Latest / Zero Shot Podcast DEMO / Revolutionizing Cancer Diagnostics with OneCell.ai V2

Transcript

Revolutionizing Cancer Diagnostics with OneCell.ai V2 · Zero Shot Podcast DEMO

• 0:00
  Mohan Utawar: And the pathologist puts it under the microscope, does the reading, and based on the pattern he or she sees, write a
• 0:08
  report.
• 0:08
  And based on that report, a treatment is given, okay?
• 0:13
  This process is called tissue biopsy, okay?
• 0:20
  speakers: This conversation follows the people, decisions, and stakes on the tape.
• 0:31
  Liquid biopsies utilize blood samples to generate reports using next-generation sequencing, making them a less invasive option for cancer diagnostics.
• 0:44
  Biopsy is the process of taking a small tissue sample from the area where you suspect that there is cancer,
• 0:51
  and further analyzing it.
• 0:53
  It's sent to the lab, that tissue sample.
• 0:57
  It's called tissue biopsy.
• 0:58
  Tissue sample is sent to the lab.
• 1:00
  The lab does the processing of the sample.
• 1:03
  A glass slide is created, okay?
• 1:10
  Interviewer: So that's the liquid biopsy aspect, right?
• 1:13
  From the sample of blood draw that you create the reports, generate the reports, using
• 1:21
  next -generation sequencing.
• 1:23
  Okay.
• 1:24
  Okay.
• 1:24
  Now, when you're doing the liquid biopsy, like, do, is this circulating, like, tumor cells or circulating tumor
• 1:32
  DNA, are they available for pretty much most cancers, or is that, like, more prevalent in some cancers versus not,
• 1:40
  or even, like, the staging?
• 1:42
  Like, maybe, like, when it goes metastatic, it's more present.
• 1:45
  Like, how does that work?
• 1:47
  What happens is cells die, some natural death and tumor cells, right?
• 1:53
  So, when they die, it spits out the DNA.
• 1:59
  That's the circulating DNA or cell -free DNA, okay?
• 2:04
  Subset of the cell -free DNA, right, from the dead cells, you know, if it is a tumor cells, the death of a tumor cells,
• 2:12
  which spits out the DNA, it's called CT DNA, okay?
• 2:16
  Okay.
• 2:17
  Information that you generate from the CT DNA, circulating tumor DNA, is most relevant,
• 2:25
  okay?
• 2:26
  But
• 2:29
  speakers: blind person touches, you know, let's say the tail and it says, oh, it's like a snake.
• 2:34
  You know, so unfortunately, the genomics, the transcriptomics, the proteomics, the methylomics,
• 2:42
  the metabolomics, these are all fields of science.
• 2:46
  But they are looking at cancer into the silos with their own lenses.
• 2:55
  Right?
• 2:56
  So the challenge here I see is that, like, usually, like, when you do, like, genomics, proteomics experiments, right, like, you have, like, a normal tissue, you have
• 3:04
  a disease, tumor tissue, right?
• 3:06
  And you're looking at, like, differences here.
• 3:07
  But then what you're collecting, that has, like, majority of that is already normal anyway.
• 3:14
  So how do you then detect the specific cancer DNA or the cancer proteins from that?
• 3:20
  Because that's a much higher level challenge than what people usually do, right?
• 3:25
  Absolutely.
• 3:25
  So how do you do that with AI?
• 3:28
  Yeah.
• 3:28
  Very, very good point.
• 3:30
  Excellent point.
• 3:31
  In liquid biopsy, I told you it's all about DNA.
• 3:35
  And you have a challenge because you start with a noisy data.
• 3:39
  You don't know whether that DNA came from a normal cell, dead cells, or a circulating tumor.
• 3:46
  I mean, the, the CT DNA, right?
• 3:49
  Yeah.
• 3:49
  And so when you start with the noisy data, again, you have to apply AI and deep analytics, more resources to basically,
• 3:57
  you know, denoise that data.
• 4:00
  Okay.
• 4:00
  The next generation technologies, particularly the one that we have invented, you are already clean,
• 4:08
  you're starting with a cleaner data because you have figured out, we have figured out, capturing that one in a billion circulating
• 4:16
  tumor cells.
• 4:17
  Okay.
• 4:18
  We already know that our starting point is a tumor cell.
• 4:23
  Oh, okay.
• 4:24
  Okay.
• 4:25
  So you're, so you have the technology to then detect the tumor cell specifically.
• 4:30
  Correct.
• 4:31
  Okay.
• 4:31
  Okay.
• 4:31
  Okay.
• 4:32
  Now, does that like, does that depend on like the kind of like surface markers there are on like the tumor cell or?
• 4:39
  Very good point.
• 4:40
  So CTC is not new.
• 4:42
  It's been there 10 years ago.
• 4:45
  Okay.
• 4:46
  Unfortunately, what happened was the sensitivity wasn't there.
• 4:51
  And the endpoint was you capture the circulating tumor cells, you count and report them.
• 4:58
  Okay.
• 4:58
  Okay.
• 4:59
  So the clinical utility was limited.
• 5:03
  You know, it told us that there is a problem.
• 5:06
  Okay.
• 5:07
  Your pet or CT scan or MRI might be clean.
• 5:12
  But if you find CTCs, that means, you know, there are some problems somewhere.
• 5:18
  Okay.
• 5:19
  So the scientist would say, oh, we found the problem.
• 5:22
  Okay.
• 5:23
  Okay.
• 5:23
  But the doctors would say, so what?
• 5:26
  My patient is happy because the pet is clean.
• 5:29
  And now you're telling me you found a CTC and you can't tell me what to do about it.
• 5:35
  Okay.
• 5:36
  So it's a very classic dilemma.
• 5:38
  Scientists get excited with what if.
• 5:41
  Okay.
• 5:42
  But the doctors are saying, so what?
• 5:45
  What if I found a CTC and the doctor is saying, so what?
• 5:49
  Unless you tell me, what is the actionability?
• 5:52
  And the science stopped there in the past.
• 5:57
  But today with the advent of next generation sequencing and with AI, okay, we are able to analyze
• 6:05
  the information in that single cell.
• 6:08
  Okay.
• 6:09
  Okay.
• 6:09
  The genomic data, the transcriptomic data, the proteomic data, and tell you and tell the doctor, hey,
• 6:18
  you know, it may be the cell coming from lung.
• 6:22
  You need a pet on basic breast, which was the original stuff, right?
• 6:27
  So without AI analyzing the cell information that is hidden in that circulating tumor cell,
• 6:35
  we would not enable the doctor to take those critical clinical decisions.
• 6:41
  Okay.
• 6:42
  So it's really the integration of information from the biology of the cell, the
• 6:50
  science part, and the AI working on analyzing the genomic data, the transcriptomic data,
• 6:58
  the proteomic data, right?
• 6:59
  So like you said, cell surface marker.
• 7:02
  You know, we know historically that if it's, let's say, HER2 marker on the surface of the cell,
• 7:09
  very likely that cell must be coming from breast cancer site.
• 7:15
  If it's an EGFR, very likely, again, no guarantees, but very likely it's coming from a lung cancer, right?
• 7:23
  Also, the way morphology of those cells look is very unique.
• 7:27
  So we use AI to analyze, you know, how the cell looks, the morphology, you know, the size,
• 7:36
  the shape, the density, the contrast, right?
• 7:39
  So there's a lot of imaging as well, then.
• 7:41
  Image analysis of the cell that we capture, right?
• 7:45
  Okay.
• 7:46
  So in addition to the multiomics, you're also doing the image analytics.
• 7:49
  Absolutely.
• 7:50
  Absolutely.
• 7:50
  Got you.
• 7:51
  Even before we do multiomics, we do the analysis of these cells, right?
• 7:57
  Okay.
• 7:57
  We do the characterization using our AI algorithms.
• 8:01
  And we give enough information.
• 8:03
  Now, one other, you know, fact that is actually very revealing is that many times the CTCs are
• 8:11
  individual CTCs.
• 8:13
  But many times you have this, you know, what we call the combinations, right?
• 8:21
  You know, you have the, you know, group of cells hanging together.
• 8:25
  Okay.
• 8:26
  Now, the risk profile of those patients, prognostic is much, much different.
• 8:31
  It's much higher risk.
• 8:33
  Okay.
• 8:34
  Again, imaging reveals those kind of things.
• 8:37
  Okay.
• 8:38
  You know, so, so there's interesting stuff, right?
• 8:41
  Now, how do we capture these circulating tumor cells, right?
• 8:45
  The CTCs, there are two methods for capturing.
• 8:48
  One is size -based, right?
• 8:51
  So, tumor cells are usually larger in size.
• 8:54
  And there are techniques, you know, micro filtration.
• 8:59
  There are techniques to really do the nano fluidics to capture those cells.
• 9:04
  That's one method.
• 9:05
  The other method is called affinity -based.
• 9:08
  So, tumor cells have certain types of proteins on the surface of cells.
• 9:14
  So, Kepcam being one very popular protein that is on all solid tumors that are epithelial in
• 9:22
  nature.
• 9:23
  Okay.
• 9:24
  You know, a transferrin in another protein that is expressed.
• 9:28
  Okay.
• 9:28
  So, what happens in the size -based capture, the cells go through a lot of, you know, pushing
• 9:36
  and pulling.
• 9:37
  Nano fluidics, you know, you're forcing this.
• 9:42
  You know, nano filtration, you're pushing it.
• 9:46
  As is, when cell leaves the tumor, okay, you know, it's in a, it was in a
• 9:54
  very comfortable position sitting in the home.
• 9:57
  And now, it has to fight the whole battle, right?
• 10:01
  Immune system really is, you know, the aggressor which is trying to kill it.
• 10:06
  So, only a few remain.
• 10:09
  They, you know, avoid all the forces against them, you know, and so they are, right?
• 10:15
  But they go through a lot of, you know, fights.
• 10:18
  They go through a lot of transitions, transformation, right?
• 10:21
  Pressure, right?
• 10:22
  So, as a result, size -based capture has not yielded great results.
• 10:28
  Okay.
• 10:29
  Affinity capture, relatively, is less stressful on the cells.
• 10:35
  Okay.
• 10:36
  You know, so that's been historic perspective.
• 10:40
  The groundbreaking science that we have, but again, capture and count doesn't tell you the whole story.
• 10:47
  So, where one cell comes in is a groundbreaking science to isolate the captured
• 10:55
  cell into a live single cell.
• 10:58
  Okay.
• 11:00
  So, traditional circulating tumor cell CTC, the end point was capture and count and give the
• 11:08
  report.
• 11:08
  For us, this is not an end point.
• 11:12
  This is the starting point.
• 11:15
  Okay.
• 11:15
  We begin with this captured and isolated live single cell.
• 11:21
  And we do the imaging modality on it.
• 11:24
  We do the NGS on it.
• 11:27
  We do the multi -omics analysis on it.
• 11:30
  Right?
• 11:30
  And so, we have a much deeper insights that enable the doctors to say what's really going on.
• 11:38
  Is the therapy working?
• 11:41
  If it is working, why it's working?
• 11:44
  If it is not working, why it is not working?
• 11:47
  And what could you do potentially to change the course of the therapy?
• 11:52
  Can we add, you know, can we increase the dose?
• 11:57
  Can we add the combination therapy?
• 11:59
  Can we change something?
• 12:01
  Or if nothing, if it is working, can I reduce the dose level?
• 12:06
  Can I give those chemo holidays?
• 12:08
  Can I, you know, do the de -escalation?
• 12:11
  Right?
• 12:11
  So, all these questions are unanswered in the minds of the oncologist.
• 12:17
  Okay?
• 12:19
  We provide AI assistant to the oncologist.
• 12:23
  Okay?
• 12:24
  Based on this scientific data that is presented, multi -omics, you know, data that is presented for a given patient
• 12:32
  at a given time point, to be able to provide them meaningful, actionable information
• 12:39
  so that they change the course of the treatment.
• 12:43
  Personalize the treatment for a given patient, which will improve the outcome.
• 12:48
  Yes.
• 12:48
  And cancer sometimes changes as well, right?
• 12:50
  So, you have the mutations, some mutations take over versus the other, right?
• 12:55
  So, in that sense, like, if it is growing, like, you know, you can detect, like, okay, what is the new mutation
• 13:03
  that's coming up, right?
• 13:04
  And then the ecologist can then change the treatment accordingly as well.
• 13:07
  It evolves.
• 13:09
  Okay?
• 13:09
  Yes.
• 13:09
  You know, and so many successful companies, you know, and I'll give you one example, Natera.
• 13:15
  It does the tissue at the baseline, at the starting point.
• 13:20
  It hones in on an information about 16 -gene PAL, and it only monitors those 16 genes
• 13:27
  longitudinally.
• 13:30
  One of the most successful companies, one of the most successful tests.
• 13:34
  But it does not address the evolution of cancer.
• 13:39
  Okay?
• 13:40
  You cannot take a snapshot.
• 13:42
  You know, it's like, you know, I've taken, you know, before leaving my home, going to the office, if I take, let's say, you know, Google Maps,
• 13:50
  right?
• 13:51
  And I take a snapshot, and I start driving, and if there is an accident, and, you know, I have to do the stuff.
• 13:58
  I can't just, you know, it cannot be a static thing.
• 14:03
  Cancer evolves.
• 14:05
  So you have to, the testing has to kind of adapt to the evolution of the cancer, you know.
• 14:12
  And I think the newer sciences, newer tests, newer diagnostics is really more reflective of the evolution
• 14:20
  of cancer, more adaptive to the evolution of cancer, and provide the tools to the oncologist, you know, that
• 14:27
  reflects the changes that happen.
• 14:29
  And that's how the outcomes are better.
• 14:32
  Yeah.
• 14:33
  So this uncoincides, right, what you described as the clinical diagnostics.
• 14:38
  So it's a clear lab model.
• 14:40
  Can you talk a bit about that?
• 14:41
  Sure.
• 14:42
  How did you get, like, certified, and what does that entail?
• 14:45
  Yeah.
• 14:46
  Good question.
• 14:47
  There are two ways a diagnostic company can offer, you know, the new innovation.
• 14:52
  Okay.
• 14:53
  More traditional way is basically, you know, your invention, you know, you go through the
• 15:01
  IP protection, and you go through a regulatory pathways, which is the 510K for an FDA
• 15:08
  clearance, right?
• 15:09
  So you find a predicate device, some similar test that has already been approved by FDA,
• 15:17
  and you establish that your test is as good, if not better, right?
• 15:23
  You know, and you do that.
• 15:24
  Now, that takes a long time, you know, two, four years, and many tens of millions of dollars before you see your
• 15:32
  first, you know, revenues, etc.
• 15:36
  The other way is a provision called LDT, lab developed test.
• 15:43
  So if you have a lab that is CLIA and CAP certified, you know,
• 15:51
  CLIA is a regulatory body, you know, that does the inspection of the labs, ensuring that they follow
• 15:59
  proper protocols.
• 16:00
  It is run by a very qualified, board certified pathologist.
• 16:07
  Okay.
• 16:08
  And if you establish and prove that you have developed a test that has a good science
• 16:15
  backing it up, good publication, you have done a good, you know, scientific experiments, quality control, and all
• 16:23
  of that stuff.
• 16:24
  Then, you know, on a limited basis for a limited target, you could offer this, you know, lab.
• 16:32
  Okay.
• 16:33
  So we have adopted a two -prong approach, a two -phase strategy.
• 16:38
  We have a CLIA CAP certified lab in Foster City here in Bay Area, okay, where we could take the samples
• 16:46
  from the patients.
• 16:46
  We process the sample and produce the results and the reports and give it back to the oncologist
• 16:54
  based on which they can basically, you know, take the actions.
• 17:00
  Our initial focus is not commercial diagnostics.
• 17:04
  Our initial focus is to provide these services to pharma and biotech companies that are doing
• 17:11
  clinical studies, clinical trials.
• 17:14
  Okay.
• 17:16
  Okay.
• 17:16
  Okay.
• 17:16
  So, you know, pharma companies are always pushing the envelope.
• 17:20
  They want newer technology.
• 17:21
  They want new science to work.
• 17:24
  Right.
• 17:24
  And so it's a shorter path, helping them accelerate their translation research, their
• 17:32
  clinical trials, gives us a shorter path to revenue and gives us an add -on confidence that our
• 17:40
  test now then can become a companion diagnostic.
• 17:43
  Yes.
• 17:44
  And eventually a commercial diagnostic test.
• 17:47
  Okay.
• 17:48
  Okay.
• 17:49
  And so tell us about, like, the information that you provide to the clinicians or to the clinical trial
• 17:56
  coordinators, right?
• 17:59
  Is that about, okay, well, there's these mutations or, like, what kind of information are you providing to them?
• 18:05
  Our OncoInsight, which is the name of our assay, is an LDT, right?
• 18:11
  So we get two tubes of 10 ml blood each.
• 18:16
  We provide the collection kits.
• 18:18
  They are available at the doctor's office.
• 18:21
  Okay.
• 18:22
  A phlobotomist goes, you know, with the patient.
• 18:25
  We draw the sample and those samples come to us.
• 18:28
  Okay.
• 18:29
  It's a one -day turnaround time.
• 18:32
  It might take two days depending on the location, etc.
• 18:35
  And then we do the processing of these samples, right?
• 18:39
  You know.
• 18:40
  So what do we do?
• 18:41
  It's a two -track.
• 18:43
  You know, the first track is we extract the DNA, you know, from the blood sample.
• 18:48
  And we have developed part of OncoInsights is a CT -DNA section of the test where we have
• 18:55
  developed a very novel 1080 gene panel that works on the liquid biopsy.
• 19:03
  And it produces the results.
• 19:05
  It tells, okay, what are the, you know, what are the genes?
• 19:08
  What are the mutations?
• 19:11
  You know, indels, fusions, very comprehensive stuff that we provide.
• 19:15
  It also provides the meaningful interpretation, not just the analysis of what genes and mutations are
• 19:23
  present.
• 19:24
  But we have created a back -end informatics comprehensive knowledge base.
• 19:30
  Okay.
• 19:31
  Which allows us to provide actionable reporting.
• 19:37
  Okay.
• 19:37
  And so we characterize based on NCCN guidelines.
• 19:42
  Okay.
• 19:43
  That what is the level 1 evidence, level 2 evidence, level 3 evidence.
• 19:47
  Right.
• 19:47
  So very comprehensive stuff that has actionability.
• 19:51
  Because of the presence of these genes or these mutations or these translocations or these things,
• 19:58
  we are able to say that most likely this is the recommendation of these drugs will work, these drugs won't
• 20:06
  work, etc.
• 20:07
  It's a very actionable report that oncologists can use it as a reference.
• 20:13
  Ultimate call is made by the oncologist.
• 20:17
  We only analyze and interpret and present the data.
• 20:23
  Based on that data, oncologists will make a call and do the stuff.
• 20:27
  Right.
• 20:27
  So the onus, that responsibility is with the oncologist.
• 20:31
  We are merely the tools to provide the information to the oncologist.
• 20:36
  The second tube.
• 20:38
  The second tube.
• 20:38
  Sorry.
• 20:38
  Please go ahead.
• 20:40
  Okay.
• 20:40
  So this is the DNA side of the story.
• 20:44
  The second tube then we extract, we capture the circulating tumor cells.
• 20:50
  Okay.
• 20:52
  Okay.
• 20:52
  And then we extract the DNA and the RNA.
• 20:56
  Okay.
• 20:56
  And the proteins.
• 20:58
  And then we run our single cell assay.
• 21:01
  And get the full analysis of what are the DNAs on each single cell.
• 21:07
  And usually a 10 ml blood will give us somewhere between 5 to 15 circulating tumor cells.
• 21:15
  That's 1 in a billion.
• 21:17
  Right.
• 21:17
  And so we are now honing around somewhere 10 cells.
• 21:21
  Right.
• 21:21
  So if we get 10 cells.
• 21:23
  Each single cell.
• 21:25
  Okay.
• 21:26
  We do the whole genome.
• 21:28
  The whole transcriptome.
• 21:30
  And a targeted proteome.
• 21:32
  So about 20 proteins.
• 21:34
  Right.
• 21:35
  And we provide a very comprehensive report of what genes, what mutations,
• 21:43
  what translocation, what fusions, what proteins.
• 21:48
  Right.
• 21:48
  You know, all of that stuff.
• 21:50
  Now, this in conjunction with the ctDNA.
• 21:54
  Right.
• 21:55
  So our Onco insights, we are not saying ctc is better than ctDNA.
• 22:00
  Or ctDNA is better than ctCs.
• 22:02
  What we are saying is that when you combine the two and give an integrated
• 22:09
  report, that is much better than any individual, you know, technology.
• 22:16
  Historically, ctDNA or selfie DNA based tests have shown that 18 to
• 22:24
  19 % of the time you have NMDs, no mutations detected.
• 22:29
  Okay.
• 22:31
  Okay.
• 22:31
  So even if the best companies, the garden, the Natera, the foundation, the tempest of the world, this
• 22:39
  is an industry average.
• 22:41
  Okay.
• 22:41
  So when you combine that with the ctC together, that thing, you know, the sensitivity goes from,
• 22:49
  you know, 70, 80 % to high 90s.
• 22:55
  Oh, wow.
• 22:55
  Okay.
• 22:56
  Because what is found in ctDNA or what is missed in ctDNA, ctCs capture it.
• 23:02
  Yes.
• 23:03
  And this is like more like usually we try and do like, you know, this multi -omics, right?
• 23:08
  Because you have different parts of the overall cell you're capturing.
• 23:14
  But now you're adding another very new angle on top of the multi -omics.
• 23:19
  You're adding the ctc angle as well.
• 23:22
  So, okay.
• 23:24
  Yeah, that's great.
• 23:25
  Right.
• 23:25
  And so this rich information that you're providing.
• 23:29
  So let's talk about like both the settings, right?
• 23:32
  Like the one is a more like the commercial setting.
• 23:35
  And then the other one is in the clinical trial setting.
• 23:37
  I said clinical trial setting, there's a lot going on right now.
• 23:40
  So how like, so based on this, what kind of decisions can be made about the clinical trial, right?
• 23:46
  Because this patient is there to get a specific therapy.
• 23:50
  So now you come up with this.
• 23:51
  So now what changes?
• 23:53
  Because either the patient is taking that therapy or they're not, right?
• 23:55
  So how does this change?
• 23:57
  Very good point.
• 23:58
  There are two separate tracks here, right?
• 24:01
  One is the clinical diagnostics, which we are doing it in India.
• 24:04
  Okay.
• 24:05
  Where the doctor wants to know what is the right therapy for a right patient at the right time point and the right dose.
• 24:13
  Right?
• 24:14
  So we provide them a full comprehensive report to help her make those decisions.
• 24:20
  Okay?
• 24:21
  That's one.
• 24:22
  The second part of that workflow, the clinical workflow is, okay, you have given the drug.
• 24:30
  Okay.
• 24:31
  Now, is the drug working?
• 24:34
  So therapy response monitoring is another aspect of it, right?
• 24:39
  And there are two pieces to it.
• 24:41
  One is the MRD, the minimum residual disease, right?
• 24:45
  95 % of the time you have taken care of.
• 24:48
  But that little 5%, right?
• 24:51
  That becomes a problem down the road, right?
• 24:54
  And so how to handle that, right?
• 24:57
  So MRD and the therapy response monitoring.
• 25:00
  That's the clinical side.
• 25:02
  Now, you ask me the second question.
• 25:05
  How is this OncoInsight assay helpful in the clinical trial setting for the pharma?
• 25:11
  Okay?
• 25:11
  So I would say three or four key points that we address.
• 25:16
  First and foremost, when you are doing the screening for patient recruitment, is this the
• 25:24
  right patient?
• 25:26
  Usually, you have inclusion criteria.
• 25:29
  But most of the inclusion criteria are very high level.
• 25:33
  Yeah.
• 25:33
  Breast cancer patient, women between the age of this and this.
• 25:38
  You know.
• 25:38
  Yeah.
• 25:39
  It's not going at the molecular level at all, right?
• 25:41
  Or the cellular level.
• 25:42
  Yeah.
• 25:43
  Right?
• 25:43
  Yeah.
• 25:43
  So that's why even when you recruit patients, you know, they don't perform very well.
• 25:50
  Right?
• 25:51
  So by leveraging assays, tests like OncoInsights, a molecular level,
• 25:59
  you have a much better fit.
• 26:01
  So patient selection is a very interesting area where we can help the clinical trials be successful by
• 26:08
  recruiting the right patients.
• 26:14
  Okay.
• 26:15
  So, they find the fertile ground.
• 26:17
  Maybe it's lung, maybe it's brain.
• 26:20
  Somewhere they go in settled tune and the secondary tumor grows.
• 26:25
  Okay.
• 26:26
  Okay.
• 26:27
  So, these are like the fugitives on the run.
• 26:30
  But
• 26:33
  Mohan Utawar: they are the microcosms of creating secondary tumor cells.
• 26:38
  Okay.
• 26:39
  And that's the root cause.
• 26:42
  Okay.
• 26:43
  So, historically, you know, it was very difficult to find those CTCs, those one in a billion
• 26:50
  circulating tumor cells.
• 26:52
  Okay.
• 26:53
  You know, and so, you know, but now the new science, the groundbreaking science to
• 27:01
  capture that one in a billion circulating tumor cells.
• 27:05
  Okay.
• 27:06
  At a very high degree of sensitivity, high degree of precision, the purity of
• 27:14
  capture.
• 27:17
  speakers: Okay.
• 27:18
  And the high degree of simplicity and lower cost.
• 27:22
  Okay.
• 27:23
  So, why you are excited about this whole field is because one cell has this groundbreaking
• 27:30
  science.
• 27:31
  And if I take this three axis, order of magnitude, higher sensitivity, order of magnitude, higher purity of
• 27:39
  capture, order of magnitude, lower cost, it's a thousand times better price performance than anything
• 27:46
  that existed before.
• 27:49
  Okay.
• 27:49
  And it seems like, so then that's going to be both from like the technical aspect as well as like the algorithmic or AIS, right?
• 27:57
  Like, are these like the both aspects going here?
• 27:59
  Sure, sure.
• 28:01
  OneCell. ai aims to democratize precision oncology and make it accessible. we discuss the integration of multi-omics data in cancer diagnostics, highlighting the need for a comprehensive approach.
• 28:17
  Mohan Utawar: data, multiomics data.
• 28:19
  Cancer is like an elephant and a six blind man touching it and describing what it is.
• 28:27
  A genomics is a science of genetics, right?
• 28:31
  It tells you something very important, but it doesn't give you a full picture.
• 28:37
  Just like a blind person touching, you know, the leg of an elephant.
• 28:41
  If you ask him to describe, you'll say, oh, elephant is like a leg.
• 28:55
  speakers: We discuss the importance of early detection in cancer treatment, emphasizing the need for timely interventions.
• 29:04
  Mohan Utawar: cancer, everything, this has to be all integrated.
• 29:08
  So the first generation of precision oncology was driven by tissue biopsy and genomics
• 29:16
  data.
• 29:17
  Companies like Foundation, they did very, very well.
• 29:20
  They established as the leader.
• 29:23
  But tissue is an issue.
• 29:24
  So second generation, the liquid biopsy, right, which is the blood -based stuff, right?
• 29:30
  You know, that came about.
• 29:32
  But what has been not fully equipped is, you know, again, that
• 29:39
  was really focusing again, you know, on the genomics and, you know, the
• 29:46
  transcriptomics.
• 29:48
  But DNAs, they don't, you know, cannot get you the transcriptomics or a little bit, but
• 29:55
  definitely not proteomics.
• 30:01
  speakers: AI is essential for analyzing multi-omics data in cancer diagnostics. Circulating tumor cells are critical for understanding cancer metastasis. we discuss the groundbreaking science behind capturing circulating tumor cells and their significance in cancer treatment.
• 30:21
  Mohan Utawar: like a needle in the haystack, okay?
• 30:25
  All right.
• 30:25
  Now, however, these are the real culprits.
• 30:30
  So, what happens is mostly, 95 % of the cancer deaths happens because of the
• 30:37
  reoccurrence, okay?
• 30:40
  The root cause of the reoccurrence, okay, is metastasis, a process where cancer spreads
• 30:48
  from its origin.
• 30:50
  It might be breast cancer, but because of the circulating tumor cells, the cells that are shed, they go
• 30:58
  run around.
• 30:58
  Most of them die.
• 31:01
  Most of them die.
• 31:02
  Most of them are killed by our immune system.
• 31:04
  But some strong ones evade the immune system and they settle somewhere else.
• 31:14
  speakers: Equally important, patient deselection.
• 31:19
  Okay?
• 31:20
  How not to recruit the wrong patients?
• 31:23
  Because one bad apple messes up the whole fruit basket.
• 31:28
  Okay?
• 31:29
  Right?
• 31:30
  And so, again, when you integrate the CT DNA and CTCs, you have a much richer set.
• 31:37
  So we could basically find better patients, you know, recruit the right patient, okay?
• 31:45
  And deselect the wrong patients.
• 31:47
  So that's one, you know, clear value proposition to the pharma companies that are doing biomarker -based chemical
• 31:54
  trials.
• 31:55
  Okay?
• 31:56
  And we know biomarker -based trials have at least two times higher probability of success from the
• 32:04
  FDA clearance.
• 32:05
  Okay?
• 32:06
  But typical biomarkers that were based on tissue was very hard, right?
• 32:11
  Because then you have to have tissue biopsy.
• 32:13
  You know, so cell biopsy or a liquid biopsy is relatively easy because you just take the blood sample.
• 32:20
  So it costs less and it's much easier to monitor and manage, right?
• 32:26
  So that's one.
• 32:27
  Yeah.
• 32:27
  And it's also faster most probably, right?
• 32:29
  It's much faster, right?
• 32:31
  Much, much faster, right?
• 32:32
  That's one.
• 32:33
  The second thing is, okay, you have now recruited the patient.
• 32:37
  You started, you have dosed the patient, right?
• 32:42
  Is it working or not working?
• 32:45
  In the today's point, right?
• 32:47
  Let's say if you're doing immunocancer therapy, whether my therapy is working or not working, I have to wait for three to four
• 32:55
  months and I have to go expose the patient to either a CT scan or an MRI or a PET.
• 33:02
  And then I would know using what they call it as a racist criteria, okay, whether the tumor
• 33:10
  is shrinking, whether it is not shrinking or whether it is growing.
• 33:15
  But look at the time frame now for every patient under study.
• 33:20
  I have to wait for three, four months before I know whether my therapy is working or not working, okay?
• 33:27
  Let's say you're targeting the drug with Herceptin, HER2, okay?
• 33:33
  And, you know, you've given the drug and now you have to wait.
• 33:36
  Now, instead of that, you know, three to four months wait, within a matter of couple of weeks,
• 33:43
  okay, if I could tell you whether the drug is working or not working, wouldn't that be a great value?
• 33:51
  Yeah.
• 33:52
  Right?
• 33:53
  How do we do that today?
• 33:55
  Okay.
• 33:56
  By capturing circulating tumor cells, right?
• 34:00
  First of all, if the drug is really working, the tumor cells should really go down.
• 34:04
  Yes.
• 34:05
  The CT DNA should really go down.
• 34:07
  Yes.
• 34:07
  Right?
• 34:08
  Okay.
• 34:09
  But you have an 18 -19 % of NMDs in CT DNA, right?
• 34:14
  So you miss out those 20%, right?
• 34:18
  But if I can capture CTCs, right?
• 34:21
  And if I say, oh, I have CTCs where HER2 is still expressed on the surface of the CTCs,
• 34:29
  that means my therapy is not quite working.
• 34:32
  Okay.
• 34:35
  Okay.
• 34:35
  And then, by doing this multi -omics analytics, if I could say, what is the reason why it is not working?
• 34:43
  Okay.
• 34:44
  And again, I am making it up, but there was a marker, let's say EGFR.
• 34:49
  So if there is an EGFR positive marker, then we know historically that HER2 will not work.
• 34:58
  Okay.
• 34:59
  You know, so you don't have to wait for three, four months.
• 35:03
  You can make those critical decisions now.
• 35:07
  And you say, based on multi -omics profile, okay, what ails combination therapy?
• 35:13
  Or can we switch over?
• 35:15
  Can we change something?
• 35:16
  Right?
• 35:16
  Or can we add something to it?
• 35:19
  Right?
• 35:20
  You know, a combination therapy to have a resistance prediction.
• 35:24
  Right?
• 35:25
  If you can predict the resistance, then actionable information will help you add some combo therapy,
• 35:32
  okay, anticipating the resistance, so that the outcomes would be better.
• 35:38
  So, you see, the combination of information, multi -omics data from a CTCs or a single cell,
• 35:46
  coupled with AI to do the prediction, is going to transform this field and make the
• 35:54
  clinical trials much more successful.
• 35:58
  Okay?
• 35:59
  Because you are taking actions proactively.
• 36:01
  You are not waiting for three months, then realizing that, oh, my drug didn't work.
• 36:06
  Okay.
• 36:07
  That's a loss of time and loss of money.
• 36:11
  Yeah.
• 36:11
  And it's also, the tumor has progressed by that time, right?
• 36:14
  By that time, tumor has progressed.
• 36:16
  Right?
• 36:17
  Yeah.
• 36:17
  And so, you're, like, this early detection of, like, the tumor increasing or tumor changing, right?
• 36:23
  So, this is, yeah, this is, I mean, this will be adding a lot of value.
• 36:27
  So, how are you thinking about, like, how can this be commercialized?
• 36:32
  Like, we talked about the clinical trial study.
• 36:34
  But in terms of, like, the clinicians using that in the U .S.
• 36:38
  And in terms of how you are thinking about commercializing, would there be any, like, pair barriers or so?
• 36:45
  So, what is your plan around that?
• 36:47
  Right.
• 36:47
  Currently, the way we are taking is help the biopharma companies.
• 36:52
  And if they use our assay to accelerate their translation research, then one path would be this assay becomes a companion
• 37:00
  diagnostics.
• 37:02
  So, that, with that, right?
• 37:04
  So, that's one path we are pursuing.
• 37:05
  Okay.
• 37:06
  Parallelly, we have initiated a number of clinical studies, you know, specific utilities.
• 37:13
  Okay?
• 37:13
  Okay.
• 37:14
  And once we get the 510ks, for instance, right, then we'll do the MALDX route.
• 37:21
  Okay.
• 37:22
  To get the reimbursement, right?
• 37:24
  Once we get the reimbursement, then we'll make this commercially available.
• 37:30
  Okay.
• 37:31
  And we will, we have two choices.
• 37:33
  Either we build our own, you know, Salesforce and the whole operation.
• 37:37
  Or we partner with larger established labs who already have the partnerships in place.
• 37:44
  Okay.
• 37:45
  And currently, what our interest is, our focus is going to be, we are the R &D.
• 37:51
  We can come up with the new assets.
• 37:53
  And we, we, our intent is to leverage strategic partnerships for commercialization of our assets.
• 38:02
  You know, in the commercial setting, we will probably strategically partner with other larger lab setters.
• 38:08
  Yes.
• 38:09
  That, that makes total sense.
• 38:10
  And now, in terms of, like, we, we talked about the, the, the fantastic science and, and technology around it.
• 38:16
  So, how, like, how do you measure the sensitivity?
• 38:20
  And, and is there, like, a lot of variation?
• 38:22
  Like, how have you, can you talk a bit about that?
• 38:25
  We go through what we call the analytical validation study.
• 38:30
  Okay.
• 38:31
  Where the sensitivity, the specificity, the repeatability, all those factors are well covered.
• 38:39
  Okay.
• 38:39
  We have on our team member, our VP of, you know, clinical development,
• 38:46
  comes with years of experience doing it with the drug companies.
• 38:52
  So, very, very regimented, very strict, very stringent, you know, criterias that we have brought into
• 38:59
  our practice.
• 39:00
  So, we do the analytical validation, then we do the clinical validation before the assay is ready for putting it
• 39:08
  into the, or shrink stuff, right, where we do that.
• 39:11
  We also do it independently with third party, independent labs.
• 39:16
  So, we were very fortunate to have a collaborator like the University of Augusta, Dr.
• 39:23
  Kohli, who has been doing this kind of stuff for companies such as Agilent, Illumina,
• 39:31
  and many other well -established companies have, you know, taken their help to do the analytical validation.
• 39:39
  We did one study where we compared our assay with an FDA -cleared assay, you know,
• 39:47
  it's called PGDX, the company which has been acquired by LabCorp.
• 39:53
  Okay.
• 39:53
  It's one of the very popular, successful, well -established assay.
• 39:57
  And I'm very happy to report to you that we showed 99 .8 % equivalency, you
• 40:05
  know, equivalent to an FDA -cleared assay done completely independently, you know, at a lab, which is the
• 40:13
  University of Augusta.
• 40:14
  And that paper will be published very soon.
• 40:18
  Manuscript is ready.
• 40:19
  We're working through the final details.
• 40:21
  But that gives us the confidence that independent lab was able to conduct an independent study with an FDA
• 40:29
  -cleared assay and we show a substantial equivalency.
• 40:34
  You know, so that becomes interesting.
• 40:36
  That's amazing.
• 40:37
  Congratulations on that.
• 40:39
  Thank you.
• 40:40
  Okay.
• 40:41
  And so now, as we talked about, you know, you talked about this third generation, right?
• 40:46
  Where is this field going?
• 40:48
  Like, let's say three, four, five years later, you're, yeah, on Queen's side, that will be on
• 40:56
  the market.
• 40:57
  What will be next?
• 40:58
  Like, where is this field going?
• 40:59
  What do you see as, like, the next, like, the fourth generation in this area?
• 41:03
  Great question.
• 41:04
  And that's the reason why we are excited.
• 41:06
  I have, our mission is to democratize precision oncology, making it actionable, making it accessible,
• 41:14
  so easily accessible across the country and around the world, and making it affordable.
• 41:21
  Okay.
• 41:22
  So, that's our vision.
• 41:24
  That's our mission.
• 41:25
  And we believe that cell biopsy coupled with AI is the transformer.
• 41:35
  Okay.
• 41:36
  Our ultimate goal is to, can this be a simple assay with a
• 41:43
  simple blood draw, be able to detect way early, and I wouldn't say displace
• 41:51
  completely, but significantly reduce the need to wait for three, four months.
• 41:59
  Okay.
• 42:00
  And look at the sensitivity of CT scan.
• 42:04
  Unless the secondary tumor grows to nine to ten millimeter, you can't detect it.
• 42:12
  Okay.
• 42:12
  A nine millimeter tumor means billions of cells.
• 42:17
  And here I am looking at single cell.
• 42:21
  Okay.
• 42:22
  You know, so if we could utilize good AI to analyze the data and show the
• 42:29
  equivalency, substantial equivalency, okay, don't have to wait for three, four months, don't
• 42:37
  have to spend, you know, thousands of dollars.
• 42:40
  It will transform the care.
• 42:43
  And particularly for survivors, cancer survivors, the people in the remission, where they have
• 42:51
  to go through once a year, twice a year, once a quarter, and go through the agonies and the pains, spend a lot of money,
• 42:58
  don't have the sensitivity.
• 43:00
  So, the celebration that my pet is clean is not very long lasting.
• 43:05
  Right?
• 43:06
  Because it didn't get detected because your tumor wasn't quite nine millimeter.
• 43:11
  But within a month after you celebrate your cleanliness, you might come back again.
• 43:17
  Okay.
• 43:18
  So, early prediction of a reoccurrence, if I could do through cell biopsy, okay,
• 43:26
  then I think we have conquered the big thing.
• 43:30
  Ninety percent of the cancer deaths happens because of the reoccurrence.
• 43:34
  Right?
• 43:35
  And, you know, reoccurrence, the root cause of it is the CTC.
• 43:39
  Right?
• 43:40
  Because of metastasis.
• 43:41
  Right?
• 43:42
  And if you can capture early and create a sensitive specific assay, which is not, which is
• 43:49
  minimally invasive, a simple blood draw.
• 43:52
  Yes.
• 43:52
  I think it will change the field.
• 43:54
  And that's our mission.
• 43:55
  And our goal in life is to impact at least a million cancer patients' lives across the country and around the world.
• 44:03
  That is fantastic.
• 44:05
  And all our best wishes to you about those.
• 44:09
  Thanks.
• 44:11
  Mohaner Terwar, co -founder and CEO of OneCell .ai.
• 44:14
  Mohan, thank you very much for your time today.
• 44:17
  Thank you very much, Amr and the team.
• 44:22
  Well, that was a great conversation.
• 44:25
  What did you think?
• 44:26
  Yeah, it was fascinating to learn about the progress being done in this area of diagnostics,
• 44:33
  liquid biopsies.
• 44:35
  Right?
• 44:35
  So we've been living through this when it was tissue biopsies and then liquid biopsies came in, and now it's liquid biopsies but only single
• 44:43
  cell.
• 44:43
  And the precision that we can have now from scientific and technology point of view, really focusing on one cell
• 44:51
  and then really figuring out from there what could be the, what's happening with the tumor.
• 44:58
  It's just fascinating how we're rapidly advancing the science and technology and, you know, which is really at the
• 45:06
  forefront of healthcare at this point.
• 45:09
  You know, it must be 20 years now that liquid biopsies kind of first emerged on the scene.
• 45:16
  And there was a lot of excitement early on about their potential, but the impact has been limited so far.
• 45:22
  To what extent do you think what one cell is doing addresses the existing limitations of liquid
• 45:30
  biopsies?
• 45:31
  So they are not only using just the liquid biopsy and then doing some
• 45:37
  regular experiments.
• 45:42
  I mean, they are doing a lot of the multiomics on this.
• 45:45
  So you talked about genomics, transcriptomics, proteomics, as well as a lot of stuff about understanding the morphology of the cell.
• 45:53
  So they're bringing in a lot of different types of analytics together using AI, using analytics.
• 46:01
  So, and there also, he talked about gene assay, right?
• 46:04
  About over a thousand genes in that.
• 46:06
  So it's really going deep into quite a bit of understanding much more about the cell, much more about the
• 46:14
  DNA, but also about the protein.
• 46:17
  So it provides a lot more information.
• 46:19
  It's also, we have to think about, I mean, what he talked about was the report that they generate, which is not just about, okay, well, this gene here and then,
• 46:27
  but then they are using NCCN guidelines to then connect this to the drug therapies.
• 46:33
  And then really talking about, okay, well, what are really the actionable mutations that we can see?
• 46:39
  Because that's the biggest problem, right?
• 46:40
  Okay, there's some mutations that you can detect in any DNA, but so what?
• 46:46
  What do you do with that?
• 46:47
  There is no drug for that, right?
• 46:48
  So now detecting these actionable mutations, but also tying them to the therapeutic options, I think that's going to be great.
• 46:54
  And then it's also, as we get more and more therapies, this is going to even, going to be more, right?
• 47:00
  Because they're going to be able to say, okay, well, these were mutations that were not necessarily actionable before, but they will be now, right?
• 47:05
  So I think that's definitely very good.
• 47:09
  And also, I mean, to me, the big thing is to be able to detect early, right?
• 47:13
  I mean, like three, he talked about three to four months.
• 47:16
  That's a huge amount of time in a cancer patient's life after diagnosis.
• 47:21
  So getting any of this information before you see that, like the tissue growing or so, I think that will be
• 47:28
  definitely a big step forward in terms of changing the therapies for the patients.
• 47:34
  And it should result in a better lifespan, higher quality of life in the long run for the patients.
• 47:40
  Mohan said this combination of applying a multi -omics approach and cell analysis to circulating
• 47:47
  tumor cells wouldn't be possible without AI.
• 47:51
  Is this a good example for the potential of AI to change clinical practice and allow us to finally
• 47:59
  realize a long -held vision of personalized and precision care?
• 48:02
  I mean, anytime you're using omics, you have to use AI because the omics data is huge.
• 48:09
  You talk about a single genome, that's huge.
• 48:11
  You're talking about proteomics.
• 48:13
  You know, you're looking at like all the proteins in the cell, right?
• 48:16
  So you definitely need AI and advanced analytics to process any omics data.
• 48:22
  And then as you're looking at combining those data, combining different types of omics, but also combining a lot of this cell data, you need the
• 48:30
  AI to glue all of that together.
• 48:34
  So, yeah, you have to have it.
• 48:35
  And, of course, in diagnostics, I definitely think that AI has a big usage in diagnostics because you're trying to figure out what's happening
• 48:43
  in this cell, and the cell has way too many components that you do need AI to do the analysis for that.
• 48:51
  The initial focus of one cell is on helping biopharm companies use the technology to improve their drug development
• 48:59
  efforts and help build a case for its test as a companion diagnostic.
• 49:05
  What do you think of that strategy?
• 49:06
  It's a good strategy.
• 49:07
  I think they are following both strategies, right, like going with the 510K route as well as this.
• 49:14
  But here, I would say definitely this gives them a lot of experience about like how it works
• 49:22
  in the clinical trial setting with how the clinicians will use this information.
• 49:27
  But also the big, I would say the big advantage there is to be able to have it as companion diagnostic.
• 49:33
  So, if they can get approved as companion diagnostic for any drug, then the test needs to be used, has to be used
• 49:40
  whenever that drug is to be prescribed, right?
• 49:46
  So, that I think will be fantastic for them commercially.
• 49:49
  And that also is just going to help in getting the general usage application.
• 49:55
  I mean, they will be generating a lot of data that will be helping with those applications as well.
• 50:02
  As you think about the various stakeholders around clinical practice, providers, payers, regulators,
• 50:10
  drug companies, and patients, what do you think about the potential for AI to become integrated into care?
• 50:18
  And what will it take to win acceptance for doing so with these different stakeholders?
• 50:22
  The way I think about it is that the physicians and patients are always very interested in understanding what is happening.
• 50:30
  Regulators, it's going to be more about, well, they're doing the trials.
• 50:34
  It's about the insurers and whether they are going to reimburse this or not.
• 50:38
  I would say one of the things is going to be, yeah, whenever you talk about AI, how expensive is it?
• 50:43
  When you're talking about next -generation sequencing, doing a lot of the proteomics analysis, that is,
• 50:50
  it's going to be, it's going to take some money, right, to do all of that.
• 50:56
  So, but then what they have to think about is that the benefits from here, right?
• 51:00
  So, at the level for the patients, but also at the population level, is this going to really then help prevent the
• 51:08
  patient from recurring?
• 51:10
  Is that going to help?
• 51:11
  Is it going to help in increasing the quantity of life, but also the quality of life?
• 51:18
  Is that going to help in maybe less therapies later on because you're able to, and the wrong therapies later on, right?
• 51:26
  Because you're detecting very early on what should be the right therapies.
• 51:30
  I think those are some of the advantages that the pairs will take into account, and then they will do the analysis.
• 51:36
  But I would say that it's going to, I mean, there's definitely the potential to have this in clinical practice.
• 51:43
  I think at some point we have to get there, right?
• 51:47
  Like we have to have these diagnostics, the early diagnostics, to be able to detect cancer early on or like the changes in cancer.
• 51:55
  Because, I mean, I always think about like diagnosis is much better than cure, right?
• 52:01
  It's, you're always much better off with diagnosis than cure.
• 52:04
  And something like diagnostics like that, it is going to require AI because cancer can, like there are so many genes that can be potentially involved
• 52:12
  in cancer that you need to have analytics.
• 52:15
  It's not going to be just one test that is going to be able to detect what the molecular mutation is.
• 52:20
  You are going to need analytics and AI to be able to detect holistically for the right therapy.
• 52:28
  It was exciting to hear and a great example of how AI can not only change drug development, but clinical practice.
• 52:36
  So, Amir, thanks as always.
• 52:38
  Thank you.
• 52:42
  Thanks again to our sponsor, Agilisium Labs.
• 52:45
  Life Sciences DNA is a bi -monthly podcast produced by the Levine Media Group with production support from FullView Media.
• 52:53
  Be sure to follow us on your preferred podcast platform.
• 52:57
  Music for this podcast is provided courtesy of the Jonah Levine Collective.
• 53:03
  We'd love to hear from you.
• 53:04
  Pop us a note at danny at levinemediagroup .com.
• 53:09
  For Life Sciences DNA, I'm Daniel Levine.
• 53:12
  Thanks for joining us.
• 53:21
  you.
• 53:21
  Thank you.
• 53:24
  Thanks.
• 53:25
  Thank you.
• 53:25
  Thank you.