Insight Molecular Diagnostics Inc. (IMDX) Earnings Call Transcript & Summary

Hello, and welcome to today's webinar. My name is Sara Riordan. I'm a genetic counselor and the Director of Medical Education at Oncocyte. I'm very excited about what we are going to discuss today, which is a novel gene expression test that assesses the tumor microenvironment to predict response to immunotherapies. This webinar is being recorded, and we will make the recording available after the webinar. Now I'd like to introduce our 2 speakers. Dr. Naoto Ueno is a tenured professor in the Breast Medical Oncology Department of MD Anderson Cancer Center, Executive Director of the Morgan Welch Inflammatory Breast Cancer Program and Clinic, and Section Chief of Translational Breast Cancer Research in the Breast Medical Oncology Department. His research focuses on inflammatory breast cancer, triple-negative breast cancer and metastatic breast cancer. Having over 6 ongoing investigator-initiated clinical trials for patients with IBC, TNBC and bone metastasis. He supervises medical and graduate students, postdoctoral fellows, junior faculty and clinical residents and fellows, resulting in their successes in both academia and industry. Dr. Ueno received the University of Texas System Region's outstanding teaching award for his contributions in 2014, and an Nylene Eckles Distinguished Professorship of Breast Cancer Research in 2012. We also have Doug Ross here with us today, who's an experienced medical diagnostic industry executive leading R&D efforts at the intersection of genomics, cardionics and diagnostics. Dr. Ross received his MD and PhD from the University of Washington and Fred Hutchinson Cancer Research Center, and trained in clinical pathology at the University of California, San Francisco. Dr. Ross' private sector career began in 2000 as Chief Scientific Officer of Applied Genomics, a company he co-founded coming out of postdoctoral training at Stanford University. Applied Genomics was acquired by Clarient, a leading national pathology reference laboratory in 2009. Dr. Ross' esteemed career in the molecular diagnostics industry also includes Chief Scientific Officer roles at GE Healthcare and Life Technologies where he launched 2 proprietary products. In 2014, Dr. Ross became a founding partner at the Bethesda Group. Dr. Ross joined Oncocyte as Chief Medical Officer in early 2020 and after a yearlong consulting position. Dr. Ross will be speaking first today about the critical unmet need to better identify responders to immunotherapies and discuss data showing the association of the DetermaIO 27-gene signature with response to immune checkpoint inhibitor therapy in non-small cell lung cancer. Dr. Ueno will then present data, first presented at ASCO 2020, indicating that DetermaIO outperforms PL-1 in predicting immunotherapy response in triple-negative breast cancer. Welcome to those of you just joining us. [Operator Instructions] Okay. Dr. Ross, please take it away.

Thank you very much, Sara. I appreciated the introduction, the kind introduction, and I'm thrilled to be here with Dr. Ueno to present some data to you and an overview of our approach to classifying response to immuno-oncology drugs. So I thought I'd start by giving you just a brief overview of our approach at Oncocyte. And I think most of you are probably used to thinking about diagnostics in the cancer space, focusing on the question of whether or not I should give targeted therapy to my patients. So much of the industry has focused on diagnostics that identify targets for drugs such as EGF receptor targeted drugs. And there's also a lot of excitement in the industry about diagnostics that look for tumor signal in blood, blood-based diagnostics. We've taken a different approach at Oncocyte. We've focused on areas where we have proprietary content for managing unsolved problems. And so I'll mention just very briefly, at the beginning of our talk, I guess, a product called DetermaRx, which is targeting the unmet need of managing early stage lung cancer, where there's a critical diagnostic decision as to whether or not to use adjuvant chemotherapy. Most of our presentation today is going to focus on what we call the DetermaIO, which is the critical question of whether or not folks should be treated with adjuvant immunotherapy, and checkpoint inhibitors being the most prominent of those. So that's the overall spectrum of where we're going. And on the next slide, this is just a brief overview of DetermaRx. And I wanted to start here because it gives you a sense of where actually the DetermaIO product might be going. So again, this is an -- this product addresses an unmet need in non-small cell lung cancer in early-stage operable lung cancer. So in principle, these patients are -- have resectable tumor. In other words, the premise is that when you take the tumor out, you would be curing the patient of a disease. But in fact, what happens in early stage, at least Stage 1 and IIA non-small cell lung cancer, is 30% to 50% of those patients have a recurrence, a very aggressive recurrence. And ultimately, it -- die of disease. And so the DetermaRx test is a test that stratifies, that identifies those subset of patients that are at high-risk of recurrence, and therefore should consider adjuvant chemotherapy. It also identifies and contrasts a set of patients that are at low-risk of recurrence and therefore, are probably safer to be treated without adjuvant chemotherapy. And the reason I wanted to bring this up is because this stratification allows you to identify, once again, the subset of patients that are most likely to respond, we have data that we're now showing you today that demonstrates that those high-risk patients, when they're treated with adjuvant cytotoxic chemotherapy, they have much, much better outcomes, 10% recurrence rate as opposed to a 40% recurrence rate and disease-free survival rate. And the way this relates to the immuno-oncology program is that if you've identified in these high-risk patients, the opportunity to respond to chemotherapy, these are folks that are candidates for novel therapy. So cytotoxic chemotherapy, cisplatin regimens are very toxic. And just at the past ASCO, a couple of months ago, the Adura trial results were released targeting earlier stage patients with EGF receptor-targeted therapies. It's really showing a very strong response to that. We think these early-stage high-risk patients are candidates for immuno-oncology therapy for checkpoint inhibitors. And if you target those patients, who are at high risk, it's much easier to power studies to demonstrate the impact of the therapy on outcomes. It's much easier to show an impact on a patient who has a 40% chance of recurrence, dropping them down to perhaps a 10% recurrence rate, than it is to start with a set of patients that are already at 15% to 20% recurrence and try and demonstrate the difference with the improvement to a 10% recurrence rate. So that's the way that we expect the DetermaIO's has to progress, through clinical trials that demonstrate its actionability in the different clinical situations where immuno-oncology drugs could be applied. So going back to the next slide. So once again, this is the spectrum of disease. We've talked about -- I talked very briefly about the DetermaRx product, which identifies at-risk, early stage non-small cell lung cancer patients. And the rest of the time for this webinar, we're going to focus on DetermaIO, which focuses on the question of whether or not we should give patients adjuvant immunotherapy checkpoint inhibitors. The next slide show sets this up a bit. Immunotherapy has come on to the scene over the course of the past 5 years or so. PD-1 and PD-L1 checkpoint inhibitors are approved for at least 13 different solid tumor types, covering up to 70 -- 150,000 patients a year on FDA-approved indications. But this one study demonstrated that even under the most favorable assumptions, only about 12% of those patients are likely to respond to the immunotherapy based upon current data. So that's a problem. We can do better. In non-small cell lung cancer, which is the most commonplace of these drugs we use to date, most patients are accessing the drug through PD-L1 staining. And in fact, most patients are getting it regardless of whether or not they have a significant PD-L1 expression. So again, we can do better than that. Immunotherapy is not completely benign, it's associated with dramatic side effects. Dr. Ueno will go into that at some level. Essentially, what it does is it takes the brakes off the immune system, which reveals latent autoimmune disease and all the -- and much of the side effects are due to the appearance of that latent autoimmune disease. And last but not least, these are very expensive drugs. And so if patients are not gaining a benefit from them, then I think it's -- that, of course, it's important to use those agents judiciously. So there's a clear need for better identification of patients that are likely to benefit from checkpoint inhibitors, and allow oncologists to use these things appropriately in patients. Next slide. This is just a little bit more detail for that -- for -- that goes behind that. These are some of the indications for which checkpoint inhibitors are approved, and their associated biomarkers. And you can see there, as an example, in non-small cell lung cancer patients, if they have PD-L1 IHC expression greater than 50%, there's only about 10% of CAMP patients that express it at that level, and they have a 40% to 50% response rate. But in patients that are greater than 1% but less than 50%, it's about 30% of patients that they have actually a very poor response rates, and less than 25% response rate. And in colorectal cancer, where patients qualified for biomarkers based upon a genetic instability, biomarker, MSI, only about 12% to 17% of those qualify and they have a less than a 50% response rate as well. So once again, a continued unmet need for biomarkers that are better at predicting response and frankly, better at predicting nonresponse. So patients can choose more optimal therapy. Next slide. So what we're going to talk about today is a novel 27 gene real-time immuno-oncology gene expression signature. And what's different about this is I've already hinted at the biomarkers that are in place today. One of them, PD-L1 targets. It is the target of these immuno-oncology drugs, one of the targets of the immuno-oncology drugs. And the other is something called tumor mutation burden or genetic instability, that's essentially the appearance of the antigens that the immune system targets. This 27 gene real-time PCR gene expression signature that we've developed is quite different. It looks at the biology of the tumor. It looks at the biology of the immune response and understands the entire tumor microenvironment to identify patients that are trying to respond to checkpoint inhibitor therapy, and potentially the mechanisms of resistance to checkpoint inhibitor therapy. And in contrast to many of the other efforts in this space, there's been a lot of efforts trying to look at the physiology of tumors, the physiology of the immune response in order to predict response to these immunotherapies, this test looks at 3 different tissue components and integrate that information to report out on checkpoint inhibitor therapy. It looks at infiltrating leukocytes, immunomodulatory infiltrates. Those -- it identifies a subset of T cells and other inflammatory cells that indicate that you're going to respond to the drug, to the checkpoint inhibitors. But it also looks at the extracellular matrix. It looks at the stroma. It looks like the wound response to the tumor, which consists of cancer-associated fibroblasts, and that -- the stroma, the wound response is responsible for repressing the response of the immune response to tumors. And in some cases, it actually looks at the tumor cells. It looks at something called Epithelial-mesenchymal transition, that's tumor cells that have become dedifferentiated, and tend to be also resistant to these checkpoint inhibitors. On the next slide, I just wanted to go through a bit of the history of this test. I'm trained as a pathologist, and I've spent most of my career developing multivariate diagnostic tests. And as a pathologist, I really look for 2 different components of what creates a good diagnostic or a good theranostic. There's really 2 things. One is, does the test classify tumors well? And the other is, does it identify a class that is clinically actionable? And at some level, those 2 things are differentiable. So in other words, you can have a very good classifier for which you've not identified a clinical action. And -- but something that's clinically actionable that doesn't perform well as a classifier doesn't perform well in the clinic. And so one of the reasons we've been enthusiastic about the test, DetermaIO, is because it has a deep history in the academic literature as a classifier. So back in 2011 at Dr. [ Stephen Ball's ] Lab at the Vanderbilt University, they originally published a 2,100 gene classifier that classified triple-negative breast cancer into 5 different tumors -- tumor types. And then, over time, it became clear that in addition to those tumor types, there were some classifiers that could modify each of those tumor types. They call them immunomodulatory, which recognized tumor-infiltrating leukocytes in the tumor. And really a combination of the mesenchymal stem-like and the mesenchymal signature, which is the part of it that recognizes the tumor stroma or the wound response to the tumor. And over time, this 2,100 gene subclassification system, by the company called Insight Genetics in Nashville, has distilled this down to a originally 101 gene classifier that recapitulated the entire classification of triple-negative breast cancer, and then recognizing the value of the immuno-oncology portion of the assay, pulled out a 27 gene component that queries the tumor-infiltrating leucocyte component and just the stromal component, and that's become DetermaIO. And in 2019, we generated the data that supports the notion that this immuno-oncology portion of the signature can identify responders to checkpoint inhibitors, and I'm now advancing that into the clinic. Just one last note is that this has been around long enough now that there's over 1,800 references in the literature to this as a classifier. So it's been in the literature as a classifier in search of a clinical application, and we now have the data in support of that clinical application. Next slide. So this is a cartoon version of what this test is doing. And I like this because it helps relate the biology, the physiology of a tumor in an immune microenvironment that this test is picking up. So the notion here is on the left side of the slide is that fortunately or unfortunately, depending on how you look at it, we all have small tumors that emerge in our bodies over the course of our lifetime, but all those small tumors are kept in check by our immune response. So it's essentially for most of our lives, we exist in a state of tumor immune homeostasis. Something happens, mostly as we get older, but sometimes earlier in life, where cancer essentially breaks through that immuno-surveillance, and now you have a clinical cancer. That for some reason, and in this case, is exhibiting the expression of PD-L1, is overcoming this checkpoint that would, otherwise, keep the cancer in check. And there's really 3 different states that have been observed around that breaking through of the immuno-surveillance. So what's portrayed here was the word HOT. Is that -- is the notion that there are leukocytes in the tumor that are inhibited by something coming from the tumor or the microenvironment. In the middle, with CAF, which stands for cancer associated fibroblasts, we're depicting that there is an immune fortification built by fibroblasts that inhibits the tumor-infiltrating leukocytes and inhibits the immune response from effectively targeting the tumor. And in the bottom, some tumors undergo a tumor transformation, and that epithelial to mesenchymal transition, where there's essentially an immune desert. There aren't as many immune cells around the tumor and since -- so we don't find a immune cells there. The DetermaIO product, we believe, is quite differentiated, and that it detects the all 3 states. It detects tumor-infiltrating leukocytes that are there that are poised to respond to the tumor, but are not active because of the checkpoint inhibition. It identifies the immune fortification, which is giving the tumor negative signals, inhibiting the immune response and it also, in some tumors, detects this immune desert, detects those with an epithelial-mesenchymal transition that simply we don't find the immune response effective there. And so in the middle, we're depicting DetermaIO, where it is shown -- it is identifying the 15% to 40% of tumors that are poised that when you add the checkpoint inhibitor, then you now get a response to the tumor. You got an immune attack. And it's also detecting the 60% to 85% of those that are resistant, that even in the presence of a checkpoint inhibitor, they continue to grow. And so we believe that those tumors that are resistant are candidates for second-generation agents that could either act independently or in conjunction with checkpoint inhibitors to overcome that resistance state, perhaps targeting those fibroblasts or that immune microenvironment, and/or tumors that develop resistance during therapy with checkpoint inhibitors are now candidates for those second-generation agents. So on the next slide, the test, once again, is a real-time PCR assay measuring the expression of 27 genes. So once again, they started as a 2,100 gene assay, which is still down to 101 gene assay, but these 27 genes efficiently measure these 3 different tumor components and the full immune response environment. And can be run in a very simple 27 gene assay that's currently run out of our CLIA LDT certified laboratory in Nashville, Tennessee. The expression of those 27 genes is interpreted with a proprietary algorithm, and the positive status is associated with the presence of an activated subset of tumor-infiltrating leukocytes, whereas the native status is associated with the presence of activated mesenchymal stroma-like gene expression pattern, that wound response gene expression pattern. Since it's recognizing the leukocytes in the tumor and the wound response, we think it's tumor type diagnostic. The leukocytes and the wound response are necessarily -- are not necessarily breast cancer versus lung cancer versus colon cancer specific. So this is translatable from tissue of origin to tissue of origin. And it's a very convenient assay, and that it runs on the standard pathology specimen, an FFP paraffin-based specimen. And it only uses 50 nanograms of RNA, plus we ask for an H&E slide to make sure the tumor qualifies. And it can be delivered with a 5-day turnaround time. This is a CLIA LDT approved test. Currently available for research use only, and so we're doing clinical development to layer in information that supports the use of this test, clinically, which -- some which we're going to show you today. Next slide. So Dr. Ueno is going to talk to you about our data in triple-negative breast cancer. I have the pleasure of presenting data that was generated in Dr. Badal's lab at the West Clinic in Tennessee, in Memphis, Tennessee. This was a study that we showed at the SITC conference back in November, an international immunotherapy conference, that -- in which we studied 71 advanced lung cancer patients that have either been treated with Keytruda or a Opdivo. About 58 of those received monotherapy, so checkpoint inhibitor therapy alone, 13 of those patients were on combination therapy, so checkpoint inhibitor combined with a cytotoxic agent, and they have a highly diverse histopathology. Most of them were adenocarcinomas as is typical for non-small cell lung cancer, but many of them were squaring with cell carcinoma hand or not otherwise specified. We had 2 different outcomes to look at on this trial, one was progression-free survival or the time from disease progression to death. But in addition, there was information about objective response, which essentially, is after 8 weeks. CT images were compared to see if the tumor lesions had either progressed or decreased in size by radiologic measurements. The next slide. This is some of the data from that trial. And this is looking at progression-free survival comparing DetermaIO, this 27 gene expression signature, to either PD-L1 IHC or tumor mutation burden, another test that's in the literature that's designed to predict outcomes to checkpoint inhibitors. And what you can see is the Kaplan-Meier plots on the right depict the outcomes either in Determa -- at the top now, DetermaIO positive patients compared to DetermaIO negative patients and it's showing you that the 1-year progression-free survival of patients, that the DetermaIO called likely responders was 78% compared to 42% for likely nonresponders. And the hazard ratio there, sort of the inverse of the hazard ratio, shows that the hazard ratio was above 3 and statistically significant. Now comparing to the biomarker which non-small cell lung cancer qualifies you for Keytruda, PD-L1 IHC, you can see in the middle plot that, in fact, the hazard ratio was much weaker. And in fact, it was statistically insignificant. PD-L1 did not perform very well in this study. And similarly, tumor mutation burden, once again, a measure of the antigenic load or the antigenic targets of the immune system, it frankly performed better than PD-L1 in this study, then again, was not statistically significant and a weaker hazard ratio of around 2 compared to around 3.5 for DetermaIO. Next slide. This is another way of looking at response in this cohort. This is objective response. Once again, looking at CT scans and measuring the tumor response over time over the course of 8 weeks to the presence of a checkpoint inhibitor or in a subset of patients' checkpoint inhibitor, which has different chemotherapy. And one of the interesting things here, and very promising things about DetermaIO, is it's showing a quantitative relationship, showing that as you go from progressive disease to a complete response, that there appears to be a quantitative relationship between the DetermaIO score and the degree of response. In contrast, in the middle with PD-L1 IHC and/or tumor mutation burden on the right, those PD-L1 performed somewhat quantitatively, but you can see that the cutoff for PD-L1, 1% IHC, is way down there at the bottom. And so although there was somewhat of a quantitative relationship, the cutoff was not performing. And really, in this case, although I showed you earlier, that TMB did have a weak but statistically insignificant relationship with objective response, it did not appear to have a quantitative relationship. And so that's -- the quantitative response, the quantitative relationship between DetermaIO and response to the drug suggests that during the course of therapy, we might be able to identify folks that are having only a partial response, and those folks may want to consider adding some of the newer agents that are coming out that are designed to boost the response. On the next slide, this is just showing you some data for those -- for the efficient autos in the audience that understand this space better. There's been a lot of controversy around tumor mutation burden in that tumor mutation burden, it predicts outcome in the patients who were treated with cytotoxic chemotherapy alone. And so it's very complicated to interpret the data that comes from combination trials. When you treat patients with both a checkpoint inhibitor and cytotoxic immunotherapy, the tumor -- the measuring tumor mutation burden, the relationship between tumor mutation burden and the outcome, can either be due to the response to cytotoxic chemotherapy or the response to the checkpoint inhibitor. In the case of DetermaIO, this is showing you some analysis of, in silico, of patients that were treated -- or sorry, from the TCGA, the public genomic data sets, showing you that in the absence of a checkpoint inhibitor, we've never seen any prognostic association. So does not adhere to predict response in the absence of any sort of therapy, and also does not predict response in the presence of cytotoxic chemotherapy. And so we believe it should be easier to interpret and demonstrate the activity of DetermaIO in mixed trials with both checkpoint inhibitors and cytotoxic chemotherapy. Next slide. This is just to give some acknowledgments to Dr. Badal's lab and the folks who did the work as well as my colleagues at Insight Genetics in Nashville, Tennessee. And last but not least, so to summarize, DetermaIO is a novel gene expression test, although it's building on literature that's over 1,800 references in the literature over the past 10 years or so. It's unique in that it accesses the biology of the tumor microenvironment to predict response to immunotherapy. And I've shown you data in non-small cell lung cancer that had outperformed PD-L1 and TMB, the currently accepted biomarkers. And that interestingly, appears to have a quantitative relationship to objective response. It's a 27 gene QPCR assay, very convenient. There's QPCR machines all around the world, although it's currently being run out of our CLIA LDT lab in Nashville. And for those of you who are from pharma companies in the audience, not only is it a QPCR assay, but it can be run on whole-transcriptome data, our RNAC full transcriptome data. And so that's just in silico, a computer -- an experiment done on the computer on existing databases is the other way that we've generated data for this test. Thanks very much.

First of all, thank you, Dr. Doug Ross. I'll now turn it over to Dr. Ueno. We're going to hold questions for the end.

Okay. So thank you very much, Sara, for the nice introduction. So my talk is going to be half updating about the landscape of triple-negative breast cancer immunotherapy. And then I'll talk about some of the IM signature and the biomarker issue that's related to this topic. So these is the 2 topics I will be covering. So here is my topics of interest. Okay. So starting from some of the review related to the immune checkpoint blockade in breast cancer and I'll start from a single-agent activity. So here are some of the examples of the clinical trial that's ongoing. But as you can see, as a single agent response rate, particular complete response is actually in the single-digit, low single digit. And even if you combine with a partial response, at most, it's about 10% plus/minus. So there are some hints of activity, and this is where people started to develop -- looking into this agent. One of the recent study is a keynote 119 study, and this is a recurring triple negative breast cancer with 1 in 2 parts systemic treatments. And this was the comparison between the pembrolizumab comparing to our investigator choice of chemotherapy that's listed here. So this study was considered as a negative study, and either this talks about CPS, which is a reflection of the PD-L1 activity, but you could see that with CPS1, you could see that there's not a clear separation of the Kaplan-Meier curve and maybe this tends slightly improvement, right? The p-value did not go below 0.05. So the important thing is that whether you look at this as a completely negative or negative study. And the refi, I wanted to mention about this study is that the treatment-related adverse event is quite different between the 2 arms. As Doug has mentioned that the side effects related to the immune checkpoint inhibitor in chemotherapy is very different. You could see clearly that the dark red area, like fatigue, nausea and diarrhea, loss of appetite, anemia, alopecia, there is a significant toxicity in the chemotherapy arm comparing December to now. And then, of course, if you look into the immunity related adverse events like hypothyroidism, hyperthyroidism, you could see a very different pattern. And clearly, you can see there's more in the pembrolizumab. And that shall give a question that they have to ask, in immunotherapy and the chemotherapy, the outcome is different. But you have to understand that immunotherapy, consistency does exist. But if you don't get this severe toxicity. The problem is that we don't know who is going to get a severe toxicity. The therapy is quite mild comparing chemotherapy. So we have to now look for the immunotherapy from the quality of life perspective. So now moving to the combination of our immune check blockade in chemotherapy and metastatic breast cancer. So the most well-known study is in IMpassion 130, and this is FDA-approved now. By the comparison between first-line metastatic triple negative breast cancer comparing to atezolizumab, plus net paclitaxel versus placebo and net paclitaxel. They have 2 primary endpoints, and it's a progression-free survival per investigator assessment and OS and also PD-L1 cell population. So this is a study. You could see that if you simply look at 1-year rate of progression-free survival, this is out with all comer, it is 23%. The other one was 17%. Now if you look at overall survival, it was 42% and 39%. And then you can see that a progression-free survival p-value is 0.0025. I'll talk about the set analysis with the PD-L1 later on. So that I just talk about generally about the biomarker. If you look the efficacy outcome, and the biggest difference is actually in the complete response rate. And it's 7% for the one with atezolizumab at 1.6%. And there was partial response, to 48 and 44, not a dramatic business, but there is definitely a lot more patients who are actually benefit to achieve a complete response with this regimen. And if you look at the side effects, you can see that there is more neuropathy in pyrexia and hypothyroidism, which is a more unique immunotherapy. So one thing that we have to remember is that immunotherapy has a lot of different toxicities, but the most focus we have is [ directed ] on the severe ones. And the severe ones is where we really like to avoid, such as pneumonitis, severe peripheral neuropathy, [ vasculitis ]. These could be actually detrimental to the patient outcome, quality of life. So there are currently 5 randomized studies, and you could see that there are 2 studies that was positive, which is the KEYNOTE 355, IMpassion 130, [indiscernible] study 119 and 131. Now you could see that the one that's clearly that was positive was noted in the first-line settings. Now IMpassion 131, just came out at a news release about 1 month ago, the data is now out there, so the detail is missing, but it did not meet the primary endpoint. Excuse me. And then going to the combination immune checkpoint blockade in chemotherapy in early breast cancer. So now moving from metastatic and going into early breast cancer. So there are currently 3 studies that's well known, and there's KEYNOTE-522 and IMpassion031, both are considered positive study. IMpassion031, the news release came in July 2020, which said that they met the primary endpoint. But that profound -- how much profound effect they had is unknown. I would talk about mostly about KEYNOTE-522 here. So before talking about KEYNOTE-522, one thing I'd like you to understand is that the readout that was used in academic setting as well as drug approval is pathological complete response. So you could divide that -- you can give any adjuvant therapy, and you could decide PCR versus non-PCR. But not only that, there is what we call Residual Cancer Burden Index and we can further divide this PCR, RCD1, 2 and 3. And why am I talking about this is because those people who achieve PCR really has the best outcome and RCD1, there's a small amount of disease left. And there are some data to suggest these could separate, but you can see the RCD1 still -- will continue to do reasonably well. But if you have any residual disease, 2 and 3, and 3 have extensive disease, you could see the Kaplan-Meier curve separates significantly. So KEYNOTE-522 study, they actually had a patient with a nearly diagnosed triple-negative breast cancer from stage 1 to 3, where they have assessment of PD-L1 and then they actually had one arm with neoadjuvant therapy with carboplatin paclitaxel followed by doxorubicin or epirubicin combined with cyclophosphamide which is commonly used as a standard of care. And then they have one arm with pembrolizumab with no -- and a placebo, and they underwent surgery, and they actually looked into the pathological complete response. The uniqueness of the study, which we still don't have a very long-term follow-up, is comparing the adjuvant therapy with pembrolizumab and placebo. So here's the outcome. The PCR rate, which is a reflection of -- correlates with event-free survival and overall survival, the one that we see pembrolizumab plus chemo at 64% versus 51%. Event-free survival was 91% versus 35%. The p-value of this PCR, which is the primary endpoint at a p-value of less than 0.001. And if you're actually with a greatly adverse event, there is a little bit more toxicity that in this study, not too much of a business working on this. The characteristic I talked to you about is that there is a different type of size of tumor noted in the environment. It's well matched. But you could see that the high rate of the PD-L1 positivity is 83% versus the 81%. And this is an event-free survival. Okay. So what's really important is how do we select this patient and how do we have an impact and how do they impact the outcome. So there are different types of the markers that we could think about. One is PD-L1 positivity, the other one is our high-end signature that we are -- be talking after this. And purely infiltration, mutation load and some people talked about CLIA [LDA]. So I think I will focus particularly on the PD-L1. So PD-L1 in the breast core, there are different antibody that's used. You can see that depending on the drug, the companion or complementary diagnostic tool is a wide range of choices. Not only that, the positivity that defined by each of the company also is very important. Currently for the assay of SP142 and more than 1%, regardless of the tumor cell membrane or tumor-infiltrating gene cell, rightly considered positive. Pembrolizumab, which I have talked about, the CPS score, it goes by more than 1%. So going back to the IMpassion study, you could see that the 2 arms of atezolizumab placebo combined with paclitaxel, by adding a PD-L1 positivity, the separation is much more clear in both progression-free survival and overall survival based on the Kaplan-Meier curve that you see here. And then taking another study in the metastatic setting, which was shown recently in ASCO that, as you can see the CPS score, basically, you could understand that the PD-L1 proclivity going up. You could see that a curved separation curve is much prominent. So you could see that PD-L1 is very helpful to select locations. Now let's look into the early breast cancer, KEYNOTE-522 study. So the interesting study of this KEYNOTE-522 is that PD-L1 positive population, and when you look at the readout of the PCR, it was 68.9% compared to 52.9%. Now the PD-L1 negative was 45.8% and 30.3%. So interesting, regardless of the PD-L1 positivity, there was a difference between the 2. And of course, the question is why, you don't have an answer. But whether, I'm currently -- because of the KEYNOTE-522, it was published in New England Journal of Medicine, that it is -- the work has recently made a new release that they are filing for their approval. And we will find out by March next year, whether this will be approved. But the question is, will they be approved based on the PD-L1 status? Or is PD-L1 data will not matter based on the data that's been presented off from the KEYNOTE-522? Now this is different from IMpassion study in the metastatic setting because currently, the indication is based on PD-L1 positivity. So PD-L1 immunohistochemical staining and some of the unique issues. So the expression's local and location dependent. Is it cancer, or they bring new cells? Is it the price? There are 4 different antibody and also the threshold, it depends on the study. And we all have to remember that PD-L1 negative like that shown in the KEYNOTE-522, there is some difference when we combine the immune checkpoint blockaded chemo. And similar phenomena, the single-agent Phase I, PD-L1 is predicted in some way to -- for a better outcome, but there were patients that did benefit from the immune checkpoint despite a PD-L1 being negative. So our study is based on the 27 gene algorithm for validation. So our study is validation of immuno-monitored gene signature while this will predict the more adjuvant immunochemotherapy in patients with primary triple-negative breast cancer. We presented this data in this year in ASCO. So as what's already been discussed. But the way original study of the TNBC typing from -- coming from Vanderbilt was based on basal-like 1, basal-like 2 and LAR, but with some of the new gene signature and then making more palatable for diagnostic is currently now having a 4 subtypes. And one of the reasons that this separation is very important, but you could see that where does IM subtype really exists. Well, you can see in the mesenchymal subside, the IM subtype is not noted, and it's probably a bit less in LAR. And so you could -- about 1 thing we have to remember, this IM signature, is something that is probably reflecting the tumor mark and environment exists in different tumor intrinsic molecular subtypes and indeed, in the different distributions, and you could see that this another study -- that mesenchymal subtype does not seem to have a lot of IM positivity and a little bit less that a minimal androgen receptor type. Now why is this important? Because we have published in the past that this may actually impact this molecular subtype, not just the IM subtype, but other BL1 BL2 could reflect the pathological complete response. As I mentioned, if you simply look the PCR, the highest PCR could be observed in -- such as BL1, mesenchymal or MSL type of things, but you can see that BL2, there's less likely respond to chemotherapy. Now these are chemotherapy does not include immunotherapy. Moving to our androgen receptor. The PCR is less that you have a lot more RCBR. So the lung cancer study that's been presented, we know the cutoff already. So the newly found IM cut point in the lung cancer will be tested in the current study. So the primary objective is to determine the predictive value of the PCR-based IO panel, which is IM of the primary TNBC with a neo-adjuvant immunotherapy. And comparing to the standard TNBC subtyping method, and we want to validate new IM cut point in the PCR-based on the TNBC with the hypothesis that TNBC patient with IM subtype has a better treatment response to [ myalgia ], in this case, the pathological complete response. So the study with turn working with Dr. Lajos Pusztai in Yale University is a 55 patient with a combination of chemotherapy with [ ipilimumab ] which is another immune checkpoint inhibitor. And then the readout for the study was pathological complete response. So we've taken the study and apply to look into one, the 101 gene original signature and to simplify the 27 gene signature and the PD-L1 expression by Immunohistochemistry stain. And this was based on RNA-seq data. So I'm going to focus mostly on the 27 gene and the PD-L1. There was not much of a difference between the 2, but clearly, the 27 gene signature performed better than the 101 gene signature. So you could see that here is the outcome, the PPR was noted in 45% of the patient, non-PCR was 55%. So the 27 gene algorithm resulted in odd ratio of 4.125. And then for standard PD-L1 using SP253 the odd ratio of 2.63. So this actually really tells you that -- this helps you to understand the profound possibility of niching the potential people that -- who may be having a PCR-based on the immunochemotherapy So here are that the sensitivity and, specificity a positive ligand ratio. And if you actually look into the PCR and PD-L1, the positive ligand ratio is 1.43 CA negative ligand ratio of 0.54. So that definitely, this gives you a sense of the meaning of the PD-L1 going on. So in conclusion, that -- what I would like to say that immunotherapy are promising, appealing due to their possible durable response and possible limited toxicity if you don't end up getting those severe, great and greater toxicity. PD-L1 is helpful, but there is limited in terms of the performance. So there is for better predictive biomarker. 30 gene PCR model showed higher bifurcated ratio and a lower negative right-through compared to PD-L1 I'd see, which indicated the new model had potential unsecured accuracy to predicting the PCR. Now the issue is that there is a lot of patients that still had -- there is a patient that despite of a negative 27 gene signature that could get a PCR. So this actually means that this could be most likely due to compounding effect of chemotherapy. As the lung cancer study shows, if you study this in a more single agent, I think you have a much more clear-cut outcome on how the ratio was at 0.3. You have to remember that, currently, there is no indication of single-agent in breast cancer but I think identifying and enriching the right patient population for immunotherapy will show some or reducing the toxicities. And of course, we wanted to let the outcome of efficacy, the one is how to deescalate by providing toxic therapy is also an important component. And then at last, I think we also need to consider from the financial toxicity perspective. I'd like to acknowledge my colleagues. And in this study, it was an effort of 4 institution. NBN, Yale University together with College of Medicine and Oncocyte supporting this one. So any questions on -- and here is my contact. So thank you very much for the opportunity to speak.

Wonderful. Thank you for that fantastic presentation, Dr. Ueno. We've seen some questions come in. You can submit yours by using the Q&A function at the bottom of your screen. But before we go to Q&A, Dr. Ross, I think you wanted to make just some quick concluding remarks.

Sure. No, I appreciate that, and I appreciate Dr. Ueno. Great presentation. So what we've shown you here today is a novel gene expression test that measures the comprehensive immune response biology. And we've shown you data in non-small cell lung cancer. That study was performed using the real-time PCR assay. So the assay that's ready for the clinic or ready for performance studies. And then the study that Dr. Ueno showed was performed using whole-transcriptome RNA-Seq data and an algorithmic interpretation of those same 27 genes. And I just want to emphasize that both of these studies were done with completely locked down algorithms, completely locked down thresholds. So these were validation studies. There was no modeling done in neither of these cohorts. And as I talked about earlier, this test essentially measures signal coming from infiltrating leukocytes and signal coming from stroma, which is tumor type agnostic, if you will. So we believe that this test is easily transferable to other solid tissue carcinomas that are listed there. And we can do some of that translation or that transference using data sets that are publicly available. And so we're interested in doing studies in all different tissue types but in also advancing drugs that are advancing in non-small cell lung cancer and triple-negative breast cancer. So appreciate the interest and happy to answer questions.

Great Wonderful. Thank you, Dr. Ross. We now have some time to take questions. [Operator Instructions]. And along with Dr. Ross and Dr. Ueno, we have a couple of other folks on the panel to help address the questions as necessary. Padma Sundar, SVP of Commercial at Oncocyte; and Rob Seitz, one of the developers of and the subject matter expert on the DetermaIO test. So I'll start by asking some questions that have already come in. And Dr. Ueno, I would like to hear your thoughts on this first, as it applies to triple-negative breast cancer, and then we'll go to Dr. Ross for his answer. So what is your view on current biomarkers for immunotherapies like PD-L1 and TMB? And why is it important to measure the tumor as well as it's immune microenvironment in assessing whether or not a checkpoint inhibitor might be effective?

Right. So I think, clearly, in a clinical setting, that measuring the PD-L1 is reasonable, and that's the compound that we use that to select the user atezolizumab. But if you understand the biology of the tumor microenvironment, PD-L1 is only one of the components and there are many very component that contributes. So I think if you look the bird's eye view, PDL is 1 but there is -- probably the tumor mutation burden and also there's other component of immunocell that exist. So of course, we could slide all the tumor and stain it and find out why the immune cells and stroma cells, and as we could look for spatial in between how far the tumor from the new cells exist. But that's really very difficult to make it pragmatic, right? So yes, there's definitely an effort from like my lab and other people are trying to understand the whole scope in a more in a detailed way. But we actually need a testing that actually gives a more holistic view. And this is where this type of genomic testing really helps to understand the entire view, which reflects not only the tumor, but it could reflect the surrounding environment. But this is just 1 way of looking at it. So we have to recognize that there other platform that probably may need to come in, and we'll see. Yes.

Just to comment on that, and I completely agree with Dr. Ueno's comments, and then I'd frame it into 2 categories. There's sort of the biologic argument that Dr. Ueno was just going into. And then there's a more practical technical argument, if you will. And the segue from Dr. Ueno's argument, 1 of the things that looking at both the tumor-infiltrating leukocytes in conjunction with the stroma or the wound response is that allows you to take the positive signal, the responsive signal, the positive signal from the tumor-infiltrating leukocytes and put it in the context of the negative signal, the inhibitory signal that comes from the stroma or the wound response. And at least the data that we've shown so far today shows you that the combination of those is superior to either one alone. We've looked at them, of course, either one alone, and the combination seems to be superior. But another more practical aspect of that is that many of these patients or many of these tumors that are being assessed are nonoperable. And so we're getting very small biopsy specimens. And there's a lot of randomness that stuck out into it to how we sample those tumors. And sometimes, they're very rich in tumor cells. Sometimes they're very rich in stroma. It can be inconsistent how well you sample the tumor and the tumor microenvironment. And to the degree that we're measuring gene expression from the 3 different compartments, I believe anyway, there's a little bit of -- or some redundancy in being able to both measure the positive the response, the negative the response, and perhaps the tumor phenotype that makes the test a bit more reliable. It's not just measuring one thing, it's integrating all 3. And if you look at our threshold, there's a lot of room between the thresholds and the average score. So that allows, I think, this to perform better across the real-world setting, where the sample that we get is not always ideal.

Great. Thank you both for those answers. And speaking of the real-world clinical setting, we have a question of how this test could potentially fit in with other testing methodologies. I'm interested in how DetermaIO fits in with NGS? At what point would a clinician order this test versus NGS?

I can address that briefly and then turn it over to Dr. Ueno. So at least in non-small cell lung cancer, and I'll leave Dr. Ueno to address it in breast cancer. But in non-small cell lung cancer, later stage non-small cell lung cancer, the first question, according to guidelines that you should ask, is if a patient a candidate for targeted therapy. So for instance, Tarceva, EGF receptor targeted therapy. The response rates to those drugs are very good. And so if the patient is a candidate by having a mutation in the EGF receptor, then they're candidate for targeted therapy. And that testing is now more and more commonly done by next-generation sequencing technology. So at the same time, frankly, we would be interested in offering an immune checkpoint assessment because in that context, if the patient is not a candidate for targeted therapy, then they are candidates for immunotherapy, and that's what the DetermaIO test would help determine. In addition, most patients treated with targeted therapy, unfortunately, have a recurrence. Those therapies are not as long-lasting -- but the response to those therapies are not as long-lasting. And in the second line, when those patients recur, than the information around whether or not the candidate -- that the patient is a good candidate for immunotherapy is highly relevant. So it will be performed at the same time as an NGS test for targeted therapy. Dr. Ueno, did you want to comment on breast cancer?

Right. So I think rather than more of a standard regulatory -- in regular clinical settings. I think there's definitely this type of testing that's very important in the clinical trial studies, and simply just measuring PD-L1 probably is not sufficient. And we all like to order this -- are we seeking all the things. But this is one of the few algorithm that's kind of validated. So putting that in the context of whatever clinical trial that you're trying to support, particularly investor vehicle initial study or if it's company study, I think it's an important way of trying to figure out what this means actually. Now I can't honestly say that, "Okay. Well, when you have a patient and we should order this," and this is not established that there is definitely more need for clinical prospective study to say that this is truly the way to determine. This is only one study at this moment. So I think stay tuned is probably the best way to say how to look at this type of approach. But clearly, what I want you to understand from my today's presentation is PD-L1 is reasonable, but I can't say it's the most optimal testing to determine anything. And then I think the complexity of the breast cancer world is if the KEYNOTE-522 does get approved, okay, the whole landscape of the TNBC treatment is going to change because, for example, if the FDA truly approves regardless of PD-L1 status, that means that there's many patients that may get exposed. And whether that's overtreating or not, that's a clinical question and tapping answer. And then I think they still could -- patient could recur because we know they do recur. And when they recur, they already are exposed to immune checkpoint inhibitor. So that means that we have to revisit and the PD-L1 may not be the right testing, right? Because now you're getting into the next phase. And that Doug had showed on the slide that maybe we have to look into have a completely different way of approaching the disease and this type of testing is where should the clinical trial should be testing. And then -- and it may guide us in terms of what needs to be done. So probably to be a little bit kind of foggy, but that's where things stay at.

Thank you. Dr. Ross, I think this is a good question for you. How might gene signature like this be used, in its current state, as a research only test? For example, would this be a useful test for patient selection for clinical trials?

Yes. I had exactly sort of the answer to -- in the question. So we are not -- just to clarify, we are not offering the test currently for clinical use. We're not marketing the test currently for clinical use. We are working very hard to advance the data in support of the test, and we're working with both independent clinical trial groups as well as pharma groups, not only in the checkpoint inhibitor space, but also in the next-generation, if you will, immunomodulator drugs. So there's a lot of drugs in development that are designed to stimulate the immune system in situations where the checkpoint inhibitors don't seem sufficient. The patient's tumor is resistant to that. And so we are very interested in, we are -- we're both doing clinical studies to advance the data that instructs how to use these -- how you use this test clinically as well as identify other agents that might benefit either from the entire signature or, in fact, components of the signature because the immune resistant part of the signature, the stromal part of the signature, we believe, is a very good candidate for these agents that are targeting the stroma, targeting the stimulation of the immune response or the overcoming of those resistant mechanisms. So it's a product that is not for clinical use yet, and is advancing through pharma and clinical studies.

Great. I think we have time for one more question, which is really related to next steps, next steps for the development of the test. Dr. Ueno, perhaps, you can discuss further research initiatives with the 27 gene algorithm that MD Anderson may be undertaking? And Dr. Ross, perhaps you could discuss the clinical development studies that Oncocyte will be pursuing?

So I think like MD Anderson currently in the moon shot study. It is a large cohort of triple-negative breast cancer. So I think having a retrospective prospective study, which has all the gene data and then show that what we have seen. So we like to show this in a larger data of cohorts. And the second phase is to really do a prospective study by selection. And whether that is going to be combined with a chemotherapy? Or is it going to be single agent? I show you the metastatic patient, and that study was negative. But if there is a much more, like a combination of the undertaking itself like PD-L1 with this study and then you niche the patient, will that help to really identify the patient which truly going to have a benefit and whether that would prevent from getting chemotherapy, but rather than giving a monotherapy -- immunotherapy. Those are kind of a potential idea where's it's going to be going.

Yes. And we're not on this call to be candid in discussing studies that are in progress. But so let me just categorize them in the types of studies that we're pursuing. As Dr. Ueno has already mentioned, doing more retrospective studies that inform exactly how to use the test, and we're very interested in non-small cell lung cancer as well as other tissue types where perhaps checkpoint inhibitors haven't been as successful with the standard biomarkers. We're interested in working with clinical trial groups that have ownership of the specimens from randomized clinical trials where we can retrospectively go back and ask the question, if, for instance, we could salvage a drug that just missed its primary endpoint, perhaps through a clinical study with the existing biomarkers PD-L1 that might have enrolled those patients into that clinical trial. And as Dr. Ueno said, we're very interested in the highest level of evidence, which is prospective studies, where you would enroll patients based upon the biomarker and randomized to the drug. And we're, obviously, talking with pharma and also biopharma, the smaller companies that are developing second-generation agents in order to move this forward through that clinical development path.

Great. Well, thank you, Dr. Ueno and Dr. Ross for your presentations, and addressing these interesting questions today. And thank you to all of you for joining us for today's webinar. You'll be receiving an e-mail that's a link to today's recording, and we certainly encourage you to share that with any of your colleagues who might be interested in this information. This concludes today's webinar. Thank you again for joining us.