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.
Today we'll focus on two of the services that we provide, mainly validation and interpretation services. In the validation service component of the presentation, we'll talk about how we facilitate the clinical validation of a NGS diagnostic assay within laboratories that are either CLIA certified or CAP accredited. We'll go through our experimental approach, show you some examples as well as talk about our experience.
Within the interpretation services segment we'll talk about the role of technology, mainly the CGW and our appearing the ex knowledge base in conjunction with human expertise, i.e. the role of varying scientists and medical directors who post process draft reports that are created by the informatics that's available in CGW. We'll walk through sematic and constitutional workflows and how the interpretive services team performs its functions there. And I'll also talk about our experiences.
So I'd like to kind of kick off the presentation talking about PierianDX and our operating principles. That is that we operate under the once faced premise that is we would like to be a comprehensive clinical lab enabler in the molecular diagnostic space. And we do that in a very comprehensive fashion with our informatics services as well as wrap around services that's shown on this next slide.
So what the one space really means is that we have an end to end integrated workflow to support molecular diagnostic testing, specifically NGS today, although other diagnostic tests will come online in the future. And that workflow really starts by ordering a test and ends with the final report that can then be put into a client's laboratory information system or medical record. The heavy lifting of that workflow is done by our software as a service to CGW that's shown squarely in the middle there. And in order to support laboratories then in a variety of other ways, in supporting precision medicine type diagnostic tests, we've created three additional services over time.
NGS Test Validation Services
First, validation services, which we'll talk about today, enables laboratories and facilitates their CLIA certification or CAP accreditation process and we'll talk through that service first. The other service that we'll talk about are our interpretation services where we provide human expertise and capability using our own tooling and technology in order to create a near final report that may be signed out by a laboratory's medical director or can be signed out by our own medical directors. Finally, although we won't talk about gateway lab services, gateway lab services really provide a very flexible approach to bringing testing in house through reference laboratory sequencing of samples across a variety of assays that are offered through gateway. And that tech only sequencing then can be processed by our informatics system and then signed out by your laboratory or through interpretive services we can provide additional support there.
With validation services there are a variety of customer benefits that are realized. The first is that we've got a very experienced team with significant molecular biology and methodological background in reviewing and analyzing tests that are planned as NGS clinical diagnostic assays. Second, we also have informatics expertise within our organization to help facilitate the CLIA certification and CAP accreditation process as you go through validation of your asset. This expertise is also then leveraged by the mass of peer [inaudible 00:09:00] deployed panels and assays that we've got within our network already today. And this ranges from small somatic cancer assays that are either amplicon based or hybrid capture based, to large even exon based assays.
And finally, with our sharing model, these validation services are really built on best of breed customer processes and customer workflows. Anything from what types of samples will be accepted as part of an assay to potentially informatics pipelines that have already been leveraged in prior settings. What's really deliberate as part of this service is clinical and computational validation of one or more assays that you plan to run within your laboratory, including table, figures, graphs, and documentation that will facilitate your CLIA certification and CAP accreditation and inspection process.
So the first thing we do is we design an experimental approach, and I'll go through the details of what that means in the following slides. We then perform the analytical validation phase or validation which is methods based. And this has been defined [inaudible 00:10:25] in the CAP checklist for next generation sequencing, which is within the molecular pathology checklist as well as in the recent publication that came out from AMP.
The second phase of validation then is diagnostic validation where assessment of clinical samples that are for the clinical indications where this assay will be used will be reprocessed within this new assay and this diagnostic specificity and sensitivity will then be calculated based on prior samples.
o the first thing we would determine are assay requirements, so what sample types will be accepted as input, what variant types will need to be detected, is this assay limited to single nucleotide variance or SNVs and small indel or will copy number alterations or variants or structural variants like fusions needs to be captured. Things of that nature will really dictate how we'll design the experiments that will support an adequate validation. Other logistical information such as how many samples will be multiplexed in a run, what instruments or instruments do you plan on using, what reagent kits instrument configurations are you planning to use are all considered in designing the experiments that'll support analytical diagnostic validation. And finally, what's the desired lower limit of detection? And this become especially important in the somatic cancer setting.
Once assay requirements are determined, then samples are selected for both analytical and diagnostic validation and then runs are designed where we optimize the run configuration to meet all of the analytical and diagnostic validation needs and those data then are processed as will be described in the following slides.
The thing ... The analytical validation phase, again, the methods based validation approach is used where gold standard cell lines with known true positive variance, such as the Genome in a Bottle NA12878, horizon diagnostics or a acrometric cell lines, [inaudible 00:13:05] cell lines, cosmic cell lines, they'll all be used either alone or in mixing combinations. And those mixes can be across cell lines or dilutions using the background genome that the cell line was based on.
Finally, in silico mutagenesis and approaches can be used in order to create additional data sets that attack the different sequence contexts, variant types, allele burden, things of that nature so that you've got a very broad set of data that you can use to essentially tune and optimize the informatics pipeline. Within the informatics pipeline, a tuning phase, multiple pipelines may be tested, multiple variant colors may be tested, either in isolation or in combination. And these are really used in this tuning phase to optimize both sensitivity or positive predictive agreement and positive-
Both sensitivity or positive predictive agreement and positive predictive value. In this phase, we also start to establish the lower limit of detection, although that's really also completed during the diagnostic validation phase.
Analytical Validation Phase
Within the analytical validation phase, a few other components are also calculated, mainly repeatability and reproducibility. So how robust is this assay across instruments, days, technicians, reagent lofts, different multiplex barcodes for the same samples across library preps, and over time? These metrics, most importantly reproducibility then, are typically then followed through in the postproduction phase, where samples are used within production runs that can be used as both measures of reproducibility as well as to evaluate quality metrics over time.
Now, within the other components here, reportable range really looks at [inaudible 00:15:20] laboratories that will decide to report all variant classifications or will report only clinically relevant variants, and based on that, the reportable ... I'm sorry ... the reference range is also described within the appendix of reports.
Diagnostic Validation Phase
Within the diagnostic validation phase, samples are selected to ensure broad coverage of genes, variants, variant types, sequence contexts, and variant allele frequency. And again, previously reported samples using a previously validated assay are typically used. Or alternatively, if there aren't enough samples that have been processed in a previously validated assay, an orthogonal assay that's also been clinically validated may be used.
Now in some scenarios where either the gene context or the variant type context or both, there are limited to no samples, in those scenarios, orthogonal validation using a previously validated assay may continue to exist in the postproduction phase until an adequate number of samples are processed for that scenario.
I want to show you some examples, and these are typically graphs. But there's associated documentation that involves a write-up of the analytical and diagnostic validation that was performed. But these are examples that show the SNV analytical performance for a range of variant allele frequencies and how that varies as a function of depth. And as would be expected at lower allele frequencies and at lower depths, sensitivity or positive predictive agreement tends to drop. Reproducibility is also shown as an example over time, and this is in the preproduction phase. These types of plots would continue in the postproduction phase as well.
These are examples of repeatability and reportable range. So examples of repeatability across runs and within runs with the same sample as well as the reportable range across four different genes and their exons, and you can see the cutoff there in this constitutional assay. Certain exons, for example, with EGFR in Exon 1 where that exon does not have adequate coverage, it'll be disclaimed.
Our experience in validation services over the past three years has been quite extensive. Approximately one in three clients receive validation services. We've performed greater than 25 assay validations to date. This ranges from, again, amplicon-based assays, hybrid-capture-based assays, which have completely different considerations. We've validated both custom-designed, i.e., client-designed assays as well as assays we've designed ourselves. Vendor-designed assays, so things like TruSight Tumor, TruSight Myeloid, the Ion AmpliSeq, the [inaudible 00:19:32], are all things that we've validated before.
Of course, we've validated across different platforms. The MiSeq, the NextSeq, the HiSeq, the Ion Torrent, the Proton, and we validated a variety of variant types as you can imagine: SNVs, indels, and structural variants.
Switching gears to our interpretation services, then, this slide really talks about again customer benefits and the services that we perform within this service protocol at PierianDx. So there are multiple customer benefits to interpretation services. One is really an issue of scale. As laboratories start to perform a higher volume of cases, they frequently find that the medical review and the variant scientist review component tends to be rate-limiting.
They can't hire people fast enough, and therefore, turnaround time suffers as a result. And our interpretive services have really been leveraged there. The very successful extent in the short time period after which it's been launched, and it's only been active for about six months as you'll hear about at the end.
Our team has years of medical experience and clinical experience in reviewing and possibly signing out reports if sign-out is a requirement. Our team also evaluates both in the cancer and constitutional context not only the clinical context of the patient but also combinations of variants that would need to be evaluated at a case level in order to generate a comprehensive report.
The service is deliver, as I talked about, with a team of variant scientists as well as medical directors who leverage the CGW in order to prepare near-final reports or final signed-out reports. And as I mentioned before, these services are very broad-based. Indeed, we support both somatic and inherited conditions, and our turnaround time is very relevant for clinical needs, namely, 48 hours with the vast majority of reports returned back in less than 24 hours.
I really want to talk about how technology is merged with human expertise in order to provide interpretation services. So we believe this combination of technology and human expertise is what will allow us to scale to respond to the volumes that customers currently do and will start doing in a variety of areas.
Clinical Genomics Worksapce
On the left side, the clinical genomics workspace, or CGW, has classification rules based on available knowledge bases that allow for automated classification of variants. In addition, a highly curated information from FDA drug labels and NCCN guidelines, ASCO guidelines, clinical trials, risk guidelines, things of that nature can be used and are indeed used within the CGW to provide a variety of therapeutic, prognostic, and diagnostic annotation data that can be used by variant scientists and medical directors in finalizing a report.
On a constitutional side, the CGW has incorporated the ACMG-based evidence codes using our knowledge base again, so there's automated evaluation of a subset of these evidence codes that facilitate variant classification per the ACMG guidelines.
And then finally, the CGW has coded rules across active and recruiting clinical trials that allows trials to be matched based on molecular findings that are present in a patient case.
Using that information as the starting point then, the PierianDx interpretation services takes over to look at existing variant classifications and interpretations that are available for a variant. This includes our extensive knowledge base of interpretations and classifications where that information is shared within our PierianDx network of over 45 laboratories today. That information then can automatically help the scientists and the medical directors to understand how this variant has been interpreted in prior cases in the same or different disease indication as the current case.
Using information then from prior clinical interpretations as well as other knowledge sources, clinical interpretations can be redrafted based on the newest available information or reclassified based on newly available information. In addition, pathway- and domain-related information can be evaluated to create more cohesive interpretations across sets of variants. Trial-matching can be made more precise based on client-specific SOPs. And variant classification can be finalized in the ACMG guidelines for constitutional reporting by processing all evidence codes using our system.
As we then walk through those review processes in detail for somatic and constitutional reporting, these concepts will then become clear. So with a CGW-generated draft report, one of the first activities that variant scientists and medical directors perform based on our interpretive services that we offer is, is the variant real?
There are a variety of quality metrics and interfaces within CGW that are used to determine whether the variant indeed is a real variant or represents a false positive based on quality information and noise that's present within that sample. Once it's determined that variant indeed exists and the determination has been made or the judgment has been made, the interpretation services team then evaluates whether an interpretation already exists within the PierianDx network.
If that interpretation already exists, then that interpretation is evaluated to determine whether a change or an updated is needed based on newest available information. And we'll talk about where the information comes from in a minute. If an interpretation does not exist, then one of the first things we look at is whether this is a real private variant using population frequency databases, and all the current databases that are commonly used are available in the CGW to do that.
We establish then the clinical relevance of a variant based on a variety of information in the somatic cancer setting based on human research evidence, databases such as COSMIC and TCGA, guideline-based data, rules that are available already within the CGW based on NCCN and ASCO as well as FDA drug labels based on literature searches that then allow us to understand whether there's compelling clinical evidence to report out on a variant but where practice guidelines have not yet been established. Molecular pathogenicity is determined based on both protein domain information and pathway information. And then silico prediction tools are considered in the final establishment of clinical relevance.
Once each variant's clinical relevance is established and interpretations are determined, the team then determines whether there is a cross-talk, a cross-genes, a cross-variance in order to then create a cohesive interpretation at a ...
In order to then create a cohesive interpretation at a case level. And that cohesive interpretation includes reviewing the patient's clinical context, prior treatments, age, race, gender, diagnosis, things of that nature, in the Sinitta cancer setting. Again, clinical trials are then suggested based on the client's specific SOP, and using that information then, the final classification and interpretations are drafted into the report.
The constitutional case review process follows a very similar methodology, with respect to determining whether a variant is real, whether an interpretation or classification exists in the PDX network on, and whether a variant is rare. In order to establish clinical relevance or a variant, the ACMT guidelines, and the evidence codes are actually processed, and [Googles 00:29:01] leverage information from prior clinical variant databases such as ClinBar, local specific databases, things of that nature. In the hereditary cancer setting, risk guidelines are evaluated using NCCN and ASCO curated information that's within our knowledge base already. Again, knowledge base, or published literature searches were used to establish other evidence codes. And a recording of that classification, and a recording of why the evidence code was coded that way, is also noted within the CGW. And again, in Silico prediction tools, as well as the patient's context, including inheritance pattern, family history's considered, in order to fill out and complete the ACNG evidence codes and to calculate the final classification in CGW.
Again, where appropriate, a crosstalk of cross-genes and variants is considered within this setting as well. And the final classification and interpretation has been updated in the report and potentially signed out if we are providing that level of service for a particular client.
This service is a very nascent service. It was launched only in January of 2017. Again, based on customer needs. We have enjoyed great success in this service already. And we currently support both somatic and constitutional assays for a wide range of customers, both academically affiliated organizations and institutions, commercial laboratories as well. And as I've mentioned before, these services are very rapid, just-in-kind, and deliver very high-quality results in clinically relevant turnaround times. Again, 48 hours, or less than 24, in the vast majority of cases.
Case Study: Moffitt Cancer Center
I'd like to end by giving a case example of how these and other services that are offered by PerianDx have been utilized by one of our clients. So Moffitt Cancer Center has, in fact, leveraged all four of these services, and that's really shown in this timeline and detailed in this slide. So Moffitt was actually our very first customer, and became our very first customer in May of 2014, where they started by using the CGW as a software-as-a-service. Now the value propositions there were multiple. There was HL7 integration to receive orders from their LIS to sending the signed-out report back to the LIS, and all the way up to the EMR. To really facilitate that clinical NGS workflow through access to the knowledge base that's available in CGW.
Very quickly, as they launched, there was a need for validation services. And so the validation service option was accepted by Moffitt for two assays, the Trusight tumor and Truseq myeloid. And we launched both of those assays within a 12-week period in parallel, by providing the validation service capability to the Moffitt laboratory. Moffitt then started processing samples in October for a period of time in late 2015, leveraged our gateway services in order to meet an expanded volume need that they had for the heme assay. And then in early 2017, Moffitt then started using our interpretive services again in demand to the growing clinical volume within their own organization.
And as you see, Moffitt has also evolved their Assay portfolio to use newer versions of myeloid assays, as well as to incorporate additional modalities such as detection of fusions through a alumina-fusion assay.
Now, Moffitt has also in the process of adopting and validating the Trusight Tumor 170 Assay, and they will be one of our first customers that will validate that assay, but there're others that are very rapidly starting to adopt that assay.