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, and welcome to today's webinar, organized by the Pathologist and PierianDx. My name is Michael Schubert, editor of The Pathologist, and I'm very pleased to be the moderator of today's online event, which will explore comprehensive genomic profiling in the laboratory. Please allow me to introduce today's speaker, Dr. Rakesh Nagarajan. Dr. Nagarajan is a founder, and executive chairman of the board at PierianDx. Trained as a physician/scientist, he has deep expertise in molecular biology and it's laboratory work flows and techniques. He serves on the College of American Pathologists molecular oncology committee, their Next Generation sequencing project team, and as Molecular Pathology's specialty inspector. He's also an adjunct professor of pathology and immunology at the Washington University School of Medicine.
Today, Dr. Nagarajan will discuss the practicalities of implementing comprehensive genomic profiling, including validation, implementing and reporting, helping to reduce validation costs, and maximize sensitivity and specificity. He'll be on hand to answer any questions that you may have at the end of the presentation. But please remember that you can submit questions at any time throughout the webinar. You will also see some polling questions appear on your screen. Simply answer those questions on your screen to clear them, and the presentation will continue. So, without further ado, let me get today's webinar underway by passing you on to our presenter. Dr. Nagarajan, over to you.
Great, thank you so much Michael and thanks to the pathologists for moderating this webinar today. As Michael mentioned, I'm the Executive Chairman at PierianDx, I also founded the company back in May of 2014. Today's webinar will focus on the following topics, so first we'll start with comprehensive genomic profiling, what the value is, what the trends are, what the adoption and the adoption rate are. We'll describe this concept of clinical management yield ie; how often is there clinically actionable alteration and associated clinical management action found wen CGP is performed. We'll talk about evidence based reporting in [inaudible 00:02:33], a variance that are resultant from a CGP, and finally we'll describe the products that we have that support LDTs or laboratory derived tests for research use when we [inaudible 00:02:51].
With that let's go ahead and jump in and talk about trends value and adoption. So, precision medicine has had a very rapid rise over the past decade plus. This includes hundreds of targeted therapies and associated needle therapies, biomarkers that are associated to those medications. Hundreds of laboratories are adopting comprehensive genomic profiling as evidenced by billing for CGP, however the majority of volume is going to appear in independent reference laboratories. That trend is changing as clinical genomics is becoming more democratized across laboratories both within the United States and globally. As I indicated, biomarkers themselves are evolving and changing, from simple single nucleotide variance to small indels to much larger structural variations such as [inaudible 00:03:59] variance in gene fusions, to tumor agnostic biomarkers like NTRK, micro satellite instability in tumor mutation burden. In addition, payor coverage is becoming much more standardized with the recent NACHA coverage determination back in 2018 of FDA approved NGS tests that are supportive of CGP.
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.
Indeed, oncology testing itself is evolving and has been evolving over the last the last decade or so, from testing single markers and small hotspot panels to much more broad NGS testing or comprehensive genomic profiling. Both the modalities are used in association with NGS to provide the best clinical management of the patient. Intramatic systems are being developed that integrate genomic and phenotypic data in order to provide the most accurate clinical management outcomes to the patient.
Now this slide really summarizes the concept that serial serial testing in cancer is vital in order to make cancer and chronic disease in this model even before there is frame cancer in a patient. Hereditary risk of cancer should be tested and monitored in order to actively manage the patient once there is a diagnosis, however of cancer. We expect over time that primary cancers are tested using CGP and that every relapse and recurrence event as well as following post treatment in a monitoring setting, and testing is performed as the disease may progress I order to tailor therapies at each occurrence and each resistance event, that we may get [inaudible 00:06:03].
Another issue with cancer itself and testing is the amount of tissue available. And, with small hotspot panels or single marker type assays that are only capable of detecting one or more biomarkers, you frequently don't have enough tissue to test all the relevant necessary biomarkers. And again, a single comprehensive test such as that offered by CGP would alleviate the need to perform single tests either serially, which would waste time, and also is not possible if there's not adequate tissue available.
Now CGP can also identify actionable alterations, and a greater number of actionable alterations than targeted assays or smaller panels. Four different studies are shown here for a range of pediatric tumors to adult tumors in advanced settings had actionable alterations somewhere between at least a third of the cases to almost 90% of the cases, and the variation there is typically dependent upon the level of CGP that's performed in the four tests, as well as the definition of actionability. Actionability can be broadly defined as matching the trials and using that as an avenue for patient management.
Now CGP is also very cost effective and has clinical utility in that there are reduced costs than non-targeted therapy use. When CGP is performed, there is an increase in the use of appropriate therapy or targeted therapy. CGP enables greater clinical trial eligibility in matching, and reduces adverse events through profiling and using drugs that are likely to work and reduce adverse effects.
In this next section, we'll talk about this concept of clinical management yield through a variety of analyses and metrics that we've performed on prior clinical patients that have been brought in our system.
I wanted to start by describing the CAP/AMP/ASCO variant classification which is becoming a standard for somatic variant classification. Namely, there are four tiers. Tiers three and four are variants of unknown clinical significance, so the lesses and benign variances respectively, while tiers one and two are variants of strong clinical significance and potential clinical significance, which sub tier based on variance or biomarkers that map correctly [inaudible 00:09:27] You have the approved therapies, are professional guidelines to those lesser tiers that are based on the size of studies and the level of consensus that those studies have found before guard the therapeutic prognostic or diagnostic indications for biomarker as related to a tumor type. A major challenge here is, how do you interpret a very large comprehensive genomic profiling assay and maintain the highest level of accuracy?
Within systems that would facilitate this type of activity, typically of you start at the very top of puddle and look at the cancer genome, and limit that to relevant genes that should be tested clinically with regard to association to targeted therapies as well as immuno-therapies, and to the biomarkers that we saw previously, [inaudible 00:10:34] 500 content that's optimized to focus on those clinically relevant biomarkers. Within our system then, that assay when it is utilized to profile an advanced solid tumor, the sample and run through secondary analysis that has been validated, a validated filter that meets the assay's performance capabilities and specifications result in a smaller number of variance that meet obsessive [inaudible 00:11:13]sensitivity cut-offs for that validation and identification of the different variant types that may be important clinically. The system then also enables automatic matching, as I'll describe in a few minutes, to relevant clinical information things like drug labels and association to guidelines such as FDA labels and GMA labels. Professional guidelines such as the NCCN and ESMO, as well as matching to clinical trials in prior medical interpretations that have been written by medical directors using our system. That information can then be used in the context of the current case to finalize each variant's classification, modify it's appropriate mass interpretation in the context of the current patient in order to [inaudible 00:12:08].
The PerianDx knowledge base itself as I alluded to in the previous slide, is composed of a number of sources starting from genome models and genomic sequence in the genome, builds 37 and 38 as well as TM and RNA encrypting models[inaudible 00:12:35] Databases that enable the interrogation of minor allele frequencies [inaudible 00:12:44]populations, as well as the frequency of variance and cancer research specimens from COSMIC and TCGA, and evidence from other clinical databases such as ClinVar. In addition, there is curated content available within the system from drug labels from the FDA and EMA. Guidelines from NCCN, ASCO and ESMO as well as active recruiting trials. Finally, there is shared medical content, and this medical content includes signed-out clinical interpretations from prior cases that automatically infer when a variant in the current patient's case matches a prior case, whether the same tumor type or not, as well as be identified an aggregated variant frequencies associated to each tumor type so that when you are [inaudible 00:13:50]your case can see how often a variant can be found in prior cases in your patient's tumor type as well as other tumor types. This knowledge base is 100% clinical focused, as rules based as I will describe and includes [inaudible 00:14:13].
The knowledge base is really executed through a rules based engine and that rules based engine [inaudible 00:14:25]a very complex genomic predicate model wherein rules can be written and fire based on variant syntax. So, following the HGVS nomenclature as well as rules that can be written on genomic coding protein coordinates ie; things like exon 19 or codon 12 variants KRAS including functional characteristics such as deleterious or likely deleterious mutations FRCA for example, and these rules can not only be written on single nucleotide variance and indels, but also gene fusions and haplo variations. Finally, the rules themselves can be written such that two variants or tow or more variants need to co-occur ina particular patient case in order for the inferences to match. And, that includes rules that can be written within a gene or across genes. So, things like EGFR and KRAS variants may be found at the same time in lung cancer for example. This rules based engine is therefore much more powerful than single variant look-up knowledge bases and have the ability to have much greater coverage of the clinically actionable G stage that TSO 500 assay covers.
In the next few slides, I go a lot deeper into that rules engine and predicate and inferencing model, wherein I describe what is called an annotation rule, and that annotation rule associates a variant to minor allele frequency databases like NOMAD cancer research databases, like [inaudible 00:16:25]GPCA, like computational predictions from things like [inaudible 00:16:30] and clinical variant databases like FLNBAR. In addition, curated rules are rules that follow that genomic fact model that I described in the previous slide, and those are then used to write rules on drug labels guidelines, trial registries, to publish literature, as well as prior cases. And the types of inferences that we store, for example are therapeutic prognostic and diagnostic inferences that the drug would be responsive or non-responsive if there is a therapeutic inference. For example, drug labels and guidelines as well as study size outcomes for the published literature in phase and location for clinical trial risks.
What happens in the actual curation of interpretations that are written by our curation team is that a variance or biomarkers are processed through our rules engine, and then automatically associated to annotation and curation rules on both variance and biomarkers are then assessed individually such that curated annotated rules are ranked. The assertions are verified by our dedicated curators, as well as through our interpretation services team that performs this type of service that I'll describe to you at the end of this presentation in the context of real patient cases in order to create a final classification interpretation that may be used by individual medical directors as they review and finalize a report themselves.
In this slide we really summarize the experience that PerianDx knowledge base has had on the genes that are reported out as part of a TrueSight Oncology 500. And what we show here is that almost a quarter of the genes have already been reported out in over 10,000 cases in our product clinical genome workspace. And therefore there is a rich set of clinical interpretations that are already available and have been written by medical directors processing part cases. Over a quarter of those genes have been reported in 5000 cases, over half in 2500 cases, almost 90% have rated 1000 cases, and all 100% of the genes have been reported out, at lest 350 cases. So, all in all great coverage of the genes that are reported out as part of TrueSight Oncology 500 and great prior experience that medical directors can leverage as they sign out their patient cases based on experience of prior medical directors.
In the next few slides, I'll talk about the knowledge based performance and the benchmarking that we've done. I'm on that knowledge base, based on prior cases.
In this first slide, what we describe is a view of about 1000 cases, four different sights where TrueSight Tumor 170 test for [inaudible 00:20:14] was validated and used in real clinical cases. And about three quarters of a million variants across those cases were then assed for specificity, sensitivity, PPV or positive predictive value, and accuracy. And that was defined as, how often did the knowledge base assign the correct classification to a variant automatically. We have the ability to track this information wherein the classification assigned to the variant by the knowledge base versus the classification of the variant as it is signed out was compared, What we find is that there's incredibly high specificity, accuracy, and positive predictive value in and excellent sensitivity and typically the false negatives and false positives in these cases are a combination of a director disagreement and some variability in reporting across medical directors which itself can be assessed in the system, as each medical director reviews and signs out the case himself of herself, as well as potentially errors directors are making in assessing the the clinical relevancy of the variant.
In this slide, we then describe what the clinical management field is using our knowledge base. And the analysis here was performed on the same set of cases that I described previously. And what we look for here is how often across those cases was at least one clinically relevant variant found, if we only utilize the FDA drug labels. If we then added professional guidelines, and finally if we then added the full medical interpretation knowledge base into the system. We also assessed how often are clinically actionable variants found in the patient's case if we limit to the patient's tumor type versus also include other tumor types. And what we find is that over a quarter of the cases require a fully available medical interpretations in order to achieve about a 95% clinical management yield or clinically actionable variant being found in over 95% of the cases when all sources are used. And what this demonstrates is that reviewing FDA approved labels and guidelines alone would result in about 2/3 of the cases having actionable findings but the other about a quarter of cases would not have actionable findings, but that's a medical interpretation ie; [inaudible 00:23:13]studies that have not yet made it into guidelines indeed have actionable findings that may be used, and that the information is available within our system [inaudible 00:23:25]medical directors.
In this section we'll talk a lot about evidence based reporting and the structure of the report that we provide with the TrueSight Oncology 500 assay. In a very illegible view of the report and it's summary is shown here. Namely, a very clinically important an critical front page biomarkers and biomarker types and numbers of biomarkers as well as a summary of variants that meet the to two tiers for the [inaudible 00:24:09]classification. A middle section of pages goes into much more detail and provides in depth interpretive content, these medical interpretations that I described and then sections that provide access to clinical trials.
Here what we've done is really zoomed in on that monthly reporting front page wherein we identify the number of biomarkers with across the different piers in a prospective CAP/AMP/ASCO classifications reporting out on biomarkers such as TMB and MSI, as well as the number of clinical trials that matched to one of the biomarkers that matched in this patient. The lower section then identifies those biomarkers of variance as well as their tiering and sub-tiering within the CAP/AMP/ASCO, as well as that biomarker's association to therapeutic and prognostic and or diagnostic information. And when there's therapeutic information for example, the drug or drugs that may be either responsive or non-responsive are summarized there.
This page shows then the in depth interpretations that are available where the variant is made according to the KRAS nomenclature including the p dot and c dot notation when available. This specific transcript, the variant allele fraction depth as are indicated through the College of American Pathologists checklist and recommendations made there, as well as the interpretation itself that provides then a very structured information in regard to gene level information to describing the general function and the biological function of the gene, the variant level information, it's prevalence in the different tumor types, the supporting evidence that was summarized in the front page. But now the textual format as well as a mixed and appropriate references and then finally therapeutic guideline information as appropriate as well as prognostic information if available.
In this third zoomed in section then shows the remainder of the report where matched clinical trials and their titles and link outs are shown, as well as variants at lower classification schools including blue exons are part of the report. There is a classification in evidence legend to really remind everybody the CAP/AMP/ASCO classification scheme and it's structure. And then finally, individualized methodology and disclaimer sections for each laboratory as it does it's own validation and has it's own specifics on methodology and legal and other organizational disclaimers that likewise are in [inaudible 00:27:44].
In this last section then, I summarize PerianDx and it's support of assays that are RUO based but are validated locally at individual laboratories using the LBT paradigm. What we do there is essentially with our partnership with Illumina wherein Illumina provides the sequencer, the assay [inaudible 00:28:23]TrueSight Oncology 500, sequencing the agents as well as protocols to process that tissue sample frequently FFPE through DNA or RNA isolation, a library prep an enrichment of sequencing and secondary analysis that runs seamlessly in all our environment or in your local environment which then is automatically connected to our cloud enabled clinical management workspace where that sequence variant as well as the BAM file are processed in order to generate QC metrics reports. And the clinically interpreted draft report that I showed in previous slides can then be reviewed and finalized within our secured web based interface in order to sign out a final report.
The CBW itself has a number of features that are critical to a high quality laboratory experience when processing, analyzing, reviewing and signing out clinical NGS tests. This includes integration with local EMR and LIS systems using HL7 as well as a rest state API that can be used to integrate these systems. The discrete API for example, enables extraction of a variance and associated information bearing discrete [inaudible 00:30:07]that can be pushed to EMRs capable of receiving that information and/or analytics and data warehouse systems for appropriate clinical quality and research purposes.
Respectively, we operate in a very secure environment using BPA tunnels as appropriate and are fully HIPPA compliant for high trust certification GDPR compliant for EU customers. I describe the clinical sharing network in great detail, but in summary shared medical interpretations that are available where you can sign out your own cases and be identified in aggregated patient variant data so that we can see how often a variant has been found in prior cases in your patients tumor type or other types. We also had a very flexible approach in order to utilize our system for example with the TrueSight 500 assay. The customer can perform secondary analysis at their local sight by sending us variant files or variant files with the BAM file [inaudible 00:31:25]metrics report, or they could start with PCL [inaudible 00:31:28] files directly and allow us to do secondary analysis in our cloud environment. We have a number of quality assurance and quality control capabilities within the system including a QC metrics report for every case that describes run exon level as well as variant level quality metrics that you can look at very quickly to determine if you need to re-run a case back from the different tissue block, re-do the library [inaudible 00:32:06] of that nature. As well as on demand reports within the system that look at quality assurance and quality control components, across a range of of reports over a period of time.
Finally, our clinical reporting capability, as I described previously [inaudible 00:32:28]report, but the clinical reporting capabilities within the system quite flexibly support post somatic cancer and germ line tests [inaudible 00:32:37]only focus on somatic cancer testing specifically with TrueSight 500 here, but the system is a single system to support clinical NGS tests whether that be a somatic germ line, a small targeted assay [inaudible 00:32:54].
Our system really is the optimum mix technology and human expertise in that our software and our knowledge base are executed through computational automation and a rules engine as I described. However, the content is developed and curated through human expertise and dedicated human curators at PerianDx. In addition, as I described previously, provide what we call interpretation services such that dedicated variant scientists as well as potentially medical directors can review cases of customers in order to review the information that the CTW has automatically matched, finalize that interpretation, an/or matching the clinical trials and provide that information back to our customer, who then may sign out their report. And that interpretation services team performs that process professional review and sign out as described by the CAP/AMP/ASCO manuscript.
In addition, we provide support to validate assays like the TSO 500 and here the validations services are comprised of a draft validation plan, which includes the number and types of samples that we would need to process including run configurations that are optimal in order to validate the assay as quickly as possible, includes steps to familiarize yourself with the assay and optimize the [inaudible 00:35:01]pipeline for that assay and frequently this is the [inaudible 00:35:05] up front, were actually like the TSO 500, we've productized that assay. It includes computational and scientific support to analyze the data coming off of samples that you've processed within your laboratory. As part of the validation, including table, figures and graphs that we review with scientific laboratory medical directors at your sight, it includes components of analytical specificity, sensitivity, reproducibility, reliability pursuant to accuracy, things of that nature. There are excessive QC calculations for reference as well as exon and hotspot coverage. Frequently dropped out exons that you may want to disclaim as part of the test that you are running. Support for sample procurement is applicable [inaudible 00:36:07] design with their report requisition. Consent forms if applicable, and finally support to create the right regulatory documentation as applicable in different regions including CAP and CLEO.
With that I really want to summarize and say that comprehensive genomic profiling is of incredibly high value, both to patients and to laboratories that are democratizing clinical genomics, as well as practicing precision medicine at their own organizations. The TrueSight Oncology 500 assay is a hallmark that in a standard at the Forbes conference on genomic profiling [inaudible 00:37:00]variants and the different types of variants that I described as well as tumor agnostic biomarkers like TMB and NSI, have excellent content and performance. We didn't really talk much about the analytical validation that's been performed. I'm on the assay, but very highly validated and ready to adopt at local laboratories where high validation can be performed and individualized [inaudible 00:37:26] can be performed.
I've also described how the combination of Illumina's sequencers, sequencing agents help TSO 500, and send their analysis pipeline out with the PerianDx workflow and knowledge base enable a streamlined sample-to-answer experience, and that includes integration with LISs and EMRs at individual hospital systems. And these are very critical components in order to enable a [inaudible 00:37:57] of agnostic laboratories to scale to the volumes that are typically needed for late stage cancer testing, followed by then a primary cancer monitoring type testing that we expect to be performed over the next 12 to 24 months.
Finally, I describe PierianDx software and specialized services including validation and interpretation services that facilitate a very rapid clinical validation and deployment of the TSO 500 assay, within your molecular diagnostic laboratory. With that, I'd like to conclude the presentation and open up the webinar to Q & A, and move to that session. I'll hand that part over now to Michael. I'm happy to answer any questions that may come up.
Thank you so much, Dr. Nagarajan for that exciting glimpse into a rapidly growing field, and for sharing your experiences. So, we've received a number of questions from audience members and we'll try to get through as many as we can in the time remaining. But, please be assured that any questions that are not answered live today will be answered by email. You'll also see some additional polling questions appear on your screen, so just like before simply answer them and clear them. So, the first question from our audience is, what kind of turn around time do you expect for one of these assays?
Great question. As we all know, turn around time is calculated in many different ways. The total turn around time to the oncologist is typically from the time they order the test to the time they get the report. The turn around time for the laboratory itself is typically from the time they receive the sample to when they sign off the report. And typically those turn around times are based on individual laboratory sites having access to specimens and being able to process them. Typically with a hybrid capture like assay, like the TSO 500, laboratories are able to achieves turn around time of 3 weeks or so total from order of tests to sign out. Individual hospital systems, the infomatics turn around time from the time when the sequencer runs complete to the draft report is generated a very small fraction of that entire 3 week cycle. Typically an over-night processing from BCO or FASQ all the way back to report.
Great, thank you. So the NHS has recently issued a new test directory that specifies [inaudible 00:41:02] that need to be tested in relation to particular disease. Is it possible to adopt these panels, or to only report on a subset of genes to produce the proper content for a report?
Absolutely. I'm glad that question came up. One of the key features in our software system is this concept of what we call a sequenced gene set versus a reported gene set. So in the case of the TSO 500 assay, 523 genes are part of the sequenced gene set. However, any number of those genes may be reported in the context of specific sub-panels or individualized tests that a laboratory may choose to report out. Indeed our laboratories today use assays like the TSO 500 to then report out on a lung set, a melanoma set, a colon cancer set and the like.
Thank you. So then how do you recommend validating tumor mutational burns?
Yeah so, very complex question and I think it is an evolving area. However, a typical validation of TMB by other laboratories have been performed with solace. Some number of samples where the TMB, a value was actually validated to XLB sequencing as well as importantly TMB's calculations that are values that are performed on the same sample across two different clinical laboratories where the other laboratory has already validated TMB. Folks typically use external reference laboratories and samples where they have TMB values in order to do that. The final component of validating TMB is by tumor type in that tumor types may have different ranges of TMB values and it's advisable that a number of different tumor types are evaluated in individually to understand [inaudible 00:43:37].
Lovely, thank you. Do you have an opinion on the utility of CGP in early stage 1 or 2 solid tumors? The NGSNCD supports coverage for stage 3 and 4 solid tumors.
Yeah, absolutely. So, I think the MCD of support very broadly for late stage cancers is entirely appropriate today. However, early stage cancers could benefit from CGP and that benefit is not only on the clinical side but also on the economic side in that as laboratories execute CGP at scale, the testing that's performed on early stage tumors in specific tumor types also requires a multiple variant types to be sequenced and reported out on. And the costs in the current reimbursement for those individualized tests is both more expensive and more time consuming, especially when tissue is limited as opposed to CGP. So, I think a number of factors need to come together in order to drive this movement to early stage cancers, and I think that we are on the path to get there.
That makes sense, thank you. How often does PierianDx update the database for clinical trials associations, and does it interact in real time with clinicaltrials.gov?
A great question. So our updates to our knowledge base is performed on a weekly basis, inclusive of clinical trials. We don't interact in real time with clinicaltrials.gov, except to link out. So within our system when a trial does match, you can link out and read additional information at trials.gov. We do weekly processing from clinicaltrials.gov data, when then screen the data that's been identified as either trials that are newly recruiting newly added as well as trials that have stopped recruiting and update that in our knowledge base, on a weekly basis.
Okay. Can PierianDx interpret whole exon sequences or even whole genome sequence data?
Great question. So, as many of you know, whole exon sequencing or whole genome sequencing can be performed both in germ lab somatic cancer settings and I think we should talk about it in each of those settings individually.
So in the cancer setting, we really haven't seen very much clinical use of whole exon and whole genome data due to cost considerations and others. Having said that, the interpretive content that we have within the CGW as I described, work with about 1000 genes that have some clinical association. That knowledge base can be used to process variants from whole exon and whole genome data to provide clinical interpretive review. On the germ line side, whole exon and whole genome data are typically used in more complex settings such as rare undiagnosed genetic disorders and CGW has capabilities to process whole exon data both in single [inaudible 00:47:46] and in trio to support things like phenotype based prioritization of variances, scoring of variance and to interrogate different inheritance models and to filter [inaudible 00:47:58] using those inheritance models in order to identify possible causes of variance and reviewed sign out. In addition in the germ line side we provide semi-automated ACMP classification reporting of variance to facilitate a variant classification per [inaudible 00:48:19]
Great, thank you. So, you provide a platform to upload FASQ files and generate VCF. The TSO 500 on [inaudible 00:48:34] 500 generates BCL files. So, are there tools to convert BCL to FASQ?
I guess there are. So we can start from BCL files and then complete that entire workflow within our environment. We can also, as I said, start from FASQ, BAM and VCF or VCF alone.
Okay. And following on from that, how long will it take from BCL files to final report generation?
A great question. So, the great limiting step that's typically a local laboratory's internet speed and access to the wide area network and the internet itself in that BCL file it's from a typical TSO 500 run can be several hundred gigabytes. So that push time is typically as little as a few hours to 4 or 5 hours followed by then a BCL conversion and FASQ to variant secondary analysis and then finally track report generation. In general, the speed of this overnight processing I think is still possible in that run to finish near the end of a workday at a site, followed by a push of that data in let's say 4 or 5 hours followed by then secondary analysis and draft reporting can still be achieved within a total of a 12 hour cycle provided that internet speed is fast enough. And that's the type of testing that we want to do with individual laboratories to ensure that.
Okay great, thank you. So, how do you call a variant clinically relevant then?
Yeah great question. So, the definition clinically relevant is defined by CAP/AMP/ASCO is how we define clinically relevant. So, when a variant fall into tiers of 1a, 2b, 2c or 2d, we call that variant clinically relevant. With that relevant strength of evidence, and that is really what the system is doing, when it is classifying that variant automatically and that is part of the total turn around time to arrive at a draft report. When our interpretation services then processes that output, we the review the information provided by the CGW with regard to what clinical interpretations have already been written, are there available FDA approved labels and/or EMA approved labels and guidelines, and what does the published literature say about this variant in order to then classify the variant as appropriate as well as if trials are available because trials and their availability affect our classification and then finalize that interpretive content [inaudible 00:52:10].
That's great, thank you. That makes a lot of sense. So how compatible is the software and knowledge base for analyzing gene expression data? For instance, her 2 and breast cancer?
Yeah, great question. So the software is really focused on DNA and RNA based variant reporting and/or biomarker reporting like TMB/NSI. Expression level data are typically not within the scope of the CGW in it's core processing. Having said that, pre NGS based methodologies and their output can be incorporated into a single comprehensive report. We're happy to talk about how we do that in a separate interaction with customers.
Thank you for that information. Do you use the TSO 500 up on base space from Illumina for the analysis?
Yeah. So the secondary analysis, whether it's performed at local sites in base space or in our cloud environment, it's the exact same secondary analysis. Pipeline code version, things of that nature. And so the application and the processing of the data will result in the same variants being identified regardless of where that file was processed.
Thank you for that. How is the patient data handled within the EU?
A great question. So as I mentioned, we are GDPR compliant. We have a data center based in Germany where all EU cases are processed, managed and stored. We follow all of the standard GDPR principles and have capabilities within the system enabling customers to download individual patient data as appropriate, delete patient data if that's been requested by a patient and therefore comes to the laboratory within the GDPR compliant sector. We operate as processors of data wherein our laboratory [inaudible 00:54:58]controllers.
That's wonderful, thank you for that information. Okay, I'm afraid we're going to have to leave it there. But again let me reassure you that any question left unanswered today will be answered after the event by email. So, thank you once again to Rakesh Nagarajan, and thanks to everyone for joining us today. I look forward to seeing you again at future webinars. Goodbye for now.