CHICAGO — The International Organization for Standardization (ISO) this month officially approved Phenopackets, a standard for sharing disease and phenotype information for diagnosing and treating rare and hereditary diseases including cancer.
The approval may be an important step toward overcoming a major barrier to the wider adoption of precision medicine: the difficulty of representing phenotypes and genotypes in machine-readable form in electronic health records, so they can be better interpreted.
"This gives us a lot of hope," Monica Munoz-Torres, a bioinformatician and genome curation specialist at the University of Colorado Anschutz Medical Campus (CU Anschutz), said at the Intelligent Systems for Molecular Biology (ISMB) conference of the International Society for Computational Biology in Madison, Wisconsin, this month.
Munoz-Torres is a lead developer of Phenopackets, which the standards body Global Alliance for Genomics and Health (GA4GH) in 2019 named one of five emerging standards central to its five-year strategic plan to support and enable responsible genomic data sharing. A speaker from clinical terminology organization SNOMED International also touted Phenopackets at the 2019 Healthcare Information and Management Systems Society (HIMSS) conference.
A large group of collaborators described Phenopackets in a paper published in Nature Biotechnology last month.
By putting detailed phenotypic descriptions as well as disease, patient, and genotypic information into uniform, machine-readable packets, Phenopackets is able to create both a "snapshot" of an individual patient at a moment in time, but it can also capture longitudinal data and produce a view of a patient's condition over the course of time, according to Melissa Haendel, chief research informatics officer at CU Anschutz and a senior Phenopackets investigator.
"You want to create regulatory and ethical validity for the clinical decision-making," Haendel explained. "We need to have a way of documenting why a decision is getting made on any given day based upon the knowledge that's available and that interpretation on that day."
For variant annotation and prioritization, Phenopackets supports the use of the Exomiser algorithm from the Wellcome Sanger Institute. On one day, Exomiser might show that a variant is of unknown significance, but that status could change the next day as new knowledge about associated phenotypes comes to light.
Besides a large number of other building blocks, Phenopackets include three elements related to interpretation: the genotype (or even a panel, whole exome, or whole genome itself), the patient's pedigree, and the interpretation. "That collection of things together, plus the interpretation of the clinician … is a document," Haendel said. "It doesn't exist yet in the sense that we're still working on getting Phenopackets into EHRs," she added.
EHRs are not always designed to represent longitudinal progression of diseases and people's health. "Figuring out how to get Phenopackets into EHRs is nontrivial because they're not designed with that longitudinal disease trajectory in mind," Haendel said.
In the meantime, the Health Level 7 International (HL7) Clinical Genomics Working Group is trying to put genotypes into EHRs and to create a continuum, she said.
The Phenopackets concept has been in development since 2014. "I think we're finally starting to converge on how these things are going to fit together," Haendel said. "We're at a point now where we have enough maturity of some of the standards, but we don't have maturity of how they fit together and how they get used together."
For their part, EHR vendors are enthusiastic about the development of Phenopackets. The new ISO recognition gives them the green light to start incorporating the standard into their technology platforms, particularly for certain use cases.
"I think it's a huge step forward," said Jennifer Ford, product manager for laboratory information systems and genomics at EHR vendor Meditech. "It really is one of the few ways that we're ever going to be able to come up with standards necessary to do clinical decision [support] in rare diseases, which is minimal right now."
"This gives us the ability to have the standard that can be leveraged by companies who create clinical decision support tools that we can then embed into the EHR to drive it," she added.
Ford has been following the development of the standard and has been in conversation with GA4GH for several years. "I think what Phenopackets is doing in the space of rare diseases is going to be one of the few ways … that we can gather the content to package together that we can then build into systems to alert clinicians," she said.
Like any new standard, it will take a few years for Phenopackets to be built into EHRs and other health IT systems, and then for healthcare organizations to use it in the real world, according to Ford.
However, this is just one piece of the puzzle. For example, she said, ICD-10 coding, which classifies diseases, symptoms, and procedures for clinical and billing purposes, is not particularly suited for rare diseases or for genomics, a sentiment shared by Haendel.
Haendel said that by putting current ICD-10 codes into the EHR for rare diseases, the job of the ontology and diagnostics communities becomes harder. "It's just another code with a label that doesn't actually have the reconciliation across all the sources," including links to medical evidence.
She noted that GA4GH is continuing to work on standards for representing variants and annotations in EHRs, in parallel with HL7's effort to express genotypes in patient records.
"You can represent the same variant multiple ways, and you can have the same representations actually meaning different variants, so the problem is how we capture the genotype in the first place," Haendel said. "There has not been a very good handshake between community drivers and standards bodies like GA4GH and vendors, but it's hopefully improving."