Democratizing Next-Generation Sequencing – An interview with Rakesh Nagarajan,MD, PhD, Founder and Executive Chairman of PierianDx - Part 1 of 2
Personalized medicine is clearly one of the most popular topics in healthcare today. What is it going to take to equip physicians, laboratories and hospital systems with the capability to make sequencing and genomic medicine available to everyone? Our guest is Dr Rakesh Nagarajan of PierianDx, who's on a mission to democratize next-generation sequencing.
Joseph Anderson, MD: Rakesh, we've been hearing so much about you in the news. You've had a lot of success so far at PierianDx, in terms of building the company and raising money. Could you just tell us - especially for folks who may not know - what exactly does PierianDx do?
Rakesh Nagarajan, MD, PhD: Delighted to do so. PierianDx is a company that supports clinical NGS and molecular diagnostics. It was founded in May of 2014 - spun out of Washington University in St. Louis, where the software and services were developed in a collaboration of the departments of pathology, immunology and genetics. Our reason for being is to democratize NGS clinically - nationally and globally - for both somatic and germ line indications and to provide supportive services to enable molecular diagnostics laboratories and health systems to practice precision medicine.
JA: What in your particular background was it that led you to start the company and what did you see as challenges or unmet needs that we were facing at the time?
RN: I'm trained as a physician- scientist, with my formal training in molecular biology. My PhD is in molecular neuroscience and transcription factors. My side training, if you will, which I did on my own, was computer science. I was a self-trained hacker at heart, and I was always trying to bring computer science into medicine and research throughout my academic career. Early in my professional career, I was finally able to do so effectively.
Late in my PhD, when the first microarrays came out, I was able to develop microarray analysis software. My career then really launched as an informatician. I decided to hang up the stethoscope and the pipette and pick up the computer keyboard and mouse to support clinical and translational research.
My experience there at Wash U, over the first eight years or so, was in supporting very broad clinical and translational research programs in the cancer center and the Clinical Translational Science Award (CTSA). Within that context, the departments of pathology, immunology and genetics, when they launched their clinical NGS program, asked my group to develop the informatics for this clinical product. And for the first time, my group had the chance to develop and support software in a CLIA-CAP-accredited environment. We were delighted to do so, because we could impact patients very directly through that model.
JA: I remember when I first met you, I was very excited because you were the first informatician that I had met. I knew there was such a need for this skill-set in the space. We were on the brink of what many thought would be a large tsunami of sequencing data.
PierianDx is a very unique name. Could you tell us where this comes from?
RN: The name really comes from Greek mythology. The Pierian Springs were where the muses bathed. The muses were the source of wisdom for science, math and the arts. The Pierian Springs were also made famous by the Alexander Pope poem, "An Essay on Criticism" (1711): "A little learning is a dang'rous thing; / Drink deep or taste not the Pierian spring." The name reflects the source of knowledge, if you will, and of course, the “DX” stands for “diagnostics.”
JA: It Sounds like a lot of thought that went into that, and I can certainly appreciate that.
You said that you want to democratize sequencing and make it available locally and globally. How exactly are you doing this and who are the customers that you're serving?
RN: We're helping customers internalize next-generation sequencing for clinical use in a CAP-accredited, CLIA-certified environment, running sequencing in their own laboratories. These are molecular diagnostic laboratories in virtually every market segment. We started with academic medical centers. NCI designated cancer centers, pediatric hospitals, large health systems with precision medicine programs, commercial laboratories, reference laboratories, and regional pathology laboratories are all customers of ours. These are all folks who are closer to the patient, if you will. In the democratization of clinical NGS, we imagine hundreds if not the low-thousands of laboratories nationally, and tens of thousands of laboratories globally, offering NGS sequencing clinically, both for somatic and germ line indications, when genomics is operating at scale in medicine.
JA: I think we all agree that this is going to play a very large role in medicine. It’s probably not something you can just adopt overnight. What are some of the challenges that these groups might be facing in establishing an NGS program?
RN: Great question. I think that the problems or challenges that folks will face at the very outset may be: creating the cost justification for setting up such a program, the capital expenditures that would be needed, the IT and informatics infrastructure that's needed, the technical laboratory personnel needed, the medical professionals needed to review findings, perform variant analysis and sign out the cases. On the germ line side, genetic counselors may be needed.
There's the aspect of, “How do I set this up?” and “What is it I'm setting up?” “Am I only setting up a laboratory or do I have an entire precision medicine program that interfaces with treating physicians to help them interpret the results so that they can act on the report with their patients?”
Once you actually commit to establishing the program and internalizing the sequencing component, one of the first things you run into is, “How do you validate an NGS assay?” Some of our very first customers were learning how to do that and we ourselves at Wash U. learned how to do that very early on. We essentially invented validation for NGS in the somatic space as well as in the germ line space. That learning, everyone doesn't need to do over and over again. One of the services that PierianDx offers is an assay validation service, where we help customers in their validation approach, their strategy, the selection of samples with a variety of variant types across genes and across diagnoses, in order to calculate things like specificity, sensitivity, accuracy, reproducibility, reliability and the like.
Once an assay is validated and customers “go-live,” they may face a different set of challenges. “Post-go-live,” they might say, “Hey I've got a complicated case here,” or “I have an interesting variant here that I'm not sure what it is.” “How do I review that in the context of read quality, alignment quality, variant allele frequency and things of that nature?”
And, obviously, interpretation itself, is a very daunting thing. Molecular diagnostics, historically, has been a single biomarker or a single variant. Whereas now, even very small panels of a dozen genes or fewer than fifty genes, can give you dozens of variants after filtering and hundreds of variants before filtering - let alone panels that are getting into the hundreds of genes - several hundred genes to five or six hundred genes - those are going to give you thousands of variants that you have to sort and filter through.
JA: When I first met you, maybe four or five years ago, the challenge to me seemed to be: “How do we handle all of this data?” - the informatics piece. But then the challenge seemed to morph into: “We didn’t realize that we would find so many variants. What is the significance of all these variants?” – variant interpretation.
RN: Absolutely. In the early days, you're exactly right. As folks started adopting NGS, many of the assays were custom built - from library design, picking the right library capture methodology and the various technologies available for amplification. What you really focused on in informatics was, “How do I create a viable bioinformatics pipeline that's analytically validated in concert with the library prep and the sequencer and the multiplexing that I'm doing on the sample types that I'm sequencing?” Secondary analysis was the focus - alignment and variant calling - in order to ensure that the variants you're calling are “really there,” i.e. the assay itself, in a methods based fashion, across the entire panel, has high specificity and sensitivity.
As the market has matured, over the past seven or eight years, there are now mature tools - either complete pipelines or individual tools - that come together very nicely, especially for vendor-based assays. Vendors, themselves, are providing secondary analysis pipelines that are close to clinical grade with some minor tweaking and validating, which can then be used in an LDT (Laboratory Developed Test) environment after analytical validation.
Informaticians like myself need not - in all cases or even most cases - build such pipelines from scratch. They come from the assay vendors as part of the offering because the customer has to be able to specifically and sensitively call variants. So you see that the focus shifts to “How do I now interpret these variants?” - which is a very important concept.
When we started with the C. G. W. (Clinical Genomics Workspace), we developed a tool that included secondary analysis, followed by annotation and interpretation. The solution we're using today is a very flexible model, where the secondary analysis is performed either externally or is wrapped inside the solution provided by the vendor. The focus that we have is around the knowledge base and the interpretive aspect of the solution.
JA: One thing that struck me, being somewhat involved with this in the early days, was that there was a large concern over sensitivity, particularly in the liquid biopsy space, where we wanted to be sure we didn't miss anything. Has that become less of an issue? Have those concerns been allayed, somewhat?
RN: Absolutely. I think that in the solid tissue space, for small sequence variants, SNVs and small indels, I think that we've been able to put those concerns to rest, fairly effectively. We can now get down to 2% – 5% allele frequency, for both of those variant types and with very high specificity. That's really very doable today.
For copy number variants and fusions, with appropriate cut offs on amplifications or fusion reads, there's very high specificity and sensitivity as well. In the circulating tumor space, we’re getting down to 0.1% allele frequency for small sequence variants.
Continued in part 2…