The below text is a transcript from the webinar. Because it is a transcript, there may be oddities that arise from the process of translating speech into text. We recommend accessing the recording, above, to gain full context.
Hello everyone, and good day. Welcome to today's Webinar entitled Report Optimization: Maximizing the Clinical Utility of Next Generation Sequencing. My name is Jeff Prescott with OncLive, and I'll be moderating today's program. Our speakers for today's program are Dr. Eric Vail, who is director of Molecular Pathology at Cedars-Sinai Medical Center, and Dr, Rakesh Nagarajan, who is the executive chairman and founder of PierianDx. In a moment, I'll hand the presentation over to our speakers, but I do want to share that today's program is interactive. We'll have a Q&A session after our presentations, and during that time I'll share questions that were provided by a pre webinar survey, as well as questions that you can submit during the presentation.
In order to do that, you can use the webinar control panel entitled questions, which is typically located on the right side of your screen. Enter your questions in the space provided, click send and double check to make sure that you're sending them to panelist. I will collect them during the course of the presentation, and share them with our speakers at the end of the program. With that, I will turn the presentation over to Dr. Nagarajan. It is all yours.
Great. Thank you so much, Dr. Prescott for having Dr. Vail and I present this very important and relevant topic in oncology and precision medicine today. I'd like to start by first discussing the agenda for this morning. I'll first start by briefly describing the evolving and changing landscape of precision medicine, and its direct interaction with oncology testing. Followed by Dr. Vail who'll first describe the utility and challenges associated with genomic profiling. He'll then cover the clinical utility and the value of genomic reporting. Then Dr. Prescott will moderate the Q&A that he described earlier. Let's jump right in, into my section. As we've seen over the past couple of decades, there's a very rapid rise in precision medicine, almost 300 FDA approved pharmacogenomic markers, as well as others in oncology.
Many approvals of drugs and medications that are now site-agnostic in immuno-oncology as well as in targeted therapies. Then a landmark parallel path FDA and CMS approval that included a national coverage policy and a national coverage determination for NGS based diagnostics in late stage cancers in a resulting comprehensive genomic profiling approach to that late stage cancer testing in solid tumors. There's also been a very rapid growth of comprehensive genomic profiling testing across a variety of market segments, including NCI designated cancer centers, IDNs that are non NCI based, as well as a community hospitals. Most of the CGP volume however, is going to a few independent reference labs. This growth in precision oncology is moving to not just a single variant type testing, but multiple complex variant types, including gene amplifications fusions, as well as integrated biomarker testing such as that scene for a tumor mutational burden and microsatellite instability.
These combinations of variants and biomarkers have specific clinical inferences that's growing and changing as the knowledge base changes. Indeed, oncology testing is evolving. If we looked a couple of decades ago, we would see single marker and hotspot panel type testing. Some of that obviously still available in the primary encounter. In the near term, we see a rise in small panels, less than 50 gene panels that are typically amplicon based reporting out on SMVs and small indels to much broader testing and a larger panel testing, where gene amplifications fusions as well as a TMB/MSI type reporting is possible. Overtime, we fully expect this type of testing to become more and more comprehensive, perhaps even to whole genome scale.
As I indicated in the earlier slides that most of the testing is done in a few reference laboratories, even though there are hundreds of laboratories today that are doing NGS testing in-house i.e as part of a health system and NCI designated cancer center, large independent hospital, things of that nature. In order to bring testing in-house however, there are a variety of challenges that need to be overcome. First and foremost, the knowledge base is very rapidly changing, so new FDA approved labels are changing practice guidelines, new publications from phase three and phase two studies, or even case reports that may change medical management. That data then needs to be curated, kept up to date. Reimbursement is one of the challenges that through FDA approval and through a national coverage determination, there is a path there. LDTs also have a path through private and the local max through LCD type approval, and a scarcity of what we call genome analysts, who actually review and interpret results from the source knowledge bases that are rapidly being curated.
At PierianDx, we support and overcome many of those challenges through an informatics solution, who we are today is a company that partners with health systems and laboratories in supporting their in-house clinical NGS testing. Where we came from is from Washington University in St. Louis back in 2011, where we developed the software technology. We're one of the first organizations to validate a clinical NGS test in a CAP/CLIA environment and publish on those results. That leadership resulted in us spinning the company out in 2014 and launching our second customer to Washington University, that is Moffitt Cancer Center, which first went live in 2014. We've continued to grow with Moffitt and many other organizations, including Cedars-Sinai, where Dr. Vail is from. Early 2019 Pierian signed a multi-year deal with the Illumina, the largest NGS sequencing vendor in the world, as well as a leader in somatic cancer assays to support cancer research and molecular diagnostics.
Today, Pierian provides a variety of services, including the software that we call the clinical genomics workspace, a complete informatics and reporting software that contains those up to date knowledge bases and alignment to variance or biomarkers that match to appropriate knowledge that genome analysts then can use to review and sign out cases. We have a variety of add on scientific and laboratory services that enable laboratories to go live, including support for validation and support for genome analysts like services, which we call interpretation services. Our medically powered knowledge base includes the largest sharing network that we know of an opt-in model, where all of our customers are sharing on the interpretive content that they're creating as they're signing out cases in their own laboratories. With that brief introduction to precision medicine and PierianDx, I'd like to now hand it off to Dr. Vail, for his part of the presentation.
Sorry, I muted myself. Good morning everybody. Thank you so much Dr. Nagarajan, and for OncLive for allowing us to give this presentation. I'm going to just go into first the utility, and some of the challenges that's associated with tumor genomic profiling, especially in solid tumors in 2019. In the beginning, the way that we detected biomarkers was rather simple, because there were only a few. We could traditionally use very limited assays. So IHC, FISH, Allele specific PCR. These are highly, highly sensitive and specific. They worked great, but they only provided a very limited set of information. Unfortunately as the requirements for more testing increased, multiple tests would be needed to get a full profile on any given tumor type. I think that lung obviously is the best example, because you have the IAS/CLC saying you need to test eight, 10, 14 different biomarkers. It's very difficult to start doing that with the single gene testing.
Not only does this increased the costs, but it increases the tissue requirements as well. In 2019 as you can see, any pathologist knows that tissue is getting smaller and smaller, and requests for testing is getting more and more. So this is really the reason why this has happened. Actually these biomarkers are a little wrong. The earliest ones were in the 1980s, which is actually hormone receptor positivity for selective estrogen receptor modulators. If you're just talking about DNA or any sort of DNA or RNA genomic changes, HER2 over expression was the first biomarker for a Herceptin, and then in 2001, we had BCR/ABL fusion. This was really the moment that I would say we cured cancer. That turned out it's a little more difficult than that.
Starting in the late aughts to the early 2010s, you'd really see an explosion of biomarkers. This is really driven by NGS profiling of multiple, multiple solid tumors starting with the TCGA, and then moving more towards other institutions including Foundation Medicine, Sloan Kettering, which have published enormous amounts of data on multiple, multiple different tumor types. What you start to see is that more and more and more tumors are getting approval, based on genomic findings. In 2017, there was the landmark decision for DMMR/MSI High for immunotherapy, which allowed for pan-tumor. Then in 2018, NTRK fusions are the first small molecule inhibitors that were approved largely for intro fusion cancers across tumor type. Just recently, we've had two approvals. One for PIK3CA mutation in breast, and one for FGFR2/3 fusions and/or activating mutations in bladder.
How can next generation sequencing help us? Well, the whole idea is that the real name of next generation sequencing is massively parallel sequencing. This just allows us to sequence a huge amount of data, and get an enormous amount of information on the tumor for relatively cheaply and quickly. The cost of a clinical, the reagent costs of clinical exomes and genomes are under $1,000 a case now. It's been commercially available for over a decade, and significantly improved in that time. The kits have gotten better, capture kits have gotten, library preparation has gotten better. Every step of the way has gotten better, especially including the informatics. So just reiterating what Dr. Nagarajan said, if you see on the right there, what we see is that we went from these hotspot panels, more disease focused gene panels and we're now in this realm of comprehensive gene panels. Even whole exome sequencing, there are a few academic labs and even some private labs that are doing whole exome and some, they're even doing whole genome sequencing on tumors right now.
Currently, the panel that we use is this 161 gene panel modified to add tumor mutation burden MSI. What's interesting now is that the technical costs is actually less than the analytical costs. When you have tons and tons of variants that are coming through, if you don't know what you're doing with it, you can get a huge amount of information that can really vary with complexity. So having expertise in this, having the ability to process and report this information is super critical. Why is it important? Well, right now there are 15 genes and MSI that are FDA approved predictive biomarkers in solid tumors. So these are the lists of them. MSI and NTRK are pan-tumor, but they're major tissue specific. When I gave a similar presentation about this maybe a year ago, that number was eight. So it's really been impressive to watch how the field has evolved so quickly. Multiple others have NCCN emerging therapy designation, or have really high levels of clinical evidence to support off-label use.
This chart is from a paper back in I think 2013, that just looked at really small gene panels on non-small-cell lung cancer and said, "Is there a biomarker with a targeted therapy? Is there a finding that doesn't have a targeted therapy, or is there no driver gene detected? Then let's look ... If we find that, let's treat these patients and see if there's a survival difference." What they found is that patients that were found to have a biomarker were treated with targeted therapy did better than patients that did not. This is a pretty significant overall survival difference. So what are some of the additional utilities with genomic profiling? Well personally, I believe that most large panels, while they're not going to find a ton of patients that have the FDA approved therapies, what you are going to find is a huge amount of patients that are eligible for genomically matched clinical trials. The NCCN, I always look to read this line, believes that the best management of any patient with cancer is in a clinical trial.
Participation in clinical trial is especially encouraged. No matter how often that guideline is published at the bottom of every single NCCN guideline paper, only 5% of patients are currently added to clinical trials. It's a little more at Sloan Kettering's of the world, but in the community it's almost zero. So there has been a real attempt, especially with the large basket trials like taper and match, to be able to go out to the community and get more patients enrolled on these trials. Germline variant detection, RSA doesn't even differentiate because we don't test tumor normal in pairs. What you can do is you can make assumptions about germline variant detection, and then ask for follow up testing with a validated assay that does that, or you can validate Sanger sequencing in-house to test blood, something like that. It's super important because of the impact for entire families.
We had a patient that came in, got sequencing and actually died before the sequencing came out. What we found was a term line BRCA2 mutation, and the patient had two daughters. The daughters were then tested and one of the daughters was positive for it and she ended up having a prophylactic vasectomy and oophorectomy and all of that stuff, this preventative maintenance that these patients get. This father got no benefit really for himself for this testing, but patients in the family got a huge benefit. Diagnostic precision is another utility of geno profiling. Many tumors are defined by their molecular events. As you could see in hematopoietic neoplasms, there's a huge classification in the WHO based on those genomic events. Right now, solid tumors don't really have that. Maybe some in sarcomas, but I think that as we move forward, we're going to start defining tumors not just by their histopathologic, IHC, all that sort of findings. We're also going to be defining them by their molecular findings as well.
Then finally, tissue of origin. We do actually a ton of paired sequencing where we look at a primary tumor from the past, and we look at a new lesion in the patient. Usually they're pretty, they correlate pretty well. Probably say there's three or four variants, there'll be three or four variants, the same variants almost. We just had a case where patient had a stage one lung cancer and then had with a KRAS mutation, and a three mutation. Then what we found are Keratin and SDK 11 mutation, and then we ended up finding in a new lesion six months later was a completely different set of driver mutations. Most likely this patient has two separate primaries, and because it was on the contralateral alum, this is a completely different stage for this patient that what would have happened if this was diagnosed as a metastatic lesion.
What are some of the challenges with genomic reporting? Well, really the first one is that you're not writing a report for one person. You're actually writing it for four different competing audiences. The clinician is your main client I would say, but you have your anatomic pathologists that are also reading this report. You have patients of course, that are reading these reports. Remember these reports, there's a lot of build up of hope, especially in the popular news media on genomic findings and patients will read every single word that you put in each reports. Then regulators, there's a huge amount of regulation that goes into this. Even without having a companion diagnostic assay, you still have to follow the same very strict regulations of CLIA. Most labs are CAP approved, so there's all sorts of requirements in that. There's huge, hugely different wants in these groups, and there's really different educational backgrounds. Those educational backgrounds don't vary just between the groups, but actually within the groups as well.
Anyone that's ever signed out a molecular report knows that you have your super users, your clinicians that you could literally just write one line just whatever variant was detected and send it out, and they wouldn't mind that you'd be able to do everything yourself. Then you have the other users that are a little less savvy, a little less up to date on this information. So you're writing this report really for them. Then there's a lack of standardization in the field. This is being updated quickly. I think that this is an issue that will start to really disappear, but genomic nomenclature can be really tricky. Especially with some of the older pre NGS genes variants part two is a perfect example. BRCA1, BRCA2. A lot of the initial variants that were described like the founder variants in BRCA1, BRCA2 in the Ashkenazi Jewish population were described in a different transcript than what is considered the canonical transcript today. So if you go back in the literature, you'll actually find that the Pdot values, the protein values are different than what you would normally report today.
A lot of the times with findings like that, what we'll try to do is put the older values in parentheses. I think this is like people that are [inaudible 00:22:58] like Cyan M2A the gene name change, we put parentheses MLL. So it's just a way of linking those things together. How we actually report the genomic change whether report the variant allule frequency, the exact copy number and amplification solutions, what exact fusion transcript. That'll differ from lab to lab, and I'm going to give some examples of that in the next few slides. Then what level of evidence should be used for clinical recommendations? It's really interesting, but molecular, an old the mentor of mine said molecular is the only field in medicine where you're allowed to make a recommendation based on an end of one study. There's a lot of, variants are relatively rare. The exact combination within groups is rare. So trying to find something that applies to this patient right now with these variants, with this disease, can be very, very difficult.
I think that it's important to not only report on the standards for the way of tiers of importance, but also in a narrative report which allows you to hedge, I want to say hedge, but allows you to provide context for the findings. Then I think that this third point is really more and more we're going towards the first [inaudible 00:24:24], whether it's to report findings or analyze findings. So are we just going to spit out some numbers and values, or are we going to actually try to analyze what we found? So match therapies, clinical trials adding guidelines, and/or large global analysis. So I think that we're moving more towards that, especially as clinical knowledge bases such as the one Pierian have allowed us to be significantly less time consuming. So let's go into one example of how different reports report the same exact FDA approved variant relatively differently. They can be confusing for the standard clinician.
I'm going to use NTRK fusions as the example. I just wanted, this first one is coming from ... These are all from large reference laboratories that the conditions would be expected to see on a regular basis. So this first one's from Tempus. You notice that NTRK fusions right here, it's showing you the fusion partner and then it's a dash. Then it's NTRK3, and then same chromosomal rearrangement. Chromosomal rearrangement and fusion, I would say the rearrangement is a more broad, but it's still a correct definition for what is going on right here. Remember the FDA label says fusion on it, and there have been reports of insurance companies denying the drug, because the test says rearrangement instead of fusion on it. So you can see that just even right here, there's already a little bit of a variance from how you'd expect to see it.
This is from Caris. Notice that in the top line, the top box, they don't report the fusion partner. They just say NTRK1, and then they say fusion detected. Then they add the drug in here. It's not until you go to the bottom that you see, here's the TPM3-NTRK gene fusion that was detected. So already, just a little bit of a difference here. This is from Labco. Labco I believe or yeah. You can see here that at no point did they include the fusion partner. It just says NTRK1 fusion, NTRK1 fusion. To be honest, this is the way that the FDA label is written. So for some people this might be easier, but it does not include what the actual fusion partner was. There's fusion written across the board. This one, so this as everybody knows this is Foundation Medicine. This is a real warning, an example of why really I believe that the whole CDx system is completely misplaced. You have this FDA approved drug with NTRK1, but because foundation locked in their CDx before the drug was approved, it is not a CDx for NTRK1.
So on the first page where you have all of these other genomic findings that have the companion diagnosis page, all you see is this really technical finding of NTRK1 TPM3. It even gives you the transcript names, which I believe that these the exons which are fused together. So it gives you all this technical information, but doesn't actually give you any of the clinical information, because they're not allowed to write it on the front. This is just going to be my little rant, but I believe that the CLIA LDT system is a far, far, far better way of allowing labs report genomic findings, especially in cancer. I think that the CDx system is completely misguided. This is the second page, and this is where you'll see these off label findings. It's very clear once you actually put it there, but it's not until the second page that you actually see it. This is something that can be missed if you don't know what to look for them.
Finally, this is actually from our lab. We report the variants of unknown significance, and I think it's just really important to reiterate that point mutation in these genes is not the same as a fusion, and amplification is not the same as a fusion, and actually has been shown to have no clinical benefit from targeted therapy. So it's important ... We actually got this call from a clinician and said, "I see it, I saw on the VUS section that there's NTRK1 point mutation. Can I give the drug for this?" We ended up like, no, of course not. I would've written it in the top line if it was there, but it's interesting because currently, the way that we, for tyrosine kinase the way that we really look for activating mutations is just to look for hotspots, right? Right now there have been no hotspots found in either the NTRK1/2 or 3 treatments for point mutations, and most like, I'm sure there's one point mutation out there that might be activating, but right now we haven't found that one. There is no indication for achieving the drug based on point mutation driven application, just only fusions.
So why is reporting important? Well on the left, we have a chart from the study for those published last year in JAMA, and was looking at, community practices in Ohio and non-small-cell lung cancer patients, whether they received just EGFR or analytic testing or some broad based genomic sequencing. The initial look at the survival data showed some favorability towards the broad-based sequencing, but when they apply more advanced statistical methods, in the end showed no benefit to it versus the routine. When you look down in the data, you see that people that received the broad-based genomic testing, only about 20 or 30% of them actually received the therapy that was indicated by said testing. So it really hammers on that point that if you don't receive the therapy, then of course the sequencing isn't going to do any benefit.
This was really confirmed when Foundation Medicine and Flatiron released their paper from the Flatiron Health Network that looked at ten of thousands of patients with non-small-cell lung cancer that had been sequenced with Foundation Medicine. You can see here that 1,100 of them had a variant that had an NCCN designated therapy, or an emerging therapy attached to it. You can see that only half of the patients ended up receiving that NCCN designated therapy, and the ones that received it, of course they did better than the ones that didn't. I think the really scary part is you look at the bottom chart, and for just the patients that had EGFR and ALK was very similar to this. So there's about what? 600 patients, 550 patients, only 65% of patients sensitizing EGFR mutations received EGFR inhibitor. These are some of the best characterized, most effective small molecule inhibitors they add years to patients' lives. Two thirds patients are actually receiving. I think the number for ALK was 70%.
When you take out EGFR and ALK, the remaining number is about 20, 30%, which correlates with that first paper as well. So there's got to be numerous reasons why these patients aren't receiving these therapies. I'm sure that insurance companies are being as difficult as possible to give off label therapies. I'm sure that there's a lot of clerical and logistical, and I'm sure some of these patients are sicker patients that maybe died for before they received the therapy, but it's not on the order of 30, 40%. So there's definitely some of these that are being missed due to just the general hustle and bustle of the practice, and not even knowing that these therapies exist for these patients. So how can we help? How can pathology help? Well, I think that there's really two parts, and they're intrinsically linked together. The first is this comprehensive analysis in reporting, utilizing EMR and crafting reports for really true personalization. Not writing just standardized templates for everything, providing updates on novel therapeutics.
So going back looking and saying, I just saw that FGFR inhibitors were approved. Do I have FGFR fusions in either bladder and other tumors, or do I have these [inaudible 00:33:26] bladder or other tumors that I can go back in and update these reports, update the clinicians. Hand matching clinical trials, both in and outside network, and I know how hard it is to get clinicians to get it at a patient's clinical trial in network. Then trying to save going outside network is even more difficult, and then explaining discordant results. Nobody wants to get a discordant result report, and then not have an explanation with it and then have to end up calling pathologist or the sign-out pathologist about why these things are different. So the second part is really outreach and educations. Both clinicians and general pathologists. So just emailing and calling, talking with them about new findings in the field, new findings in their cases, new findings that are relevant to them. This really builds personal relationships and trust in the laboratory.
It allows people to give you much more leeway to do what you think is fit towards the reporting. I really believe that this enhances patient care, because when clinicians have trust in the laboratory, they have trust in the findings, which means they have trust in the medicine. That's just something that can really help all the patients. So how do we report? We don't have any bells and whistles where I don't have a graphic designer to do this. The report is black and white. It's very, just a bunch of tables. What I believe is the most important is that the first page has everything really important for patient care. It has the summary of the results with therapies and guidelines. What we'll see here is like, so this patient, the young age female had EMR4 out fusion. So we said these are the drugs that are approved for it. What therapies would be the outside indication, resistance to therapies, any guidelines and any clinical trial opportunities.
Underneath that we have an immunotherapy section, so it's TMD now and includes a MSI as well. Then underneath that is the thing that I think that makes our report look different is this molecular pathology interpretation. Using the EMR, we're able to summarize the clinical status in this case, and it provides an in depth personalized analysis. Inside this as well, we give a summative report of what the actual guideline is. So here's the guideline from the NCCN for this patient. Whether or not, so we say that [inaudible 00:36:04] is preferred for these patients. Additionally, if there's a really relevant clinical trial that we believe will help this patient, obviously this one does have it, but we're able to talk about this here. What's interesting enough is that this patient also had a FISH, was FISH negative. Before the clinician came to me, I had already written up this report and so I called the clinician. So basically FISH has a 70% sensitivity for ALK.
It's funny, this is one of those things that even I didn't know before this case. I really think that RNA based NGS sequencing is going to really just completely take over from FISH for fusion detection. Then the remaining three to five pages, it's not a super long report, has everything else. So we added some clinical trial details, some variant details. The VUSs that we detect and in all the other regulatory requirements. So disclaimers, methodology, what's in the tests, et cetera. It's funny, I know that none of these things are ever read by anyone, because actually when NTRK was approved, we had tons of clinicians calling us and saying, "Do you do tests for NTRK?" It's like, "Yeah, it's been in the genes tested for the last year and a half." So, this is definitely something that has to be included in the report, but doesn't add a ton of additional value.
So let's just jump into the different examples of how we can help with the reporting. The first one is explanation of rare variants. As we do more and more sequencing, we're producing more and more rare variants, and clinicians are often pretty unprepared to evaluate them. This is where I really see the need for descriptive reporting and direct consultation. It's funny because a lot of the times these will have prior studies that are describing, because prior studies are more hotspot based and don't really detect, don't cover the same thing. So it's not that they were a false negative, it's just that they didn't have the coverage of that region and EMR is crucial in these cases. So the example of this is, the recent example is we had a 49 year old male, with metastatic bone and brain non-small-cell lung cancer. He was diagnosed at an outside sister hospital. They had done EGFR by Sanger, so Sangering all of the exonic regions, BRAF and PD-L1.
That was all negative at Outside Hospital. They intended to do a FISH panel for ALK and ROS, but they were unable to get it done because of lack of tissue. So lucky for us, it came to us for repeat biopsy just to actually complete the molecular profile. Our commission said you might as well do the next generation sequencing panel as well while you're doing this. What we found was that the patient had EGFR kinase domain duplication. So this is duplication of Exons 18-25. This is a rare EGFR variant. It's less than 1% of currently reported EGFR mutations, but because we weren't looking for it before, I bet you as with all of the EGFR activating variants that this will slowly increase over time. There is very sparse clinical evidence on it, because it's so rare, but the few reports that we saw suggested that there was sensitivity TKI therapy. I think the best data was on afatinib, but there really wasn't anything on newer TKIs.
This was all discussed with findings with ... This was also discussed with the commission prior to the tumor board, but also during the tumor board. I think the important thing here was explaining the discrepancy with the prior EGFR studies. So they said the other prior EGFR study was wrong, said no, the priory EGFR study only covers the exonic regions of the DNA. If you're not sequencing CDNA or the RNA, you're not going to find this if you're just sequencing the Exons. You're only going to find this if you're sequencing the deep NTRKs for the break points. This is just an example of how different testing methodologies we'll find different things. Eventually, Osimertinib was chosen because of its better brain penetration. The patient started therapy in his head, complete symptom really. So went from basically being in a nerve-racking pain to essentially being completely free of pain, and now has radiographic shrinkage of all of the brain and bone mets.
This next one is like kind of related to this before, and this one is just really, I love when something like this happens. It's the reason that I keep a list of interesting cases. So because molecular therapeutics is advancing so rapidly, novel therapies are continuously being approved. Often, the therapies that work on a target on one term tumor type work at another. BRAF is a perfect example with I think colon cancer being the exception, not the rule. So what we see is that when we have activating [inaudible 00:41:17] or something like that, generally those patients even if they have a different tumor type, will respond to the inhibitors that are designed for the initial tumor type. Really no patients should miss out if available. This is an example from a 73 year old female with GBM, and we have a very, very busy neuro oncology department. So this was actually one of the first fusions that we found, and was found to have an FGFR3-TACC3 fusion. It's about two to 5% of GBMs.
At this time there was really no targeted therapies. There was just the match trial, but as you know match trial requires you to receive standard care. So she ended up receiving that care, which was TMZ radiation, but unfortunately she developed aplastic anemia and was no longer eligible, because her blood counts were too low so she was considering hospice. Then the FGFR TKI Erdafitinib was approved for FGFR2/3 mutated urothelial carcinoma. We basically said, we queried our database we said who else have we seen with either of these mutations, or these fusions. We found this patient and actually 100 GBM patient with the same fusion. We reached out to clinicians and amended the report to include this as off label therapy. Let's see if the one transition I put, no, of course not. So it's supposed to have a little animation that shows if you put the Erdafitinib. We updated the report and right here we put Erdafitinib as the FDA approved therapy outside indication.
Interesting enough, the patient was able to recently obtain from compassionate care, and began treatment. So this is a real success story from the molecular profile and laboratory. Clinical trial matching. I think in those areas that is really the most contentious, I know that a lot of laboratories don't do this one because it's relatively time consuming to do it right, and there's really no easy way to automate it currently. Most patients have variants with an open related clinical trial, but a lot of the high level clinicians say if we're not going to use it, why do you even supply it? I have had other clinicians come to me and say, "We do use this." So I think that supplying for the broadest possible audience is the right decision. So how can we actually make this higher yield though? Here, once again, creating them takes a ton of work. Easily searchable automatable databases help, still is a very large manual portion.
What we try to do is we almost do two tiers of service. So the first tier is we match clinical trials based on the variance, roughly based on the tumor type exclusion material, all that sort of stuff. Try to make sure that we didn't recommend a pediatric trial for an adult, stuff like that. So relatively simple matching. Then for trials that I or the other signup pathologist we believe in, we really go a little more in depth. So we go through the EMR, we find out, see if the patient has any exclusion criteria. A lot of the time, symptomatic brain mets, stuff like that. Whether or not these patients will be eligible for the trial, even looking at blood values, stuff like that. Then what we'll do is if they're still eligible, we'll put that trial actually in the molecular pathology interpretation and in that narrative summary at the top. We'll actually write a little line about it and say, like this trial showed great data that we really think that this would be appropriate trial for this patient.
Obviously in system is a big plus, but even in the local area, we try to do this for them as well. An occurring example is this trial for EGFR or ERBB2 exon 20 insertions. We've been very heavily referring patients to trial at a local hospital that's not at Cedars, that uses this TKI poziotinib that has shown really great efficacy comparative to these 0% response rates that I see with normal TKIs right now. I'm going to do two little topics now about things that we're transitioning, and things that we are thinking about into the future. I'm going to start with tumor mutation burden. Really the interesting thing about tumor mutation burden and the belief I have in tumor mutation burden is that while right now billing is a contentious topic, because the billers and Medicare and everybody will basically say, "Well, you don't need to do a 5,100 gene panel, 200, 300 gene panel to detect these 15 biomarkers. You can do a much smaller panel, and bill it the 81445 rate down at the 81455 rate."
That's not true with tumor mutation burden. So tumor mutation burden can only be detected by large panel sequencing. You see that the gold standard is a tumor normal exome, but most panels are actually tumor only for cost reasons. You can see here that the sweet spot really starts to begin after one mega base. So this paper actually recommend something between 1.5 and three mega basis of sequencing coverage, and just that's usually in the range of three, 400, 500 genes. The reason for this is to be able to differentiate between the low mutational groups and the high mutation groups. Usually high in a foundation is at 20, I believe right now, but most labs are somewhere around 10 mutations per mega base, five to nine mutations per mega base. For intermediate low, less than four. The idea is that if you have high return mutation burden, you will respond to immunotherapy. If you have a low, you probably won't. Intermediate, maybe, but this is really a one size fits all approach.
Listen, this is what we're doing right now, but does it really translate from tumor to tumor? Is it the best way to do this reporting? I think not. So these two charts, so I'll go over the one on the left first. You can see that this is just charting objective response rate to immunotherapy by tumor type. The median number of coding somatic mutations per mega base, so how high the tumor mutation burden is on average. What you see is that there's a pretty good correlation. Tumors that have a high tumor mutation burden respond better to immunotherapy than tumors that do not. However, it's not perfect. I think the R square was about 0.75, and the authors used some statistics to basically determined that tumor mutation burden was about 50, 55% of the reason why tumors respond to immunotherapy. The rest is tissue specific. So maybe it's giving a straight value isn't the most important thing. Maybe it's important to give the value in relation to that tumor type.
You can see that from this study from Cornell, if you look at overall survival data, patients with the top 10% within their histology did amazing. Patients within the top 20% did pretty well, but the bottom 80% did much worse than these groups just after receiving immunotherapy. So a lot of labs have started to switch over to a system where they report not just a value, but a certain percentile in that tumor type. What this requires however, is either A, standardization of tumor mutational burden reporting across all labs, which there really isn't right now. Or B, a huge data set of your own to pull from, so that you can compare apples to apples. So for smaller labs, this can be very difficult. I think that as we move forward and we get that standardization, we will eventually see the ability to do this on a much larger scale. Then the second, a little ending point is this topic of visibility of reporting. I think it's really important to not miss the fact that reports are often PDF that gets just shoved into the dreaded media tab.
They're thrown in there, no one ever sees them again. They're not discreet, they're unsearchable. They're disconnected from the histopathology. They basically make it so that there's a huge likelihood that your report that you spent all your time working on, never gets looked at by a single soul. How about it was routed to the surgeons' message box, instead of the oncologists' message box. So the oncologist never note, gets an alert that it got received. There's all these situations where this can be really quite problematic, and this can lead to situations where the molecular results are never reviewed. So we're actually working on multiple different solutions, to try to see if we can increase the uptake and use of these targeted therapies. The first is a direct discrete data integration with the EMR. So an ability for the genomic reporting system to essentially dump the genomic data into the EMR, allowing the oncologists to edit with a click into their notes, so that it's always in there in the oncologists note. So they have the ability to review it.
That's a huge task and it requires a ton of work by the EIS, or the IT department of your hospitals and that can be slow. So a bridge that is something along the lines of a molecular consult note. So instead of putting this data in a PDF in a report, instead you take that first page data. So the findings and the molecular pathology interpretation, and you stick it as a note inside the EMR to increase that visibility. The third idea that we're thinking about is actually EMR alerts. So you take drugs like, you take non-small-cell lung cancer with the EGFR. So really, really high level evidence things, and you put in the chart. Every time that they go to order say chemotherapy or something like that, you'll have a little alert that says, "This patient has this alteration, and we notice that these drugs, one of these drugs was not prescribed. Are you sure you would like to do that?" Of course, you'd have it as probably initially as a soft stop, but there are institutions that actually start putting these in it's hard stops as well.
So these are all ways that informatically and with the collaboration with the clinicians, you can really allow this data to be better uptaked and better utilized. So finally, I just want to summarize really the high level points of the talk. Genomic profiling can provide very high value for both patients and laboratories. Molecular oncology are reporting currently and for the foreseeable future is highly, highly complex. Differences in the way that we report, just small language differences can have really real world impacts. Standardization and integration with the EMR can help mitigate these challenges, as I said before. I didn't talk about it quite enough, but when in doubt, just pick up the phone. Talk with the clinicians, explain the findings. Allow that relationship that you built to really help translate the data. So with that being said, I think we'll move to question and answer. I think we have five, 10 minutes that we go over.
All right. Thank you very much Dr. Vail. Hello everyone, this is Jeff Prescott. So I know and I appreciate everyone that has been sending questions in. If you do have a question, send it over. We do have limited time and I do have quite a few things to ask, so I will start this. Dr. Vail, would you talk a little bit about what the standard flow or reiterate the standard flow and process from an ordering physician to the results presentation? In particular, can you speak to a little bit about maybe how practitioners can work with handling administrative challenges which may be present in ordering tests?
Okay. So this is actually a topic that we're really trying to work on right now, especially around the idea of pre-authorizations. I know that we currently haven't quite solved this, so let's just go through the process. Right now, we have two different processes depending on tumor type. We have some reflex orders that are ordered automatically. So look, non-small-cell lung cancer, every case gets sequencing including stage one. We actually were beyond guidelines there. We seek what's every single case in non-small-cell lung cancer. Every single case of metastatic colorectal cancer, every single case of high grade Cleo neoplasms, and then for low grade neoplasms, we'll do it on physician requests. Then the other ones are more on physician requests, but depending on which group you're in, this could be a little different.
In general, the way that it works is the clinician, unless it's a reflex order, the clinician will either call us up, write an email. We have the ability to order inside epic. There's 1,000 different ways that they can do it, but once that gets done, then either I or the signup pathologists, usually me, will pick out which block to do the testing on. We'll end up doing the testing, which will go through testing and then reporting. To the administrative challenges, especially towards the billing side of things, I think as an academic medical center, we have the ability to eat a lot of this testing that a lot of private labs don't really, or at least smaller private labs. The large private labs obviously have a huge ability to do this, but the smaller private labs don't really have that ability to eat a lot of the costs of testing as part of development research, that sort of thing until the reimbursement market really stabilizes.
I've been talking to some of my colleagues and there's been a couple of great ways that they've got around this billing issue. I know one of them, I'm not going to say who's who, but I know one of them uses a dedicated staff member inside the pathology lab, who all they do is pre-authorizations for next generation sequencing testing. They've basically gone from a pretty normal horrible reimbursement rate to 80, 90%, because they've been able to have this sort of navigator that literally does everything from the clinician except push the button, to be able to get the pre-authorizations. That person has more than paid for themselves over the lifetime. So I think that having the pathology laboratory take that on, because one, they have the expertise. They know why these tests would be indicated. They have the ability to do this, and they have the ability to tie this in with the testing itself. The clinicians are just super, super busy. So this allows us to work symbiotically to get the testing from patients.
All right. Next question, what would you consider some perhaps ethical and legal considerations when informing, or if you ... Informing family members of genomic variants that have medical implications?
Obviously, you can't inform a family member of a genomic variant if the patient hasn't allowed that family member, that's basic [inaudible 00:58:50], right? Once a patient has expired, that's out the window. So that case that I was talking about, about that germline variant in that patient, that's a huge ... We talked with the ethics about it and they said absolutely. That's something that will impact that patient without having any impact on the current deceased patient.
Okay, super. Can you elaborate a bit more on what is meant by reporting protocols being superior through the CLIA LDT pathway, versus the CDx pathway?
So, it's higher than just reporting. It's flexibility of the assay. So listen, I get why everybody wants companion diagnostics. You get paid, right? That's a great thing. It's what makes, I think what's the thing that every dean of every medical school has ever said is that without margin there's no mission. So it's just the reality of the situation that we live in, but with that ability to get paid, you get a black box. You get no ability to change your assay without huge amounts of investments in time. You have no ability to really ... We run a couple FDA proven molecular tests and you get a result. That's it. You really get no good QC data, you get no ability to question your findings. We've had false positive findings on this that we've caught. How many have we missed? We have no idea, because there's no ability to actually look inside the assay. So I really strongly believe that that the CDx pathway is not the correct pathway for molecular testing, but I'm getting the DFTH stronger than me.
Got you. Would you say that all solid tumors regardless of stage, should have their cancer comprehensively genomically profiled?
The researcher in me of course, says yes. So I think that to ever get the information that we need to go down the path of solving cancer is to literally get millions of tumors sequenced. The way that we do that is we sequence every tumor. As costs have gone down, I think that it starts to become more and more feasible. To be honest, two or $3,000 in the grand scheme of a cancer patient's care is a drop in the bucket. I think that it's something that is being denied now, not because of lack of clinical benefit, but it's being denied now because the drugs that are associated with this testing are so expensive that the insurance companies don't want you to find the associated biomarker. The real end point is that right now is there clinical utility in testing patients with low stage disease?
Probably not, but there's a neoadjuvant trial going on right now for patients with EGFR surgically resected, EGFR mutated non-small-cell lung cancer. How else are you going to, if it's surgically resectable, that means you shouldn't be testing it in the first place. The initial data that was looked at for whether or not neoadjuvant or adjuvant EGFR TKI worked EGFR mutated lung cancer, use EGFR amplification as one of the biomarkers. It's completely outdated. It doesn't really apply to the current knowledge base that we have right now. I think you're going to find that in a lot of tumors that actually providing these therapies as an adjuvant or neoadjuvant will prove survival. That's just my educated guess, so I think there will be utility in the future.
Okay. When dealing with someone that already has a diagnosis, or have been treated in further down the line, does it make sense to go back and tests a tumor that's already been actively treated? That makes sense?
So you're saying, should I go back and test the original tumor, or should I test the metastasis? Is that what you're saying?
I think you should test both. I think that the most comprehensive information you can get is both, because once again, we had seen separate primaries where they're completely discorded. This is something that can have real implications on both prognosis and treatment. Maybe I'm biased because I run a sequencing laboratory, but I think that the answer is to sequence as much as possible. I know the director of translational genomics that are in our institution [inaudible 01:04:05], he has a couple of papers showing the difference in heterogeneity in both lung cancer and in colorectal cancer, and the impacts that can have on therapy selection. So these things, even with tumors that are obvious metastasis from the same primary, I think it's helpful to sequence as much as possible.
All right, thank you. So one last question and then I think our time is up. We did have a few folks asking about rare variants. Can you talk a little bit about how you may have found them, and what do you think clinicians should know about rare variants?
It depends on the level of evidence supporting that rare variance as a biomarker of some sort or another. The way that we find rare variance is just you sequence a lot of cases. If you broadly sequence a lot of cases, you're going to find a lot of reverts. If you look at any chart of the most common mutations in any particular cancer type, usually P53 is up at the top at like 60, 70, 80, 90%. Then it will be like CDKN2A like 30%. Then everything else will drop off and be like five, 10, 15, three, two, one, and there'll be 30, 40 genes, one, two, 3%. This is the way that cancer is going to be solved, where it's going to be a death of a thousand cuts sort of thing. So the only way that you get that information is to do more sequencing, and then to actually act on that sequencing with therapeutic treatment, and to see if those patients respond to those therapies.
With that kinase domain duplication, it was rare, but it was common enough to have some therapeutic case reports written. We see rare variants that we're pretty sure are pathogenic, [inaudible 01:06:11] site that is completely conserved in all vertebrates, a different mutation at the same site causes pathogenicity. There is the area's depleted for missense mutations, and the whole gene is depleted from missense mutations. You just get this whole [inaudible 01:06:32] that's saying that this variant is pathogenic and it's driving the disease, but there is for that specific one, there is zero biochemical functional analysis done. So you have to then hedge and you have to say basically these are all the reasons why I think this is pathogenic. These are all the reasons why I think that this has some value for this patient, but there is no direct biochemical analysis on this and it could possibly be wrong. That's side of this ... You can only go so far. You can only go as far as literature that take you.
All right, very good. Thank you very much, Dr. Vail. Again everyone, on behalf of our speakers, Dr. Vail and Dr. Nagarajan, this is Jeff Prescott with OncLive. I appreciate everyone's time today. Enjoy. Take care, and thank you.