Episode #6 –Using genomics to accelerate drug development

Ellen Sukharevsky  0:29  

Hello and welcome to the BC Platforms podcast. My name is Ellen and I will be your moderator today. BC Platforms is a global leader in providing a powerful data discovery and analytics platform as well as data science solutions for personalized health care. BC Platforms enables cross functional collaboration with our global federated network of data partners. Today's podcast will focus on the topic of using genomics to accelerate drug development. This podcast sheds light on industry challenges and expectations as well as innovative approaches to collaboration. The discussion is led by Tõnu Esko, BC Platforms SAB chairman and Vice Director of the Institute of Genomics University of Tartu, where he also holds a Professor of Human Genomics Position. He is head of the Estonian Biobank Innovation Center, and focuses on public private partnerships and innovation transfer. Dr. Esko is also a research scientist at the Broad Institute of Harvard and MIT. He acts as one of the senior leaders for Estonian personalized medicine programs and serves as scientific advisor for several companies. 

Our speaker today is Iris Grossman, who has been on the leading edge of personalized medicine, big data analytics and biotech R&D for 20 years. She is Acting CSO of Amide and advises other biotech in her field of expertise. She has held leadership positions, or consulted for top big pharma companies like GSK, Eli Lilly, Takeda and Teva, where she was VP Head of early stage development, startup and consultancies. She specializes in leveraging high dimensional genomic insights and E-health as engines of therapeutic discovery, development and Lifecycle Management across a wide range of modalities and therapeutic areas. Iris also serves as SAB member of BC Platforms, Tipa Biobank, Eleven Therapeutics and CGRP Diagnostics. She earned her B.Sc. in Medicine at the Technion, and a PhD in Medicine, Population Genomics and Pharmacogenomics for the Technion, co-mentored with the Weitzman Institute. Next, she was a postdoc fellow in Personalized Medicine and Population Genomics at Duke University. Now, I will hand over to speakers for a brief intro to begin the discussion.

Tõnu Esko  2:40  

Hello, everyone. My name is Tõnu Esko, and I'm happy to be just through this very exciting topic.

Iris Grossman  2:47  

Hello, I'm Iris Grossman. Thank you, Ellen, for the intros and for inviting me to this exciting talk.

Tõnu Esko  2:54  

So without further ado, let's jump in. So today we are discussing how the use of genomics could accelerate drug development. And recently there have been a couple of very exciting papers showing that when the early stages of drug discovery is based on human genomics evidence, it seems that it's often four times more likely that the truck will reach the market. So I think that's a very clear indication that there is a lot of value for the industry in the big file banks and their data repositories. And that's the topic that we are discussing today. So Iris, what do you think are the benefits of using big data in drug discovery? As a first question from my side?

Iris Grossman  3:45  

Thank you. Yeah, so I devoted my entire career to this space. So I'm obviously an avid supporter. I'm very fortunate to evolve together with the field over these last 20 years, and been very fortunate to work with and in collaboration with field experts in pharma, and all the other stakeholders, whether in academia, biobanks and other stakeholders. I think it's an amazing, amazing ecosystem that we've built and I'm really excited to be part of this journey. I can tell you, I can share a couple of ideas and examples of where it's already standard of how we operate, both in big and mid size pharma biotechs, as well, an area where I've been operating most recently. So it's very much the standard of how we think about the pipeline, many therapeutic areas start with basically the hypothesis behind why pursue a certain target, and then certainly putting together the validation, critical path, the way we call it, it hinges on human genetics, which is very rewarding. So we think about everything from the very early stage of what we know about patients suffering from that particular subtype of the disease, what is known about rare and common variations. And from that point, my entire pipeline, at each one of the milestones for decision making, I always incorporate human genetic data. So the pipeline is peppered basically with the must have input from what we know about human genetics. And when we have gaps in knowledge, we go and seek them. So that's when we pursue cohorts or biobank collaborations in order to know more about these patients, and both increase our confidence in the frequency of the target or pathway that we're targeting, as well as alleviate risk, potential safety effects. And then this is what I refer to until now, is the first layer, the way I would relate to it, which is the sequence variation. But that is, you know, that's the grammar on top of that, for us to gain confidence. And those targets or those genetic variations, we need to incorporate knowledge about functionality. And here, there's a whole slew of omics technologies that seem to mushroom literally every day and give us insight about the functional effects of those variations, particularly relevant in non coding regions, which are pretty much 98% of the genome. You know, when I started my career, we used to call it junk DNA, nobody knew what it's good for. We all know by now that this is really perhaps at least as important certainly as we think about pathways and interactions between genomic regions. So we incorporate all of that. I want to make the point that even though today I'm focusing about undiscovered and preclinical, it is important to start thinking about perhaps the most downstream utilities very early on in discovery, companion diagnostics, lifecycle management. You literally start writing your TPP, target product profile, as we call it an industry, you create an early TPP, literally at nomination of a program, because you really want to have a really good overview of what is known what is unknown, where do you need to create. Sometimes we need to create our own registries and invest four years in developing the right cohorts and potential recruits very early on in the program. Of course, this data will inform inclusion and exclusion criteria, and so on. Now, before I'll pause, and we'd like to hear your thoughts too, I want to make the point that what I shared until now was basically how we built these R&D processes around small molecules and biologics, which is what we've had for 20 years. But this human genetics know how has yielded novel modalities, gene therapy, cell therapy, msos, you know, again, we have many, many new modalities every every year, those are in and of themselves, first yielded by the Human Genome Project, but secondly, require human genetic understanding, even just for the QA QC of the product. I can tell you that working with gene and cell therapy companies very early on, you need to qualify your product by what you know, of course, not the donor about the acceptor, the population, potential immunogenicity, and targets of integration. So human genetics is only starting to impact our therapeutic development in ways that we never even dreamt off. It's really exciting.

Tõnu Esko  9:14  

Thanks, Iris, it was an excellent broad overview of the topic. And I fully agree that genetics is basically the A and O of every project that involves humans. And while the human genome is super complex, and I think no one dreamed that it's going to be that complex to understand what the heck is going on. But at the same time, you know, there is so much information to be gained. And I mentioned all this, the basic drug discovery process for those new modalities of cell therapies, using CRISPR in vivo, so we can inject innovation directly into the functioning organism and just with one shot cure to patients. So I think this all is very, very fascinating and something that literally seems like science fiction, it's already there. And I also agree that the sequence variation is extremely important and I keep on dreaming how we could get everyone genetically profiled and use that, not just in science, but also in healthcare, and coming up with new therapies, but also having those other modalities like RNA sequencing, proteomics, metabolomics, to making sense, or understanding the functional consequences of the sequence variation. And that's what we are trying to steer biobank to get as many layers on top of the genetics and on top of the health data to really unlock the power of fusion. So you gave a really nice broad overview, but do you have, for the listeners, some more in depth examples, where the genetics when available in large numbers, or in very specific patients could actually be helpful in the drug discovery or more broadly, in the field of locating the medicines to the market? 

Iris Grossman  11:18  

I think that probably the example that everybody listening to us is aware of, at some level or another, is the contribution of Human Genetics and Genomics to fighting COVID. I think that the entire global ecosystem really has been at its best sharing from the very early first sequence of the virus itself, which the databases and databases that accrue all the mutations to date, and not only just depositing mutations, but actually annotating in real time, the functionality, the specific implications to clinical manifestation, really an amazing effort that in normal times, would have taken us years to produce and curate. So this is incredibly encouraging, I think for the entire field. Hopefully, we won't need another crisis to push us to even better performance. And so that's one example and it affected the approval of antibodies. There are multiple repurposing programs going on that seem to be quite promising. So a lot going on that we have only sent the tip of the iceberg. What I thought I'd share in this context is a little bit, maybe a look under the hood, at how we think about early discovery programs in terms of where human genetics, and genomics bring in value for decision making. For many of these programs specifically, I cannot share specific names, but I think I can give enough context for people to relate to. So I would say that the foundation of a program that goes into R&D would be defining a well articulated rationale, we call it the therapeutic hypothesis. And it sounds simple. At the end of the day, it actually is a statement that incorporates several lines of evidence that we either have, or we absolutely need to obtain in order to have a viable program. And so you start this before you've done anything, it should be what you basically stay to the work to the world or to your internal world - this is what I'm going to try to do, this is the value proposition, and these are the types of experiments that will let you my stakeholders know that I'm progressing in the right direction. So first and foremost, we like to add disease and subtype genetics. And we'll look at rare and common variations, both in terms of causality as well as pathways aggregate pathways within specific cell types. So here is where the functionality data comes into play. Because of course, except for tumors, the sequence is consistent across our body. However, the expression levels are different and different genes explain different cell types even within the same organ, there might be very different components with a very different mixture of gene expression. And so we try to aggregate everything we can from either external or internal data sets that give us basically a map of what we know about that disease? What do we know about the target organ or organs, even if it's a systemic disease, it might be that the source of the aberration is in a single organ, like liver disease. Liver is often the manufacturer of genes that affect other organs, brain, heart, and so on. And then one of the early types of annotation that is very important to pay attention to is how does the variant affect the disease I'm interested in. What I mean by that is, if it's a susceptibility drive, it may be very important for curative disease, for curative therapeutics, gene therapy, cell therapy, but less so for disease modifying modalities. In that case, you want to see gene dose effects that are associated with progression, severity, age of onset, certain phenotypic characteristics that are correlated with your therapeutic goal. So these are very important, distinguishing points. And many times we don't have a lot of information about that. This is where we really need to get to collaborate with biobanks ideally once with longitudinal data sets or connect with academics that have well curated longitudinal cohorts. And that might be a very early go no go decision for some companies. Many companies then look at different types of functionality layers. So it used to be that we said well, we do transcriptomics and that would be RNA sequencing or earlier qPCR. But this is actually a very crude way to think about functionality even at the transcriptomic level, because steady state, RNA bog RNA is a very crude measurement. Today, we can look at nascent transcriptomics, and decay rate and we can look at specific single cells and now we have some cellular in situ sequencing. There are many, many ways we can go into great detail to understand or at least infer that likely interaction between genetics function, then proteomics and downstream, systemic effects. The downside of going into such great detail of understanding of a particular marker and a particular disease and a particular cell type in a particular set of omics readouts is that we become very vulnerable to the specific system that we're studying the particular technologies, each one has their own biases and blind spots, and they're very expensive. And so it's very difficult to generalize. And so what we do and again, early on in drug discovery and preclinical development is continuously try to elucidate the main driver of each one of these signals, develop essays that are more robust, and that we can use in larger or in additional sound systems and additional sample types, and try continuously to be more generalizable, and kind of tie it back to the clinical and patients. So this is one of the challenges and frankly, exciting parts of doing this type of research in drug development. The other two important points to mention are effect size. So it is quite rewarding when you do have data sets that tie between genetics, maybe some expression, either RNA, and or protein, and some functional readouts. Depending on the disease, you'll have different types of functional readouts, downstream is when you can infer effect size. So one very simple example is haplo-insufficiencies where it's very obvious that 50% effect size would have clinical relevance. So there are tricks like that that we're trying to use. This is in a single gene scenario, but once you think about a pathway or interactions, you can think about ways that help you hone in to what is the effect size that you're expecting to see and gives you guidance as to whether your dragon development is actually making the fact that you needed to or you need to stop. And early attrition is a good thing. Last point I'll make is safety and lifecycle management, which you can infer from FiuA’s and similar types of analysis and give you, again, both confidence that when you tweak that gene or protein, you're not going to induce unwanted effects, or that at least they're tolerable within the context of that particular disease, and also start to have an idea of what are the other opportunities for that type of drug, maybe in a different regimen, maybe a different modality. But that target has other potential, which helps you gain interest in developing a whole organization around that type of disease and targets. So these are, again, elements that we think about very early on, especially when you need to think about a hoard portfolio, and you must make decisions, well, this program gets the go, this one maybe is put on hold, because resources are always limited.

Tõnu Esko  21:22  

Yeah, I really like your last idea that, when you have identified the target using a multitude of data approaches, then you use the FiuA’s for example, it's like genome wide association studies, very big one variant or one region variants and try to understand where you can identify other therapeutic areas for the same trunk. And I think it's a very often overlooked aspect of those big G or the use of the big data to really broaden the use of the variance. So that's really, I think, really, really interesting. And I think also like all this the use of genetics in finding the right patients, it's very well known that for them in oncology, and edifying, you know, the tumor for the right patients, we can prescribe the right medicines, but I think it goes way beyond that. And I personally think that very soon we will have almost every product that will have some kind of indication where it is either for efficacy or side effects or dosage information. And I think it's only good because we don't have an average patient. Everyone is special, in one way or the other. So that's, that's, I think it's very, very important. So I've highlighted that, you know, that the COVID has been an example for us, like how to collaborate and the bank very nicely highlighted that we hopefully don't need another crisis to open up the data sets and, and collaborate more. Because in our field, especially looking from the biobanking, and datasets side, it's very much if the data is siloed. And even more, the intellectual property or the know-how is siloed. So the industry may be open to collaborating with academia, but I think it's quite hard to collaborate between the industries. So if we can now in the podcast give recommendations to the stakeholders, recommendations to the industry, recommendation letters to the academia, recommendations to the data custodian. So what would be your recommendations in the light of what we have discussed to the stakeholder?

Iris Grossman  23:43  

Yes, so this is an area I'm very passionate about. Because I think that, again, the stakeholders have done tremendous work over the years generating UK Biobank. And the list is very, very long and impressive. And each one of these data sets has its pros and cons. So we definitely need more. And to do more, we need all the stakeholders to continue to be engaged. And I think that's one aspect of this. The other is we're under utilizing these resources, I would say, because they're so expensive to generate. It's mostly Big Pharma that end up in a consortium, they can't even generate them on their own, which makes a lot of sense. So there is a consortium. Big Pharma is doing again, amazing job collaborating with each other and generating those data sets. However, even though the aggregate statistics are shared, often broadly, or can be accessed by smaller players. There's limited access certainly to the data sets that have very fine clinical longitudinal information. And I think that there is room for a new type of consortium to be generated, where the biotechs, which are really the cutting edge innovators in this ecosystem of biopharma. They have the novel analytics tools, they have novel modalities, for interrogating or even experimenting on some of these samples, but they don't have access to the whole genome sequencing data from these big data sets. And so we have a bit of a disconnect. And even though Big Pharma may end up buying an asset or a biotech, there will be this disconnect between what biotech had access to and all of a sudden now gets potentially access to all this data that may have been pivotal to decision making early on, as I explained earlier, how we we form a disease program or a therapeutic program. So I see a way for all stakeholders to benefit, I think that it's very important that there's still incentives. It's not just open source, but I see a lot of benefit for all sides to open up access in certain conditions and terms. And potentially biobanks can play a wonderful role here and ensure that the way data flows, ensuring certain milestones are met. And, of course, privacy and the critical security elements. So I see this as an area that would really open up innovation and in a big way for our industry. And I look forward to seeing how the ecosystem relates to that. And what I would say is that I would think about non-traditional players as well. So some of the diagnostics companies, even some CRO’s are have a lot to contribute in this context. So I would not be restricted. And then the last point, I would say that pharma has fortunately been collecting, and using genetic data and samples for genetic analysis, for well over 20 years, at least in my experience. It's true that early on our consents were not in broad for a whole host of reasons. But there are ways to anonymously create anonymous IRB approvals in retrospect, and certainly for the last 10 years, consents have been very broad. And there's, again, a way to use this data for the benefit of additional programs, additional players, and I would think it would honor the contribution of patients who participated in the clinical trial because they will contribute to progress in that field. So I think that is another area where we are at the maturity level, that ethics as well as our ability to maintain, again, security and privacy and so on, allows us to be a little bit more daring.

Tõnu Esko  28:14  

And really, you know, really approved that consent and privacy and security must be the essence of every or any study that we do. But at the same time, this consent management, for example, meets, it's also information. I think the way how we can reopen the old consents when they have been more limited, but I think there is room also for this dynamic consent for the participants that they are, you know, happy with one type of study, but maybe not with another one, but the consent never would block the participant from, you know, contributing their data for the for the common good. And I really liked your approach and potential role for bio banks that they could be not just open the data and, and be open for innovation, but could be the middleman for this new type of collaboration where we're being behind around the same table, you know, the bio banks, the data biotechs with technology, and then, you know, big industry with this very comprehensive programs to try the new drug discovery and making sure that new medicines reached market. So thank you very much Iris for participating in a very lovely and insightful discussion. And now I wouldn give it back to Ellen to go through this podcast series.

Ellen Sukharevsky  29:35  

Thank you, Tonu and Iris for joining and to everyone for listening. Speakers, do you have any final comments?

Iris Grossman  29:41  

I just want to thank you both. It's, as you can see, an area I'm very passionate about. I believe that the field has come a long way in terms of quality, quantity and ethics. And so I expect the next 20 years to eclipse the last 20 years since the Human Genome Project event started. 

Ellen Sukharevsky  30:00  

Great, thank you for tuning in to our podcast. To connect with our company and learn more email sales@bcplatforms.com or visit our website bcplatforms.com. Thank you so much, and we hope to stay connected with you.