In the fall of 2014, Boston-based Partners Healthcare and the Tucson, Ariz.-based diagnostic IT solutions company Sunquest Information Systems announced a joint collaboration with the aim to accelerate genomic-based medicine. At the core of the relationship is a strategic investment by Sunquest into GeneInsight, Inc., an IT platform company owned and developed by Partners over the last decade. The platform streamlines the analysis, interpretation, and reporting of complex genetic test results.
Partners is a majority shareholder in GeneInsight, and when the 2014 announcement was made, officials from the organizations noted that Sunquest and Partners have a shared vision for clinical genomics, with the goal of developing a next-generation genomic information system and knowledgebase that will speed the advent of precision medicine. “With this relationship we will be able to bring new genomic knowledge to treating clinicians much more rapidly, advancing the field of personalized medicine and ensuring that our patients, and patients worldwide, will benefit from the most up-to-date genomic information available at the point of care,” Anne Klibanski, M.D., chief academic officer at Partners HealthCare, said at the time of the collaboration.
Now that the tool is fully developed, Partners recently announced that it is looking to sell GeneInsight to hospitals and labs around the world, “hoping to tap into the emerging market for genomics and generate more money to support operations and research at Partners,” according to November report from the Boston Globe. The report noted that Partners will sell GeneInsight to Sunquest, which then will market the technology. Partners will earn royalties from the sale, the Globe said, adding that the global market for genetic sequencing is growing at double-digit annual rates and is expected to reach $2.8 billion by 2018, according to consulting and research firm Frost & Sullivan.
Indeed, the technology has been developed to integrate genomic information into the pathology workflow, and into hospital and health system electronic medical records (EMRs). Integrating this genetic information into the practice of medicine in a robust and timely manner is challenging, however, as the complexity of the genomic interpretation process allows for only a small fraction of data to be accessible to researchers and clinicians, according to experts at Partners.
To this end, Sandy Aronson, executive director of information technology at Partners, recently spoke with Healthcare Informatics about these challenges, how the partnership is working so far to improve access to genetic data at the point of care, and what the future holds for integrating genetics into healthcare. Below are excerpts of that conversation.
What is the concept behind the GeneInsight platform?
The central concept behind GeneInsight is that we, as a community, have constantly gotten better and better at the sequencing technology that underlies identifying variance in patients. For a long time we have known that the interpretation of those variances is a very critical step that needs to be managed well in order for all of this sequencing technology to yield significant benefits for patients. When genetic tests are launched, within the research enterprise there is a discovery made that links a gene or a set of genes to a clinically important fact where clinicians would want to be able to determine which of their patients have genetic variations that links them to his fact. This could be a predisposition to a sudden cardiac death, for instance.
When that research occurs and papers are published, and there are decisions made by clinical labs that say, ‘Okay, this is significant enough where we should launch a clinical test,’ and the labs then start sequencing patient genes. When the research enterprise says that a gene is linked to a clinically relevant fact, what they’re actually saying is that they have identified variance within a gene. When you sequence patients, you’re literally reading all of the letters associated with the relative parts of a gene, and you’re going to find variations that weren’t initially seen in the research. This is happening in the clinical context and you have to decide what to do with them. Through the clinical process itself, you will likely develop better data that may help you refine and better understand the interpretation and significant of those variances.
Can you give examples of what researchers are looking for to help clinicians?
We do an inherited heart disease panel, and there you have some instances where sudden cardiac death runs in the family. Sometimes you will have really healthy people who will experience sudden cardiac events and die, and there is a genetic predisposition to it. Underlying that, there are a large number of different genes, any number of which can contain a mutation that puts a person at risk for this. If you can identify within a family the specific mutation that is putting that family at risk, being passed down through generations, or not passed down, once you have that knowledge, you can go and look at each member of that family and say if he or she is at risk or not at risk. Either you’re just like the general population, or you’re at risk like your previous family members were.
Another example I can give is in the form of having a cancerous tumor. The question is what genetic anomalies are driving that tumor’s growth? The reason you want to know that is that drugs often target specific pathways. When people talk about personalized medicine drugs, the side effect profiles are more narrowly focused when dealing with a patient with specific pathways. Those drugs are effective when the pathway they target is driving the cancer; they are not effective when that cancer is growing because some other pathway is active. Looking at whether you have variances that can be targeted by these drugs, or that have disabled tumor suppressors that may make other drugs more effective, helps you figure out what treatments are likely to work for this patient.
When genetic tests are run, samples are sent to labs, and labs determine what variances are present in the relevant regions of patient’s DNA that are associated with the question they are being asked to answer. They identify variances and then write a report that explains what’s known about the implication of that variance, called interpreting the variance.
In order to do that, you have this problem of knowledge management surrounding the variance. So GeneInsight is a suite of applications, and two apps in the core of that suite, are GeneInsight Lab for labs, and GeneInsight Clinic for clinics. Within the lab app, there is a knowledge base, and it enables labs to interrelate information on the tests that a lab offers, the genes that are covered by those tests, the variances that have been identified in those genes, and the state of knowledge in the form of classifications, linking those variances to disease states or pharmacogenomic effects, and references underlying all these things.
On top of the knowledge base, a geneticist can make configurations so it has templates that pull information from the base so they can organize themselves down into concise patient-specific, information-dense drafts about what is known about those variances in that patient. Drafts are then viewed by geneticists, changed if necessary, and reports are then sent out to treating clinicians.
Also, for the parts of this report that aren’t totally specific to a particular patient that involves the knowledge of genetic variation and the implication of genetic variance, the lab will make updates if there is a need for it, and then they redraft the report for them. That process saves them time, as the knowledge will be used to draft future reports. It also creates a continuous learning environment within the lab. Each report coming through is creating an opportunity to understand whether there is enough additional knowledge about this variance to change its interpretation.
Now, what are clinicians doing with this information at the point of care?
When you learn something new about a variant, you can go back and look and see where that variant was previously reported. When a clinic is using the GeneInsight Clinic app, the report is organized inside the technology at Partners, and integrated within the EMR—both our Epic one and our custom one, so it provides a patient genetic summary. I would argue the most important aspect of it is that for patients that were previously tested, if a variant is reclassified by the lab, those clinicians receive alerts telling them something new has been learned about this patient.
Alerts are graded, and high ones go to the clinicians, while others are bashed. Based on these alerts, a doctor may want to consider updating their treatment regimen. This is new in healthcare—there is no new visit by patient, or no new test ordered, but just an update that has occurred on the variant, totally independent from the patient. We have a process whereby that information can be interjected into the clinical flow, and the patient can be benefitted, which is unlike what we have seen historically. For instance, you might tell the patient that he or she has a variant of unknown significance, and you don’t know what it means. Or you have a more definitive result like a pathogenic variant, and now you have to be aggressive. Again, this extends the continuous learning environment to the clinic.
Now, as time passes and more people get seen with this variant, you can better understand whether it segregates with disease in families, and maybe additional studies happen that lend more evidence to this variant being more pathogenic or benign. Information accumulates and it reaches a threshold where the lab changes its classification of the variant. This provides an opportunity to change the patient’s profile.
Integrating genetics into healthcare is immature to date. How do you see this changing in the future?
We do feel that we have achieved a tight integration with this. One of biggest issues out there is reimbursement. There is no reimbursement associated with updating a clinician with new information that emerges under a previously tested variant—even though it’s potentially critical to the care of patient and the future of medical costs in terms of getting the diagnosis right. We decided what was needed was a system that was capable of providing those alerts without costing geneticists any time.
I think genome sequencing will become increasingly important. Instead of sequencing a certain set of a patient’s genes associated with a condition, what will happen instead is you will assay all of the genes one way or another, and you will only interpret the set associated with the relative condition. That will make it more of an informatics exercise. The potential to reinterpret that sequence as a patient develops new indications over their lifetime.