Genetics Podcast -74
===

[00:00:00] Patrick: Hi, everyone. Welcome to the genetics podcast. For this last episode, before the holidays in 2021, we're bringing you a special episode. That's a crossover episode with Oleksandr from the personalized medicine podcast. In this episode, we're going to talk through 6 major developments in 2021, Oleksandr and I each brought three to discuss. These range from cancer testing, newborn sequencing, AI, and machine learning.

We're going to talk about a number of different things. It was really great. Oleksandr is a super smart guy. He knows lots about personalized medicine and I learned a ton through the episode. Hope you enjoy it. If you have any feedback on the episode or on the podcast as a whole, as always, please feel free to email us at podcast@sanogenetics.com. If you're enjoying the podcast, please just share it with a friend. That's a great way for us to spread and get more people listening. You can also leave a review on one of your favorite podcasts players without further ado. I'm going to turn it over to the episode and I really hope you enjoy.

Thanks again for listening and happy holidays.

So excited to do this crossover episode, we're going to get right into it. I'm going to ask Oleksandr actually to just give a quick intro to his, it's not necessarily his number one. We haven't ranked these in any order, but number one on his list of big developments in 2021. And I'm going to ask a few questions and we'll kick the discussion off over to you, Oleksandr.

[00:01:22] Oleksandr: I think that that was an interesting exercise, right? To just reflect back on the year and try to see what were the main developments, the main highlights of of the year that his passion for personalized medicine for genetics for the broader field of precision medicine. And I think the one development that I think we both want to pick for this year would be the launch of the massive program, the massive screen in program for cancer by grail and NHS in UK so-called gallery or Gillary test and which aims to detect cancer, essentially presymptomatically. So I think what is very unique and what is very special about this program, that for the first time, That will be a really large trial.

I think about 140,000 patients will be recruited into it over the course of next few years, to determine essentially whether with the liquid biopsy test, we can detect cancer before the symptoms actually kick in. And If that, if that works out, I think we will probably look back at this as a watershed moment in the history.

Right. We don't know if it will work out. We don't know what the efficiency of this test would be, but shell, it worked out that would be a major game changer because if we can detect cancer early, obviously their spectrum of treatments that is just available for, for those patients would be completely different.

Right. And then. Possibility of positive outcomes, it will be much, much higher. So I think that is for me, for sure. One of the, one of the top developments in this year. And I'm very, very curious to see how it will go and how it will develop over the course of next few years. 

[00:03:13] Patrick: Couldn't agree more. I wonder if maybe you could break down what the bear cases or those who suspect this won't work. If it doesn't work, why, why wouldn't it work? And, and what could be the stumbling block for this? From your perspective? 

[00:03:29] Oleksandr: Yeah. Good question. I think, well, to be on the skeptical side, right? This might not work for, for several reasons. I think the. My biggest fear, at least that it might be not specific or not sensitive enough. Right. So what grail has proven before that they are actually able to detect cancer based on the cell free DNA floating in the blood from those cancer cells, but they could do it in the patients that have already manifested symptoms.

Right? So cancer in that group of patients was already diagnosed by other methods. Now the big challenge at the big question, Is do we have a sufficient quantity of that cell-free DNA in the, in those patients that can be detected for manifestation of any symptoms. And, and that's, I think the biggest question mark, right?

If, if it's, if it's still the case, if we can detect that cell-free DNA, if it is there, then it's great. And I think if, if that barrier is overcome, everything else should fall into place. And then another point maybe that I see that is more on the social ethical spectrum of of things is will patients like to know about possibly having cancer or not in advance, right?

If the test is sensitive, but maybe it's not specific enough, let's say to returns false positive results and not interbank do 30% of the cases, then we start asking those ethical questions. So shall we stress out patients that may be not having cancer in the first place with those false positive results?

I'm curious to see how this will develop. 

[00:05:11] Patrick: Yeah. Do you have any sense in the trial how, how deep they're going to measure those sorts of things? Because this is one of the things that struck me as a big challenge of it is that they're probably measuring things like healthcare utilization from an economic perspective to say when those false positives come in and they will, what is the cost of all that additional screening? Because this has been discussed for a long time in germline breast cancer and, and you know, other kind of inherited cancer screening, but it strikes me as a really tricky thing to measure comprehensively.

Do you have any, any sense of how they're measuring that or what what the end points besides, you know, false, positive, false, negative, but to, to measure actually the economic impact as a whole. 

[00:05:54] Oleksandr: Yeah, to be honest, I don't know that in detail and how, how grail wants to test this. So I would hope that at least since they are doing this in collaboration with NHS, they should be addressing those issues.

And given just the scale of the trial and the number of patients that they plan to recruit, that would be really important. I think what they are doing right, at least from, from the outside perspective is that they are really focusing that on the patients that previously haven't been diagnosed with cancer.

I think that enrollment criteria is the patients that haven't been diagnosed with cancer for at least three years prior to, to inclusion to the trial. And that also don't show any symptoms that might be related to any type of cancer. But I agree with you, that would be really interesting to see how they can actually define that.

The status will have economic and health benefit to the patient and to the healthcare system. 

[00:06:49] Patrick: Yeah, I'll definitely be following a closer one of the things that struck me at the time, this was announced about the coverage was, was also, there were a lot of people on taking the negative angle on this about is, is this where the NHS should be spending?

It's scarce resources. And I I'm, I'm obviously like you quite a techno optimist and like this kind of stuff. And one of the stories developments we're going to talk to is probably in the same kind of categories this, but I did think that was an interesting take where there were some people saying, you know, should, should we not allocate these resources to many of the more myriad immediate term problems, but, but obviously there's a case to be made on the other side, which is that this is an enormous and growing challenge that requires a moonshot type solution or set of thinking, at least. So I'm, I'm unbalanced pro running these kinds of experiments in the NHS, but it's just, it's just interesting, good to see that they are stepping forward and taking these, these big swings that not everybody necessarily is.

[00:07:50] Oleksandr: Yeah, definitely. And Patrick, you know, this better, and maybe I have actually a question to you in this regard. Why do you think UK is on the forefront of these things? And it's not just about this program. I think when we look at COVID and let's say the genetic screening the genomics screen that has been done in UK, it's pretty much unparalleled to what has been done in any other developed countries.

So what is special about your case of why you are so successful in pushing those innovative programs forward? 

[00:08:18] Patrick: Yeah, it is a really good question. I think. And I'm, I'm in no way an expert on this, but from what I can see living here, it's seen very much as actually a big economic engine of the country. And so they put out the UK puts out industrial strategy top down, focus on.

Personalized medicine, genetics. So it's actually not just seen through the lens of how can we use this to better health, but it's seen very much as, as part of the UK, is USP a unique, you know, unique selling point or value proposition on a global scale. We should be at the forefront of this. So I think that's a big driver behind it.

And it's, it's the reason why they do a lot of these things. I think there's also some serendipity in that. David Cameron had a, I had a child that was affected, I believe by brain cancer. And personally was very motivated to put a lot of money into genomics when it was quite early in its development and, and funded programs like the a hundred thousand genomes program. And I think some of these early bets, you know, as, as you know, have compounding effects where if the decision is made right or wrongly at that time, then you go down that path because, because you've made that decision. So I think a couple of really good decisions, 5, 6, 7, 10 years ago.

We actually, we had the CEO, and I'm one of the founders that UK biobank on the podcast and they, they started, they started almost 20 years ago, I think actually more than 20 years ago now. And only about 10 years ago, did they get large scale funding to do, to do genomic sequencing and, and first sheet of typing and then sequencing?

So a lot of it has been many years in the making, but then big top-down decisions from the government and and other sources to fund these large projects. Perfect. 

[00:10:03] Oleksandr: Perfect. So I'm curious, what's what's on your list. What would you is as one of the biggest developments of 2020? 

[00:10:09] Patrick: Yeah, this is another NHS one where it's not going to be hopefully to UK centric.

But this is a good follow up on this. This has been recently announced that the NHS in genomics thing that are working together on a very large scale newborn, whole genome sequencing screening program. And, and I hesitated on this a little bit because there are many countries that have been. Doing it in large large-scale testing for newborns has been expanding.

But this is the first time that we've seen whole genome sequencing announced on such a scale. There have been programs. So this is a big program, baby seek out of Harvard in the U S that has done a lot of amazing work on what parents expect to receive back the ethical implications and the psychological implications, also the health implications.

But again, I think it's just so powerful when an institution like the NHS gets involved in these kinds of things, because if it's successful, There's a way for it to actually be translated across an entire population. So I thought that's what was so interesting and special about this. They're having announced too many details about exactly what they'll be screening for, but it's, it's looking to be a, you know, a significant range of initially I suspect rare, inherited severe and treatable probably disorders that there'll be screening for, but it gives a platform to start to test new things you know, and hopefully roll them out on a national scale in the UK. And I imagine others will follow suit as as the program has been successful. So I thought that was a really interesting and, and pretty big development that I, I certainly wasn't expecting, actually, if you would ask me at the start of the year, 

[00:11:47] Oleksandr: Yeah, definitely. And do not, if the idea would be just to do the whole genome sequencing, exome sequencing, is there sequencing of specific, specific last year that are related to specific genetic inborn disorders? So what is the. The end game there. 

[00:12:05] Patrick: Yeah. So they're doing whole genome sequencing. They, and they'll basically be applying an in silico panels.

So they're going to sequence the whole genome, but they're not, I actually can tell you, they're certainly not going to report on things like APOE four status that might tell you about Alzheimer's risk. They will report on I'm sure. Things like SMA and many other they'll report on. I'm not sure exactly what the flow will go.

Probably these. These participants will already have gone through a heel prick test and, and come through without any considerations or issues there, but they haven't actually announced what they're going to test for, but they're going to essentially, in theory, be able to test for anything that a whole genome can cover.

It's really just a question about what they actually will report on it. And I understand they've made the decision to not report on late, later in life. Things obviously APOE 4 is a particularly challenging one to discuss, but even some that may be a little bit more straightforward, like, like BRACA.

I imagine they will not be reporting on those, given that it's a 30 to 50 year timeline before any of those children or their parents could, could make any action prophylactic action on that. 

[00:13:21] Oleksandr: And I guess in this case, The storage of the data will remain with the authorities who will do the testing.

Right. So overall patients in theory, over the long-term get access to their full genome sequencing. 

[00:13:39] Patrick: Yeah. So in, in theory, absolutely. My understanding is it'll become part of genomics England's library. They actually used the metaphor, which I think is great. They call it a reading library. So. A lending library is somewhere where you can take books out with you.

But a reading library is somewhere where you have to go and read the books there. So they have a trusted research environment where academic researchers, pharma researchers can go in and analyze the data. And, and I, I, my understanding is that participants will consent to be part of this program. In the UK, you can make a subject access request. I believe it's called and actually request access to your data. But I think it's a really operationally challenging thing for them to do because it's such a large file. I don't think they love to get tens, tens, hundreds, thousands of of these requests. But I do think there will be a day, certainly within the next five to 10 years, where, where participants will have access to the data, because it's so unique where you only need to test once and you can have a lifetime of interpretation. I think it will absolutely go that way. 

[00:14:41] Oleksandr: Perfect. Yeah. And what do you think how this can improve our understanding of some of the rare diseases that are maybe even more than let's say the average road diseases that we speak about?

Like a SMA, a something that happens maybe in one in a million cases. What, what type of developments would you expect to see happening on that front? 

[00:15:01] Patrick: Yeah, it's a, it's a really good question where one of the things that I think is interesting about this is it's actually very tricky to do. Kind of truly population representative large-scale genetics cohorts, almost all of them are really heavily ascertained.

So in my PhD, I worked on a project focused on rare neurodevelopmental disorders and childhood disease, and we knew the way that project was structured. You only get into that research study if you've, if you, your child has a rare neuro developmental disorder. So it's hard then to make any extrapolations about what the population frequency of a of a particular diseases you can with some clever statistics, but one of the powers of this, actually I think if they plan to just camp out in a maternity wards and try to enroll everybody, who's interested is my understanding of the recruitment plan.

And, and it'll be probably something like a third of all babies that they need to enroll in a given year to hit the targets. So they actually should get a really very representative sample. Of, of the population and a true rate of some of these rare diseases. Cause there's a lot of open questions about penetrance and having one of these, having a genetic variant that shows up in people with a particular rare disease, does it always cause that disease or are some people actually avoid having the disease due to another protective variant?

I think they should be able to help answer some of those questions in a much more robust way than, than it's been possible previously. 

[00:16:31] Oleksandr: Perfect. Yeah, it's great to hear 

[00:16:34] Patrick: it is. It's a it's exciting times. All right. Let's I think we should transition here. I'm going to ask you for your number two on your list, which is a really interesting one. I'm looking forward to talking about it over to you. 

[00:16:46] Oleksandr: Yeah, so I think the one that, that I, I want to pick also for this year is the progress that have been doing in the field of autologous stem cells and stem cell therapies in general, I think stem cells kind of entered our life with a discovery of IPCs.

I believe it was 2007, 2008. There was a lot of buzz, a lot of hope at the moment. Didn't turn to clinic that fast or as fast as we wanted it to happen due to several limitations in how we essentially manufacture those cells, how we differentiate them into the appropriate tissue and how we maintain the safety of the, of those induced pluripotent stem cells and prevent them from becoming cancer cells in the human body.

But I think there are a few diseases where IPCs in general, stem cells are slowly making their way to clinics. And one that I want to pick for this year specifically would be Parkinson's disease. If you look at Parkinson's disease and why Parkinson's disease is particularly important frontier for, for stem cells, there is no a hundred percent efficient therapy, right?

There are several medications that can be taken to some extent alleviate symptoms. There is deep brain stimulation, which has proven to be quite effective, especially in the early onset of the disease. But nevertheless, it doesn't treat the disease completely. After some point in time, patients will still deteriorate to the worst state and eventually death.

And this is a really sad, so what can be done? We can try to rejuvenate or replenish that dying source of, of neurons to, to essentially save the patient and make sure that the brain restores its normal function and symptoms of Parkinson's disease essentially disappear. The challenge there is, I would say twofold.

The first one we've pretty much overcome over the last few years is how to deliver those cells. The good thing about Parkinson's disease in comparison a lot of other neurodegenerative disorders, you know, exactly which cells you need to replace, you know exactly the space in the brain you need to place the cells into.

So it is a very complex type of surgery, but this in the very, very precise and very defined type of surgery. And over the course of time, we actually learned how to deliver those cells to that very specific point of the brain. Now, the second challenge is obviously how to get the cells and how to differentiate them from neurons.

If you work in your biology, you know, that neuronal culture has a, probably one of the most difficult cultures to, to maintain in the dish, let alone take those cells out and then reimplant them into the body. So the progress that that has been made, and there are a few very exciting clinical trials being published this year.

One of them would be from the blue rock. This is the company that essentially develops allogeneic stem cell therapies. They take pluripotent stem cells from healthy donors. They differentiate them into, into the department energic neurons and they re-inject them in the human brain. They've published a very exciting study this summer on the safety and the efficacy of that approach in the first patients. But I think where are the true holy grail coming to the grail again in this space would be the autologous stem cells. And the reason is that with allogeneic transplantation, you still have the probability of quite severe immune response and overall difficulty into adjusting of L's implanted cells to adjust to the environment of the, of the host brain.

The developments that we've seen this year, a common mean like from Aspen neuroscience the company headed by Jeanne Loring and I had a pleasure hosting her on our podcast. In the beginning of this year, they're really showing very, very promising results in the way they can take, take the patient's own cells.

Do you differentiate them to IPC state and they develop a very cool sophisticated or not. Maybe that's sophisticated, actually, it's it's Allegan tests, genetic tests to test for possible teratoma formation. So they have a way to screen the cells to separate the pool of ITCs that is actually not dangerous for the patient.

And then very robust protocols to differentiate those two neurons. And they're planning to launch that into clinics either next year or the year afterwards. And this is Samsung that I'm very, very much looking forward in the future. 

[00:21:18] Patrick: Yeah, I think it's excellent. I am also, I'm very fascinated about the allogeneic versus autologous debate.

It seems like there's a little bit of a a debate going on within the field of ultimately both of them will be. I'm sure it will be useful and important, but it seems like there's a little bit of a question of which one will ultimately win the day. And, and I think you gave a good overview there of essentially autologous is taking somebody's own cells, reprogramming them in some way, and then transplanting them back into them.

Whereas Allah genetic allogeneic is so-called off the shelf where you've taken effectively a donor population of cells and find the one that's the closest match. I'd be really interested in your thoughts on if you had to pick one horse to bet on in that race. And it may be a neuroscience let's say to just constrain things a little bit.

It sounded like you were saying autologous maybe is is your favorite and I'd be happy to take the other side for the sake of debate. I'm curious, which. Which one do you think has, has the most staying power 

[00:22:20] Oleksandr: Since I'm the host of personalized medicine podcasts, I have to pick autologous. I don't really have a choice in this race.

Right. But like all jokes aside, I think both of them will definitely have their place in the market. And I think that will depend a lot on the condition that is being treated. And how easy is it to manufacture those cells and make them safe for the. We will see different developments for different diseases.

And I don't even think it would be easy to generalize. It let's say for neurobiology or for cancer, for any specific type of condition or indication. In general, I am a fan of autologous stem cells. There is a, I don't want to say misconception, but probably bias towards thinking that autologous stem cells are necessarily much more expensive to produce.

They probably are at this moment of time. The advantage, however, of autologous IPCs is if you have established protocols, how to do, how to first collect the cells, differentiate them and then differentiate them in the target tissue. You need normally the minuscule number of those cells to essentially treat the patient.

And you also don't have to maintain those large banks of let's say, allogeneic cells and make sure that they are of the right quality, that they don't accumulate say cancerogenic or other types of mutations over time. Which is also a challenge for phylogenetic therapy. So I think it will depend on the specific indication on the specific type of treatment.

There are certainly a lot of benefit in allogeneic therapy when we speak about the CAR t-cells I think there has been a lot of progress in that space, but I'm also curious to see how this will develop over a longer period of time. 

[00:24:04] Patrick: Yeah, I definitely share your point that the primary concern, it seems like is how you scale up manufacturing or, or the, the process.

But I don't think we're there yet with either autologous or allogeneic, both have their, their own challenges, but it seems like if the manufacturing or scalability challenge can be solved, then it's clearly superior for obvious reasons that if you can use someone's own cells. But I, I know, I know I don't know enough about it to really speak to 'em it's too much of an expert, but I understand that it's a big challenge ahead to actually get that get that process to be scalable in a cost effective way.

[00:24:46] Oleksandr: Absolutely. And look there again, a lot of startups who are trying to tackle that manufacturing challenge. We also had another client from Selena on our podcast, I think developing the differentiation protocols to achieve specific type of cells is hard, but then scaling that up is probably even harder just because you need to control for so many different parameters in those very complex environment.

And it's not an easy task, so there's plenty of room too, for developments in that space manufacturing and quality control, and then actually release of that product as the final drug for the patient. 

[00:25:26] Patrick: As far as you're aware is Parkinson's the first disease. Where is it the furthest along in terms of autologous stem cell therapies and getting to the clinic, or are there others that are, that are further along or, or, or in a similar. In a similar ballpark. Yeah. 

[00:25:43] Oleksandr: Good question. I don't want to be around here. So I would honestly answer that I'm not aware of other diseases that are more progressed in terms of autologous cell usage well, maybe if you look at the autologous cells or stem cells. In the broader sense, you can argue that some of the cell-based cancer therapies are based on autologous precursor cells.

So those won't be IPCs per se, but those will be autologous cell therapies. But in terms of truly differentiated somatic cells then differentiated into, into the target tissue, I think Parkinson's would be one of the, one of the first ones to, to, to have those types of therapies. 

[00:26:26] Patrick: I'm also not aware of any others, maybe one of the listeners who knows more about this than us.

If there are others, they'll, I'm sure they'll email us or or otherwise let us know. 

[00:26:35] Oleksandr: Definitely let us know. Great. Patrick, I know that your number two on the list is also very, very exciting. Also concerns those emerging. Therapists. So why don't you tell audiences? What was the second biggest? 

[00:26:53] Patrick: Yeah, it's, it's not exactly a first in and of itself, but it's it's I think a significant continuation of what is going to be a trend for a very long time, which is extremely effective one-time therapies.

Initially I think most of these were for very rare diseases like SMA. However, there are many, many more examples. I think coming down the pipeline in particular too, that I wanted to highlight. Was a gene editing of PCSK nine that was initially done in primates. So this is a there there's a set of genetic variants in this gene that will cause familial hypercholesterolemia.

So high, high cholesterol, and it was shown that a onetime at base editing could be made in primates and there'd be durable. At least my understanding is at least. 10 months potentially or longer of of durable cholesterol lowering. And in humans there was, this was not gene editing, but in humans there was a an transplant of pancreatic islet cells into a type one diabetic person that appeared to essentially fundamentally.

And I think in this case, it's, they, they measured for at least 90 days, but to have a transformative effect there. So these are both two very common conditions. And obviously I think we're going to continue to see this in rare diseases, but it seems like as a category these one time you know, maybe they're not one time, maybe they need to be done every couple of years, but shifting from chronic management into something that's much closer to one time, I think is obviously an amazing development from a technological standpoint. But I'm also really personally interested in how the business models around healthcare need to evolve, to cope, to handle these kinds of things, which many of the rare disease treatments have been infamous for how expensive they are, but yet they do, they do deliver as much value to, to the patient's and family's lives as is on the sticker price.

But our system has not yet figured out how to pay for those at scale. 

[00:28:47] Oleksandr: Yeah, I agree. Like if you look at the price tags, for example, for gen smart, I think today the most expensive drug that is available there on the market to treat SMA that that looks scary at first . Sight, but what do you think can be the past forward there?

How can we bring those therapies? To the broader market and make them available for everybody who needs them. 

[00:29:11] Patrick: Yeah. I, I've done a little bit of thinking about this. I know there'll be smarter people and in our audiences, who've, who've thought about this a little bit more, but there's a big, there's two big differences in health, economic or business models between the U S where I grew up and the UK where I live now.

I think the UK is in a better position because of the vertical integration of the system that the NHS can take a very longterm view on my health and say, if it's worth spending a million pounds for one dose, now, if it saves us far more than that and saves your life over a longer period of time. So I think they've got, not an impossible chance to solve that problem. But I think there are still significant challenges there simply because of the cost of some of these therapies. And, and I know if you do the back of the envelope calculation, if rare to see if therapies were available for every rare disease, then it would affect. And they were the price point that the few that are available today are it would bankrupt the NHS. But I think that ignores the you know, the, the fact that prices are going to come down and competition is going to enter the market over the long run. So I actually think vertically integrated systems national health care systems will be able to figure that out, but it's going to take some time where it's more challenging is in the U S and other places where health insurance dominates.

And I had somebody explain this to me, and it made a lot of sense after they explained it to me. But basically you're not guaranteed to be with your health insurer for any period of time. I mean, you, you may, they may expect you to be with them for five years or so on average, but if you move to a new state or or, or just simply decide to switch, they have a re a much bigger challenge, which is if they've paid for that million dollar treatment and then two years later, you leave and you go to someone else, then, then there, their models break down. So the health insurance companies actually have to figure out how to develop some sort of credit system at large to account for the people who are swapping between and, and the way that value works out over the long run.

And so that's a much trickier problem, I think, to solve it in terms of how this works. And I think it's probably. We'll, we'll delay the adoption in some of these cases, just because it's going to be a little bit more challenging to figure out how to pay for it. 

[00:31:27] Oleksandr: Yeah, definitely. And Patrick looking at those genetic inborn diseases where one time treatment can be applicable.

What are the other indications where you expect these type of therapies to emerge over the next few years? 

[00:31:45] Patrick: I have a, I have a personal bias towards the long tail of rare neurodevelopmental and other childhood diseases. One of the big challenges here is it's it's assumed. I mean, and there's some, some works backs up, but that there's a window of opportunity where you need to be able to treat the child in order to have, in order to actually have an effect that if brain development has already been going for many years and the child becomes an adolescent or a young adult, it may be very difficult to actually reverse some of those changes.

But if we combine the previous discussion we were having about early screening and detection with advances in base editing, gene therapy, other approaches, then I can, I can see a really I'm very optimistic about a future where diseases can be detected far earlier and actually treated in that window of opportunity.

There've been a few papers that have showed in, in diseases. I believe like Rett syndrome and some others that that window of opportunity is actually, it's not really narrow. It's potentially wider than we think there may be a few year window where you can actually make meaningful differences, but probably every disease will be the same.

And that work needs to be done to figure out which ones, which ones need to be treated when. Perfect. Great. Now the next one on your list. I'm excited to talk about because I, this is completely not on my radar. I had I had no idea what this was and I did a little bit of reading beforehand, but I'm excited to hear from you and ask you a few questions.

So I'm going to turn it over to you for the next one, to talk about glyco proteomic markers, and explain to us what that. 

[00:33:23] Oleksandr: Yeah, I guess this is like an orphan highlight right on our list. I love like a biologist. I must confess and big disclaimer. This is something that my wife has been working on quite extensively in her research career as well, and why I like it.

So let me first break this down. So. Glycol proteomics, glycol biology allows us to measure different subset of biomarkers. That is not really that common in modern diagnostics. If you think about diagnostics, you think about DNA. So you would either use PCR or sequencing to detect specific DNA or RNA sequences, or you would use some sort of immunoassays or mass spec to examine proteins.

Right? These are your two classical type of biomarkers. You might also throw metabolites on top, proceed have some metabolomics, but that's, that's essentially it. So when we speak about omics, it's always a genomics proteomics metabolomics, but there is this beautiful world of glycoprotein comics or glycolmics.

That allows you to measure glycosylated versions of the proteins and the reason why it hasn't penetrated our routine diagnostics so much so far, is that because those glycans, those glycoproteins are notoriously hard to study. So if you take, for example, mass spec, which has progressed tremendously over the last 20 years and help us to understand essentially proteins and how, how they work.

It's really hard to measure glycosylated proteins. It's really hard to measure sugars. Although those are relatively simple molecules. If you compare that to DNA again and proteins, they don't like in mass spectrums, they don't fly well in mass. 

[00:35:08] Patrick: So did they just slip through the net somehow and they'll show up. 

[00:35:11] Oleksandr: They slipped through the net and because of the simplicity in their structure. Right so if, for example, you have glucose mannose, it's the same chemical formula and they would generate very similar ions. Right? It's it's hard to distinguish those. And then on top of that, you don't really know the sequence of that glycan tree on top of the protein in the first place, because if you have DNA.

You already have sequenced banks, you know, like what ..The specific sequence of that gene should be. If you have a protein, you know what the specific sequence of that protein should be because you know, the gene that encodes food, but you have no idea how that oligosaccharide tree on the top of the protein would look like, but that oligosaccharide tree can be extremely important and can be extremely different between healthy cells and cancerous cells and the development that I've picked as a third one for, for this.

Would be the first approved at da approved glycoprotein based test for cancer detection. And this is actually the test for, for the pelvic cancer that is developed by the company called intervene. It's a San Francisco based company. Co-founded by professor Bertozzi whom I, a big fan of and admirer of her work.

If she listens to this podcast, somebody in our audience knows her. Please let me know, because I've been trying to get too far along. Yeah, it's just, she's essentially the godmother of the field. And essentially what this DAS allows to do is to again, detect the cancer much, much earlier. That test also allows to distinguish between benign and malignant type of tumor, which is also not that triggered with a conventional bio markers.

And they have a full pipeline of other cancer indications where that type of approach can be used in the future. So I'm very, very excited about that and obviously a little bit biased as well. 

[00:37:00] Patrick: Absolutely. So how does the test, how does how does the technology work in, in the way that mass spec and other methods miss these pieces?

What, what are they able to do that allow them to, to catch them? 

[00:37:14] Oleksandr: Yeah. So essentially it's a lot of trial and error and trying to understand how, how to deconvolute those signals that are coming from the, from the sugars on the sugar part of the of the glycoprotein. It's also involves quite complex chromatographic step, right?

So you would have to sometimes cleave those sugars, run them on a different columns. Try to understand the retention times. How does it differ between different combinations of those oligosaccharides. So then you build the libraries. So you kind of create the empirical knowledge of what the signal should look like.

Then you compare those to those databases and demonology that your generate the better your detection system gets. But I think the essence of this test was just to try to understand, okay, what type of oligosaccharide tree would give you what type of signal. 

[00:38:04] Patrick: And then effectively, is it the case that knowing these glycoprotein markers. Help to differentiate. Is it treatment strategy for the pelvic and ovarian cancer? Is it tells you something about russian and what is it, what does it tell you? 

[00:38:21] Oleksandr: Yeah, from my understanding, in terms of pelvic cancer, it's mainly earlier diagnosis. The first thing. And the second one is distinct ability to distinguish between those benign and malignant tumors, which is not the trivial for some of the other indications that they have in pipeline.

I believe it will be possible to determine the severity of the disease because you don't only, it's not only just a yes or no type of answer, you might have different type of those oligosaccharide trees on top of the protein. And then. Let's say relative expression levels of, of specific specific glycoproteins if you will, can also determine the severity of the disease and determined. 

[00:39:04] Patrick: Great. What I really liked about this one was I'm I'm blanking on who the famous person who said this wasn't and I'll also probably get the three in the wrong order, but hopefully I get the spirit, right. Which is a new, the, the way that science progresses is technology data and new hypotheses in that order, it's not often people think, oh, we've got a new idea.

And then we go out and test it. But actually the thing that comes first is a new technology that helps you uncover something new. And in this case, it's to see, see where, see where we haven't seen before. So technology data, and then hypothesis, and then the loop starts again. So it's always, I always love to learn about new, new views and to do parts of biology.

Obviously, sequencing has been that that's been the story of sequencing for the last 20 years. The amount of, of discoveries that have come out of just the technological change, but it sounds like there's a, this is happening here as well. Something I should spend a little more time on. 

[00:40:02] Oleksandr: I completely agree with you on, on this approach. And I think that ties very nicely to the last point that that you wanted to bring up. And I am sure we could have talked about this topic for the whole episode. Yes. So why don't you tell our audience, what did you pick? 

[00:40:19] Patrick: Actually, that's a really good such a good segue because it is a new technology and potentially transformative Uber technology, not just to help to our whole world, but the peace. Yeah. The, the, the development I wanted to bring, which again, isn't necessarily a single event, but is a continuation of a trend is, is there's been some really big advancements in AI machine learning and drug discovery.

And those of you who know me know that I'm not really a, I'm not a kind of complete AI ML fan boy, that that's all I I talk about. But I think, I think it's a useful tool, but there have been. In particular, a couple of big stories that stood out the first and foremost is the open sourcing of of deep minds, alpha fold.

That enables us, you know, us, obviously I haven't downloaded it yet because I don't spend that much time coding anymore, but anyone who reasonably knows their way around a machine learning model can now download this incredible machine learning model for doing protein, folding and predicting the way proteins fold.

And, and I think that this will allow a thousand flowers to bloom over the next couple of years. And, and it's remarkable that they've actually made the decision to open source it and not keep it internally to do whatever they do. But I think they recognize that the. Opportunity for having an impact by open sourcing, it was, was so massive.

And there are a couple of other kind of smaller stories around this in particular, some of the leading AI and ML based drug discovery companies like re like recursion and benevolent AI are striking some pretty significant deals with large pharma. And obviously these deals can always, always need to be taken with a grain of salt because they'll, they have payments that long long in the future should the molecule succeed. But but the fact that people are starting to take these approaches much more seriously, I think there is an opportunity for these kinds of approaches to understand data that we're collecting in a way that that is impossible for even an army of humans that are trolling over it to do some I'm very optimistic, cautiously optimistic.

Of course, there's lots that still needs to be done to get these aI generated drugs into the clinic. And I think the detractors will argue that many of these AI generated drugs are actually could have been discovered by people anyways. But nonetheless, I think there's a lot of interesting things happening there.

[00:42:38] Oleksandr: Definitely. And Patrick, you already alluded to this, it's probably the field that has so much hype in it. So how do we consciously separate wheat from chaff here and understand what actually might have technological and clinical benefit in the future and what just put smokes in the eyes. 

[00:42:58] Patrick: I've this is probably pretty straightforward and, and, you know, maybe obvious, but I think it's two things, really.

Number one. Be honest about the benchmarks that we're comparing it against. So are we making discoveries using this new technology that are truly transformative and humans couldn't do. I think this is with alpha fold no one can argue against this. Right? There's the, the, the you'll hear even the most. You know, credentialed x-ray crystallographers that have said things that took me a decade or, or that we've never been able to crack have been solved by this algorithm. So I think it, number one is what are we comparing it against? And is it clear? They demonstrably better where I don't think the evidence that, you know, the preponderance of evidence, isn't quite there for many of the origination of new molecular entities or, or drugs.

But I think the second piece is just time. Time will tell. Do these, you know, do these things deliver do these models deliver out-sized results compared to what other have. And, and I think there are many companies that are getting you know, have considerable backing behind them, of people that believe they can.

And so I think it will work itself out in the next five years or so either they will produce significant results or they won't. But I think we're seeing, we're certainly seeing encouraging signs that none of them are going away anytime soon. 

[00:44:15] Oleksandr: Great. And speaking about business and like where the money flows in this space.

So where do you see the future of, of that type of AI drug development do expect large companies to either acquire some of those smaller players and then fully integrate them, or you expect, let's say the whole new class of pharmaceutical companies to emerge. That will be mainly AI driven. And we'll kind of pump that early assets into the large pharma 

[00:44:43] Patrick: Yeah, it's such a good question. I, I suspect that a lot of the power will actually be in the hands when someone invents a, a paradigm shifting technology or something that's sufficiently new. I think of Maderna as an example, and they've obviously benefited tremendously from the success that they've had and in developing the COVID vaccine, but they have a very novel tech platform that they can go in a million different directions. So they have the option, I think, to build a complete standalone pharmaceutical company. I think this the same as the case with many of the AI based ML based drug discovery companies, if they're able to build something that allows them to generate new. Know, new new medicines, repeatably, and at scale, I don't see any reason why they couldn't go and do it themselves.

What you know, and my understanding this is, is many people in this role to know much more about this. But my understanding is the reason that most companies go through the acquisition process to, to be acquired by a large pharma is because they believe that either. They're going to be able to get to scale and distribution much more quickly by doing that because those companies have the incredible global reach manufacturing.

All of those. All of those other things, or because they're a company that's actually just developed a single asset or set of assets and not a platform technology or something that, that could just repeatedly generate new assets. So I think for companies that have built a, that the metaphor I often hear is, is have you, is your company at golden egg or is it a golden goose?

If you've, if it's a goose that lays eggs, then those companies will be able to do it themselves. I think if they choose to. But if it's merely an egg that you've licensed out of a university or, or something like that. Then, then probably they'll continue to sell those eggs to large pharma who, who will help to get them to scale and get them into the hands of patients and doctors that can benefit from them.

[00:46:42] Oleksandr: What do you see as a next kind of big development in terms of AI based drug discovery. So we've seen, obviously it was alpha fold. A lot of things happening in the protein structure deconvolution now was a recursion and their deal was Roche. Do you think it's indication specific game or what, what will be kind of the next wave of a few targets coming out of there. 

[00:47:09] Patrick: Yeah. I, I think there are a number of different applications, even just for alpha fold itself. That could be pretty transformative. So one is actually in the diagnostic space, which is, we have so many challenges in the whole genome and exome sequencing of, of variants about no insignificance.

So you know, a patient has cancer. Right, that runs in their family. You sequence the BRCA one gene and they have a variant in there, but it's not one of the hundreds now, probably thousands of known variants that in that gene that causes breast cancer is, and the question is, is that is that genetic variant pathogenic?

Is it, is it cancer causing or at least predisposing or not? There've been some amazing work. One-off looking at genes like BRCA one that actually are able to systematically. Perturb or mutate every base of the gene and ask the question of, does this change the function, does this change the function, does this change the function?

And actually you build up a library of every single change possible, and you do this in a cell line and why this is so powerful is because you'll never see all variants in humans, but you can do it systematically in the lab and basically build, build a library and ultimately a model. What I think things like alpha fold will enable us to do is to not just do that one by one, but to do that in a computer model to say, if we change the DNA by one base, run the program and ask what happens to the protein and then make a prediction and, and to how that impacts human health.

And ultimately you need all the, you need the, you need enough observations of human health, to be able to feed that back into the top of the model, but I'm really interested to see if it can be applied to, to improve the way we diagnose. And essentially it's a way to simulate things you've never seen before in the population using, using a model that can give you a best guess of, of what it's likely to do when you do see it.

[00:49:05] Oleksandr: Yeah. Fantastic. And let's, let's, let's watch that very closely and let's hope that all of these things will happen rather sooner than later. 

[00:49:13] Patrick: I'm sure somebody out there is working on it right now. 

[00:49:16] Oleksandr: Definitely. 

Perfect. Yeah. It has been an exciting year for sure. For personalized medicine and for our podcasts, I believe as well.

Definitely. This was fantastic conversation. We should definitely repeat it next year. 

[00:49:30] Patrick: Agreed. Yeah, no, it was a lot of fun. I think in early 20, 22, we should do this and try to predict. What's going to happen. And then at the end, it's 2022, we can get together and, and realized how wrong all of our predictions were and what we, what happened that we didn't guess what happened.

[00:49:47] Oleksandr: Exactly. There is nothing more fun than looking back at what you said. What you said before and just, just calls. Great, Patrick. Thanks so much for, for doing this crossover. That's been a pleasure. That was fun. I hope everybody in our audiences enjoyed this episode as well. If you liked it, let us know.

We'll try to do this more often as well. If you think that we miss something really, really important that happened in the field in this year, then please let us know on LinkedIn or Twitter. We'll we'll happily share this with our audiences because. Obviously, we don't have the full oversight of this ever-growing field.

Right. It's, it's hard to catch up. Was there ever a that's going on? 

[00:50:27] Patrick: It is. And, and we did have a couple more waiting around. In the backdrop that we prepped, what we decided to cut it, just just to make sure that we, we didn't bore you all with having too many of these, but if everybody liked it, then, then let us know.

And if not you can also let us know. Thanks. Oleksandr was really a pleasure. I enjoyed it and have a, have a happy holiday as well. Hope if you're traveling anywhere that you get there safely and things don't get canceled and it stay safe as well. 

[00:50:53] Oleksandr: Perfect. Thanks Patrick. Same to you and happy holiday, season two.

Rest well. Yes. 

[00:50:59] Patrick: Thanks.