Clinical Cancer Genomics Gives Bioinformatics Businesses A Boost
This article was originally published in Start Up
Executive Summary
New products are combining genomic information with drug and clinical data to help oncologists make better treatment decisions. Buoyed largely by demand in oncology, is the time ripe to build lasting companies on bioinformatics platforms?
Start-up companies have tried and failed to forge sustainable businesses around informatics in the past. That may now be changing.
In the last month, three Boston-area companies have launched products or announced programs aimed at analyzing and interpreting genomic information for molecular profiling, all targeted wholly or in part at the clinical oncology market. Three factors are rapidly opening this new large market for bioinformatics: the ability to obtain and sequence patient samples faster and more cheaply, the increase in the number of available targeted drugs and diagnostic tests, and a growing roster of clinical trials grounded in genomics.
On May 30, Foundation Medicine Inc. (FMI) launched FoundationOne, a test that identifies molecular alterations in a patient’s tumor and matches that profile with relevant targeted therapies and clinical trials. A day later, N-Of-One Inc. began roll-out of a stand-alone cancer informatics software package called PrecisionWorks to hospitals, and also said that FMI was using the platform to conduct the initial curation pass of the public domain data that goes into FoundationOne. Then on June 11, Knome Inc. kicked off an early access program for installation of its clinical genomics software package, knomeCLINIC. Although they vary in scope and capabilities, all three offerings aim to assist oncologists’ decision-making on treatment options, combining genomic information with relevant data on drugs and data and clinical research.
Bioinformatics, the computational means for assembling and analyzing a genome, has for some time been a tool in search of a market around which to build a sustainable business. “Even five years ago, if bioinformatics tools were as mature as they are today, I’m not sure [they] would have had any impact on the relevance of all the clinical information,” says Michael Pellini, FMI’s CEO. The same is true for next-generation sequencing. “It would still have been a research tool because there was no way to intervene from a therapeutic standpoint,” he says.
But with the availability of new targeted drugs and molecular diagnostics, the pieces are now coming together. The convergence is putting a premium on data interpretation – the final element of the computational process. Data interpretation includes identifying the relevant mutations, the drugs that target them, and the clinical trials available to patients that include those drugs (which, among other things, allows the drug costs to be covered). It also includes a search of the literature for studies that have identified what each gene does and how its protein products work together in a given pathway. With that information, genomic mutations can be decorated with a map of the known pathways, especially those involved in cancer, allowing physicians and patients to make the most informed decision on where to hit that pathway with a given drug. “You might have several mutations in that pathway but you want to take the most strategic one,” says Elaine Mardis, PhD, co-director of the Genome Institute at Washington University. Typically the furthest along in the pathway is the one to identify as a targetable mutation, she says. “It’s the last bit of interpretation that’s necessary to really give back an informed response to the oncologist.”
To lead today, large health care systems need to be able to provide this kind of molecular profiling strategy to patients. “They have to introduce it and they have to do it quickly,” says Christine Cournoyer, CEO of N-Of-One. Cournoyer was formerly CEO of Picis, a provider of clinical decision support now a part of UnitedHealth Group’s OptumInsight. Picis started out providing support in high acuity areas of hospitals – the ER and ICU – then evolved into clinical decision support at the point of care. “We were introducing knowledge for the physicians and making that available so they could better treat patients,” she says.
In many respects, N-Of-One is another evolution of that model. Its platform is designed to provide the ability to link up-to-date knowledge about molecular targets – in this case, biomarker information, be it genomic or proteomic, culled from any diagnostic test platform. This roadmap identifies the key biomarker sets that should be evaluated, taking into account the clinical data around an individual patient’s tumor and their medical history. It also aggregates that information in the context of disease type to outline the relevant testing on a case-by-case basis.
N-Of-One has mostly worked directly with patients and oncologists thus far. Now, however, the company believes it can launch into the provider market. “We can scale with this offering,” says Jennifer Levin Carter, MD, President and CMO. N-Of-One will continue to work with patients directly but will focus its go-to-market efforts on partnerships with diagnostics companies (FMI, for example) and a direct strategy into the large integrated health networks.
The type of data may be novel, but importantly, the clinical market is accustomed to buying clinical decision support systems, Cournoyer says. “They’re already purchasing it as software, as a service, which is one of the reasons we want to sell it that way. It’s a business model providers are comfortable with.”
With the high cost of cancer diagnosis and treatment, providers are looking for more accurate diagnoses and more cost-effective treatment strategies. “This plays exactly to the pressures they are facing today,” says Cournoyer. “We help them deploy these solutions, rather than their having to hire their own scientists and do their own curation, and get them off and running in 30-60 days.”
Knome also sees a market ripe for a distributed, scaled-up clinical informatics offering. It is in discussions with almost 20 clinics, according to Martin Tolar, MD, PhD, President & CEO. “There’s tremendous interest from people who need to implement these technologies across the industries and utilize them in the clinic right now,” he says. The availability of sequence information at a large scale, a sufficient amount of information to analyze for disease and drug response and having patients in the clinic they can actually advise, and a decision to go from a service model to a software business model, were all contributing factors, Tolar says. (See (Also see "Illumina Goes Consumer " - In Vivo, 1 Jun, 2009.).)
After starting out in consumer genomics, the five-year-old start-up shifted and began applying its genomics-based platform to academic research and drug development. Yet while half of Knome’s current revenues come from projects with pharma, the clinical market “is overshadowing everything,” says Tolar. “In the academic lab we started with five, maybe 10 samples, now we can get to a hundred, a thousand. In pharma it may be hundreds or thousands. But in clinic, it’s tens of thousands and beyond. Even now, they really aspire to have tremendous throughput.” Knome intends to install its informatics platform in hospitals, where it can be used for clinical services. Although its platform is not cancer-specific like those of N-Of-One and FMI, it expects pathology to be the initial entry point into hospitals, along with departments of laboratory medicine.
“Pathologists are probably the most suited to manage the flow of this information,” says Tolar. To bridge into clinical applications, Knome is enhancing its platform to be a physician reporting system, with all the pieces needed to generate a report that meets regulatory requirements for sign-off and that can be integrated into the electronic medical record and an institution’s clinical intelligence network.
Knome understands first-hand the differences between applying genomics in R&D and applying genomics in the clinic. Research takes a disease population and a normal population and looks for variants responsible for the disease and drug development looks for the mechanisms of drug response. But in the clinic, information already generated helps deliver the most precise answer to a physician in a specific situation. “You have to be able to drill down fast and show how you come up with not only the data supporting a decision but also the variants,” says Tolar. “If you’re making a decision in breast cancer, for example, you absolutely have to know and show what’s going on, that you have a true genome.”
All three companies and others including West Coast brethren Personalis, a year-old start-up out of Stanford University and Omicia Inc., which has been deliberately developing genome analysis tools since 2002, acknowledge the importance of providing data interpretation in a context appropriate for oncologists, pathologists and patients. In addition to the promise of high-quality information, each has its own differentiators. FMI provides a seamless and meticulous integration of the entire testing process; Knome claims the benefits of a whole genome analysis; and N-Of-One a patient-centeredness and an inside-out awareness of hospital infrastructure.
“The interpretation piece is necessary because the notion that a pathologist or oncologist is going to readily incorporate genomics into their bag of tricks is just highly deluded,” says Mardis. “It isn’t going to happen soon, but in the meantime people are going to want these tests so doctors are going to have to find a way to make these tests available and understand them themselves.”
That said, even with a growing demand for these tools, companies must be mindful of the cliques and silos within the medical community – a potential barrier to adoption. “Four years ago when we started N-Of-One, there was no infrastructure for this at all,” adds N-Of-One’s Levin Carter. “Our goal from the beginning was to build…the type of infrastructure that breaks down the barriers of fragmentation and drive solutions to the point of care.” Indeed, the success of a company offering complex algorithms and computational expertise may depend as much on its human touch.