0:47
Super excited to be here and chat a little bit about Sapient’s multi omics offerings and how these can accelerate discovery both from a target ID as well as drug development, particularly in oncology therapeutics.
1:12
So as mentioning, we'll chat a little bit this morning about multi omics and how this can propel next generation cancer therapeutics.
1:49
So just as we think about biomarkers in oncology, certainly this has been an area of great innovation over the last two decades or so.
1:56
And when we really think about oncology as a whole, our basic view of this has been grossly sort of altered by these multi omic technologies.
2:05
And so if we think back just 20 years ago or so when I was in medical school, our traditional view of cancer was one that's really based on histopathology, whereby there are three classifications for lung cancer based upon how this looks on histologic slide.
2:17
And we would designate this as adeno, squamous, or large cell lung cancer.
2:22
Well, over the subsequent decades, we've seen with the discovery of molecular pathology, particularly in the genetic space, how this has evolved in the early 2000s with the identification of KRAS and EGFR.
2:33
And now when we think about lung cancer today, there's over three dozen subtypes of lung cancer, each of which is designated by its own mutational spectrum, each of which we know response to a completely different therapeutic.
2:45
And in doing so, we've been able to leverage these genomic biomarkers to really personalise our understanding of lung cancer.
2:52
Now traditionally, much of this work has really been centred on genomics, meaning both DNA as well as RNA.
2:59
And again, this has been incredibly helpful advancing our understanding of these diseases, but they've also been somewhat limiting in our therapeutic applications.
3:08
And the reason for that is that genetics largely represents a minority fraction of attributable disease risk.
3:15
In the oncology space, for instance, this is estimated to be somewhere on the order of 10%.
3:19
Thinking through host genomics and how that may influence development of disease and for other disease states all the way up through autoimmune disease, it may be in the 20 to 30% range.
3:29
Irrespective of the underlying disease type, it's still minority.
3:32
And in the most cases, your genes are set from the moment of conception.
3:36
And so these are not altered through your lifetime.
3:40
When we think about how we may be able to go beyond the genome, this is where really multiomics has had an enormous amount of impact in thinking through beyond just DNA and RNA to proteins and particularly now proteoforms looking at variants of protein.
3:53
So not only the protein itself, but the individual isoform of the protein, whether it be a post translational modification such as glycosylation or phosphorylation, a variant including non-canonical variants or splice variants, et cetera.
4:07
And then also as we extend this beyond just proteins in large molecular to also small molecule designees such as metabolites and lipids.
4:14
Now when we think about those entities that are to the right of the slide here, these reflects not only endogenous processes, but also exogenous processes, influences such as diet, lifestyle, environmental factors, et cetera, all of which can influence disease risk.
4:29
And in being able to measure those forms that are to the right here, we begin to understand the underlying functional mechanisms that drive disease states.
4:38
These are highly dynamic, which means they can be used for diagnostic purposes both in understanding the disease process as well as monitored in real time to understand therapeutic response.
4:48
And most importantly for the sake of this talk, these really begin to provide some key actionable targets that are central to oncology therapeutics.
4:57
This is particularly important as we think about how oncology targeting has evolved over the last several decades.
5:04
Traditionally again the druggable genome was listed only as or limited to only several 1000 proteins within the whole genome here and with over just under 1000 or so protein targets which have been approved for FDA therapeutics with a long tail.
5:19
Today, particularly with the new drug modalities whether it be CAR T therapies as we heard about from John or protein degraders, T cell engagers, CRISPRs, radio ligands, ADC's et cetera.
5:30
The spectrum of potential targets has greatly expanded from under several 1000 proteins to now the entirety of the underlying human protein, whether it be 20,000 proteins or over 100,000 various proteoforms.
5:42
And this represents now a significant opportunity for novel target identification.
5:50
Again, when we think about how target identification has largely been done in the past in the oncology space, it's largely been limited to use of genomic and transcriptomic measures as surrogate measures to understand potential protein expression as a means for identifying targets.
6:05
And this is somewhat limiting for many potential applications.
6:09
And that when we think about the relationship between the underlying transcript level of a gene, for instance, as well as its actual expression, it's protein expression within the tumour, what you see is often times to the right here where there's very limited relationship and it and honestly looks like a cloud of randomness.
6:28
If you look at the distribution across all proteins and all RNA species within a tumour, you see the figure here on the left whereby the R-squared value is centred on zero.
6:37
So there's very much within the state of disease very limited correlation between the amount of RNA for a particular target and the amount of protein expression, which is why it's absolutely essential to be able to measure these proteins directly.
6:51
Now traditionally the limitation has been really and how do we begin to measure these protein targets and how do we begin to understand these potential markers of disease.
7:00
And this has changed massively over the last several years, particularly over the last 18 months due to innovations both in sample processing, underlying mass spectrometry instrumentation and underlying software.
7:10
Now that allows for distributed based computing.
7:13
And what these technologies have been able to do is to greatly expand the coverage for those proteins that we measure.
7:19
Traditionally this has been in the hundreds of range.
7:21
This is now extended to thousands and 10s of thousands as I'll show you in just a moment in a particular bio sample, our throughput has greatly expanded from dozens of samples to a day to now thousands.
7:32
And this really allows for hypothesis free discovery, particularly using non-targeted multiomics including now with AI classifications.
7:41
This just gives you a view of our laboratory Sapient operates worldwide, but we're based in San Diego where our physical labs are located.
7:48
And you get the sense very quickly for the amount of mass spectrometry infrastructure and robotics we have in place.
7:54
And what these technologies enable us to do is to take any biological specimen, whether it be a liquid specimen such as blood, urine, CSF, tears, eye fluid, saliva, et cetera, or solid sample such as a tumour, a tissue, cells in a dish, media, et cetera.
8:07
And in any of those biological specimens.
8:09
Now we can measure thousands of proteoforms, metabolites and lipids across virtually any sample matrix, whether it be human or preclinical.
8:17
We leverage a number of different mass spectrometry technologies for our high throughput work with limited typically to the Bruker timsTOF line of instrumentation.
8:25
And these are very much amenable to precision targeted guided approaches where you can be done in a discovery manner where you're measuring thousands of these species, or in a very targeted manner where you're limiting yourself to very key proteins of interest.
8:38
This allows us now to capture multi omics at great breath and in depth, far greater than was previously possible.
8:44
And to give you an idea of what these numbers look like, for instance, for discovery proteomics, now we can assay over 12,000 protein groups in a tumour or tissue within liquid samples such as plasma, CSF, urine, et cetera.
8:58
We're on the order of five to 6000.
9:01
And these allow measurement not only the protein itself, but the actual isoform of the protein, the proteoform, including post translational modifications and variant proteins, not only in the free floating proteins, but also an exosomal and EV proteins.
9:15
For metabolomics and lipidomics, we can assay more than 15,000 metabolised lipids in a biological specimen.
9:21
We also have specialised multiplex cytokine and chemokine assays that leverage technically called NULISA-Seq that allow us to measure 350 plus different markers.
9:30
There we provide services for bulk and single cell RNA sequencing, particularly as a means for enabling proteomics based discovery as well as targeted assay development for difficult to assay proteins.
9:41
Where why we can develop these targeted assays. To give you an idea of the workflows.
9:46
Each of these systems, whether it be around proteomics or metabolomics and lipidomics leverages unique sample processing in the sake of proteomics based approaches.
9:56
We're using a novel nanoparticle system that allows us to extract again free floating proteins, but also exosomal and EV based proteins, whether it be from plasma, urine, tissues and tumours et cetera.
10:08
We're using state-of-the-art mass spectrometry with various types of chromatographic systems and then specialised data on the back end and software that allows us to reconstruct protein levels, their variants or metabolites.
10:21
In the end.
10:21
What this allows us to do once again is to broadly capture various protein isoforms, PTMs and very high throughput or to assay thousands of metabolites and lipids in very high throughput across biological specimens.
10:35
These approaches have been applied across the entirety of the drug development pipeline from very early on with target identification all the way through initial screening.
10:44
We do quite a bit of high throughput proteomic based drug screening including in the targeted protein degrader space, preclinical studies for both forward and reverse translation as well as in translational and clinical studies for understanding patient stratification, dosing, target engagement and ultimately companion diagnostics for responder versus non-responder status.
11:05
Now to give you an idea of some of the data and what this actually looks like, here is some of the data in a small number of individuals and plasma samples.
11:14
Using these plasma proteomic workflows, we can capture more than 5400 different protein groups that includes multiple different protein isoforms.
11:23
So the total number of isoforms is much higher.
11:26
And what's particularly special around the types of nanoparticle capture and hold systems that we're using is that they are really quite robust and that they're not influenced by other additives within the sample.
11:39
They use a glueing mechanism whereby once there's an interaction between the nanoparticle and a protein that is stuck in place and that means you don't see this drop out.
11:47
That is typical with nanoparticles whereby 1 sample is 5000 proteins, the next one is 500 proteins.
11:52
And so we're able to keep consistent levels of measure across large numbers of samples.
11:57
You can see across a cohort here of about 50 or so samples, there's a minimum of 4500 proteins per sample and then across the entirety where we're in the 5000 to 6000 range.
12:10
Now proteomics is special and that it allows us not to only capture single proteins for targets, but to also understand biological processes and pathways that may be important for disease onset or drug response.
12:22
And so even from the plasma proteome itself, there's over 100 different signalling cascades.
12:27
They can be assayed as shown here on the left.
12:30
This includes everything from torque signalling and complement signalling to various interleukin and inflammatory pathways, notch signalling, et cetera.
12:39
And not only again once understanding signalling cascades, but also understanding drug response.
12:45
And so among the several 1000 proteins captured within the plasma proteome, there are several hundred of which represent approved drug targets or emerging drug targets, which are quite important for understanding pharmacodynamic response.
13:00
Now as part of these operations, in order to enable sort of discover using these technologies that we've also built very special internal data assets that we term our human biology database that now represent over 50,000 biological specimens from over 10,000 individuals with longitudinal clinical outcomes and which we've been able to perform these multi omic measurements.
13:21
This data along with all the clinical information has been amalgamated in a centralised database which can be leveraged as part of validation for a biomarker or to better understand a biological process that may be at play with a drug response.
13:35
In addition to the plasma work, we've also built an extensive tumour database that now extends to over 1000 tumours plus 15 normal human tissues across common cancers as you see here on the right.
13:48
And this has really been particularly important in aiding our clients and identifying new potential drug targets.
13:54
And so shown here is just some data from a gynaecologic cancer relative to normal adjacent tissue.
13:59
As you can see on the volcano plot, not surprising, there's thousands of proteins that are differentially expressed in a tumour sample.
14:06
We can cross this against all normal tissues and begin to understand those proteins that are unique to the cancer itself.
14:12
And what's particularly exciting when you look at known and emerging drug targets that are shown here in orange, there's a number of common drug targets that we've all heard about, FOLR 1, EPCAM, et cetera.
14:22
But there's also dozens of additional proteins that you see at the edges of the volcano plot here that are as differentially expressed for which they are not currently therapeutics that are under development.
14:34
And these really represent new potential drug targets that are being pursued by a number of clients together with Sapient.
14:42
Now these really here are describing canonical proteins.
14:44
When we extend this to non-canonical proteins of the dark proteome, meaning those variants of proteins that are not typically sort of described in the human genome, this becomes even more differential and you can really begin to find specific proteins that are unique to the tumour state.
14:59
One here is this one such example by how we begin to leverage this not only for target identification but also for drug development and in rescuing of failed therapeutics.
15:09
And so this is a programme in ADC together with the top five pharma company in which an antibody drug conjugate therapeutic had failed in a phase one studied to the lack of efficacy relative to placebo.
15:20
And by leveraging these plasma proteomic tools, particularly those around unique proteins that were in this particular tumour type, we were able to identify a pharmacodynamic response biomarker for the singular agent.
15:32
And when you look on the right you subset those that received a therapeutic according to those that had a pharmacodynamic response versus not that you can see that these two survival curves begin to display significantly.
15:43
And in doing so, you can begin to subclassify individuals who respond to a particular therapeutic for those who do not.
15:49
And so this has now been developed as a companion diagnostic along with this particular therapeutic.
15:56
We'll go ahead and stop there.
15:58
Time allotting.
15:58
Happy to take any questions.
16:00
We're at booth 78, so I would invite anyone who's interested to please stop by.
16:04
Would be happy to chat.
16:05
I realise this is a rather quick presentation, but happy to really talk to any of the technical details that may be of value to anyone.
