How Translational Research Can Impact the Development of Precision Medicine

0:07 
Hello everyone. 

 
0:07 
Thank you for joining us. 

 
0:08 
It's great to have you with us. 

 
0:10 
Welcome to this monthly science exchange, which is part of Oxford Global's Biomarker series. 

 
0:14 
Today's session will be on the topic of how translational research can impact the development of precision medicine. 

 
0:22 
We'll start the session off with a presentation from Mohamed Hassanein, who's the Director of Translational Medicine at Pfizer and is a distinguished expert in drug development with a focus on oncology, immuno oncology, immunology and gene therapies. 

 
0:36 
With over 15 years of experience, he has notably contributed to the development and approval of three groundbreaking drugs, including cemiplimab for cutaneous cell carcinoma. 

 
0:46 
As a pioneer in biomarker strategic advisor, he has supported the drug development for immunological and inflammatory and rare diseases. 

 
0:54 
As Chair of the AAPS Biomarkers and Precision Medicine community, he advocates for precision medicine and its integration into medical practise. 

 
1:04 
His recognition and impact on the pharmaceutical industry underscore his commitment to enhancing global healthcare through translational research. 

 
1:12 
After the presentation, he'll be joined by Daniel Strasser, the Senior Director, Head of Translational Biomarker Group at Idorsia. 

 
1:21 
Daniel holds a Master's degree in Pharmaceutical Sciences and a PhD from the Institute of Molecular Pharmacy at the University of Basel. 

 
1:28 
With 17 years of experience in translational science, he has concentrated on the sweet spot, bridging the research with clinical development. 

 
1:35 
In his current role at Idorsia Pharmaceuticals Limited, he drives the biomarker strategies to facilitate the successful translation of preclinical drug projects to clinical development. 

 
1:46 
With a focus on inflammatory, autoimmune and immuno oncology projects. 

 
1:52 
He established collaborations with clinical experts to decipher disease mechanisms and drives the implementation of computational methodologies to understand the match between drug candidates and patients. 

 
2:05 
Throughout the session, you can send any questions you may have through the chat for our panellists. 

 
2:11 
But for now, over to you Mohamed. 

 
2:14 
Thanks so much, Tom. 

 
2:15 
And I'm really excited to be a part of this June instalment of the science change with primary focus in Biomarkers. 

 
2:23 
And the topic of today, as Tom mentioned, will be primarily about the main question that always come to mind for all of the drug developers, especially ones working the biomarker, how to translate basic science discovery to a tangible biomarkers that really drive the drug development and implement precision medicine approach to the other treatment to different patient population across different therapeutic areas. 

 
2:54 
So I'll kick off this with just highlighting the main definition of a translation medicine, precision medicine. 

 
3:01 
So we can all be on the same page of what those terms means and how to leverage their finding to feed each other. 

 
3:10 
So translational research at the simplest possible way is defined as the applying basic science discovery to improve human health. 

 
3:20 
That could be achieved by taking basic science discovery from preclinical models and the models for instance, and translated it to enhance the drug development through the clinical trial and potentially future clinical practise as well through diagnosis, prognosis and tailoring treatment for specific patient population. 

 
3:45 
Precision medicine is at its simplest possible definition from the National Cancer Institute in the US is using the personal information from any given individuals, whether it's a molecular characteristics such as genomic, proteomic, transcriptomic or environmental or lifestyle information towards preventing, diagnosing or tailoring treatment to this particular person. 

 
4:14 
So in the next few slides, I will just highlight some case studies or some evidence of how translational medicine can really derive or changing the treatment paradigm of a specific group of inflammatory diseases, which is inflammatory bowel diseases that are right for this particular approach, but it's not there yet. 

 
4:37 
So just quickly inflammatory bowel disease. 

 
4:40 
For those of you who are not familiar with it, it's a group or a spectrum of chronic inflammatory disease that impact the gastrointestinal tract, primarily the large colon: The known enterology for it is not really fully understood. 

 
4:56 
However, it's believed to be a complex interplay between inflammatory, genetic, environmental and microbial system in the gut. 

 
5:05 
All and all the clinical manifestation of those IBD diseases are really painful and quality of life altering and I just highlight the four main ones here that you can leave it to your imagination. 

 
5:21 
Abdominal pain, diarrhoea, rectal bleeding and weight loss and all of those symptoms are result of the severe damage of the lining of the colon. 

 
5:33 
I'm not going to go through the details of difference between the two main clinical phenotype which is the Crohn's disease and ulcerative colitis for the interest of time. 

 
5:40 
But although they share the same symptomology, they differ in where the impact in the location of the impact in the GI tract, the histological features a little bit different. 

 
5:54 
Pathologically also they are different, but the complication if not treated can lead to severe abscesses, in case of the Crohn's disease and it can ultimately lead to severe weight loss and developing even of colon cancer. 

 
6:11 
So the current treatment paradigm of IBD is what I call a trial and error. 

 
6:17 
The first step when a patient come to the clinic, it's really the goal is to relieve the symptoms and that could be done by a variety of immunosuppressant regimen of steroid and others. 

 
6:33 
And the goal here is to relieve the symptom from the patient. 

 
6:37 
The second objective is to make sure that we reduce the inflammation by the virtue of monitoring the extent of inflammation in the gut and also monitoring other inflammatory biomarkers such as or the most famous CD active protein CRP and faecal calprotectin, which is in neutrophil activation biomarker that is well known as a surrogate for inflammation. 

 
7:05 
Later on, the patients treated by a variety of therapeutics which I'm going to detail in the next slide. 

 
7:11 
But the goal here is to if the patient did not respond to this initial treatment, we start to head with more specific targets such as antiTF alpha and others to really reduce the inflammation of the gut and hopefully lead to release of the symptoms and ultimately healing of the gut. 

 
7:31 
Now if the patient did not respond to all of those first line and second line therapies, they have to go through the same process again in a reiterative manner. 

 
7:41 
So you can imagine the time and the cost lost for the patient going through this futile treatment paradigm. 

 
7:51 
So we've got to be a better approach. 

 
7:52 
So just to quickly highlight the therapeutic landscape of IBD as we stand today, there is 3 mechanism of actions that for the targeted therapies apart from just the steroids and amino acylate. 

 
8:07 
One of them is or the most famous one is targeting the inflammatory cytokine by neutralising antibodies such as TNF alpha and highlighted here three of those infliximab, adalimumab and Golimumab. 

 
8:22 
And this way we reduce the downstream inflammatory response in the gut. 

 
8:27 
Other approaches including inhibiting specifically IL23, for instance, IL12, which derive the inflammatory pathogenic T helper cells. 

 
8:42 
And this way also we kind of reduce the inflammation by suppressing this path. 

 
8:50 
Downstream of that, There's other class of drugs or the second class of drugs which is JAK inhibitor and this aimed at basically suppressing or in blocking the signal downstream of the cytokine receptor interaction. 

 
9:04 
And this way also we prevent the downstream inflammatory cascade. 

 
9:08 
And the third most common mechanism of action is by inhibiting the trafficking of the pathogenic immune cells to the gut by blocking its receptor that binds to endothelial cells that lining the colon. 

 
9:23 
Regardless of all of this treatment, the response rate is still really very low. 

 
9:29 
We're talking about anywhere between 15 to 20/25% response rate. 

 
9:34 
That leaves the vast majority of the patients basically without any treatment options. 

 
9:40 
And the main reason for this I would say suboptimal response rate is that the disease is really complex. 

 
9:48 
There's a great deal of heterogeneity, intra interpatient heterogeneity in this disease. 

 
9:53 
So there's multiple pathways  that might be at play in a different patient and that lead to not all the patient respond to specific target therapeutic that basically are driven by any of those treatment options that are presented to you. 

 
10:11 
So the conclusion from this is that there is a dire need, unmet medical need for developing novel therapeutic based on targeting specific molecular pathways and that those basically driven by maybe multiple types of biomarker can also lead to tailoring treatment to those right patient population and potentially also predicting response for each types of treatment. 

 
10:43 
Now in 2024, the repertoire or the toolbox of the precision medicine tools has expanded dramatically. 

 
10:52 
So we are not looking at just a single biomarker in tissue or in circulation. 

 
10:58 
Rather we have a plethora of tools that we can implement to really augment our understanding of the molecular underpinning of the disease, understand the mechanism of action, tailoring treatment and predicting response. 

 
11:13 
And those tools are just highlighted. 

 
11:14 
The main ones that apply for a lot of immune diseases and other diseases, but some of them are specific to the IBD, gut microbiome, multiomics, genetics, proteomics, transcriptomics, metabolomics and many other omics that have been investigated right now and also as an individual platform or in tandem with each other in an integrative way. 

 
11:41 
And I think Daniel will talk about this later on. 

 
11:44 
Also immune- profiling for the different cell phenotypes and serum biomarkers, molecular endoscopy, the gut microbiome, which is an emerging and interesting area of research. 

 
11:58 
And finally, the tool that bring all of this together in integrative manner to provide a more comprehensive understanding of the disease and the mechanism of action, which is the newcomer right now. 

 
12:12 
AI machine learning models, which I believe some of us or a lot of us already start to experiment with. 

 
12:19 
And of course the benefit as I mentioned, you will be able to tailor the right treatment to the right patient population, predict the response, lower the cost from the futile treatment by making sure the right patient received the right treatment and not going through this trial-and-error things. 

 
12:39 
So there have been many cases and the evidence has been accumulating in the last, I would say 20 years from the genomic part, transcriptomic gut microbiome, immunophenotyping, molecular endoscopy and others on biomarkers that really in small studies if you will, but still evidence that can drive or enhance our understanding of the disease biology lead to a better prediction of response and can be used also to diagnose the patient population. 

 
13:14 
I'm not going to go through each one of those, but just highlighted a couple for the interest of time today. 

 
13:19 
So one, the first evidence here that presented to you came from the phase two clinical trial in Crohn's disease where an anti-IL23 antibody known as brazikumab has been used. 

 
13:32 
And in the right hand side here you see that the treatment arm here has a good response rate by the virtue of the CDAI score, which is an improvement in the Crohn's disease outcome compared to the placebo with the dotted line. 

 
13:49 
Now IL-22 has been identified as a response biomarker for a macrodynamic biomarker. 

 
13:57 
And you can see here in the treatment group with the solid line, there is a time response in the active treatment by the virtue of reducing the inflammatory IL-22. 

 
14:08 
But there is no response or lower response in the placebo group beyond just the pharmacodynamic utility of this particular serum biomarker. 

 
14:17 
Also from this phase two clinical trial that showed that just looking at the baseline levels of IL-22 stratifies patients in terms of responders and unresponsive responders based on the specific threshold of the IL-22 concentration in the serum. 

 
14:42 
So I'm just highlighting here the one that showed to have the best responders, or I call it the super responders are people with high level of IL 22. 

 
14:55 
The threshold is 15.5 picogram per ml comparing to the other lower thresholds here you see the response rate is much lower and comparing to placebo is the same way. 

 
15:06 
So that's just one case study showing that just using one biomarker can be used not only to monitor the response but also at baseline before any treatment can predict response time. 

 
15:20 
And of course additional clinical validation needs to be carried out to confirm this one. 

 
15:25 
Another case example just to wrap up this piece here is the finding that in infiltrating plasma cells in the colon biopsies should be a good predictors of response of TNF alpha. 

 
15:43 
So the higher the abundance of those plasma cells CD3, CD38, in the colon, the lower the response rate for TNF alpha. 

 
15:55 
So just think about this application in the clinical setting that will be very helpful for physicianS when patient come to say, ok, you have biopsy and you have this biomarker the plasma cells in the gut. 

 
16:08 
If you have high abundance, you're unlikely respond to TNF alpha. 

 
16:13 
So that's spared the patient the cost and the time and the agony of going through treatment. 

 
16:18 
So just one highlight here those this those type of precision medicine approach is not really an outlandish idea in the immune or autoimmune diseases. 

 
16:31 
So has been right now been implemented and there's a test here called PrismRA in the rheumatoid arthritis where they have a panel of 18 biomarkers in tandem with clinical features and other serum biomarkers. 

 
16:48 
They make a composite and the physician shown in real world data that when they implement this predictive test, which is tell you the likelihood of not responding to TNF alpha. 

 
17:01 
And you see the outcome is much better here in the dark purple when they implement this test versus patient who just all comers here. 

 
17:08 
So that's basically I think the trend right now that you're trying to implement in other inflammatory diseases and I think the IBD is going to be probably the next one. 

 
17:19 
Just to conclude this piece here and hand it over to Daniel that I hope I convince you there is many lines of evidence supporting the utility of using biomarkers to basically drive the precision medicine approach for patients stratification and for tailoring treatment patient also for deciphering a disease heterogeneity and predict response to different therapy. 

 
17:49 
And I think the last slide here is that I think the future direction is that implementing those maybe multiomics, a type of approach could be a way to go for this disease. 

 
18:00 
Also using other machine learning tools to integrate the data from the different platforms, seroproteomics, genomics, transcriptomics to really have a deeper understanding of the disease and be able to develop models that can help the drug development process and maybe tailoring combination therapies for a better clinical outcome. 

 
18:29 
So with that, I will hand it over to Daniel to carry this further. 

 
18:33 
Daniel, thank you very much. 

 
18:47 
OK, you can see the screen. 

 
18:50 
Good. 

 
18:51 
So first of all, I'm excited to be here to have this discussion around precision medicine and want to say first that my thoughts and statements that I make, they are mine and might not necessarily represent the opinion of my employer. 

 
19:08 
We were very ambitious to carve out four discussion points we wanted to cover. 

 
19:14 
One being translating the mechanistic understanding from bench to bedside. 

 
19:20 
Another one, understand disease heterogeneity and investigate drug response based on molecular phenotyping, then going from single to more multi-dimensional biomarkers. 

 
19:30 
And that of course requires them also using for example AI tools to not only develop but also to test the hypothesis that we have. 

 
19:42 
Now we have fleshed out questions to these four different discussion points, but I would start with one of the first questions we received from the chat. 

 
19:52 
And unfortunately I don't see the chat anymore at the moment, Mohamed. 

 
19:56 
But I think I remember that the question was where have where do we have the success stories? 

 
20:01 
And I think you showed one on the IBD I think there that's definitely one from the immunological space. 

 
20:08 
There are others like for example also in SLE, there is a lot of heterogeneity in the disease, but there is of course also a lot of attempts to identify the patients. 

 
20:22 
That respond to treatment in the clinical trials. 

 
20:24 
So maybe Mohamed, do you have any other good examples in oncology? 

 
20:30 
Yeah, well, I mean in oncology, obviously the examples are so many to count and probably people are listening to this and of course Daniel, you know them very well and just highlighted, you know that oncology has been really the poster child for precision medicine for I would say the last 15-20 years at least. 

 
20:50 
And of course targeted treatment using a specific molecular phenotypes or mutations such as EGFR, bulk fusion and many others has been the way to treat those patients who have bear this particular mutation. 

 
21:10 
So that's a clear example. Obviously in immune oncology right now several trials have been shown to a successfully targeted patient for instance with specific genomic or genetic phenotypes as MSHI mismatch repair mutation in variety of cancer and there obviously people know that there's Keytruda and others drugs immunotherapy has been approved based on that. 

 
21:40 
So there is a great deal of a paradigm shift in the oncology space from a tissue based or location based treatment to a molecular base or tissue agnostic way. 

 
21:53 
So the story has been a pretty much a success in immunology but I think right now in other diseases such as rare diseases, for instance, the story is the same. 

 
22:05 
I mean you need to have a precision in medicine approach to make sure that the patients that you are developing for instance, a gene therapy for this particular deficiency in a specific enzymes and of course Duchenne muscle dystrophy is one of them. 

 
22:21 
Potentially Gaucher disease is another one where there is a clear and targetable genetic deficiency that gene therapies try to mitigate. 

 
22:32 
So those are two therapeutic areas. 

 
22:36 
IBD is coming up. 

 
22:38 
I show also example of rheumatoid arthritis in my last slides and there is there's others also on the way, but those are just the start. 

 
22:48 
I think we are in the verge of even expanding this to other familiar areas. 

 
22:54 
So overall one could add that it has worked very well when there is like one factor. 

 
22:59 
So it's very clear. 

 
23:00 
I mean for a lot of the rare diseases there is one vector and that's the vector you look at and that's then also driving of course the precision medicine which in oncology is also looked for. 

 
23:11 
But I think as soon as there are multi dimensional factors, there are different disease driving pathways that we need to understand that maybe also are not just based on one measurement. 

 
23:22 
And then I think it gets very difficult. 

 
23:24 
And of course, that's in immunology is one of the indication areas where there is a lot of overlapping pathways that drive and maybe at the beginning of the disease initiation and maybe an early phase is driven by another pathway or maybe by two other pathways. 

 
23:42 
And then later it separates into two different pathways. 

 
23:45 
So I think the this not only heterogeneity for their disease, but also for the phase of the disease where we want to treat complicates here the precision medicine success, I would say. 

 
23:59 
So Daniel, there's other question trickling in. 

 
24:01 
Do you mind me throwing you one more or do you want to continue with your presentation? 

 
24:05 
Sure. 

 
24:07 
So one I think common questions or a common theme here is that what are the most beneficial tools in the precision medicine that we can use? 

 
24:20 
I'll allow you please take a crack at this and I'm happy to comment further. 

 
24:28 
I think at the moment when it's about, of course, what is very beneficial is if it is very clear. 

 
24:36 
So I think analysing the gene of maybe disease cause is an easy way to do it. 

 
24:47 
I think there are more complex technologies that are helpful for more of the complex and multifactorial diseases. 

 
24:58 
And talking about RNA sequencing, I think that's going to change quite a lot in the field that we can sequence the patients before, maybe before treating and during treatment and then maybe even later to see what changes, what do they look like at the beginning. 

 
25:14 
I think that has been very beneficial. 

 
25:21 
I think also because it's very easy to sample, we know how to sample RNA that it's stable. 

 
25:29 
And when we talk about proteins, it's much more difficult to make sure that you get your protein of interest in the correct or in the intact form that you want to quantify. 

 
25:39 
So I think here definitely the RNA sequencing technology tool has been very beneficial. 

 
25:50 
But, and that's I think is what is hampering at the moment the progress. 

 
25:53 
The more data we have, the more we need to embrace the data, analyse it and integrate it. 

 
26:00 
And I think that's for me at the moment. 

 
26:04 
The, of course, also the focus, but that's of course what is maybe hindering a bit on the success of using this tool. 

 
26:14 
What is your thought? 

 
26:16 
Yeah, I know this is a great thought actually you cut right through it. 

 
26:21 
I think my kind of general comment is that one size does not fit all right. 

 
26:27 
You cannot use you said like for instance, this particular tool going to work in every disease area in for every mechanism of action. 

 
26:35 
It really depends examples, right. 

 
26:38 
You mentioned rare diseases. 

 
26:39 
So rare diseases for instance, if you are talking about some sort of metabolic deficiency, you may be, yeah, you will look at mutations for instance of this particular gene as a way to diagnose and make sure that you stratify patients that most likely respond to your drug. 

 
26:57 
But end of the day, you will need also to measure the enzymatic activity of this particular patient population. 

 
27:04 
I'm just going to mention Gaucher as one of them that's the lack of enzyme that breakdown certain types of sphingolipids. 

 
27:13 
So end of the day, you need to also check the phenotype. 

 
27:15 
So you need to measure the enzymatic activity. 

 
27:18 
Make sure that those patients also not only have a the genetic deficiency, but they have also deficiency in the enzyme activity itself. 

 
27:29 
Because some patient may carry the gene, but they do not have a really a symptomology or they do not have a deficient the disease. 

 
27:39 
Maybe the 30% is enough. 

 
27:41 
So you may need multiple tools to drive your drug development, either patient certification, drug monitoring or predicting response. 

 
27:52 
One size doesn't fit all. 

 
27:54 
You have to think about understand the biology of the disease and make sure that you have the right tool to these right questions, right. 

 
28:03 
So that's kind of my kind of overall type of response to that. 

 
28:13 
So the other question, Daniel, you want to tackle this or you want to carry on? 

 
28:19 
I see the chat now again. 

 
28:20 
Yes, OK, good. 

 
28:22 
Yeah. 

 
28:22 
So the next question that came in via chat is what technological advances in tissue phenotyping are you particularly looking forward to? 

 
28:33 
I have to say, I have been, I mean, looking forward to seeing more of the tissue phenotyping using RNA. 

 
28:47 
So to actually sequence on the spot in the tissue for the pathways that you're interested in. 

 
28:53 
I think this is also something that's going to, we already have it, we already do it, but I think it's not yet fully embraced. 

 
29:03 
I would say it's definitely not used in the clinics a lot, definitely not in immunology, I would say. 

 
29:10 
But the spatial transcriptomics for me is definitely a technological advancement that will put a lot of multiplexing into the tissue phenotyping to help us understand the different pathways. 

 
29:25 
I don't know, Mohamed, anything else you're looking forward to in the tissue phenotyping in the future? 

 
29:30 
Yeah, No, I think you hit the nail on the head, Daniel. 

 
29:35 
I think the spatial omics in general are, I think this is the future of really translational medicine because you can gain a lot from this and you can address the ultimate question about the tissue heterogeneity or whatever this is you're studying by tracing back those molecular phenotypes to a single cell in a particular tissue in particular space. 

 
30:02 
So that's going to be a really game changer. 

 
30:05 
But I think in addition to transcriptomics, people have been looking at other omics, if you will, in this epidomics, proteomics, although the technology is not as mature yet as transcriptomics because it's at the start. 

 
30:20 
But I think it's getting around. The other one that I would mention right now that probably some of you already work in this 3D imaging. 

 
30:28 
All right. 

 
30:28 
So we have been looking at 2D imaging taking a slice and say, OK, this is a representative of the tissue heterpgeneity, you know, that's not necessarily true. 

 
30:41 
So there are technologies out there. 

 
30:43 
And maybe I can share some of those with you guys later on looking at the 3D imaging. 

 
30:51 
So there's technologies out there using 2 photon microscopy and others that look at the different slices of the tissue by having some clearing type of reagents that take out the connective tissues and that allow the microscope to go through the depth of the tissues and really reveal novel information you wouldn't be able to see in a 2D type of thing. 

 
31:18 
So that's another exciting technology, cytof is another one that potentially can give you a high resolution of the molecular making of the tissue. 

 
31:35 
So that I think those are three main technologies and ultimately also, as Daniel mentioned, the multiomics really putting the whole picture together in an integrative fashion to get a deeper understanding of the disease. 

 
31:52 
And I think with this type of technology comes again, the complication of data when we did hundreds of samples before, let's say blood transcriptome, we can now do 10-20 of spatial transcriptomics and have to integrate all the data. 

 
32:12 
And the more of this data we have and the more complex actually to put the picture together. 

 
32:19 
And of course also complex to make them a biomarker, because the ultimate aim is then also to use it later to diagnose or to predict response. 

 
32:31 
And of course, this all requires then the use of sophisticated artificial intelligence models to all to not only, and I think that's one of the discussion points, I think we have to start with the discussion points. 

 
32:47 
So one of them is actually using AI to develop and test the hypothesis. 

 
32:50 
I think this will become more and more important when we move on with these Multiplex technologies. 

 
32:58 
And let me go to the first number of questions that we had listed, not going to go through all of them. 

 
33:05 
We had one that we wanted to start with as to how to make sure mechanistic understanding is translated to clinical studies. 

 
33:12 
And it fits for me a bit into another question that we got in the chat. 

 
33:16 
The question there was most frequent translational issues and how to overcome them. 

 
33:20 
So I think we are in the right spot to discuss that. 

 
33:23 
And I think one of the, when we talk about the issues that we have is first of all, to have a very solid understanding of the mechanism. 

 
33:37 
That's what as a pharmaceutical company that we understand what our drug candidate is doing molecularly, phenotypically, maybe even. 

 
33:48 
And then to move that understanding that is put together usually in a research environment to make sure that this is then bridged into clinical studies. 

 
34:02 
Because that's where we generate in the clinical data to understand if there is a link between our mechanistic understanding and what our compound, what our drug candidate does and how this then maybe actually also leads to identification of a responder population and then later predictive biomarker. 

 
34:21 
So I think a lot of difficulty there is that it's a long process. 

 
34:25 
It starts in research and it ends actually at the very end maybe when the drug is on the market or even later. 

 
34:33 
And it's also very important that our mechanistic understanding that we generate in research is as human relevant as possible. 

 
34:43 
So that we are not chasing animal model artefacts or cellular artefacts that we have generated in research. 

 
34:50 
But that we have a high understanding, a high level of evidence for our, for the human relevance of our mechanism. 

 
34:58 
And that we then make sure that we test this in the clinics. 

 
35:01 
And there I think one of the difficulties is of course, that clinical trials have to also adapt to some of these new technological advances that we have not only in how to actually measure biomarkers, but also how to analyse them. 

 
35:21 
So I think it's also these are for me, the critical points 1 needs to take care overcome in the organisation that you can at the end have a successful, I would call it biomarker strategy or precision medicine strategy. 

 
35:38 
So Mohamed, what do you think? 

 
35:40 
Yeah, I think you hit all of the notes, Daniel, perfectly. 

 
35:44 
And I think that also kind of cover some of the question that have been raised about the challenges of transmitting the basic discovery to precision medicine and how you overcome them. 

 
35:56 
I think one thing I would comment on it just in general is that there is no foolproof way of ensuring success. 

 
36:05 
There's going to be, I mean, there is a reason the success rate or attrition rate of a drug is like, you know, one in 10, right. 

 
36:15 
This is the reason why one drug out of 10 actually makes it to the clinic and benefit patients. 

 
36:20 
And that's basically, you know, basically give you an idea why some of the finding that we see in the pre-clinical phase of the drug discovery does not translate to the clinic. 

 
36:33 
So the chance of failure is always there. 

 
36:38 
And I think we tried to mitigate some of this by choosing, as you mentioned and the right clinical model, really have a deeper understanding the biology, have any types of make sure that we do all of the right things to make sure that there is a high success rate. 

 
36:57 
But again, that does not warrant things. 

 
37:00 
And sometimes the mechanism of action is not successful and the disease itself that's absolutely fine. 

 
37:05 
I think even the failure, it's a good opportunity to learn for the next assay. 

 
37:10 
For instance, why this assay didn't work? 

 
37:13 
Well, it's not hitting or it's not modulating for instance, the critical disease pathway underneath the target itself. 

 
37:21 
And maybe you're not doing that. 

 
37:22 
And then we learn from this and we go back and what I call it reverse translation, go back to the preclinical from the clinical say, OK, this now doesn't work because we could not modulate those important disease pathways. 

 
37:37 
Now we need to design a better drug that maybe can modulate this in a better fashion or maybe different modality or something, right. 

 
37:47 
So I think it's a helpful to make sure that we do those analysis to inform not only the forward translation, but also the reverse translation once reached the current. 

 
38:01 
And what is in your view a good time point or maybe a critical time point to start to think about this like when should we start to build a strategy here? 

 
38:15 
And it's not about just building a strategy and then doing the trial, but when does the company have to start to generate the data? 

 
38:24 
I think as early as possible. 

 
38:25 
In general, once you have a think lead candidate, that's the, I think the maybe the sweet spot as you mentioned, where we say, OK, we have a lead candidate, we think it works this way. 

 
38:40 
This is the hypothesis. 

 
38:41 
What type of data we need to think about, what type of data we need to generate to build the confidence in this particular mechanism of action and increase our chances of success and make sure that it's really modulate the path, the right pathway that's relevant to this particular disease. 

 
39:01 
I think even at this point, we need to think about what types of disease indications we need to think about beyond the first one that we think about what type of that can maybe expand. 

 
39:12 
If it doesn’t work in an indication, maybe it will work for specific indication. 

 
39:17 
I'm sorry, maybe it work for other indications where you know, maybe prior exposure to other drugs will help or you know, so all of those questions I think should probably start as early as the lead candidate selection and even prior to that, in my opinion. 

 
39:35 
Would you then do that for all the projects or would you kind of prioritise based on specific criteria? 

 
39:40 
Where do you put more or less effort? 

 
39:42 
It's maybe good to know because we all usually work on the resource constraints, money constraints. 

 
39:49 
So how do we decide where to put the efforts? 

 
39:55 
Greg, this is like the $10 billion question, all of us trying to think prioritisation. 

 
40:05 
I think it really it. 

 
40:07 
I'm going to give you the genetic answers, a case by case. 

 
40:10 
You know, if you're working in ecology with a specific mutations that impact 1 or 2% of population for specific type of cancer. 

 
40:23 
Then I think it's a clear cut case and then you know, what need to do to get the minimum data package, technical data package to move forward to the clinic, you know, But if you're looking for more like a wider disease, like an autoimmune disease, RA, for instance, that situation has been different because you may want to think about the disease axis, for instance, or potentially molecular pathways that impact this disease that you want to modulate. 

 
40:55 
And you have to look also other factors outside the biology, which is what are the competitive, what is the competitive landscape of this particular disease area? 

 
41:06 
Before you invest the money and the time and the effort to develop a drug, what's your success of or what's your chances for you to distinguish your drug from everything else out there, right? 

 
41:20 
Is it the biology? 

 
41:21 
Is it the side effects? 

 
41:23 
Is it the cause? 

 
41:24 
Is it, what is it exactly? 

 
41:25 
So you need to factor all of those external factors in addition to the biology and the mechanism of action when you make a decision or you try to develop a biomarker strategy for your assay. 

 
41:42 
Excellent. 

 
41:42 
Yes, I fully agree. 

 
41:45 
And I think they and a very important point is actually your competitors, because your competitors can be your collaborators on some aspects, right? 

 
41:56 
And one of the questions is from the chat is also asking what are the mechanisms of collaborations? 

 
42:04 
It says with pharma, but I think even between pharma because for complex diseases to really work out new biomarkers is something that actually companies best maybe do together to put the resources together and try to. 

 
42:24 
And I think then we move to the next one understand disease heterogeneity, because this disease heterogeneity is important for all of us. 

 
42:33 
And of course the drug response we want to look at is maybe similar, but maybe not. 

 
42:39 
So the company, the propriety information still is what does our mechanism do. 

 
42:46 
But I think on the disease heterogeneity, we can collaborate between companies pre competitively or also with of course other players in the field. 

 
42:58 
So what do you think about these collaborations? 

 
43:02 
Yeah, Daniel, this is another good point that you raised right now, I think I wholeheartedly agreed that we need more collaborations to really tackle those very difficult questions like really understanding the disease biology and heterogeneity across multiple disease and therapeutic areas. 

 
43:20 
I think there's many success stories and many initiatives out there and the different indication, I just highlighted a few obviously the cancers atlas genome, it's one of those you know, success stories where people put all of their data, molecular data or some of their molecular data in I think 20 or 30 different tumours and the people research from around all can tap into that to really the develop hypothesis or test hypothesis. 

 
43:50 
So that's one example. 

 
43:51 
And of course in the immunology space, there is multiple consortia where people also collaborate and provide data from their internal asset utilised data obviously so the people can collaborate across different companies. 

 
44:09 
3TR profile and many other consortium immunology space has been ongoing for a few years now towards that goal. 

 
44:20 
But I think we need to build at least in non oncology, I'm going to say a portal of some sort where some of those unanimous data can be shared across the globe, not only for the companies were participating in this particular consortia here. 

 
44:38 
And then I think, you know, there's some legal issues with that. 

 
44:41 
There's consent issues. 

 
44:43 
There is other stuff that need to be resolved regulatory issues, of course, beyond the science, but that's something I think we need to work harder on for sure. 

 
44:57 
And I think also TCGA is a good example of RNA expression. 

 
45:00 
Again, there's something that is easily done across many sites where samples are taken and that I think it also shows again how important this type of data is to characterise the tumours, but also to identify biomarkers in the future. 

 
45:22 
I think a very good example for example, immunology or for disease in general is the open target initiative. 

 
45:31 
They also just announced an IBD focus there because I think they realised how heterogeneous IBD is that there is a lot of interest to better understand the heterogeneity in there. 

 
45:46 
The open target is an example where there are multiple pharma companies coming together for actually working around matching targets with disease, which is maybe less about just disease heterogeneity, but still brings all this data together that can also be then used and integrated with the data and that you've generated in house. 

 
46:09 
But I guess when we talk about this disease heterogeneity and using a lot of these multi-dimensional or these large data sets or RNA sequencing when we go into the clinic. 

 
46:21 
So what are the implications for actually the clinical studies, the for the designs for the feasibility when we come with our, let's say AI developed hypothesis. 

 
46:35 
So how does that then how does this work with this interface with the clinical studies? 

 
46:46 
Yeah, I mean, yeah. 

 
46:47 
Well, that's a great question. 

 
46:48 
I think that's another great question that you put here that we are always struggling with. 

 
46:53 
I mean, I think that the critical design is a huge issue that includes multiple stakeholders. 

 
46:58 
But I could think of, you know, in oncology, for instance, people embark upon basket trials and umbrella trials and all this kind of good stuff. 

 
47:06 
Because you wanted to know as quickly as possible if this drug works in multiple indication or what indication the drug will work on. 

 
47:15 
That's not feasible or not really possible. 

 
47:20 
And you know, much more difficult to do in other therapeutic areas such as autoimmune diseases because of, you know, cancer. 

 
47:29 
You have the standardised wave monitoring and shrinkage, right, that can apply to multiple tumours. 

 
47:37 
So you can quickly use this to monitor if your drug works or not. 

 
47:41 
That's not the case in immunology. 

 
47:44 
Each immune disease has its own clinical readout if you will, or measurement that defines success or efficacy. 

 
47:53 
Nonetheless, I think the one area that I think it's going to be really helpful is that maybe the biomarkers can really be the guidance or maybe a network of biomarkers or multiomic signatures can maybe drive what indication we should target rather than start from the histology based type of indication and see what happens. 

 
48:21 
I think that will be really helpful improvement and potentially paradigm shifting concept that we can use to really expand indication and expand patient populations that can benefit from a particular drug by looking at what pathways that this drug is actually modulated and look at the commonality. 

 
48:45 
And there is examples from this an immunologist space, for instance, depending that from the region around it's 1 success story that has been this is IL-4 IL-13 receptor pathway and has been shown to work and I think 5 or 6 different indications based on the type 2 inflammatory biomarkers, right. 

 
49:07 
So that's one example of immunology like almost a precision medicine approach that have been used to identify new indications agnostic of the tissue of origin or the location of the disease. 

 
49:25 
And there is more are coming also. 

 
49:27 
I mean TNF alpha has been used to obviously multiple indication as well, but there is more tailored ones are also in the pipelines from different companies aimed at again using the molecular underpinning of the disease or pathways to derive or expand the reachability of a specific drug and inclusive success in multiple patient population. 

 
49:54 
So to me those are I think the potentially more the low hanging fruit for this. 

 
50:04 
Yes. 

 
50:04 
And I think immunology, one of the difficulties sometimes are the endpoints. 

 
50:09 
And I think that's when biomarkers can help you to say, you know, you're doing something that goes in the right direction. 

 
50:17 
Although you're super heterogeneous endpoint like a score that is calculated out of 15 different factors, it doesn't really show an effect. 

 
50:25 
But I think you can start to understand what you're molecularly doing. 

 
50:29 
And you might based on that also then enrich your population in your trial to see maybe the response is large enough to be then successful. 

 
50:40 
So II mean, biomarkers is evolving. 

 
50:45 
It's something that has to evolve over your development. 

 
50:50 
And I think it can end with the precision medicine outcome doesn't have to, but can. 

 
50:57 
And yeah, that's a great point. 

 
51:00 
And I cannot help but comment. 

 
51:01 
And this is actually is the great point here that the surrogacy can we use this biomarker if we have a strong surrogate biomarker biomarkers, can we use this actually to potentially predict response and maybe be the base for potential phase two, phase three pivotal trial based on this rather than doing each clinical endpoint. 

 
51:27 
That’s a great point. 

 
51:33 
Then we also wanted to touch on single versus multi-dimensional biomarkers. 

 
51:38 
I would say we wrote down why multi-dimensional. 

 
51:42 
I think it's kind of clear that many of these diseases we discussed are so heterogeneous that you likely need to look at different dimensions. 

 
51:53 
And I think there are very different dimensions. 

 
51:55 
Of course, I think one of them is let's say we have something that we can measure like gene expression, but it might also be something that the physician subjectively actually identifies. 

 
52:07 
So there might be different dimension that you have to combine or that might help you to easily make a prediction. 

 
52:19 
Sometimes it's also easier because those different dimensions like an assessment from a physician might be already accepted by the clinical community. 

 
52:33 
So it's already there. 

 
52:34 
You don't even have to do much validation. 

 
52:36 
It's there. 

 
52:37 
If it's stable, you can use it and combine it maybe with something else. 

 
52:42 
So I think the multidimensional can be an approach to be to make it feasible. 

 
52:48 
But of course, I think it's mainly coming from the fact that those diseases are very heterogeneous. 

 
52:53 
Sometimes it's not one disease, sometimes it's just patients put into a disease and they are very different diseases at the end. 

 
53:03 
And I think then the question is maybe a bit, and if we go into these multi dimensional biomarkers, like how do we validate and qualify them? 

 
53:11 
Because we are now putting a lot of dimensions together, maybe some that are measurable and some that are not so measurable, so what are your thoughts on this Mohamed? 

 
53:18 
One huge topic and that actually almost needs one entire webinar or even multiple webinars to discuss frankly,. my quick thoughts are, you know, I think we can follow the NIH or this USNIH best type of paradigm. 

 
53:42 
You know that the two stages of validating or qualifying biomarkers, right is, you know multiple steps you have to have like we stand the rise and save doing that. 

 
53:58 
You know, if this biomarker is a use for decision making basically using a fit for purpose approach based on the context of views, the degree of validation needs to be done. 

 
54:11 
So I think the analytical validation as well. 

 
54:15 
And then make sure these assays are, you know valid. 

 
54:18 
And then the second one is the clinical validation itself and retrospective and prospective trials to make sure that this biomarker performed as advertised and is associated with the clinical endpoint of interest, right. 

 
54:32 
So again, I think there's a distinction between a biomarker that we use for discovery or for internal decision making versus a biomarker that we are planning to use to drive the drug development and potentially share with regulatory authorities as a proof of concept of more proof of mechanism diagnostic. 

 
54:56 
Those are different tiers that were on different types of validation. 

 
55:03 
Yes, in light of time, we'll move to the last point because we cannot have the webinar without AI. 

 
55:10 
So the last one using AI to develop and test hypothesis. 

 
55:17 
And I think one first question, can patient heterogeneity be better captured with AI? 

 
55:23 
And I personally believe yes, because the multidimensional, the multidimensionality is very difficult fast to grasp, to actually grasp, to make a hypothesis as to develop one and then also to test it, of course. 

 
55:39 
So I would say yes, we need to integrate it with all the complexity that we have, maybe not for all the diseases, but for the ones where we are struggling. 

 
55:46 
I think this is going to help a lot. 

 
55:53 
I see they have one question. 

 
56:00 
How much again we are already seeing large scale multimodal cell and disease modelling with AI. 

 
56:05 
How much can these AI generated disease models contribute to mainstream drug discovery development? 

 
56:10 
Yes, I mean, that's exactly the process that we have to go through because you might have the models that do stratify your disease, right. 

 
56:22 
So you might have a model that says there are three types of patients. 

 
56:27 
And of course what one can use this model to actually stratify the patients. 

 
56:35 
You can use the model to retrospectively look at, of course, also the response rate to see if maybe these models identify a responder population for your drug. 

 
56:45 
That's a way of doing this. 

 
56:46 
Then of course, if that would be the case, then you need to and you want to move on. 

 
56:54 
You need to follow up and then you need to make these disease models. 

 
56:59 
You have to follow the companion diagnostic trajectory potentially, if that is what you want to do, if it's something that it is used more already, maybe it's already used in the clinics. 

 
57:11 
There is already a use case for something else, let's say to predict flare in the disease or to predict a severity. 

 
57:20 
Then of course it could serve as a complementary diagnostic. 

 
57:23 
So principal it's already there. 

 
57:24 
You can use it and you have a higher response of your drug in that population. 

 
57:29 
And then it's more a question of how do you make sure that then the physicians have access to that technology, to that model. 

 
57:39 
And I think that's of course then the difficulty if it is, yeah, I don't know. 

 
57:45 
I mean that's maybe the difficulty, but maybe also the easiness. 

 
57:48 
Because I think if you can have something like this on a system available for all the physicians, then it's again easy to, it's maybe easier to have access to that than to generate a lot of data at the physician side. 

 
58:05 
So I think that's a way how it can contribute, I would say. 

 
58:11 
But then of course also we as the industry want to have models that are built based on our Amway because that's again then we are they're not testing splitting by disease, which might be relevant, but we're testing for our mechanism. 

 
58:26 
And of course, that would be in the way how we develop those models and use them in clinical trials. 

 
58:32 
Mohammed, anything to add on this one? 

 
58:34 
Yeah, well, I would love to add more because I agree with you on everything. 

 
58:39 
But I think when the top of the hour here and I know that a lot of people will be starting to drop off. 

 
58:45 
So I just wanted to thank you Daniel personally for really rich discussion and bringing great points. 

 
58:52 
I hope the people who listen to this will have benefited a little bit and we are more than happy, I guess personally and maybe Daniel is to answer any further question down the road as well. 

 
59:03 
So thank you, Daniel, and thank you, Mohammed. 

 
59:12 
Thank you so much, Mohamed and Daniel for spending the time to share your knowledge with us today. 

 
59:17 
It's very much appreciated. 

 
59:18 
And thank you everyone for joining us. 

 
59:19 
We hope you can join one of our monthly site changes in the future. 

 
59:23 
And I hope you enjoy the rest of your day. 

 
59:25 
Thank you very much.