years. We can certainly imagine, at least, a risk calculator that includes smoking status, the cholesterol numbers, all of those things, and in addition throws in a polygenic risk score to help refine and increase the accuracy of a risk calculation. That’s something that’s currently being studied today. And where we can use what we already know medically and biologically about heart disease to throw into the risk calculation.
When we’re dealing with an embryo, we have no other clinical information to go on. We don’t know if that embryo is a smoker – I hope it isn’t a smoker yet! So, all we would have is the risk score, all we would have is the genetics. So that, again, that limits us to just the bare essentials of what we can learn from genome-wide association studies. Even within the realm of genetics, things like rare variation, or ultra-rare variation, which is something I also study in a separate part of my research, are left out. Currently, what we capture in polygenic risk scores is common variation, which is variation that is spread throughout the general population at a frequency of genetic variants that are in 1% or 5% or more of the general population. Whereas a lot of the action, in terms of disease risk, happens at the realm of very rare or even private genetic variation that occurs de novo when the embryo was formed. And that will never be incorporated into a polygenic risk score.
FLG: Another aspect of it as well is with monogenic disorders. We’re still not amazing at diagnosing these either. Some people go into their 30s or 40s and they still don’t have a diagnosis, and these are rare monogenic disorders. So, should we not focus on these where we potentially know more information?
Todd: So that’s sort of like the UK model that I was mentioning before. That is sort of like looking for your keys where the light is under the lamppost. There are certain things that are well understood about monogenic disorders and those can be clearly specified. But it is true that there is still a large amount that is unknown about monogenic disorders, and there are many patients that, you’re right, have a very complex clinical picture and they still can’t figure out what the underlying genetics are. So, it is certainly not the case that monogenic disorders are simple. But what we can say is that certain components of monogenic conditions are currently well enough understood that they are actionable. So, I think there we would get into the issue of clinical actionability. And I think that’s sort of where the rubber meets the road in the clinic – what can science tell the clinician and tell the patient with a high enough degree of certainty that it is clinically actionable? And so, like in the UK, they take a cautious or conservative approach – where we have a high degree of certainty, they consider that to be clinically actionable. And in the US, that’s sort of turned on its head – whatever you can do and whatever you can get away with in the clinic, that’s the action that you take. And it’s sort of a libertarian approach of caveat emptor.
FLG: Say if tomorrow, we perfect this technology and it’s really accurate. When do we start to have the discussions about where we draw the line? Because obviously, you were talking about mental health and things like that. Where do we draw the line on what diseases we would incorporate into embryo screening? When should we start having those conversations? Because I feel polygenic embryo screening for me has come out of nowhere.
Todd: Yeah, sorry to interrupt. The reason I interrupted is that we need to have those conversations yesterday! That’s when we need to have those conversations. I mean, absolutely, there’s so many implications at the social level and at the ethical level. The word eugenics gets thrown around, and that word sometimes can shed more heat than light – even if you take away that word and you don’t worry about it being utilised by government to do wicked things, there still could be potentially wicked implications in society if we screen out certain conditions. Then that adds a stigma potentially to individuals who are still carrying those conditions. This is something that has already been discussed in the disability community with respect to monogenic disorders as well as chromosomal abnormalities such as Down syndrome. This has been discussed in many countries. 90 to 95% of families that receive a prenatal diagnosis of Down syndrome choose to terminate the pregnancy and the Down syndrome community has discussed what the implications are for those who are born with Down syndrome and what that means in terms of stigma.
And in this field the movie Gattaca is often talked about. And I think, even though it’s a science fiction movie, and there’s a lot that is fanciful about that movie, the idea of someone who is born with conditions that some might perceive to be preventable brings about a potential for stigma or shame that absolutely needs to be addressed at the level of societal and ethical discussions and needs to be addressed today. I mean, you even see it with COVID. People sometimes say, ‘Well, I tested positive for COVID.’, and other says, ‘Well, what did you do?’ There is a stigma on the person who tested positive for COVID. Was it their fault somehow? So, absolutely. This raises the questions that you’re posing, Shannon, yeah.
FLG: I feel mental health is already such a taboo subject. And over the years, it has gotten a little bit better – people feel they can speak out. But you would hate, if we rolled this out and you screened to remove schizophrenia or depression, that this would create a further divide and people then don’t want to speak out.
Todd: I’m really glad that you mentioned that in particular, because mental disorders and psychiatric diagnoses often are more stigmatising than other forms of medical illness. 50 years ago, the word cancer was unspeakable and people called it the C word and so forth. And that stigma has slowly eroded over time. But in mental illness, there’s still a very large societal stigma around them. And I agree – I think I would be remiss if I failed to mention the fact that the concept of eugenics not only arose in the context of mental illness and in the context of intelligence and trying to breed out of the population individuals who were at the time, 100 years ago, referred to as ‘mentally defective’. Not only did eugenics emerge out of the fields of mental illness and mental issues, but the entire field of genetics itself actually emerged in some ways out of this, certainly the field of statistical genetics. All of the founding fathers of statistical genetics were actually eugenicists who were attempting to breed out mental illness and low intelligence out of the population 100 years ago or more. And this is something that really needs to be grappled with and reckoned with by our field and any of us who consider ourselves geneticists. This is not incidental to the history of genetics or a misapplication of genetics, but it is part of the origin story of the field of genetics. And in order to avoid those horrible and horrific implications, we need to address those head on. And again, in conversations that should have started yesterday, before we proceed any further in understanding how to apply these.
FLG: Yeah, because as we mentioned, there are commercial companies who are already doing this. And I feel like the conversations have not really happened. You’d hate for it to go any further without having these important conversations. How do you think we can ensure that there is transparency and engagement with the general population within these discussions?
Todd: Thank you for asking that. The burden is really on us as a scientific community. I’m struck by the fact that the clinical and ethical implications of CRISPR genome editing, which are far more complicated to implement in the clinic, have been much more broadly discussed in the scientific community, in the general population and by the political regulators and so forth. When the so called CRISPR babies were born a few years ago in China, that was literally almost earth-shattering news. And it sort of stopped the world, scientifically speaking, and urgent sets of regulatory decisions were made. And we feel that it’s up to the scientific community to do better, with respect to polygenic embryo screening.
In one of our recent papers, we just did a quick literature search in PubMed and we saw that hundreds of papers had already been published on CRISPR embryo editing. And literally fewer than a dozen papers on polygenic embryo screening had been published. And if you screen the PubMed search even further, where you force the word ethics into the search, you still get 100 or more hits in the CRISPR literature, and you get literally one hit in the polygenic embryo screening literature, and that was our own paper. Now, that’s gotten a little bit better, there’s a few hits now, since we published that. But literally, it’s two orders of magnitude difference. And that’s on us as scientists to do better and to raise the awareness in the general public. And that’s why I thank you so much for hosting me here and getting an opportunity to address your audience today.
FLG: I definitely feel like it has flown under the radar. When I found out, I started having these ethical discussions within my own household. So, just getting it out there is so important to try and kickstart these conversations.
Todd: Yes. And the more that I learned about it, and I think about this every day now, my thoughts and my understanding of the ethical implications have evolved tremendously as well and are still evolving now. So, I think there are very easy knee jerk reactions that one might have on either side of the discussion. I think that it just takes a discussion to really get at the nuances and the subtleties, and it takes scientific research to provide data to really inform those conversations. So that’s why we were glad to recently be funded by the National Institutes of Health and specifically the National Human Genome Research Institute in the US to do more research to be able to inform those discussions.
FLG: I have been writing pieces about how genomics is being applied in different countries and the different capabilities around the world. If this gets rolled out, what impact do you think this will have on the disparities that already exist? Because in many countries, they only screen for three or four conditions, compared to the UK and US who screen for many more. Could this potentially have a knock-on effect on the already existing disparities?
Todd: It certainly would. The phenomenon of the rich getting richer, would certainly apply here. This process would not be cheap. So, I think how much of a benefit somebody would derive from that is still up for debate or for research. It’s possible that someone might be paying a lot for a very modest amount of return. But there’s no question that because it would be expensive, so it would feed into all of that. It also raises a whole issue of who is seeking IVF. A lot of people who seek IVF now are individuals with fertility issues, where even getting one healthy and implantable embryo is hard to do. And if a baby is born, then issues of what they’re polygenic risk score is secondary or not even on the table. But if this raises a whole new set of people who seek IVF, specifically in order to obtain these scores – that really changes the game. And so that’s something that we need to understand as well – not only how would it change for current patients, but for future patients who aren’t even currently considered patients.
FLG: Yeah, I think that is going to raise a whole other set of ethical questions because IVF is already really expensive. And then this could be going down the route of – I don’t want to say – designer babies and things like that, with the rich being able to do it and other people not.
Todd: Right, because at the technical level, it’s not cheap, but it’s not prohibitively expensive to do this if you’ve already started the process of IVF. If you’re seeking to start the process of IVF, just in order to do this, then that raises a whole other set of issues. And it also raises potential risks that are also not well understood. What are the risks of IVF to someone who is otherwise healthy and who otherwise might conceive without IVF? It’s still not well known what are the potential health risks of doing IVF, taking embryos, pulling cells out for genetic testing, and then implanting the embryo again and delivering that baby, compared to just the natural process of conception and birth. Could there be hidden risks, either to the mother or to the baby, in just undergoing that set of procedures, even if you can afford it? So that needs to be weighed in the whole cost benefit analysis as well?
FLG: Yeah. We also don’t know what impact it will have on evolution. But that’s a whole other discussion! What do you think the future of this area will look like?
Todd: Well, one concern that I have if you want to get into the issue that you raised with evolution (I think evolution itself is a little far down the road), is that if different societies or different nations don’t agree on how this should be utilised, or if there were perceived to be sort of a genomic arms race across nations. That might sound a little fanciful or a little farfetched, but that would probably be a source of considerable concern. But in the short run, I think that my interest as a scientist, is just getting as much understanding and information as possible and understanding how to communicate that to doctors and to patients. I think the biggest risk, and I think we see this again with COVID, is that developing scientific knowledge is hard, but communicating it is even harder. And that’s why companies such as yours are so important. And here we are 18 months into the pandemic, and I think if you took 10 people on the street and asked them to just explain how COVID works in layman’s terms – I’m not talking about getting into a deep molecular understanding of it – but even just in layman’s terms, how do you catch COVID? I don’t know how many out of 10 would be able to explain it to you in ways that makes sense. And that’s our failing. And we have to do better.
FLG: Yeah, definitely. Communication is just so important. Thank you so much for joining me today, Todd, it’s been really interesting. I could talk to you for the rest of the day about this, but I think this just highlights not only how far we’ve come in genomics in terms of progression and advancements, but also how important as we said it is to communicate with people and get these important discussions going. So, thank you so much for joining me. It’s been great.
Todd: Thank you so much for hosting me. I’ve greatly enjoyed the conversation.
FLG: Thank you.