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.
Julia Karow: The title of today's webinar is NGS Roundtable: Ask the Experts about the Future of Clinical NGS Testing. The sponsor of this webinar is Pierian DX. Our Roundtable participants today are Dr. Allie Grossmann. Assistant Professor of Pathology at the University of Utah and Medical Director of Surgical Pathology and Molecular Oncology at ARUP Laboratories. Dr. Karl Voelkerding. Professor of Pathology at the University of Utah and Medical Director of Genomics and Bioinformatics at ARUP Laboratories and Dr. Rakesh Nagarajan, Chief Biomedical Informatics Officer at Pierian DX, who will be motorizing the panel discussion.
Julia Karow: Today's webinar, we have an interactive format driven by questions, you, our audience, provided through a pre-webinar email survey, as well as questions you enter through the Q&A box. We have structured the event as follows. Rakesh will introduce our first topic emerging scientific and clinical trends in NGS Testing and Allie will then give a short presentation about liquid bio-C testing for cancer, as well as next gen sequencing to predict immuno therapy response. This will be followed by a panel discussion based on your previous submitted questions. After that, Karl will introduce our second topic NGS Clinical lab operations and quality management and our panelists will discuss that topic based again on questions you submitted through our survey. Following this, I will moderate a short Q&A session to address additional questions submitted during the webinar.
Julia Karow: I will now hand it over to Rakesh, please go ahead.
Rakesh Nagarajan: Thanks Julia and GenomeWeb for hosting this webinar. It's a true pleasure moderating this round table session with professional colleagues of mine or real leaders in the cancer and constitutional clinical genomics testing space. We've organized this webinar into two segments. One really a scientific and clinical segment focusing on merging trends and a second regulatory and quality management issue segment, both obviously tying to Next Generation Sequence based testing. Without further ado, I want to introduce Allie Grossmann from the University of Utah and ARUP Laboratories will kick off the first session where we'll be discussing NGS based testing to detect cell-free DNA and to make immunotherapeutic decisions in the oncology setting. Allie?
NGS Testing of Solid Tumors
Allie Grossman: Today I would like to emphasize right at the bat that what I'm going to talk about is really emerging testing in the solid tumor oncology field and I would like to emphasize this is emerging and very few of the tests and platforms and strategies we're talking about today are actually in the clinic. A couple of them are. Giving, I want to just acknowledge my colleague John Phipher who couldn't be with us today. He helped compose the content here and had a lot of input and I'm sorry he can't be with us but I'll do my best to represent his input.
So, first, Liquid Biopsies. So, we're talking about testing for mutations in solid tumors with cell-free, circulating tumor DNA. It's a really exciting topic with quite a bit of potential clinical utility. I'll try to touch base on really very basic general strategies, what the limitations are right now and what they might be in the future and who are the patients that might benefit the most. Next slide.
So, first, what is a potential clinical utility of liquid biopsy? And I want to emphasize the word potential. There are several. This could be used, so, taking a blood drop from a patient can be used to establish therapeutic targets, to detect residual disease after therapy, even monitoring treatment response, to detect relapse or progression, detect development of resistance mutations, and potentially even limit more invasive, higher-risk, more costly procedures, potentially limit costly high resolution imaging and what ... go ahead, next slide.
What might be some strategies for interrogating these somatic alterations that we might find in the blood. There are two general strategies. One is you can go after very specific known mutations with a targeting mutation detection and let me emphasize, there are a lot of different methods to do this and each have their strengths and weaknesses. I've listed just one here that is popular right now and has made it into the clinic and the technology that we use here at ARUT, it's called digital droplet or emulsion PCR. The other strategy would be to look at a whole genome, a whole exome or a broad, or even a small hot spot panel but multiple targets within an NGS panel and I've listed just a smattering of references. For our audience, there are a lot of papers on this topic and this is the papers listed here are by no means comprehensive. Next slide.
It's really important, I think this is probably one of the most important slides here. It's really important to talk about limitations of liquid biopsies and probably the most important concept is from when using NGS panels to interrogate multiple mutations from liquid biopsy specimens, there is some really special bioinformatic pipelines that are going to be required for this and these are emerging and I would refer our audience to a recent paper in nature biotechnology that I have listed here. A group from Stanford has an excellent publication but this concept of using barcodes to eliminate false positive from sequencing artifact is really important. These are emerging bioinformatic and nucleic acid chemistry approaches and we're going to see more and more papers on this. There are some publicly available pipelines and software tools that can be used for this but again, this is emerging and another concept that's really important in terms of limitations is the sensitivity. The sensitivity at this point, in general, appears to be less than testing the solid tumor directly. I think that's going to change overtime and I think sensitivity's going to improve but right now, it appears to be less sensitive overall but that's going to depend on the tumor burden of the patient and we see this in multiple publications and multiple studies where depending on the tumor burden in the patient, your ability to detect very low frequency mutations in a cell-free DNA fraction can vary widely. Next slide.
So, where this might have the most clinical utility, I think that's going to be in lung cancer and the reason is because lung cancer has multiple actionable targets and yet we are required to test for these targets on very limited tissue from lung biopsies. So, you have multiple target that need to be interrogated and they compete with each other. In addition, pathologists need the tissue for a diagnostic work up. So, some of the tissue gets used on immunohistochemistry for example and when that's established, then mutation detection and a biomarkers for immunotherapy begin and though right now EGFR and ALK are part of the standard routine testing for lung cancer, however there are many other mutational actionable targets to go after that are emerging with actually drug data linked to them and then recently, there's been a big demand for PD-L1 biomarker testing and these are immunohistochemical stains and there are two FDA approved biomarker [inaudible 00:10:16] for PD-L1 from two different companies that go with two different drugs. Recently, in this fall, the FDA approved the drug Pembrolizumab/Keytruda in the first line setting for lung cancer.
So, this is for lung cancer that is EGFR and ALK wild type and at this time, this is the only immunotherapy that's FDA approved for newly diagnosed lung cancer patients and there's been a huge surge in test ordering for the companion diagnostic that goes with that and that is the PD-L1 22C3 immunohistochemical stain and again, this competes for tissue with all these other markers that need to be tested and it presents a real challenge for the proceduralist, the radiologist and pulmonologist who are sampling the biopsy. There's the demand for them to get more tissue, which is hard because they're at risk for the patients. It puts a lot of pressure on the pathologist who has to very carefully manage this tissue to accomplish everything the oncologist needs to take care of the patient and that brings up the topic of immunotherapy. Next slide.
So, what's really exciting right now in oncology and I'm sure you all have heard about this is immune checkpoint blocking for cancer patients and I think probably the pre-tumor types where this has been most of the data is in melanoma, lung cancer and [inaudible 00:12:02] but there are two pathways that have FDA approved drugs on the market. I'm sure you all have heard of this. For the CTLA-4 pathway and the PD-1, PD-L1 pathway and I'm not an immunologist but I can just briefly give you an overview. So, the CTLA-4 inhibitors like [inaudible 00:12:25] interfere with T-cell activations and so, antigen presenting cells, so APC's present tumor antigens, novel two antigens to a T-cell and the T-cell becomes activated to attack those tumor cells that are expressing those antigens and this is where mutations come up. Somatic mutations create what's called neoantigens. So, tumor antigens. So, once that T-cell is activated, it's going to be searching out those tumor cells to kill them but the tumor cells are sneaky, they have mechanisms of inhibiting the T-cells and that's where the PD-L1, PD-1 pathway comes into play.
So, PD-L1 is the ligand, there's also PD-2 and PD-L1 is expressed on the surface of tumor cells and that's where we can visualize that and see it and quantitate that with immunohistochemistry and that's the FDA approved biomarkers for PD-L1 but PD-L1 expression on the tumor cells will interact with the PD-1 receptors on activated T-cells and suppress T-cells and it helps the tumor evade the immune system and so, they're, again, FDA approved drugs such as Pembrolizumab and [inaudible 00:13:51] that inhibit this pathway. They're PD-1 inhibitors, they're also PD-L1 inhibitors that are in clinical trials. Next slide.
So, the challenge in the field is because not all patients actually will respond to immunotherapy and there are some toxic side effects from immunotherapy. The challenge in the field is coming up with biomarkers that predict who will benefit the most from these agents. They're expensive agents and again, have toxic side effects and we would like to minimize the use on patients who aren't going to respond. Again, I'm not an immunologist. The immune system is very complicated. There are a number of review articles written by immunologists and oncologists that address this issue. I would refer our audience to an issue from Science Magazine from May that has a number of articles addressing this and there's a quick editorial by some leaders in the field and the title of that editorial is called the Cancer Immunogram. I would refer them to that but just briefly let me touch on some of the really interesting data out there.
So, we talked about mutations leading to tumor antigens, neoantigen. Well, it turns out and there's some excellent data in melanoma that overall, global mutation load or neoantigen load can correlate with who will respond to either anti-CTLA-4 therapy or anti-PD-1 therapy. So, that's really excellent news and it's definitely a global mutation load is going to be something we need to keep an eye on in terms of biomarker development. What was also very interesting was scientists had really hoped that by sequencing neoantigens in many patients who are receiving these therapies that recurrent neoantigens that were common among patients that correlated with response would be found but unfortunately, that was not the case. Nevertheless, overall mutation load has persisted both in melanoma and lung as correlated with response. Next slide.
And I would refer you to there are several papers actually in lung cancer that show this that the nonsynonymous mutation burden that is associated with objective response and clinical benefit and profession-free survival in patients treated with anti-PD-1 therapy and again, this concept of mutation load, mutations being the new antigens, there are multiple ways to interrogate. I would say mutation load is a surrogate for neoantigen load and you can measure neoantigens by RNA transcriptome data, you can use a surrogate, which would be mutation loads by DNA. The tumor types that have the highest global mutation load appear to be lung and melanoma cancers as well as a couple of other types but that may be why we're seeing great responses in these patient populations. Next slide.
So, the question has come up. Can NGS, gene panels substitute for whole exome sequencing or transcriptome sequencing in terms of biomarkers in predicting who's going to respond to immunotherapy? Well, there is some data out there that NGS panels and the mutation load detected in these panels do correlate with clinical responses and have some predictive power that is comparable to full exome sequencing but I would say there are a lot more studies that are needed. In general, we don't really have any standards for how this ... for the wet chemistry, for how these essays are done. We don't have any standards or the bioinformatic pipelines. We don't know what the absolute thresholds are for high intermediate or low mutation burden and how that exactly correlates with predictive power. So, there's a lot of work that still needs to be done. Next slide.
And this is just another set of data that basically suggests that comprehensive genomic profiling can predict response to immunotherapy. I would say that what I have observed is that and the literature reflects this is that global mutation load and PD-L1 expression are not the only biomarkers that appear to correlate with responses. We know from a variety of publications that certain oncogenic mutations will activate PD-L1 expressions for example and influence the patient's response to immunotherapy. We know that resistance mutations can develop and interfere in gamma signaling pathway that will inform how a patient is going to respond and so, certainly NGS, whether it's full exome sequencing or gene panels can be informative but how all that comes together with other data on for example, tumor infiltrating lymphocytes, how, what populations of immune cells are present and other [sidealitic 00:20:14] expression markers that have been linked to response to immunotherapy. How all that is going to come together in a cohesive story that we can package into a biomarker panel is still yet to be determined. Next slide.
Rakesh Nagarajan: Great, thanks Allie. I'm going to now open up the Q&A part of this segment and we'll get right to the first question and these questions came in as part as a pre-webinar survey. So, can you comment on the limitations of circulating tumor DNA testing with regard to clinical setting? And you've covered a little bit of this in the introductory slides but perhaps for specific tumor types and all three slides. But perhaps the specific tumor types and also, as a validation characteristic like lower limit of detection as expressed by either allele frequency, or number of copies, and sensitivity.
Allie Grossman: Sure. I mean, these are important questions that they are difficult to answer. The challenges of the cell-free DNA is that, first of all, the amount of tumor cell-free DNA that is in every patient is gonna be highly variable.
And that variation depends on tumor burden on the patient and how well the tumor's diffused. It might depend on tumor type. It looks like in glioblastoma, it may be harder to detect because of the blood-brain barrier, perhaps. It may be harder to detect cell-free DNA sometimes.
It also can depend on the contribution of the healthy cell population to the amount of cell-free DNA that's mixed with tumor DNA in the peripheral blood.
And so you could imagine that if you have a patient who is in the midst of a massive inflammatory response that there's gonna be a lot more shedding of cell-free DNA from non-tumor cells, and that's gonna dilute the [inaudible 00:22:28] fractions from somatic events.
So really, in terms of assay validation, anyone who attempts this is gonna have to validate their assay over a wide range of intuitive DNA. And the lower limit of detection will vary over the range of the input in what's going on in that patient. And there are some simple ways to think about this.
You can estimate what your lower limits of detection might be. And we know, for example, that genomic loci are represented in one nanogram of DNA. There's about 330 copies of each genomic loci.
And so 10 nanograms, in our experience, 10 nanograms of DNA is pretty reproducible extraction yields from cell-free DNA from cancer patients. And that would be about 3300 copies of DNA.
You had a hundred percent conversion of all of those copies in your assay. Was either the PCR prep, or the library prep, or whatever technique you're using, you'll get 3300 copies.
Well, if you wanna detect a .1% in allele fraction, that would be calling unique reads in three copies.
So this is an example of how the amount of input you're starting with can really vastly affect your sensitivity and your limit of detection.
But again, anyone who wants to validate these tests are really gonna have to validate a wide range of DNA inputs.
So the other thing that's a real challenge is the size of the DNA fragments. Recovery of those very small DNA fragments, like a 140 plus base pairs, is technically challenging and probably where a lot of the effort needs to be invested in development and validation.
And I think the cash we may have a question coming up, that's really into this. In the next slide perhaps?
Rakesh Nagarajan: Yeah, that's right. [crosstalk 00:24:52] So I think we could talk about the experimental methods and I think you've hit on size selection and I think you could expand on that here, as well as the molecular barcode issues that you've brought up.
Allie Grossman: Yeah. So what experimental methods are recommended for purification? Well, I understand from my colleagues who are true nucleic acid biochemists. They tell me that, frankly, the cryogen kit for circulating nucleic acid is probably the best kit for extraction.
They've tried the automated methods and they haven't been so good. So I think their recommendation is, right now, this cryogen kit. I don't generally want to try to promote any one company but I would have to say that would be their favorite.
In terms of library prep, I think the most important concept here and, again, I'm hearing this from my colleagues who are nucleic acid biochemists who designed these assays and the really smart bioinformaticians that they work with, is that the wet chemistry is the library prep.
And the bioinformatics pipeline need to be developed together. And this is where those barcodes come in. So there's this concept, this term barcode molecular family.
These barcodes have to be added as part of the library prep. And those barcodes are really important for eliminating errors and false positives. And that's probably the most important point in addressing this in terms of maximizing sensitivity and specificity.
Rakesh Nagarajan: Great. That's perfect. And I think the reference that you've put on the initial slide use the molecular barcode family to really increase sensitivity, and also reduce or polish error rates that were normally seen as part of oxidative damage in the library prep. So I think both methods, at least in that manuscript, worked well.
So the last question is, really, what are the regulatory requirements for cell-free DNA testing? We can limit this to circulating tumor DNA testing but I suspect the answer'll be the same.
Allie Grossman: Yeah. And I actually had to call my colleague, Dr. Voelkerding, on this. I am not aware of any regulatory requirements. I think that should be determined, at least for circulating tumor DNA.
I know that the FDA, for example, and TAP have an interest in regulatory requirements for NIPTs or prenatal testing and Dr. Voelkerding can comment on that a little bit more. Would you like to speak up, Karl?
Dr. Voelkerding: Yes. Can you hear me? Yes, hi. This is Karl Voelkerding. Currently, through the College of American Pathologists, there are laboratory requirements that are articulated in a series of documents that laboratories need to meet in terms of checklist requirements if they're performing noninvasive prenatal testing on circulating cell pre-DNA, but specifically, for the detection of fetal aneuploidy.
At this point in time, there are not yet requirements through the College of American Pathologists for laboratories performing cell-free DNA specifically, but they are in the process.
The respective sub-specialists within the College of American Pathology committees are beginning to look at this topic because it is an emerging topic specifically what types of controls and requirements would laboratories need to meet if they're employing cell-free DNA for clinical diagnostic purposes.
Rakesh Nagarajan: Perfect. Thanks, Karl. So I wanna spend just a few minutes on immunotherapy questions. This slide really summarizes the majority of the scientific or clinical questions.
So, can you comment on the assay performance when you're using NGS to make immunotherapy decisions? So, for example, is there published reliability and error rate and how does this compare to immunohistic chemistry?
Allie Grossman: Yeah. I actually don't think there's adequate published data yet on using NGS to make actual clinical decisions. I think that this is still in the research field. In tumor boards we talked about this. We may have a report in front of us that the lab performing the test has commented on this.
This appears to be a hyper-mutated tumor. However, I would just remind everybody that we really don't have enough data yet to know what those thresholds, those cutoffs, really should be.
And so we're kind of feeling around in the dark at this point. In contrast, for immunohistic chemistry, we have two FDA-approved assays and one of 'em was performed in the clinical trials, the 22-c3 biomarker for benralizumab.
That performed really well enough, at least in lung cancer, to actually make it as a companion diagnostic test. Is it perfect? No. It's actually very difficult to interpret.
The FDA and the pharmaceutical company and the [inaudible 00:31:04] have put forth a test, through clinical trials, that is a subjective assessment.
You're using immunohistic chemistry to estimate quantitative levels of pedia-1 expression on tumor cells and that's the challenge. And I would say I don't think that that's gonna end up, in the long run, being the only test we have available.
And I wouldn't be surprised to see it combined with NGS-based assay and other assays that look at expression levels of cytolytic pathways and antigen presentation pathways.
I think all of that is coming. And it'll be really interesting to see what combinations of different interrogations and different modalities they'll use to come up with tests that are more informative than what we have now.
But in terms of NGS, the data is very much emerging and we just need to keep tabs on this.
Rakesh Nagarajan: Great. Thanks so much, Allie. I think we're gonna stop that for a segment. I wanna move on to the second component. In our experience that Pierian has been very similar with our partners, they're starting to ask about more and more comprehensive assays including those that use circulating tumor DNA testing or looking at tumor mutation load and I think this will be a ripe area in 2017 and I really expect our partner sharing network where we're able to amalgamate data across customers to perhaps feed into the knowledge that's gained by looking at targeted panels and predicting tumor mutation load.
So with that, I really wanna move to the second segment. I wanna introduce Karl Voelkerding from the University of Utah and ARUP Laboratories. Karl will be discussing proficiency testing for clinical NGS applications. Karl?
Proficiency Testing for Clinical NGS Applications
Dr. Voelkerding: Yes. Hi. Good morning and good afternoon. I want to, first, just provide for the audience some general background.
Clinical Laboratories, specifically in the United States that provide clinical testing, are required to be certified through, ultimately, the federal branch of our government through a legislation that was passed in 1988, referred to as the Clinical Laboratory Amendments of 1988, often just referred to as CLIA, CLIA 1988, or CLIA '88.
And this is a requirement. There are several sub-requirements within the CLIA legislation and one of which includes that CLIA-certified laboratories are required to participate in proficiency testing for the tests that they offer for clinical diagnostic purposes.
And essentially, proficiency testing is the external performance indicator of a laboratory's quality and ability to accurately generate diagnostic results.
Now, through the years, one of the organizations that has come to the forefront in terms of providing proficiency testing has been the College of American Pathologists.
And the College of American Pathologists, or CAP, has what is referred to as Deemed Status through the Center for Medicare and Medicaid Services and the Center for Medicare and Medicaid Services oversee administratively the CLIA program.
And so the CAP through this Deemed Status has the ability to accredit laboratories, that is if a laboratory is set up and it wants to offer diagnostic testing, it needs to undergo an accreditation process and then subsequently participate in proficiency testing.
And so the CAP both accredits laboratories and administers proficiency testing for a wide variety of diagnostic testing inclusive of clinical chemistry assays to immunoassays to mass spectrometry-based assays to a variety of genetic-based assays. Next slide please.
So one of the critical challenges that's perennially present in proficiency testing is how to actually develop the material and actual samples that will be sent to clinical laboratories for proficiency testing.
And this is very true for all areas of testing and it's been especially challenging to develop these types of materials and samples to provide proficiency testing for clinical next generation sequencing-based diagnostic testing.
And this is further challenged by a very rapid and growing diversity of the different types of diagnostic assays that are being based on next generation sequencing.
The first assays that were being offered by clinical laboratories included those that were focused on the detection of germline variants or inherited disorders. This area has expanded from sickle gene-based testing by next generation sequencing through multi-gene panels now to exome and whole genome.
So this has expanded considerably over the past five years. Also within that time frame has been the introduction by multiple laboratories of gene panels for the detection of somatic variants in both solid tumor and hematologic malignancies.
And relevant to the presentation by Dr. Grossmann, I think we are going to see on a near-term horizon more and more exome- and/or genome-based sequencing on solid tumors and hematologic malignancies, both for diagnostic purposes of the primary genetic variation but also to provide these types of information for correlation of neoantigen load as reflected by sequence variation at the genomic level. Next slide.
So in order to develop proficiency testing materials or, essentially, samples that would be sent to laboratories in a blinded fashion that they would then process through their laboratory protocol and report back results to the College of American Pathologists, specifically for next generation sequencing, we have been leveraging the concept of a methods-based proficiency testing.
And this is really to asses the laboratory's ability to detect a spectrum of germline or somatic variants in clinically relevant genes. And this reflects the primary analytical goal of this type of testing.
Historically, most proficiency testing in the area of genetic applications was what we referred to as analyte-specific. So, for example, a sample would be specifically tested for the cystic fibrosis gene, with the cystic fibrosis gene being the analyte.
Or we would send out samples specifically for KRAS or NRAS testing with those two genes being, respectively, analytes.
The challenge is is when a lab has a 50 to several hundred gene panel for somatic variation, you are now dealing with a diversity of genes, a diversity of variant spectrum from single nucleotides to insertions and deletions.
So the main question is can the laboratory detect the variation as opposed to any one specific variation in a specific analyte. So you have to have a broad test to determine router capability of the assay, hence the concept of a vapid-space proficiency testing.
This can be augmented by inclusion of samples that have specific variants that you want to make certain a laboratory can absolutely detect. For example, a mutational hotspot, a somatic variant mutational hotspot. Next slide.
So, we undergo an extensive process to develop proficiency testing materials for next generation sequencing. For example, in the area of germline variants, we have focused on taking human cell lines that have been appropriately consented by the donor and then characterizing them by subjecting them to exome or whole genome sequencing, or both, and then doing a very thorough bioinformatic characterization to essentially identify the variants that are present in the sample. Then that provides us a population of sequence variation from which we can establish a proficiency test.
In the area of somatic variant proficiency testing, one of the challenges we have there is to try to essentially replicate the variable or varying allele fraction or percentage of somatic variants at different levels. For example, a 50% allele fraction of a variant versus a 5% allele fraction because this is the type of range that laboratory because this is the type of range that laboratories are actually diagnostically testing for. So therefore, you need to both test the laboratory for it's ability to detect the variant but also to detect the variant at a relatively low allele fraction.
So the lower limit of detection of a laboratory needs to be assessed as well. And in this case, we're utilizing cell lines that can be genetically engineered specifically to reflect those types of variations in both sequence and allele fractions. Next slide.
So, in 2005, the College of American Pathologists launched the first proficiency testing for next generation sequencing and this was specifically a program or a proficiency testing program that would encompass laboratories performing both gene panel testing for a variety of inherited disorders to exome and genome sequencing.
And so this particular proficiency testing program is characterized by laboratories receiving 10 micrograms of extracted DNA from this highly-characterized cell line and then the laboratory receives a list of information of up to 200 chromosomal positions that they can interrogate depending upon their assay.
So perhaps a laboratory is doing a multi-gene panel for a particular disorder and in the 200 selected chromosomal positions, they may actually test for seven or eight of those positions. So that's what they focus on. They perform their assay using the DNA, take it all the way through their bioinformatics pipeline, identify variants, and then report back to CAP any variation or if the chromosomal position is a reference or wild type position, they report that information back to CAP.
Is a variant detected or not? What is the zygosity of the variant? And then if a variant is present, they're requested to describe the variant using standardized nomenclature from the Human Genome Variation Society.
And this particular program has been in effect since 2005 and now we're in almost at the end of 2016. Next slide.
Then subsequently, two additional programs for proficiency were introduced specifically for somatic variants. And in the top panel of this slide, it just shows the catalog description of one of the two programs.
And this first one atop is specifically for laboratories performing solid tumors. And you can see in the additional information below in the paragraph, you can see the list of genes that this particular proficiency testing program is relevant to and will assess.
And this particular program, laboratories receive physical DNA samples, they are asked to process them through their sequencing platform, targeted enrichment for example, followed by their bioinformatics and then, again, to report back to CAP what they've identified the zygosity and what is the description of the variant.
In the lower panel is essentially a correlate to this which is a proficiency testing program for a variety of genes that are involved in hematologic malignancies. And you'll see those genes listed in the additional information.
And it's important to note that, biologically, there is overlap in some of the genes between the solid tumor panel proficiency testing and hematologic malignancies. For example, p53 kit, for example.
And one thing additional on this is that proficiency testing is usually administered twice per year so you'll see, on the right-hand side, it says program information, the amount of DNA, how many samples, and then, in this case, shipped out twice per year for laboratory so that they can meet their CLIA requirements. Next slide.
So, we do have, though unfortunately, we have some very critical limitations of physical proficiency testing materials. One, they are very expensive to produce and maintain, especially cell line-based material. Ultimately, they have a genomic limitation of information, if you will, so you ultimately cannot replicate the diverse spectrum of variants that a laboratory may encounter during clinical diagnostic practice. And so any one sample will be limited in terms of the type of variation that is present in the sample.
And also, these are limitations so one approach that we're taking to this limitation is actually to turn to what we're referring to as in silico proficiency testing which is based on taking DNA sequence files and actually introducing pathogenic variants or variants of uncertain significance, perhaps, depending on the scenario, enter the sequence files and then providing those to the laboratory and then they can assess their bioinformatics pipeline.
Now, there are lot of variations on the theme that could potentially be pursued by this but in the next slide, I will discuss the first in silico proficiency testing product that we have available for somatic variants.
The approach we took on this is to focus on two popular commercially-available cancer panels. One offered by Illumina, the TruSeq Amplicon Cancer Panel, one offered by Thermo Fisher, Iron Torrent, which is the AmpliSeq Cancer Hotspot Panel.
And in this setting, these are what is provided to the laboratory, and it's important to note that the laboratory has to be performing either of these two assays for this proficiency testing to be relevant to them, is that they receive sequence files, so either FASQ files and/or BAM files, and within those files, a variety of variation has been introduced into the relevant genes that are captured by these methodologies, captured and interrogated.
And then the laboratory takes these FASQ and/or BAM files and then introduces them into their bioinformatics pipeline, takes that through their bioinformatics process, identifies the variation, and reports that back to CAP.
And so these are programs that are now ongoing, and there will be more to come in the next year and onward from CAP, and again, to try to meet the diversity of the growing applications. Next slide.
Rakesh Nagarajan: Right. So Karl, I wanna thank you for a good introduction in the proficiency testing as well as that for NGS and it's great to hear about these [inaudible 00:51:03] products. I'm gonna jump right in to the first question.
So far, the proficiency products that have been released by CAP have been educational in nature. Can you talk about when this will become graded and can you also talk about in silico products that go beyond the defined assays that you described in the previous slide?
Dr. Voelkerding: Sure. Thank you for that question. So when the process to develop our proficiency testing typically follows this format. There's the idea and conception of what would it'd be the proficiency testing product as generated by individuals that do volunteer work and serve on committees for the college.
And then, essentially, a pilot project is conducted based on the concept to see if it can be successfully piloted through volunteer laboratories, a small subset of volunteer laboratories. And then, if that looks promising, it is then launched as an educational proficiency testing that laboratories can subscribe to.
Important to appreciate in educational proficiency testing is that the laboratories that participate are not graded specifically and therefore, they will not be penalized based on their performance for that proficiency testing.
The CAP reviews these educational proficiency testing surveys, they always occur in the first year of the launch of a proficiency testing program. It can certainly extend into a second year until there's sufficient data accumulated by CAP in terms of how are the laboratories handling this new PT product. Are they technically being successful in handling it logistically? Are the results that are coming back in sufficiently robust that it warrants that we can convert this to a graded proficiency testing program?
Of the four products that I just described, the germline product that was first launched for germline detection of panels within panels exomes or genomes has been converted over to a formally-graded program at this point in time. The other three programs are still in their educational phase.
Now, in terms of what's next, specifically with regard to in silico proficiency testing, what we are trying to work for, our goal is to work for further developing what I would refer to as our NGS bioinformatics infrastructure within CAP to allow us to do truly-customized in silico proficiency testing. What I mean by truly-customized is that a laboratory in the future would submit their sequence data files generated through their own physical samples to CAP. Those data files would then be in silico mutagenized and then sent back to the recipient laboratory. So they would either submit data files from samples that they had sequenced or CAP would provide them a sample, a physical DNA sample, the sequence, and then they would submit those. So we're looking at different options in that regard.
And then the real value of this kind of customize is that the laboratory might have a 200 gene panel for a specific area of inherited disorders that they want to be assessed on and they would provide us the genes, they would provide us information about their technical sequencing. We would receive that information and then we would introduce sequence variant content into those files that would be returned back to the laboratory. That way, we can begin to accommodate the wide diversity of assays that are being utilized by different laboratories. Everything from different size somatic variant cancer panels to different germline panels to exome and genome. So that's what on the horizon over the next several years.
Fantastic, Karl. And I think this matches really well with our own experience that Pierian were, even within a small subset of partners, were seeing over 30 unique assays, many of them customized.
Rakesh Nagarajan: And so I think the custom in silico PT is really going to be of value to organizations like us and labs that participate in clinical NGS. With that, I wanna now hand the conversation over to Julia to manage the questions that were posed by audience members during the webinar. Julia?
Julia Karow: Thank you, Rakesh. So, just before we go into the Q and A, we would just like to ask our attendees to take a moment after the webinar has ended to take our exit survey to provide us with your feedback.
So now let's start with the general Q and A. So here's a question for Allie. How does tumor heterogeneity complicate the use of liquid biopsies for diagnostics?
Allie Grossman: I think that's an excellent question and that's a very difficult question to answer. Tumor heterogeneity will always be a thorn in our side no matter what kind of assay we're doing, whether we're testing directly or indirectly.
The hope is that what is circulating in the blood would be more representative of the whole burden of the tumor that a patient may be harboring.
There certainly is data out there of high but limited concordant of mutation between primary tumor and metastasis and different metastases.
I think we're gonna still have a problem of tumor heterogeneity with liquid biopsies. Again, you could have some sites of disease, brain metastasis for example, if a patient has a lung, liver, and brain metastasis, what we see in the blood may only represent what's in the liver and the lung because of blood-brain barrier issues and clonal heterogeneity of metastasis.
I think the data is still emerging on that and we need a lot more studies. But I anticipate tumor heterogeneity's always going to be a challenge.
Julia Karow: Alright, thank you. And other question for Allie. What is the difference in terms of clinical utility between liquid biopsy and solid tumor testing for identifying germ line mutations?
Allie Grossman: For identifying, I'm assuming that the question pertains to germ line variants that may be pathogenic that may be identified in tumor normal paired sequencing, I'm guessing. [crosstalk 00:58:57] Yeah, there's probably many ways to interpret this question.
So, again, I honestly don't think there's enough data on that yet. In cell-free DNA, you can certainly see different fractions. So cell-free DNA from normal, healthy cells, the data is beginning to show that those are a little bit larger fragments, maybe 160 base pairs, whereas tumor DNA may be closer to 140. But you'd have to have a way of distinguishing germ line from, maybe that's gonna be size, maybe that's gonna be [inaudible 00:59:43] patterns, but you would have to have a way of distinguishing germ line without directly interrogating the buffy coat.
You could do that with a blood sample, you can split out plasma from a buffy coat and pair them that way, and then that way you'd have a lot more confidence about what you're sequencing. But the ability to distinguish cell-free DNA from tumor from healthy, normal cells that are also populating cell-free DNA is still emerging.
We have this issue with directly testing the tumor as well. The only way to be truly confident about a germ line variant that you might be suspicious of when you've tested the tumor alone is to actually go back to either at simultaneously or retrospectively go back after the fact and evaluate the patient's germ line with a separate sample that's non-tumor. I hope that answers the question.
Julia Karow: Yeah. Thank you, Allie. Unfortunately this is actually all the time we have today for questions. So, with that, I'd like to thank our guests Allie Grossmann, Karl Voelkerding, and Rakesh Nagarajan. And also our sponsor PierianDx.