Advertisement

Electronic Health Records and Genomics

Perspectives from the Association for Molecular Pathology Electronic Health Record (EHR) Interoperability for Clinical Genomics Data Working Group
  • Alexis B. Carter
    Correspondence
    Address correspondence to Alexis B. Carter, M.D., Children's Healthcare of Atlanta, Laboratory Administration, 1575 N.E. Expressway, Atlanta, GA 30329.
    Affiliations
    The Electronic Health Record Interoperability for Clinical Genomics Data Working Group of the Informatics Subdivision, Association for Molecular Pathology, Rockville, Maryland

    Department of Pathology and Laboratory Medicine, Children's Healthcare of Atlanta, Atlanta, Georgia
    Search for articles by this author
  • Lynne V. Abruzzo
    Affiliations
    The Electronic Health Record Interoperability for Clinical Genomics Data Working Group of the Informatics Subdivision, Association for Molecular Pathology, Rockville, Maryland

    Department of Pathology, Wexner Medical Center, The Ohio State University, Columbus, Ohio
    Search for articles by this author
  • Julie W. Hirschhorn
    Affiliations
    The Electronic Health Record Interoperability for Clinical Genomics Data Working Group of the Informatics Subdivision, Association for Molecular Pathology, Rockville, Maryland

    Medical University of South Carolina, Charleston, South Carolina
    Search for articles by this author
  • Dan Jones
    Affiliations
    The Electronic Health Record Interoperability for Clinical Genomics Data Working Group of the Informatics Subdivision, Association for Molecular Pathology, Rockville, Maryland

    The Ohio State University Comprehensive Cancer Center, James Cancer Hospital and Solove Research Institute, Columbus, Ohio
    Search for articles by this author
  • Danielle C. Jordan
    Affiliations
    Association for Molecular Pathology, Rockville, Maryland
    Search for articles by this author
  • Mehdi Nassiri
    Affiliations
    The Electronic Health Record Interoperability for Clinical Genomics Data Working Group of the Informatics Subdivision, Association for Molecular Pathology, Rockville, Maryland

    Department of Pathology and Laboratory Medicine, Indiana University School of Medicine, Indianapolis, Indiana
    Search for articles by this author
  • Shuji Ogino
    Affiliations
    The Electronic Health Record Interoperability for Clinical Genomics Data Working Group of the Informatics Subdivision, Association for Molecular Pathology, Rockville, Maryland

    Brigham & Women's Hospital, Boston, Massachusetts

    Harvard Medical School, Boston, Massachusetts

    Harvard T.H. Chan School of Public Health, Boston, Massachusetts

    Broad Institute of Massachusetts Institute of Technology and Harvard, Cambridge, Massachusetts
    Search for articles by this author
  • Nimesh R. Patel
    Affiliations
    The Electronic Health Record Interoperability for Clinical Genomics Data Working Group of the Informatics Subdivision, Association for Molecular Pathology, Rockville, Maryland

    Department of Pathology, Rhode Island Hospital and Alpert Medical School of Brown University, Providence, Rhode Island
    Search for articles by this author
  • Christopher G. Suciu
    Affiliations
    The Electronic Health Record Interoperability for Clinical Genomics Data Working Group of the Informatics Subdivision, Association for Molecular Pathology, Rockville, Maryland

    Department of Pathology and Immunology, Washington University School of Medicine, St. Louis, Missouri

    Institute for Informatics, Washington University School of Medicine, St. Louis, Missouri
    Search for articles by this author
  • Robyn L. Temple-Smolkin
    Affiliations
    Association for Molecular Pathology, Rockville, Maryland
    Search for articles by this author
  • Ahmet Zehir
    Affiliations
    The Electronic Health Record Interoperability for Clinical Genomics Data Working Group of the Informatics Subdivision, Association for Molecular Pathology, Rockville, Maryland

    Department of Pathology, Memorial Sloan Kettering Cancer Center, New York, New York
    Search for articles by this author
  • Somak Roy
    Affiliations
    The Electronic Health Record Interoperability for Clinical Genomics Data Working Group of the Informatics Subdivision, Association for Molecular Pathology, Rockville, Maryland

    Department of Pathology, Cincinnati Children's Hospital Medical Center, Cincinnati, Ohio
    Search for articles by this author
Published:October 14, 2021DOI:https://doi.org/10.1016/j.jmoldx.2021.09.009
      The use of genomics in medicine is expanding rapidly, but information systems are lagging in their ability to support genomic workflows both from the laboratory and patient-facing provider perspective. The complexity of genomic data, the lack of needed data standards, and lack of genomic fluency and functionality as well as several other factors have contributed to the gaps between genomic data generation, interoperability, and utilization. These gaps are posing significant challenges to laboratory and pathology professionals, clinicians, and patients in the ability to generate, communicate, consume, and use genomic test results. The Association for Molecular Pathology Electronic Health Record Working Group was convened to assess the challenges and opportunities and to recommend solutions on ways to resolve current problems associated with the display and use of genomic data in electronic health records.
      The use of genomics in clinical medicine is expanding rapidly. Although many laboratories are implementing highly complex technology to perform next-generation sequencing (NGS), the utilization of genomic data is substantially hindered by a lack of standards in analytical pipelines, data interpretations, and reporting. As genomic data become more widely available, health care providers struggle to interpret genomic results because they are complex and highly variable between laboratories in scope of testing performed and degree to which variants of unknown significance are reported. They can also be difficult to locate in electronic health records (EHRs) and can be even more difficult to compare across different tests and panels over time. Genomic testing results are most often stored and displayed in EHRs as either flat text reports or portable document format (PDF) files only and without storage of discrete result data. Both of these report formats facilitate result communication and documentation, and PDFs can support hyperlinks to additional information. However, because discrete data cannot be easily extracted from these flat reports, neither format is optimal for correlating NGS results with other data that may be critical to patient care, such as cross-checking variants between different reports on the same patient, alerts for therapy that is contraindicated on the basis of the patient's genome, or cross-referencing the variant profiles across patients. Furthermore, because individual variants are buried within a text report, aggregate variant profiles, which gather variants from different assays on the same patient onto a single view, are generally not possible without significant effort and may result in errors associated with extracting discrete variant data via natural language processing.
      The Association for Molecular Pathology (AMP) has long recognized the importance of informatics to the success of genomic medicine. As the use of molecular pathology and genomic testing has increased, so have the complaints regarding the significant lack of functionality that necessitates the use of paper records and/or manual transfer of data to and from instrument software, laboratory information systems (LISs), and EHRs. In late 2019, the AMP Board of Directors called for the formation of an Electronic Health Record Working Group with expertise in molecular pathology and clinical informatics to recommend solutions to the AMP Board of Directors on ways to resolve current problems associated with the display and use of genomic data in EHRs. After it was convened in early 2020, the working group performed a comprehensive examination via environmental scan of current problems and barriers to the display and use of genomic data in EHRs. Multiple problems were described by the working group members, which were subsequently categorized and independently ranked using a form of multivoting (Agency for Healthcare Research and Quality, https://digital.ahrq.gov/health-it-tools-and-resources/evaluation-resources/workflow-assessment-health-it-toolkit/all-workflow-tools/multivoting, last accessed December 8, 2020). The results of the discussion and a description of current state are included in this article.

      Scope and Structure of the Document

      The AMP working group had a narrow scope of activities, which are reflected in this article. The working group recognizes that all components of software and systems used to generate genomic test results for patients are important and has structured the document from test ordering through data generation and result display.
      LISs as well as other laboratory instrument software, for example, are critical to laboratory operations because they have many other functions in addition to those in EHRs. These include, but are not limited to, housing nucleic acid quantity and quality, aliquot-specific information, reagent calculations, batch information, various pipeline output files, variant databases, and a myriad of other functions. However, the scope of this working group was to specifically identify challenges with the use of EHRs for genomic data.
      Both germline and somatic variant types were included in the discussions, as were all molecular pathology and genomic testing methods. These include traditional assays, such as PCR, RT-PCR, Sanger sequencing, restriction fragment length polymorphism, Southern blots, gel electrophoresis, NGS, karyotyping, in situ hybridization, fluorescence in situ hybridization, single-nucleotide polymorphism microarrays, oligonucleotide microarrays, array comparative genomic hybridization, and others. The use of these methods in various types of laboratories was also discussed with regard to challenges in EHRs, including molecular and genetic laboratories; cytogenetics laboratories; and laboratories performing fluorescence in situ hybridization, tissue typing, histocompatibility, and immunogenetics (human leukocyte antigen testing). A brief discussion of distinct requirements for reporting of genomic data for infectious organisms is also included. Figure 1 presents a graphical representation of the current overall challenges and opportunities at a high level, which this article describes in detail. The reader is advised to refer to the figure throughout this article for reference.
      Figure thumbnail gr1
      Figure 1Workflow and data transformations required for a genomics-ready electronic health record (EHR). Orders are screened for payer coverage and correct coding, and adjusted for the appropriate next-generation sequencing panel or gene set on the basis of presumptive diagnosis within the laboratory information system (LIS). Data generated by the bioinformatics pipeline undergo base calling, alignment, variant calling, and automated and/or manual annotation before final molecular geneticist/pathologist interpretation. This interpretation is supplemented by clinical decision support (CDS) rules (supported by external and internal databases) to drive the subsequent diagnostic workup and therapy options. A critical enabling feature is structured storage and display in the EHR of variants, gene fusions, and copy number alterations/chromosome abnormalities. API, application programming interface; BED, Browser Extensible Data.
      This article is organized by describing the caveats of the working group scope, other efforts in this area, the challenges and opportunities for genomics in EHRs in order of workflow (orders to results), and interoperability (Table 1).
      Table 1A Summary of Challenges and Opportunities
      ChallengesOpportunities
      EHRs are not yet ready to send accurate, coded, and appropriately granular clinical history, signs, symptoms, family history, and other broad sets of data elements to laboratories without generating significant burden on providers.Develop stakeholder consensus-derived and standardized methods to apply accurate concept codes to the clinical notes, signs, symptoms, and family histories as well as specimen locations, such that necessary information could be sent by the provider using an automated or semi-automated import into the electronic order for genomic testing.
      EHRs lack sufficient information about genetic test orders and generally do not have discrete variant result data to facilitate appropriate test ordering and utilization.

      Absence of standardized genomic variant data structures from the LIS to the EHR.
      Establish a consensus standard for minimum discrete data required to define a genomic variant in an EHR.
      Areas where the Association for Molecular Pathology (AMP) is actively engaged with subject matter expert working groups addressing identified challenges and opportunities. Future recommendations, along with accompanying educational offerings, are being developed by the EHR Working Group and other AMP working groups to address these community needs.


      Establish a consensus standard for minimum discrete data needed to define a genetic/genomic test order.
      Interoperability standards for genomics are currently limited in several ways, including the use of syntax with limited hierarchy (HL7 version 2.x) and inadequate coding systems for genomic orders and results.Before mandating changes to interoperability requirements, governments and regulators should carefully review the cost and burden to laboratories as well as the safety of existing coding standards that are currently inadequate for genomic data. Such work should occur after establishment of a consensus standard for minimum discrete data required to define a genomic variant.
      Genomic reports between laboratories are variable in structure and content.Develop recommendations for content and structure of genomic reports that support providing the established consensus standard for minimum discrete data required to define a genomic variant in an EHR.
      Areas where the Association for Molecular Pathology (AMP) is actively engaged with subject matter expert working groups addressing identified challenges and opportunities. Future recommendations, along with accompanying educational offerings, are being developed by the EHR Working Group and other AMP working groups to address these community needs.
      Implementation of molecular pathology and genomics professional societies multi-organizational consensus standards and guidelines for variant nomenclature, hierarchical result structure, and genomic report formats has been modest.Encourage laboratories and EHRs to support and implement molecular pathology and genomics professional societies multi-organizational consensus standards and guidelines for report content and structure to standardize structure and content between laboratories.
      Areas where the Association for Molecular Pathology (AMP) is actively engaged with subject matter expert working groups addressing identified challenges and opportunities. Future recommendations, along with accompanying educational offerings, are being developed by the EHR Working Group and other AMP working groups to address these community needs.


      Professional organizations should consider further developing consensus report structures that use sound principles of human factors engineering and usability.
      There are no standards for how to display aggregated variant data over time and between sample sources, tests, and laboratories that keep clinical context and associated interpretation intact.Standards for aggregation of genomic data over time and between sample sources, tests, and laboratories should be developed that keep clinical context and associated interpretation intact.
      Consensus guidelines on best practice for requesting and providing reclassification of variants is not currently available.Establish consensus guidelines on best practice for requesting and providing reclassification of variants.
      Areas where the Association for Molecular Pathology (AMP) is actively engaged with subject matter expert working groups addressing identified challenges and opportunities. Future recommendations, along with accompanying educational offerings, are being developed by the EHR Working Group and other AMP working groups to address these community needs.
      Genomic reports are difficult for medical personnel and patients to understand, and in the United States, patients have the right to immediately access their genomic test reports on request.Professional organizations should consider further developing consensus report structures that use sound principles of human factors engineering and usability that enable understanding by most patients.
      Absence of national and international standards for CDS rules.Establish evidence-based international recommendations for CDS in genomic test orders and in drug orders impacted by genomic results for safety and consistency in practice with a prerequisite minimum discrete genomic variant data consensus standard established.
      Currently, textual data from genomic reports are exported and released to authorized parties, and analysis of such data is limited by its highly variable format and lack of structure.Establish and integrate appropriate, safe, and functional international standards for interoperability and data retrieval.
      Current technology lacks sufficient functionality to ensure that release of genomic data to appropriate health care organizations, research studies, or clinical repositories requires and receives informed consent from the patient or legal guardian, and/or institutional board review.Develop technology and cybersecurity functions in EHRs that ensure that genomic data are released as authorized after informed consent from the patient along with tools to educate patients about informed consent.
      Current coding and interoperability standards are not adequate for genomic data.Develop safe, complete, and accurate standards for coding and interoperability of genomic data, per the future established minimum discrete variant data standard.
      CDS, clinical decision support; EHR, electronic health record; HL7, Health Level Seven; LIS, laboratory information system.
      Areas where the Association for Molecular Pathology (AMP) is actively engaged with subject matter expert working groups addressing identified challenges and opportunities. Future recommendations, along with accompanying educational offerings, are being developed by the EHR Working Group and other AMP working groups to address these community needs.

      Caveats of Working Group Scope

      This AMP working group recognizes that no EHR has all of the capabilities desired by either this working group or others. The purpose of this document is to provide a framework for future work by this working group and others for guidelines and standards to help improve EHRs in the future.

      Other Efforts to Define EHR Genomic Specifications

      It is important to start by recognizing and briefly describing other efforts in the area of enabling or facilitating the use of genomic data in EHRs, some of which are long-standing, and which have attempted to tackle the difficulties of accurate and standardized transfers and displays of complex data, including genomics, in the EHR. A collaborative effort on the standard representation and integration of genomics into EHRs was initiated as part of the Roundtable on Genomics and Precision Health at National Academies of Sciences, Engineering, and Medicine during 2013 to 2014 (Institute of Medicine of the National Academies, https://www.nationalacademies.org/our-work/roundtable-on-genomics-and-precision-health, last accessed June 4, 2021). This initiative, “Displaying and Integrating Genetic Information through the EHR” (DIGITizE), transitioned from the National Academies to the Health Level Seven (HL7) Fast Healthcare Interoperability Resources (FHIR) Foundation to continue its mission. The DIGITizE initiative produced a limited number of documents and models regarding pharmacogenomics, laboratory connectivity, and clinical decision support (CDS; National Academies of Sciences Engineering Medicine, https://www.nationalacademies.org/our-work/digitize-displaying-and-integrating-genetic-information-through-the-ehr-action-collaborative, last accessed October 2, 2020).
      The Electronic Medical Records and Genomics (eMERGE) Network, initiated in 2007, is a NIH-organized and funded consortium of US medical research institutions that combines DNA biorepositories with EHRs for large-scale, high-throughput genetic research in support of implementing genomic medicine
      • Wiesner G.L.
      • Rahm A.K.
      • Appelbaum P.
      • Aufox S.
      • Bland S.T.
      • Blout C.L.
      • Christensen K.D.
      • Chung W.K.
      • Clayton E.W.
      • Green R.C.
      • Harr M.H.
      • Henrikson N.
      • Hoell C.
      • Holm I.A.
      • Jarvik G.P.
      • Kullo I.J.
      • Lammers P.E.
      • Larson E.B.
      • Lindor N.M.
      • Marasa M.
      • Myers M.F.
      • Peterson J.F.
      • Prows C.A.
      • Ralston J.D.
      • Rasouly H.M.
      • Sharp R.R.
      • Smith M.E.
      • Van Driest S.L.
      • Williams J.L.
      • Williams M.S.
      • Wynn J.
      • Leppig K.A.
      Returning results in the genomic era: initial experiences of the emerge network.
      (National Human Genome Research Institute, https://www.genome.gov/Funded-Programs-Projects/Electronic-Medical-Records-and-Genomics-Network-eMERGE, last accessed October 2, 2020). The consortium brings together investigators with a wide range of expertise in genomics, statistics, ethics, informatics, and clinical medicine from leading medical research institutions across the country. One of the major goals of the eMERGE network is the development of methods and best practices for integration of genetic clinical testing results and other genomics information with clinical data in the EHR, enabling clinical decision support systems, and improving accessibility and usability by clinicians for optimizing patient care. During phase I to III implementation, the eMERGE network included 68 electronic phenotype algorithms and >100,000 participants with genomic data deployed across nine study centers and two genotyping facilities. Returning genetic clinical results has been implemented across the network, including data related to pharmacogenomics and genes associated with breast and colorectal cancers. A recent report summarizing the network's experience on return of genetic results to the EHR at the participating sites noted heterogeneity and real-world problems associated with returning genetic testing results. The report emphasized the need for developing new mechanisms to support return of genetics results and focusing on required elements to achieve that goal
      • Gottesman O.
      • Kuivaniemi H.
      • Tromp G.
      • Faucett W.A.
      • Li R.
      • Manolio T.A.
      • Sanderson S.C.
      • Kannry J.
      • Zinberg R.
      • Basford M.A.
      • Brilliant M.
      • Carey D.J.
      • Chisholm R.L.
      • Chute C.G.
      • Connolly J.J.
      • Crosslin D.
      • Denny J.C.
      • Gallego C.J.
      • Haines J.L.
      • Hakonarson H.
      • Harley J.
      • Jarvik G.P.
      • Kohane I.
      • Kullo I.J.
      • Larson E.B.
      • McCarty C.
      • Ritchie M.D.
      • Roden D.M.
      • Smith M.E.
      • Böttinger E.P.
      • Williams M.S.
      The Electronic Medical Records and Genomics (eMERGE) network: past, present, and future.
      (eMERGE Network, https://emerge-network.org/about-emerge; National Human Genome Research Institute, https://www.genome.gov/Funded-Programs-Projects/Electronic-Medical-Records-and-Genomics-Network-eMERGE, both last accessed October 2, 2020).
      More recently, Sync for Genes is described as a technology-based effort to gather health data from individuals in the United States and make that information available to researchers in a standardized format. One of the tools cited for use by this effort is the HL7's FHIR as a method to standardize the health data, including genomic data (The Office of the National Coordinator for Health Information Technology, https://www.healthit.gov/sites/default/files/sync_for_genes_report_november_2017.pdf, last accessed October 2, 2020).
      The American College of Medical Genetics and Genomics has issued a points to consider statement on genomic information within the EHR,
      • Grebe T.A.
      • Khushf G.
      • Chen M.
      • Bailey D.
      • Brenman L.M.
      • Williams M.S.
      • Seaver L.H.
      The interface of genomic information with the electronic health record: a points to consider statement of the American College of Medical Genetics and Genomics (ACMG).
      defining the scope of genomic data and raising important points regarding access to data and social justice. They emphasize that genomic data should be looked at more broadly, as they include interpretation of testing results documented in clinical notes, excerpting of results into different areas of the medical records, and possibly direct-to-consumer testing results that end up in the EHR. In addition, it is important for patients to have access to genomic data in a Health Insurance Portability and Accountability Act–compliant manner, with the ability to continue to receive results as they are updated and to transfer those results to other health care facilities (optimally achieved electronically through EHRs with the utilization of standards). Finally, because of unique features (such as the predictive nature of results for later risks, impact on family members, changing societal perspectives, and the dynamic nature of knowledge that allows for reinterpretation of previously reported genomic variants), providers, EHR vendors, and health information exchanges should develop mechanisms to protect sensitive genomic information in the EHR.
      • Kannry J.M.
      • Williams M.S.
      Integration of genomics into the electronic health record: mapping terra incognita.
      ,
      • Warner J.L.
      • Jain S.K.
      • Levy M.A.
      Integrating cancer genomic data into electronic health records.
      Several medical institutions have made efforts to work with EHR vendors and laboratories to operationalize some of the American College of Medical Genetics and Genomics guidelines
      • Warner J.L.
      • Jain S.K.
      • Levy M.A.
      Integrating cancer genomic data into electronic health records.
      ,
      • Lau-Min K.S.
      • Asher S.B.
      • Chen J.
      • Domchek S.M.
      • Feldman M.
      • Joffe S.
      • Landgraf J.
      • Speare V.
      • Varughese L.A.
      • Tuteja S.
      • VanZandbergen C.
      • Ritchie M.D.
      • Nathanson K.L.
      Real-world integration of genomic data into the electronic health record: the PennChart Genomics Initiative.
      and address some of the challenges addressed in this document.
      The Clinical Genome Resource (ClinGen) has an EHR Working Group that aims to ensure that the ClinGen resource is designed to be accessible to providers and patients through EHRs and related systems
      • Heale B.S.E.
      • Overby C.L.
      • Del Fiol G.
      • Rubinstein W.S.
      • Maglott D.R.
      • Nelson T.H.
      • Milosavljevic A.
      • Martin C.L.
      • Goehringer S.R.
      • Freimuth R.R.
      • Williams M.S.
      Integrating genomic resources with electronic health records using the HL7 infobutton standard.
      ,
      • Overby C.L.
      • Heale B.
      • Aronson S.
      • Cherry J.M.
      • Dwight S.
      • Milosavljevic A.
      • Nelson T.
      • Niehaus A.
      • Weaver M.A.
      • Ramos E.M.
      • Williams M.S.
      Providing access to genomic variant knowledge in a healthcare setting: a vision for the ClinGen Electronic Health Records Workgroup.
      (ClinGen-EHR Working Group, https://www.clinicalgenome.org/working-groups/ehr, last accessed November 9, 2020). In 2018, ClinGen was the first entity recognized by the US Food and Drug Administration (FDA) as part of its public human genetic variant database program (ClinGen-FDA Recognized Human Variant Database, https://www.clinicalgenome.org/about/fda-recognition; Recognition of ClinGen Expert Curated Human Variant Data Decision Letter, https://www.clinicalgenome.org/site/assets/files/3978/approval_q181150_letter_jpnd_final.pdf, both last accessed December 28, 2020) for variant curation of germline variants for hereditary disease, where there is a high likelihood that the disease or condition will materialize given a deleterious variant (ie, high penetrance). Although such recognition does not constitute marketing clearance or approval, data from FDA-recognized databases would generally constitute valid scientific evidence that can be used to support the clinical validity of genotype-phenotype relationships that may be used as assertions in a premarket submission. Variants curated by expert panels according to the FDA recognition letter are marked as reviewed by expert panel and FDA-recognized database. There are numerous other projects for improving the overall usability of genomic data, which are too numerous to describe and which have not been specifically referenced herein because they are not specific to the usability of genomic data in EHRs.
      In comparison to those efforts described above that have specifically addressed challenges and opportunities of genomics in EHRs, this article presents a more in-depth view of the technical challenges for each portion of the workflow, and high-level summaries are presented at the end of each section for reference.

      Challenges and Opportunities for Genomics in EHRs

       Orders: Improvements for Electronic Orders

      Laboratories often receive electronic laboratory orders from EHRs, particularly when the laboratory is internal to the health care organization that houses the EHR or when they have an HL7 interface with the EHR. However, orders for genetic testing are often devoid of important clinical history, specimen type, and diagnosis information relevant to the requested test. Unlike other laboratory tests, the amount and breadth of information that may be possible to include on even one genetic test can be staggering. Data elements, such as the patient's observed phenotype (affected versus not affected), status of the specimen submitted (germline unaffected tissue versus tumor versus somatic overgrowth region), and tumor status (primary tumor versus metastasis), are easily added to orders now because the build is relatively small and easy to execute. However, suspected disease condition, clinical history, clinical features, family history, tumor type, and specimen source location would each have to have large numbers of options available. This makes build in most EHRs difficult if not impossible, and it generates unacceptable burden on physicians to manually fill out these forms. Automated entry of the required data elements is a future and lofty goal, but currently this is unattainable in most EHRs because these data elements are not captured in a coded and structured manner in the patient's clinical notes in most if not all EHRs. Problem lists are generally unhelpful because they typically only contain confirmed high-level diagnoses and may also be out-of-date. Some reference laboratories have extensive electronic forms for data entry of clinical history available on their websites for exomes and other tests. However, significant time is required to fill out these electronic forms, and some also lack comprehensive coverage of the concepts needed for the patient. There may be inadequate search functionality, and there may be no place to free text in necessary clinical history when it cannot be found in the form. Compared with simply sending a copy of the latest clinical note and/or pedigree on the patient with the specimen and requisition or filling out a simple free-text field on clinical history in the electronic order, it is not surprising that genomic orders in EHRs have not generally been built in this manner. From a data analysis and aggregation perspective, it would certainly be preferable for laboratories to receive this information in a structured and coded manner, but this will not be possible until EHRs are better able to capture accurate and granular data on symptoms and course and until coding systems are able to comprehensively include this information in an unambiguous and accurate manner. Coding systems, such as Logical Observation Identifiers Names and Codes (LOINC, https://loinc.org, last accessed January 21, 2021) and Systematized Nomenclature of Medicine Clinical Terms (SNOMED-CT, https://www.snomed.org, last accessed January 21, 2021), are challenging to use, especially in the United States. For example, it is not certain whether LOINC or SNOMED should be used for clinical history information from a regulatory perspective. LOINC has foundational deficits from a data aggregation perspective because of its lack of structural hierarchy of concepts. It also lacks sufficient granularity to reliably distinguish between different tests that are not semantically interoperable.
      Summary: EHRs are not yet ready to send accurate, coded, and appropriately granular clinical history, signs, symptoms, family history, and other broad sets of data elements to laboratories without generating significant burden on providers.

       Orders: Improvements Needed in Utilization Support

      Ordering genomic tests in EHRs poses challenges that, if not properly managed, can lead to unnecessary or redundant testing as well as delays in performing the correct test with subsequent delays in diagnosis and therapeutic intervention. Concerns over genomic test utilization have been driven, in part, by the high cost of these assays, leading many health care organizations to devote considerable time and effort on appropriate test use and strategies for order review as cost-saving measures.
      • Riley J.D.
      • Procop G.W.
      • Kottke-Marchant K.
      • Wyllie R.
      • Lacbawan F.L.
      Improving molecular genetic test utilization through order restriction, test review, and guidance.
      • Desai K.
      • Hooker G.
      • Adeboyeje G.
      • Kachroo S.
      • Sen S.S.
      HSR20-083: real-world utilization and coding variability in medical claims for next-generation sequencing (NGS)-based diagnostic tests among cancer patients in the U.S..
      • Hsiao S.J.
      • Sireci A.
      • Pendrick D.
      • Freeman C.
      • Yang J.
      • Schwartz G.K.
      • Mansukhani M.M.
      • Carvajal R.D.
      • Oberg J.A.
      Clinical utility and reimbursement for expanded genomic panel testing in adult oncology.
      In addition, with multiple internal and external laboratories performing comprehensive genomic profiling of cancer specimens, difficulties in locating and managing the clinical orders and reports can lead to the performance of multiple high-cost tests with similar or, worse, confusing results and possible depletion of tissue specimens. Because of the complexity surrounding ordering of genetic tests, many laboratories require review by an expert in genomic testing before tests are performed or sent out. Such human review is expensive and not reimbursable, but it has been found to be cost-effective in reducing unnecessary testing.
      • Riley J.D.
      • Procop G.W.
      • Kottke-Marchant K.
      • Wyllie R.
      • Lacbawan F.L.
      Improving molecular genetic test utilization through order restriction, test review, and guidance.
      ,
      • Desai K.
      • Hooker G.
      • Adeboyeje G.
      • Kachroo S.
      • Sen S.S.
      HSR20-083: real-world utilization and coding variability in medical claims for next-generation sequencing (NGS)-based diagnostic tests among cancer patients in the U.S..
      Thus, better CDS is warranted to help ordering providers know when testing has already been performed. When testing has not been performed, better CDS would help providers be able to appropriately contrast and compare different genetic panels by the genes included, types of variants detected (and not detected), and diseases detected along with their clinical and analytical sensitivities. CDS is admittedly a hot topic in many areas of medicine, and some EHRs have included the ability to download and use applications developed by third parties for some types of CDS. However, the lack of fundamental and standardized discrete variant data in most EHRs currently makes use of such applications premature for genomics.
      EHRs have a whole host of clinical data that is not present in an LIS. These can include medications, provider notes, testing from other laboratories, knowledge of whether a medical geneticist has been consulted, and other findings that may support or call into question the use of the test being ordered. Because cytogenetic, molecular, and genomic test results are often reported to EHRs as flat text files or PDFs, discrete variant data are often unavailable to an EHR for ordering support. Without such data, the ability to utilize clinical data in conjunction with variant data for important CDS functions, such as germline pharmacogenomic and cancer-associated drug-variant checks, cannot occur without cumbersome interventions that are hard to sustain over the long-term.
      Summary: CDS in an EHR requires discrete genomic data elements for best test utilization and patient care.

       Data Transfers: The Existing HL7 Standard and the FHIR Effort

      Any improvements in the handling of genomic data by the EHR must recognize the preexisting design and constraints of LISs, EHRs, and data transfer methods. Architectural redesign and software development efforts to improve functionalities, such as those listed below, may be hampered by difficulties of making large changes in EHRs while patient care is ongoing.
      The long-standing international HL7 standard, initiated by the ad hoc HL7 International group in 1987, has promulgated standards for the exchange of clinical and laboratory data (HL7 International, http://www.hl7.org/index.cfm, last accessed December 8, 2020). This structure, which is basically a defined message format, serves as the basis for most data transfers between the LIS and EHR currently. Shortcomings of this standard have been recognized related to difficulties in data integration from disparate sources and difficulties in handling of hierarchical data.
      Most recently, HL7 has developed a framework, termed the FHIR, to modernize the approach to data transfer, including genomics (The Office of the National Coordinator for Health Information Technology, https://www.healthit.gov/sites/default/files/sync_for_genes_report_november_2017.pdf; HL7 FHIR, https://www.hl7.org/fhir, both last accessed December 8, 2020). FHIR adopts contemporary web technology [eg, Representational State Transfer–based application programming interfaces (APIs)/web services], rather than a proprietary API. It consists of multiple linkable and extendable data structure specifications called resources, which covers different health care scenarios such as patients, conditions, and clinical observations and reports. These resources may be tailored by standardized sets of constraints called profiles. SMART (Substitutable Medical Applications & Reusable Technologies) on FHIR Genomics leverages resource extension to handle multiple types of clinical genomics data, including wrappers for genomic data files [eg, Variant Call Format files (VCFs)], messaging services for genomic laboratory results, and document services for interpretative genomic reports.
      • Mandel J.C.
      • Kreda D.A.
      • Mandl K.D.
      • Kohane I.S.
      • Ramoni R.B.
      SMART on FHIR: a standards-based, interoperable apps platform for electronic health records.
      ,
      • Alterovitz G.
      • Warner J.
      • Zhang P.
      • Chen Y.
      • Ullman-Cullere M.
      • Kreda D.
      • Kohane I.S.
      SMART on FHIR genomics: facilitating standardized clinico-genomic apps.
      FHIR can be used as a stand-alone data exchange standard, but can also be used with existing standards. Sources of data can be from EHR systems or separate sequencing systems.
      • Alterovitz G.
      • Heale B.
      • Jones J.
      • Kreda D.
      • Lin F.
      • Liu L.
      • Liu X.
      • Mandl K.D.
      • Poloway D.W.
      • Ramoni R.
      • Wagner A.
      • Warner J.L.
      FHIR genomics: enabling standardization for precision medicine use cases.
      Although FHIR may seem to be the answer to interoperability woes, the only US federal requirement to use FHIR is for communication between EHRs and third-party applications accessing EHR systems using APIs. For communications between laboratories and EHRs, EHRs are mandated by law to use HL7 version 2.5.1 for most other operations under Promoting Interoperability (The Office of the National Coordinator for Health Information Technology, https://www.healthit.gov/topic/meaningful-use-and-macra/promoting-interoperability, last accessed February 17, 2021), formerly known as Meaningful Use through the US Department of Health and Human Services 2015 Edition Health Information Technology Certification Criteria, 2015 Edition Base EHR Definition, and Office of the National Coordinator (ONC) Health Information Technology Certification Program Modifications (The National Archives and Records Administration, https://www.federalregister.gov/documents/2015/10/16/2015-25597/2015-edition-health-information-technology-health-it-certification-criteria-2015-edition-base, last accessed January 21, 2021). The differences in both syntax and semantics between HL7 FHIR and HL7 version 2 are immense, and significant cost will be incurred by vendors and laboratories to switch to FHIR for interfaces between laboratory instruments, laboratory information systems, and EHRs.
      Summary: Despite the improved ability for FHIR to handle data transfer of hierarchical and genomic data, its implementation has been limited in laboratories. There are no legal mandates for communications between EHRs and laboratories to use FHIR, unlike the older version of HL7, which is required for use by certified EHR technology in the United States (The Centers for Medicare & Medicaid Services, https://www.cms.gov/Regulations-and-Guidance/Legislation/EHRIncentivePrograms, last accessed December 8, 2020).

       Discrete Variant Data: The Foundational Architecture of Enabling Genomic Medicine

      The single most important architectural component that would facilitate the use of genomic data in EHRs is precisely what most, if not all, EHRs lack: the ability to consume and store individual variants discretely in the database in a scalable, standardized, and reliable way. There are conflicting opinions regarding the scope of genomic data that should be stored in an EHR, with opinions ranging from storing everything (including raw uninterpreted and unannotated data files, to make these files more accessible to research) all the way to restricting content to textual or PDF reports so that variants can only be viewed in the context of their interpretation. It is the opinion of this working group that the signed-out report should continue to be part of the medicolegal record and that any discretely stored variant data that transfer to the EHR in parallel with the formatted report must have appropriate patient-specific and specimen-specific annotations and interpretative context with them such that the variant is consistently interpreted in its appropriate and complete context. An analogous example is cancer synoptic data that displays within a pathology report. While the cancer synoptic contains relevant items with regard to diagnosis, grade, margins, and staging, the full report and interpretation often contain additional information and elucidation that can help clinicians put the overall diagnosis into clinical context. Although a variant can certainly be annotated with appropriate references to individual variant entries in online databases, these databases cannot replace a standardized variant data structure for clinical purposes because they lack the required information that is specific to the patient and specimen.
      One tool that is needed for release of properly annotated discrete variants to an EHR is a standardized clinical grade VCF file as output from laboratories to EHRs. So-called clinical grade VCF files have been discussed and published,
      • Lubin I.M.
      • Aziz N.
      • Babb L.J.
      • Ballinger D.
      • Bisht H.
      • Church D.M.
      • Cordes S.
      • Eilbeck K.
      • Hyland F.
      • Kalman L.
      • Landrum M.
      • Lockhart E.R.
      • Maglott D.
      • Marth G.
      • Pfeifer J.D.
      • Rehm H.L.
      • Roy S.
      • Tezak Z.
      • Truty R.
      • Ullman-Cullere M.
      • Voelkerding K.V.
      • Worthey E.A.
      • Zaranek A.W.
      • Zook J.M.
      Principles and recommendations for standardizing the use of the next-generation sequencing variant file in clinical settings.
      but the standards are not widely used, in part because they are not required by any regulatory or laboratory accreditation agency. In addition, these clinical-grade VCF standards lack requirements for annotation, clinical context, left versus right variant alignment, and interpretative data that are necessary for accurate interpretation and use. Specimen type, date of collection, genomic coordinates, and human reference genome version, as well as other information, are critical to interpretation of tumor-associated variants. The authors note, however, that there is no single file that captures all these elements. The Human Genome Variation Society's (HGVS’s) c. and p. nomenclature,
      • den Dunnen J.T.
      • Dalgleish R.
      • Maglott D.R.
      • Hart R.K.
      • Greenblatt M.S.
      • Mcgowan-Jordan J.
      • Roux A.F.
      • Smith T.
      • Antonarakis S.E.
      • Taschner P.E.M.
      HGVS recommendations for the description of sequence variants: 2016 update.
      as well as the verified classification of the variant at the time of reporting (eg, clinically significant or benign); type of origin, if known (eg, somatic, germline, or germline mosaicism); and, for cancer specimens, the estimated tumor percentage, tumor diagnosis, monoallelic versus biallelic alterations, assessment of haplotype phase, copy number variants, and the presence of subclonal populations with variants indicating resistance or poorer prognosis are all important. Metadata associated with interpretive context for reported variant(s) should be readily available in the EHR. Identical variants can have different effects and clinical meaning for different patients, depending on the clinical context. For example, the clinical significance of BRAF c.1799T>A (p.Val600Glu) is dependent on the tumor type. Similarly, genotype-phenotype correlation can evolve in response to scientific studies, and a variant's clinical interpretation may be modified over time in a parallel manner. For example, EWSR1 fusions were historically associated with Ewing sarcomas, but were subsequently found in a large number of different tumor types, including epithelioid carcinomas. Some variants, such as pharmacogenetic variants, are not clinically significant until the patient is exposed to a medication or other environmental stimulus. EHRs and other systems need to be able to do more than to just display genetic variants; they need to facilitate the use of and access to the contextual data for a set of genomic alterations to improve health outcomes.
      All of the standard file types for NGS (eg, FASTQ, uBAM, BAM/CRAM, and VCF and its variations) lack sufficient requirements for patient and sample identification that are necessary to ensure that variants are associated to the correct sample and patient (The VCF Version 4.2 Specification, https://samtools.github.io/hts-specs/VCFv4.2.pdf; Sequence Alignment/Map Format Specification, https://samtools.github.io/hts-specs/SAMv1.pdf, both last accessed June 4, 2021). Essentially, none of these file standards, which were developed for research, require sample identification or patient identification to be embedded in the text of the file. Similarly, the issue of coordinate alignment also needs to be rectified. While the left (5′) alignment is used by multiple VCF standards, the HGVS nomenclature uses right (3′) alignment. Some bioinformatics pipelines have special software in place to detect and resolve alignment conflicts that can result in incorrect results on a report, but such software is not available to everyone or may be incorrectly implemented. Continuing the work of the CDC to improve the rigor of the clinical-grade VCF standard, including the alignment issue, is recommended. Adoption of such a standard by regulatory and accreditation agencies will facilitate its use.
      Even if an international clinical-grade VCF file were to be realized, there are additional barriers in translating VCF data into HL7 version 2 messages that further discourage its use with this particular standard. Genomic data inherently have hierarchical relationships that are poorly represented in a flat messaging system, such as HL7 version 2. At the other end of the spectrum, work being done by the HL7 genomics working group is focused on numerous attributes per variant, each with its own set of LOINC and SNOMED-CT codes. The high level of complexity stymies implementation, standardization in real-world use, and may not address the most critical factors for proper interpretation by physicians, other health care providers, and patients. Several groups have also raised concerns regarding the safety of LOINC codes for laboratory interoperability for common tests that have much less complexity than genomic tests.
      • Stram M.
      • Seheult J.
      • Sinard J.H.
      • Campbell W.S.
      • Carter A.B.
      • De Baca M.E.
      • Quinn A.M.
      • Luu H.S.
      A survey of LOINC code selection practices among participants of the College of American Pathologists Coagulation (CGL) and Cardiac Markers (CRT) proficiency testing programs.
      • Carter A.B.
      • de Baca M.E.
      • Luu H.S.
      • Campbell W.S.
      • Stram M.N.
      Use of LOINC for interoperability between organisations poses a risk to safety.
      • Stram M.
      • Gigliotti T.
      • Hartman D.
      • Pitkus A.
      • Huff S.M.
      • Riben M.
      • Henricks W.H.
      • Farahani N.
      • Pantanowitz L.
      Logical observation identifiers names and codes for laboratorians.
      There is also controversy in the laboratory community on which variants should be included in the report. Some laboratories do not report variants of unknown significance, rare, and/or private variants, whereas others do. Most, if not all, laboratories do not report variants that they have interpreted to be benign polymorphisms, including single-nucleotide polymorphisms, deep intronic or intergenic noncoding variants, or false positives (artifacts). Like variants of unknown significance, there is also controversy about whether quality control, quality assurance, or other parameters should be reported and discretely stored in an EHR. Arguably, differences in the quality or quantity of nucleic acid or in the tumor percentage between samples could have an impact on interpretation if these parameters were at or just outside of threshold.
      Another parameter that should also be represented in reports, as well as discretely in the EHR at test level, is the scope of the analysis that was performed for an individual test. This enables physicians, including pathologists and geneticists, to easily compare different tests that were performed on the same patients. The scope of the analysis includes the regions of the genome that were analyzed, as well as the types of variants that were in scope for detection. For example, Browser Extensible Data files indicate the regions of the genome that were included in the analysis for reporting and, often more important, those portions that were not (Regents of the University of California, https://genome.ucsc.edu/FAQ/FAQformat.html, last accessed December 8, 2020). It surprises some health care providers to discover that fully sequenced genes are, in fact, not fully sequenced, and discrepancies between different tests that both covered the same gene(s) can be due to differences in the breadth and uniformity of coverage of those genes. In addition, Browser Extensible Data files may not be a completely accurate representation of the ability of an assay to detect all variants within the Browser Extensible Data file coordinates because of included regions that are difficult to sequence (eg, GC-rich areas and homopolymer regions). Some genes also have known variations in untranslated and deep intronic regions (eg, ANKRD26 and TERT) that may not be detected by standard sequencing of exons and short flanking intronic sequences. Along the same lines, many important variants are large-scale deletions, duplications, and whole or partial chromosome gains or losses in subpopulations of cells, which may not be detected by NGS assays without sophisticated sequencing chemistry, and level of detection required for minimal residual disease may not be transparent or sufficient for the health care provider's needs unless other specialized techniques are used (eg, unique molecular indexes) and specialized bioinformatics algorithms (error-corrected variant calling)
      • MacConaill L.E.
      • Burns R.T.
      • Nag A.
      • Coleman H.A.
      • Slevin M.K.
      • Giorda K.
      • Light M.
      • Lai K.
      • Jarosz M.
      • McNeill M.S.
      • Ducar M.D.
      • Meyerson M.
      • Thorner A.R.
      Unique, dual-indexed sequencing adapters with UMIs effectively eliminate index cross-talk and significantly improve sensitivity of massively parallel sequencing.
      (UC Davis Genome Center, https://dnatech.genomecenter.ucdavis.edu/faqs/what-are-umis-and-why-are-they-used-in-high-throughput-sequencing/#:∼:text=UMI%20is%20an%20acronym%20for,%E2%80%9D%20or%20%E2%80%9CRandom%20Barcodes%E2%80%9D, last accessed December 8, 2020). Closely related to this is the need for including information about a molecular test's performance characteristics and limits of detection. Similarly, including information about any confirmatory testing that was performed for the variants reported is also important, especially for assays that sequence at a lower depth of coverage. Inversions, fusions, and methylation changes will not be detected by standard DNA NGS without additional library preparation and bioinformatics pipeline algorithms. The methods used in genomic testing are also important to the scope of the types of variants that can be detected by the assay. Because HGVS c. and p. nomenclature is dependent on the human reference transcript and annotation database (eg, RefSeq, https://www.ncbi.nlm.nih.gov/refseq; Ensembl, https://useast.ensembl.org/index.html; Locus Reference Genomic Sequences, http://www.lrg-sequence.org/index.html, all last accessed September 8, 2021) that were used, these transcript accession and version numbers (eg, NM_004333.6 for the BRAF gene) should be made available to providers and patients along with the variant(s). The use of different transcripts for the same region can cause differences in the ability of a laboratory professional and/or clinical provider to interpret results or to compare results with online genetic variation databases, particularly when the transcripts used or their versions are not reported with the results. Even when these are reported, a transcript reference sequence represents one transcript, often the longest or canonical transcript, and this may not reflect the alternative splicing, distinctive transcription start site, or other functional difference occurring in the patient. RefSeq, for example, also does not incorporate legacy numbering schema that can help clinicians compare older published data with current variant nomenclature.
      • MacArthur J.A.L.
      • Morales J.
      • Tully R.E.
      • Astashyn A.
      • Gil L.
      • Bruford E.A.
      • Larsson P.
      • Flicek P.
      • Dalgleish R.
      • Maglott D.R.
      • Cunningham F.
      Locus Reference Genomic: reference sequences for the reporting of clinically relevant sequence variants.
      ,
      • Dalgleish R.
      • Flicek P.
      • Cunningham F.
      • Astashyn A.
      • Tully R.E.
      • Proctor G.
      • Chen Y.
      • McLaren W.M.
      • Larsson P.
      • Vaughan B.W.
      • Béroud C.
      • Dobson G.
      • Lehväslaiho H.
      • Taschner P.E.M.
      • den Dunnen J.T.
      • Devereau A.
      • Birney E.
      • Brookes A.J.
      • Maglott D.R.
      Locus Reference Genomic sequences: an improved basis for describing human DNA variants.
      There is an ongoing collaborative initiative, Matched Annotation from National Center for Biotechnology Information and EMBL-EBI (MANE), that is aimed at harmonizing the transcript definition across databases and versions. However, the harmonization process is still ongoing for several clinically important genes (MANE, https://www.ncbi.nlm.nih.gov/refseq/MANE, last accessed September 14, 2021). Consequently, this group recognizes the additional need for consensus agreement on which specific transcripts to use for clinical patient care, although for genes with variable splicing and transcriptional alteration, comparison of different transcripts may need to be reported. In the absence of such agreement, the inclusion of transcript accession and version information is critical to prevent misinterpretation or medical error.
      A more structured and standard definition of what constitutes a full variant description is needed, such that the variant could be separated from its report, viewed independently in another piece of software, and understood in the clinical context of the patient. The minimum elements required for full variant description would need to be defined, and these data elements would need to be unambiguous and consistent. This could be difficult to define for complex variants and pharmacogenetic variants, where the name and annotation of haplotypes and/or diplotypes can impact the clinical interpretation, but nonetheless it is a critical component to be able to reconcile variants between test results performed at different laboratories on the same patient and same or related sample types. Such standardization would help ensure consistency in how variants are reported between laboratories and tests producing overlapping assays.
      Another challenge in getting genomic data into EHRs is the lack of a standard genomics data model to facilitate the ability of a laboratory to send discrete data to multiple different EHRs with minimal to no revision of their data feeds. Genomics modules are being developed within EHRs, but they often lack public documentation of their schema, thus necessitating custom interfaces with each different EHR via many trial-and-error attempts. This is not only an interface maintenance nightmare but is at risk for erroneous mapping across interfaces, which could lead to loss of formatting a readability at best and lost or wrong data reaching the patient at worst. By contrast, most EHRs accept textual and/or PDF reports, and this is probably one of the biggest reasons why this is the most common genomic result format. Text reports, including PDF and/or rich text format files, have tremendous benefit for clinical genomics, because they may allow formatting, such as italics for gene symbols and clickable hyperlinks to external websites as well as tables and other ways to organize and present the data for readability. However, EHRs and LISs should also allow for the parallel transmission of discrete data as a solution to the separate problems described in this report. Furthermore, decisions on how to standardize the linking of text reports and discrete genomic results to nonmolecular reports on the same specimen source, such as microbiology and anatomic pathology, would need to be made.
      Summary: Transfer and storage of standardized variant data structures from the LIS to the EHR is the key enabling feature for future use of genomic data in the EHR. The structure should be a standard between internal and external laboratory sources of genomic data.

       Display of Genomic Test Results: Providing Sufficient yet Usable Data

      The formats used for genomic test reports are highly variable between laboratories and sometimes even within a single laboratory. Some variations in report formats may be due to technical limitations in either the LIS and/or EHR (eg, inability to italicize gene symbols, inability to display anything other than American Standard Code for Information Interchange (ASCII) text, or lack of tabular format), which can generate further intrareport variability between systems. EHRs also may lack the ability to link a genomic report with relevant other clinical history, such as family history or pedigrees, or with reports on the same specimen on which genomic testing was performed, such as microbiologic cultures, surgical pathology reports, and hematopathology reports. Despite the availability of several published consensus standards from professional societies on the format and content of molecular and genomic reports,
      • Richards S.
      • Aziz N.
      • Bale S.
      • Bick D.
      • Das S.
      • Gastier-Foster J.
      • Grody W.W.
      • Hegde M.
      • Lyon E.
      • Spector E.
      • Voelkerding K.
      • Rehm H.L.
      Standards and guidelines for the interpretation of sequence variants: a joint consensus recommendation of the American College of Medical Genetics and Genomics and the Association for Molecular Pathology.
      • Li M.M.
      • Datto M.
      • Duncavage E.J.
      • Kulkarni S.
      • Lindeman N.I.
      • Roy S.
      • Tsimberidou A.M.
      • Vnencak-Jones C.L.
      • Wolff D.J.
      • Younes A.
      • Nikiforova M.N.
      Standards and guidelines for the interpretation and reporting of sequence variants in cancer: a joint consensus recommendation of the Association for Molecular Pathology, American Society of Clinical Oncology, and College of American Pathologists.
      • Gulley M.L.
      • Braziel R.M.
      • Halling K.C.
      • Hsi E.D.
      • Kant J.A.
      • Nikiforova M.N.
      • Nowak J.A.
      • Ogino S.
      • Oliveira A.
      • Polesky H.F.
      • Silverman L.
      • Tubbs R.R.
      • Van Deerlin V.M.
      • Vance G.H.
      • Versalovic J.
      Clinical laboratory reports in molecular pathology.
      • Ogino S.
      • Gulley M.L.
      • den Dunnen J.T.
      • Wilson R.B.
      • Payne D.
      • Lowery Nordberg M.C.
      • Gong J.Z.
      • Krafft A.E.
      • Uphoff T.S.
      • Donahue P.
      • Hunt J.
      • Garrison G.
      Standard mutation nomeclature in molecular diagnostics: practical and educational challenges.
      laboratory report formats continue to be as unique as each individual human's genome, making them difficult for many health care providers and patients to read and interpret.
      • Haga S.B.
      • Kim E.
      • Myers R.A.
      • Ginsburg G.S.
      Primary care physicians’ knowledge, attitudes, and experience with personal genetic testing.
      • Nightingale B.M.
      • Hovick S.R.
      • Brock P.
      • Callahan E.
      • Jordan E.
      • Roggenbuck J.
      • Sturm A.C.
      • Morales A.
      Hypertrophic cardiomyopathy genetic test reports: a qualitative study of patient understanding of uninformative genetic test results.
      • Macklin S.K.
      • Jackson J.L.
      • Atwal P.S.
      • Hines S.L.
      Physician interpretation of variants of uncertain significance.
      • Recchia G.
      • Chiappi A.
      • Chandratillake G.
      • Raymond L.
      • Freeman A.L.J.
      Creating genetic reports that are understood by nonspecialists: a case study.
      • Farmer G.D.
      • Gray H.
      • Chandratillake G.
      • Raymond F.L.
      • Freeman A.L.J.
      Recommendations for designing genetic test reports to be understood by patients and non-specialists.
      The College of American Pathologists requires correlation of pathology report with molecular findings (College of American Pathologists, https://www.cap.org/laboratory-improvement/accreditation/accreditation-checklists, last accessed January 21, 2021). Many laboratories are starting to incorporate clinical trial information into textual reports, but in some systems, it takes some degree of manual effort for health care providers to get additional information about the specific trials referenced. Having hyperlinks to clinical trials in the report and based on the patient's genomic parameters would help facilitate this review, especially because eligibility for clinical trials is limited and because the list of clinical trials is constantly changing. For pharmacogenetic results, links to appropriate databases with more information for health care providers and pharmacists about the meaning of certain variants in the context of the patient's overall medication history and current regimen would help in determining alternative strategies when necessary.
      Pathologists and clinical molecular geneticists may focus on ensuring that all necessary content is available in the report, but this information is not always presented in a manner that the end user can read or use. Usability is a term that is commonly used in the clinical informatics community and refers to the effectiveness, efficiency, and satisfaction with which users achieve goals in a particular environment
      • Zahabi M.
      • Kaber D.B.
      • Swangnetr M.
      Usability and safety in electronic medical records interface design: a review of recent literature and guideline formulation.
      (Healthcare Information and Management Systems Society, https://www.himss.org/resources/what-user-experience-healthcare-it; National Institute of Standards and Technology, https://www.nist.gov/programs-projects/health-it-usability, both last accessed December 28, 2020). Multiple studies have documented that health care providers find genomic reports difficult to read and use,
      • Haga S.B.
      • Kim E.
      • Myers R.A.
      • Ginsburg G.S.
      Primary care physicians’ knowledge, attitudes, and experience with personal genetic testing.
      • Nightingale B.M.
      • Hovick S.R.
      • Brock P.
      • Callahan E.
      • Jordan E.
      • Roggenbuck J.
      • Sturm A.C.
      • Morales A.
      Hypertrophic cardiomyopathy genetic test reports: a qualitative study of patient understanding of uninformative genetic test results.
      • Macklin S.K.
      • Jackson J.L.
      • Atwal P.S.
      • Hines S.L.
      Physician interpretation of variants of uncertain significance.
      • Recchia G.
      • Chiappi A.
      • Chandratillake G.
      • Raymond L.
      • Freeman A.L.J.
      Creating genetic reports that are understood by nonspecialists: a case study.
      • Farmer G.D.
      • Gray H.
      • Chandratillake G.
      • Raymond F.L.
      • Freeman A.L.J.
      Recommendations for designing genetic test reports to be understood by patients and non-specialists.
      which may be due to lack of expertise in genomics, reports crafted without good usability principles, or both. To date, there are few articles focused on the usability of genomic reports.
      • Solomon I.B.
      • McGraw S.
      • Shen J.
      • Albayrak A.
      • Alterovitz G.
      • Davies M.
      • Del Vecchio Fitz C.
      • Freedman R.A.
      • Lopez L.N.
      • Sholl L.M.
      • Van Allen E.
      • Mortimer J.
      • Fakih M.
      • Pal S.
      • Reckamp K.L.
      • Yuan Y.
      • Gray S.W.
      Engaging patients in precision oncology: development and usability of a web-based patient-facing genomic sequencing report.
      • Gray S.W.
      • Gagan J.
      • Cerami E.
      • Cronin A.M.
      • Uno H.
      • Oliver N.
      • Lowenstein C.
      • Lederman R.
      • Revette A.
      • Suarez A.
      • Lee C.
      • Bryan J.
      • Sholl L.
      • Van Allen E.M.
      Interactive or static reports to guide clinical interpretation of cancer genomics.
      • Westendorf L.
      • Shaer O.
      • Pollalis C.
      • Verish C.
      • Nov O.
      • Ball M.P.
      Exploring genetic data across individuals: design and evaluation of a novel comparative report tool.
      • Marsolo K.
      • Spooner S.A.
      Clinical genomics in the world of the electronic health record.
      Patient groups are interested in standardization and usability of their medical data, and the Association for Molecular Pathology and other professional societies are involved in these efforts (Common Cancer Testing Terminology, https://www.commoncancertestingterms.org, last accessed April 6, 2021). More research is definitely needed in this area.
      Summary: Genomic reports between laboratories are variable, and even when they include complete information, they may be difficult for providers and patients to understand, due to either lack of genomic expertise or usability or both.

       Display of Aggregated Results across Different Tests

      Patients may have multiple genetic tests over time. It is common for such tests to contain overlapping targets; also, they are performed at different laboratories using different methods with different sensitivities and specificities for variant types and disease conditions. Each laboratory test report is most often a stand-alone text document or PDF, and most if not all EHRs lack the capability to compare and contrast variants across different laboratories and test panels. Not only does this make it difficult for providers to see a more global view of test results for the patient (alias combined or integrated displays of results), but this also makes it difficult to determine why two different laboratories or test panels would have obtained different variants for the same gene. Although these discordances are most commonly due to differences in the scope of testing performed (panel gene or gene segments included) or, in the case of cancer testing, the tissue type, amount of tumor, intratumoral heterogeneity, tumor progression, or interferences from clonal hematopoiesis of indeterminate potential,
      • Jensen K.
      • Konnick E.Q.
      • Schweizer M.T.
      • Sokolova A.O.
      • Grivas P.
      • Cheng H.H.
      • Klemfuss N.M.
      • Beightol M.
      • Yu E.Y.
      • Nelson P.S.
      • Montgomery B.
      • Pritchard C.C.
      Association of clonal hematopoiesis in DNA repair genes with prostate cancer plasma cell-free DNA testing interference.
      it can also be due to true discrepancies resulting from false positives and false negatives, which are important to detect and resolve before acting on the information. Because the presence or absence of a particular variant is highly dependent on the scope of testing, context, false results, or errors, alerting providers to conflicting results in an EHR will be challenging and, if poorly implemented, could be unnecessarily alarming and cause delays in care. Another challenge is that, for variants that do not have unequivocally established clinical significance, different interpreters may have different interpretations of the clinical significance of the same variant.
      Amalgamating a patient's variants into a single view is a laudable goal, but it must account for several important factors to be useful and safe without being misleading or confusing. First and most important, any variant present in an aggregated view must be linked to the interpretative text associated with that variant in its source report. This is unrealistic until the current text-based reports have some underlying structure to facilitate accurate links. There are commercial companies and other endeavors who market their ability to scrape data from reports or to request text reports in an eXtensible Markup Language (XML; W3C, https://www.w3.org/XML, last accessed January 21, 2021) format in an effort to aid in clinical decision support in EHRs. However, the full consequences of this need to be assessed. Attempts to use optical character recognition combined with natural language processing and even artificial intelligence without adequate oversight and validation will likely generate inaccurate associations that could result in patient safety issues. Second, providers should be able to filter and compare the list of the patient's variants on the basis of the specimen source and the date of specimen collection. Not only does this help with the analysis of a changing variant profile that can be seen with somatic variants in cancer, but it is also important in cases of mosaicism, whether such mosaicism occurs constitutionally as a result of genetic variation in early development (eg, somatic overgrowth syndromes) or iatrogenically from bone marrow transplant or another procedure. Third, discrete variant data are easily linked to online variant knowledge bases and repositories to assist with clinical decision support. However, the curation and, consequently, the accuracy of such databases vary, and changes can occur at any time (see below). These databases may include variants associated with a particular phenotype, pathogenicity, or pharmacologic impact for which insufficient data exist to make these determinations. Similarly, variants may be listed as pathogenic, which are actually benign polymorphisms, and vice versa, and the risk of these is higher with rare population-specific and/or private variants. Polymorphisms that have possible pharmacogenetic impact may not be identified as such in a database that is focused on cancer or inherited disease. Similarly, in the absence of clinical data, such as medication history, relevance of these findings may not be noticed. Links to possible available clinical trials for specific variants, genes, and conditions are also useful but again may be misleading, depending on how the links are generated and the limited information available in the clinical trial database (eg, links to adult clinical trials may not be useful for pediatric patients). Some of these databases are freely available, whereas others require a membership/subscription. Some databases may not be up to date or adequately maintained, and it is also possible that some databases will become obsolete because of lack of funding. EHRs and health care organizations are urged some degree of caution in determining which variant-specific links will be provided from within the EHR. Fourth, anytime the number of variants that come into an amalgamated view increases to the point where a provider cannot focus, requests for being able to apply electronic filters will soon follow. Should an EHR decide to allow variant data to be filtered on the basis of certain criteria, then the EHR should ensure that i) it is obvious that filters have been applied, ii) each filter applied is specifically listed, and iii) the end user can turn off each filter on demand. Finally, but important, aggregating variants across different molecular test results for a patient can only be possible if the variants can be accurately cross-referenced or looked up, necessitating the need for standardized variant representation, as described in the prior section.
      As with any laboratory test, EHRs must also be able to accommodate addended or corrected results for a previously reported sample. For clinical and medicolegal reasons, EHRs and LISs will supersede the most recent report with the corrected/amended report while ensuring that the prior version of the report is still accessible for review. Such amendments or corrections can occur when new information is received (eg, the scope of testing was expanded after new clinical history was received), when a variant is reclassified on the basis of new scientific data (see section below), or when a pre-analytic, analytic, or postanalytic error occurs. Providers must be able to easily determine when information in an amalgamated view is new (addended) or corrected, including the ability to see when a variant has been downgraded in interpretation as well as upgraded. This presents some challenges for laboratories as most if not all laboratories do not report benign variants that lack pharmacogenetic impact as well as challenges for those laboratories that do not routinely report variants of unknown significance. Providers must also be able to easily get to the source of the variant information, including the laboratory, testing platform, level of sensitivity of the assay, and variant types in scope of the assay.
      Professional societies in molecular pathology and genomics have released multiorganizational consensus guidelines for variant nomenclature, hierarchical result structure from most clinically significant to least, and genomic report formats.
      • Richards S.
      • Aziz N.
      • Bale S.
      • Bick D.
      • Das S.
      • Gastier-Foster J.
      • Grody W.W.
      • Hegde M.
      • Lyon E.
      • Spector E.
      • Voelkerding K.
      • Rehm H.L.
      Standards and guidelines for the interpretation of sequence variants: a joint consensus recommendation of the American College of Medical Genetics and Genomics and the Association for Molecular Pathology.
      ,
      • Li M.M.
      • Datto M.
      • Duncavage E.J.
      • Kulkarni S.
      • Lindeman N.I.
      • Roy S.
      • Tsimberidou A.M.
      • Vnencak-Jones C.L.
      • Wolff D.J.
      • Younes A.
      • Nikiforova M.N.
      Standards and guidelines for the interpretation and reporting of sequence variants in cancer: a joint consensus recommendation of the Association for Molecular Pathology, American Society of Clinical Oncology, and College of American Pathologists.
      However, implementation of these guidelines has been modest. This has had the unfortunate result of a wide variety of reporting formats with an even wider variety of variant nomenclature and the actual data included. For example, some cancer genomic reports subtract out variants suspected or confirmed to be germline in origin and may not make it clear that such subtraction has occurred. This can cause difficulty if one or more of the germline variants is associated with a hereditary cancer syndrome and/or the variant is also somatically acquired in the other allele. Algorithms to subtract germline variants may lack the sophistication required to detect this. The examples go on, with variation in how laboratories address clinical sensitivity, analytical sensitivity, detectable lengths of insertions/deletions, the copy number variants that the assay can detect, the scope of testing, and more.
      Discrete variant data also lend themselves to being uploaded into crowd-sourced knowledge bases. However, EHRs and health care organizations should not permit such activities to occur from within an EHR, as this is best left in the hands of the laboratory that produced the data. Similarly, laboratories should use caution in uploading sequences from patients who have not given informed consent to have their data shared because genetic information is classified as protected health information in the United States and may also be protected in other countries (US Department of Health & Human Services, https://www.hhs.gov/hipaa/for-professionals/faq/354/does-hipaa-protect-genetic-information/index.html, last accessed January 21, 2021).
      Given the ongoing severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) pandemic and the recent detection of new and highly contagious SARS-CoV-2 variants, the need surrounding reporting of the genomics of infectious organisms and agents has never been more important. Although the working group did not focus on nonhuman nucleic acid testing, many of the items discussed in this workflow would reasonably apply to both human and nonhuman nucleic acid test results in EHRs. However, there are several distinct additional functions that are critical for the use of genomic data from nonhuman organisms and agents in EHRs. First, microbial genomes are less likely to have international standards for use and, as has been shown with SARS-CoV-2, new variants can arise quickly. A more standardized and rigorous system of reference microbial genomes is needed, which can rapidly adapt to the development of new strains. In addition, genetic testing for microorganisms requires integration with infection prevention and infectious disease modules in EHRs to enable infection prevention monitoring and early intervention, as well as contact tracing. Federal reporting requirements for SARS-CoV-2 during the pandemic have exposed serious limitations in the public health infrastructure of the United States. Each state has its own requirements for which test results to report and can vary significantly on the mechanism of reporting (fax, manual data entry into an online portal, secure file transfer protocol, or electronic laboratory reporting using HL7). This generates barriers for EHRs, LISs, and other systems that need to report to multiple states.
      Requirements also exist related to structuring genomic data by the existing taxonomy of microorganisms, while still allowing flexibility for new and renamed organisms as well as changes to taxonomy structure. Another important difference from human genome analysis is that instead of one reference genome used for humans, there are as many reference genomes for microorganisms as there are microorganisms. Like humans, this complexity is further increased because there can be multiple different transcripts for the same genomic region of a single organism, thereby making cross comparison of the results from various microbial genomic assays challenging. When compounded with the fact that these assays may be repeated on the same patient more frequently than a germline or even somatic cancer assay, the need to compare results of different assays for the same patient, as well as across patients, becomes more critical and challenging at the same time. It is the hope of this working group that a group with expertise in microbial testing, bioinformatics, and clinical informatics will investigate issues with reporting genomic data to EHRs for microorganisms.
      Genomic results should also be integrated with results from other forms of pathology and laboratory tests within the larger context (ie, big picture) of an EHR. One of common issues is discrepant nomenclature of genes and gene products in pathology reports.
      • Fujiyoshi K.
      • Bruford E.A.
      • Mroz P.
      • Sims C.L.
      • O’Leary T.J.
      • Lo A.W.I.
      • Chen N.
      • Patel N.R.
      • Patel K.P.
      • Seliger B.
      • Song M.
      • Monzon F.A.
      • Carter A.B.
      • Gulley M.L.
      • Mockus S.M.
      • Phung T.L.
      • Feilotter H.
      • Williams H.E.
      • Ogino S.
      Standardizing gene product nomenclature-a call to action.
      This is best exemplified by the NKX2-1 gene, whose gene product is nearly ubiquitously described as TTF-1 (thyroid transcription factor-1) in surgical pathology reports, whereas there exists the completely unrelated gene TTF1 (official name, transcription termination factor 1). As a consequence, a given patient may have results of TTF1 two-copy deletion (or large truncation mutations) and TTF-1 overexpression in the same tumor. To avoid ambiguity and confusion, it is recommended that EHRs standardize their databases such that each discrete gene in the database be referenced by its Human Genome Organisation Gene Nomenclature Committee (HGNC)–approved symbol and that each gene-associated protein be named with use of HGNC-approved symbol followed by protein at a minimum. Additional columns to reference the full descriptive names of each as well as the HGNC identifier are also recommended [eg, NKX2-1 protein (TTF-1) or NKX2-1 protein (HGNC:11825, TTF-1)]. Having said this, it is important to note that HGNC gene symbols as well as HGVS nomenclature are updated frequently. Although HGVS nomenclature is versioned, HGNC gene nomenclature is not, and individual gene symbols can change without notice. Having defined release cycles for HGNC and HGVS nomenclature updates as well as timestamped versioning for HGNC gene symbols would help EHRs keep reference tables appropriately up to date.
      Summary: The EHR should provide functionality for clear longitudinal display of data and a way to compare gene and gene region content for each NGS panel. A tiered approach, presenting the complexity of genomic data with a dynamic and integrated report, is the preferred approach. Such integrated reports should include hyperlinks to additional curated content for specific variants.

       Display of Reclassified Variants

      There are at least four discrete rationales for reclassification of a variant in the EHR. First, an error may have been discovered, which requires an amendment of the report. Second, additional previously unknown clinical history or other information may become available that impacts the interpretation of the variant (eg, all affected but no unaffected family members have the variant previously classified as a variant of unknown significance, thus altering the bayesian risk level for the proband). Third, a provider may request a reinterpretation of a specific variant or of an entire report. This is not uncommon after a patient is seen by medical genetics and/or a genetic counselor, or when a patient has progressive disease when such progression was unexpected (eg, cancer). Fourth, the laboratory may engage in a systematic process-based re-examination. Such re-examinations may occur at regular intervals (eg, re-examination 4 to 6 months or year after the initial report is issued) or may be triggered at the time when significant new variant information, gene information, drugs, or clinical trials become available that would affect therapy options.
      Because information about the significance of genomic variants is changing at breakneck speed, yesterday's variant of unknown significance may be tomorrow's therapeutically actionable variant. Some laboratories already have processes whereby all previously reported results are reclassified when new information is discovered, without being requested by the health care provider (ie, systemic reclassification), and most, if not all, laboratories have a process whereby an individual report may be re-examined for possible reclassification following an external trigger, such as physician request (ie, trigger-based reclassification).
      Another rationale for regular reinterpretation is that EHRs can now easily allow end users to link to outside databases for every reported variant. An end user could find data in one of these outside databases that does not match with the professional interpretation provided by the laboratory. In addition to the high degree of variability in the accuracy of such databases, there is also variability in the published literature regarding variants, phenotypes, and conditions. Interpreting genetic and genomic results is a skill that requires expertise, knowledge of the details of the test being performed, as well as clinical data. If an EHR allowed changes to be made to laboratory-generated data or reports that did not originate from the laboratory, there could be significant consequences to patient safety, especially if a report or variant interpretation is changed from accurate to inaccurate. It can also unfairly jeopardize the trust that clinicians place in the performing laboratory and have adverse consequences for federal, state, or local laboratory compliance and accreditation. Laboratories are responsible for ensuring that test reports and data reach the test report destination accurately and completely, and they cannot do this if the results are changed by entities outside the performing laboratory. Federal Clinical Laboratory Improvement Amendments requirements for amending laboratory results may also not be followed. Last, such changes have been known to prohibit or cancel subsequent amended results generated by the performing laboratory from posting to the EHR because of technical configuration issues in LISs and EHRs, which misunderstood an amended result from the laboratory as not superseding a change within the EHR. Instead, EHRs may provide a mechanism to contact the source laboratory to either place orders for a professional reinterpretation or inquire about the interpretation of a specific variant. Professional review of any reinterpretation is essential because the need to consider the full clinical context for every revised call precludes automated reinterpretation. Reinterpretation is as important as the original interpretation, and sign-out of the report, especially given that life-and-death decisions, family planning decisions, and decisions about chemotherapy, bone marrow transplantation, and other therapies with a high risk of toxicity and complications are being made on the basis of the results.
      When a variant is reclassified, an updated report from the laboratory is required, and the ordering provider should be notified according to laboratory policy. Although most pathologists and laboratory professionals who sign out genomic reports have an aversion to the word amendment because of the medicolegal implications, in clinical informatics and in terms of HL7 interfaces, any update to a report that includes changes to the prior interpretation is technically an amendment, even when the later version of the report is incorporating information that could not have possibly been known at the time the report was initially signed out. When the amended report is re-issued to the EHR, any changes must be readily visible, including the date and reason for the change.
      Any variant that was previously reported to the EHR, but which has subsequently been reclassified, must be sent again to the EHR with the date, updated classification, and reason for the change. It is unsafe for a variant to disappear from view because it was changed into a class that the laboratory would not routinely send to the EHR (eg, variant downgraded from variant of unknown significance to a benign polymorphism). Some laboratories may prefer to issue a completely new report with a new case number for reclassifications for billing reasons, as well as to avoid amending the original report. However, it is problematic from a clinical perspective if the original report is still visible in the EHR and there is no indication that a reclassification has been performed and is available elsewhere.
      There are some complications to reclassification when considering how to display variants in an EHR. Provider-requested reclassification may result in a new report rather than an amended report from some laboratories. This is especially true if additional genes were interrogated or requested as part of the analysis. Clinical Laboratory Improvement Amendments has requirements for amendments and corrections that must be followed in the United States, and it follows that these requirements must be met in an amalgamated view as well as for the individual report for safety reasons [Centers for Medicare & Medicaid Services Standards and Certification: Laboratory Requirements (42 CFR 493), https://www.ecfr.gov/cgi-bin/text-idx?SID=1248e3189da5e5f936e55315402bc38b&node=pt42.5.493&rgn=div5, last accessed January 22, 2021]. Variants may be reclassified as somatic instead of germline and vice versa, and EHR displays must make such changes obvious because of their clinical significance. This stresses the importance of making it obvious whether paired tumor-normal sequencing has been done in cancer genomic testing.
      There are billing and financial implications of reclassification that are beyond the scope of this article. There is a weakly supported Current Procedural Terminology code for reinterpretation, which has marginal reimbursement. Having said this, orders for reclassification should be clear as to the reason why the reclassification has been ordered.
      Summary: Consensus guidelines on best practice for requesting and providing reclassification of variants is a top priority. The different rationales for reclassification and their impact on clinical management and laboratory workflows need to be carefully considered.

       Display of Genomic Results to Patients

      The Health Information Technology for Economic and Clinical Health act of 2009 (US Department of Health & Human Services, https://www.hhs.gov/sites/default/files/ocr/privacy/hipaa/understanding/coveredentities/hitechact.pdf, last accessed December 29, 2020) put meaningful use and subsequently promoting interoperability into place. This federal law required that laboratory data and other health information be sent to patient portals so the patients could access their data. Because of the sensitivity of genomic results and/or the presence of more restrictive state laws, many health care organizations either elected not to send genetic information to patient portals or required providers to manually release the results to the portal. The complexity of genomic reports is another factor to consider given that there have now been multiple studies showing that many physicians lacking experience in genetics are not able to adequately interpret the results.
      • Haga S.B.
      • Kim E.
      • Myers R.A.
      • Ginsburg G.S.
      Primary care physicians’ knowledge, attitudes, and experience with personal genetic testing.
      • Nightingale B.M.
      • Hovick S.R.
      • Brock P.
      • Callahan E.
      • Jordan E.
      • Roggenbuck J.
      • Sturm A.C.
      • Morales A.
      Hypertrophic cardiomyopathy genetic test reports: a qualitative study of patient understanding of uninformative genetic test results.
      • Macklin S.K.
      • Jackson J.L.
      • Atwal P.S.
      • Hines S.L.
      Physician interpretation of variants of uncertain significance.
      • Recchia G.
      • Chiappi A.
      • Chandratillake G.
      • Raymond L.
      • Freeman A.L.J.
      Creating genetic reports that are understood by nonspecialists: a case study.
      • Farmer G.D.
      • Gray H.
      • Chandratillake G.
      • Raymond F.L.
      • Freeman A.L.J.
      Recommendations for designing genetic test reports to be understood by patients and non-specialists.
      Therefore, these reports are not easy for most patients to understand, either; anecdotal reports from consumer genetics services illustrate this issue.
      In June 2020, the 21st Century Cures Act (National Archives and Records Administration, https://www.federalregister.gov/documents/2020/05/01/2020-07419/21st-century-cures-act-interoperability-information-blocking-and-the-onc-health-it-certification, last accessed December 8, 2020) became effective and greatly expanded the mandate to release laboratory and other test results to patients, essentially making it illegal for health care organizations to withhold health information from patients and patient portals unless the health care organization could demonstrate that the results put the patient at risk for harm (eg, releasing genetic test results for osteogenesis imperfecta that include suspicion of nonaccidental trauma in the clinical history to a child's patient portal for the suspected caregiver to see). Currently, there appears to be some variability in the interpretation on the allowable length of a delay between the release of genomic and other sensitive laboratory and pathology results by the laboratory to the health care provider and the release to the portal or to a patient, and compliance with both federal and state law in states that require genetic counseling for genetic testing results (eg, Georgia) will be difficult or impossible. Regardless, this new federal law on information blocking, which requires almost all results, including genomic and genetic results, to be released immediately to the patient on verification as of May 2021, may reignite conversations about how to generate patient-friendly genomic reports, but it is uncertain how this will occur without increasing the burden on laboratories.
      Summary: There are distinct regulatory requirements and information needs for patient access to genomic results, as compared with providers, that should be supported in the EHR.

       Clinical Decision Support Rules

      Previously, CDS has been mentioned with regard to giving providers easy access to information about discrete variants listed in an EHR. However, there is a whole other area of CDS in EHRs that is used widely for drug-drug alerts and drug-allergy alerts. Drug-variant alerts are used in a limited number of health care organizations, typically highly specialized instances that are not scalable. However, once a discrete variant data standard is made available and used, it is not unreasonable to assume that drug-genome alerts will be far more widespread. The potential use cases for properly implemented drug-genome alerts include, but are not limited to, alerts to physicians when prescribing chemotherapy that has been shown to be ineffective for tumors with a particular variant profile, medication that is contraindicated in certain genetic conditions (eg, inborn errors of metabolism), and medication being prescribed to a patient with pharmacogenetic variants severely affecting the metabolism of the drug. As with any clinical decision support rule, they should be passive (ie, allow the provider to make their own decision), be transparent, and contain links to the original source of the medical evidence supporting the rule. Obviously, clinical decision support rule should only be running on discrete variants that are accurate and current, especially in the case of amendments, reclassification, or tumor progression.
      However, without national and international standards for such CDS, each health care organization will be left to generating its own rules, much in the same way that sepsis algorithms are being developed currently, which may lead to discrepancies and unanticipated errors.
      • Schinkel M.
      • Paranjape K.
      • Nannan Panday R.S.
      • Skyttberg N.
      • Nanayakkara P.W.B.
      Clinical applications of artificial intelligence in sepsis: a narrative review.
      Similarly, several third-party commercial vendors have developed CDS tools that integrate with EHRs, LISs, and/or raw NGS data; however, because of lack of standards, the solutions are inconsistent in their implementation, increasing risk for divergent outcomes.
      Summary: Discrete genomic data will facilitate scalable drug-genome alerts and other forms of CDS, and international recommendations for CDS are needed for safety and consistency in practice.

      Interoperability

       Cross-Functional Requirements and Standards

      To realize the full benefits of genomic data in the EHR, more standards must be developed that use existing standards as the foundation where possible and appropriate. There are already international standards for variant nomenclature and for gene symbols
      • den Dunnen J.T.
      • Dalgleish R.
      • Maglott D.R.
      • Hart R.K.
      • Greenblatt M.S.
      • Mcgowan-Jordan J.
      • Roux A.F.
      • Smith T.
      • Antonarakis S.E.
      • Taschner P.E.M.
      HGVS recommendations for the description of sequence variants: 2016 update.
      (Human Genome Organisation Gene Nomenclature Committee: Resource for Approved Human Gene Nomenclature, https://www.genenames.org, last accessed December 28, 2020). The American College of Medical Genetics and Genomics and Association for Molecular Pathology have developed standards and guidelines for categorization of germline variants, including copy number variants,
      • Richards C.S.
      • Bale S.
      • Bellissimo D.B.
      • Das S.
      • Grody W.W.
      • Hegde M.R.
      • Lyon E.
      • Ward B.E.
      ACMG recommendations for standards for interpretation and reporting of sequence variations: revisions 2007.
      ,
      • Riggs E.R.
      • Andersen E.F.
      • Cherry A.M.
      • Kantarci S.
      • Kearney H.
      • Patel A.
      • Raca G.
      • Ritter D.I.
      • South S.T.
      • Thorland E.C.
      • Pineda-Alvarez D.
      • Aradhya S.
      • Martin C.L.
      on behalf of the ACMG
      Technical standards for the interpretation and reporting of constitutional copy-number variants: a joint consensus recommendation of the American College of Medical Genetics and Genomics (ACMG) and the Clinical Genome Resource (ClinGen).
      and the Association for Molecular Pathology, College of American Pathologists, and American Society for Clinical Oncology have both guidelines for somatic variant classification and guidelines for reporting genomic variants in cancer.
      • Richards S.
      • Aziz N.
      • Bale S.
      • Bick D.
      • Das S.
      • Gastier-Foster J.
      • Grody W.W.
      • Hegde M.
      • Lyon E.
      • Spector E.
      • Voelkerding K.
      • Rehm H.L.
      Standards and guidelines for the interpretation of sequence variants: a joint consensus recommendation of the American College of Medical Genetics and Genomics and the Association for Molecular Pathology.
      ,
      • Riggs E.R.
      • Andersen E.F.
      • Cherry A.M.
      • Kantarci S.
      • Kearney H.
      • Patel A.
      • Raca G.
      • Ritter D.I.
      • South S.T.
      • Thorland E.C.
      • Pineda-Alvarez D.
      • Aradhya S.
      • Martin C.L.
      on behalf of the ACMG
      Technical standards for the interpretation and reporting of constitutional copy-number variants: a joint consensus recommendation of the American College of Medical Genetics and Genomics (ACMG) and the Clinical Genome Resource (ClinGen).
      HL7 has standards for data transfer of health information, but the HL7 Genomics Working Group standard is still maturing and not yet ready for use given its complexity and dependence on LOINC codes (HL7 International, http://www.hl7.org/special/committees/clingenomics, last accessed December 8, 2020).
      Summary: Effective use of genomic data in EHRs depends on integration of appropriate, safe, and functional international standards for interoperability and data retrieval.

       Reporting Genomic Results to Outside Organizations

      Discrete variant data in EHRs may also help facilitate reporting of results to outside organizations and agencies. The coronavirus disease 2019 (COVID-19) pandemic brought about huge changes in public health reporting requirements with which laboratories have had to comply on short notice. It is reasonable to assume that reporting of laboratory test data in general to public health agencies will see significant changes in the post-pandemic era. Release to public health agencies is typically governed by federal and state laws and often does not face the same restrictions as reporting results to other organizations.
      Release of discrete variant data to outside health care organizations that are taking care of the patient requires informed consent from the patient or legal guardian, and release to research studies or crowd-sourced clinical repositories requires institutional review board review and/or informed consent from the patient or the patient's legally authorized caregivers unless a legal data use agreement is in place. Currently, textual data from genomic reports are exported and released to authorized parties, and analysis of such data is limited by their highly variable format and lack of structure. However, once a discrete variant data standard is developed and adopted, more standard extract, transform, and load procedures may be used by organizations to rapidly upload the data into their own databases for examination. EHRs are starting to support APIs and the HL7 FHIR standard, which may help facilitate not only the export but also the import of genomic data from another organization. As previously stated, there are multiple data elements that would have to be present and available on each variant to ensure that important information, including required Clinical Laboratory Improvement Amendments test report information, such as the laboratory that generated the data, is easily available to the providers taking care of the patient. EHRs may decide to be able to export data via a Consolidated Clinical Document Architecture (Office of the National Coordinator for Health Information Technology, https://www.healthit.gov/sites/default/files/c-cda_and_meaningfulusecertification.pdf; https://www.healthit.gov/topic/standards-technology/consolidated-cda-overview, both last accessed December 8, 2020).
      Easy transfer of data brings about many conveniences, but also risk. As with any transfer of sensitive information, such as health information, cybersecurity measures are important. Health care organizations and EHR vendors must pay close attention to any third-party software involved in the transfer of information, including any risk associated with the vulnerability of open-source APIs and proprietary APIs using outdated and nonsupported operating systems and other components. Patients requesting that their data be exported to a third party should be made aware of the risks as data become more portable. The integrity of the data from sender to recipient is also incredibly important, especially for variant data files that have been known to be truncated in transit.
      Summary: Discrete genomic data with appropriate interpretation attributes will facilitate interoperability between systems and organizations. Patient education and informed consent as well as cybersecurity measures are paramount to keeping patients informed and their data safe from unauthorized access.

      Conclusion

      In summary, there are many challenges that lie ahead on the road to truly interoperable genomic data in EHRs, as described in this article in detail and referred to at a higher level in Figure 1. The purpose of this article is to outline some of the major challenges identified by this working group and to educate readers about the complexities of genomic data and risks associated with improper display, format, or use.

      Disclaimer

      The Association for Molecular Pathology (AMP) Perspectives are developed to be of assistance to laboratory and other health care professionals by providing guidance and recommendations for particular areas of practice. The Perspectives should not be considered inclusive of all proper approaches or methods, or exclusive of others. The Perspectives cannot guarantee any specific outcome, nor do they establish a standard of care. The Perspectives are not intended to dictate the treatment of a particular patient. Treatment decisions must be made on the basis of the independent judgment of health care providers and each patient's individual circumstances. The AMP makes no warranty, express or implied, regarding the Perspectives and specifically excludes any warranties of merchantability and fitness for a particular use or purpose. The AMP shall not be liable for direct, indirect, special, incidental, or consequential damages related to the use of the information contained herein.

      References

        • Wiesner G.L.
        • Rahm A.K.
        • Appelbaum P.
        • Aufox S.
        • Bland S.T.
        • Blout C.L.
        • Christensen K.D.
        • Chung W.K.
        • Clayton E.W.
        • Green R.C.
        • Harr M.H.
        • Henrikson N.
        • Hoell C.
        • Holm I.A.
        • Jarvik G.P.
        • Kullo I.J.
        • Lammers P.E.
        • Larson E.B.
        • Lindor N.M.
        • Marasa M.
        • Myers M.F.
        • Peterson J.F.
        • Prows C.A.
        • Ralston J.D.
        • Rasouly H.M.
        • Sharp R.R.
        • Smith M.E.
        • Van Driest S.L.
        • Williams J.L.
        • Williams M.S.
        • Wynn J.
        • Leppig K.A.
        Returning results in the genomic era: initial experiences of the emerge network.
        J Pers Med. 2020; 10: 30
        • Gottesman O.
        • Kuivaniemi H.
        • Tromp G.
        • Faucett W.A.
        • Li R.
        • Manolio T.A.
        • Sanderson S.C.
        • Kannry J.
        • Zinberg R.
        • Basford M.A.
        • Brilliant M.
        • Carey D.J.
        • Chisholm R.L.
        • Chute C.G.
        • Connolly J.J.
        • Crosslin D.
        • Denny J.C.
        • Gallego C.J.
        • Haines J.L.
        • Hakonarson H.
        • Harley J.
        • Jarvik G.P.
        • Kohane I.
        • Kullo I.J.
        • Larson E.B.
        • McCarty C.
        • Ritchie M.D.
        • Roden D.M.
        • Smith M.E.
        • Böttinger E.P.
        • Williams M.S.
        The Electronic Medical Records and Genomics (eMERGE) network: past, present, and future.
        Genet Med. 2013; 15: 761-771
        • Grebe T.A.
        • Khushf G.
        • Chen M.
        • Bailey D.
        • Brenman L.M.
        • Williams M.S.
        • Seaver L.H.
        The interface of genomic information with the electronic health record: a points to consider statement of the American College of Medical Genetics and Genomics (ACMG).
        Genet Med. 2020; 22: 1431-1436
        • Kannry J.M.
        • Williams M.S.
        Integration of genomics into the electronic health record: mapping terra incognita.
        Genet Med. 2013; 15: 757-760
        • Warner J.L.
        • Jain S.K.
        • Levy M.A.
        Integrating cancer genomic data into electronic health records.
        Genome Med. 2016; 8: 113
        • Lau-Min K.S.
        • Asher S.B.
        • Chen J.
        • Domchek S.M.
        • Feldman M.
        • Joffe S.
        • Landgraf J.
        • Speare V.
        • Varughese L.A.
        • Tuteja S.
        • VanZandbergen C.
        • Ritchie M.D.
        • Nathanson K.L.
        Real-world integration of genomic data into the electronic health record: the PennChart Genomics Initiative.
        Genet Med. 2021; 23: 603-605
        • Heale B.S.E.
        • Overby C.L.
        • Del Fiol G.
        • Rubinstein W.S.
        • Maglott D.R.
        • Nelson T.H.
        • Milosavljevic A.
        • Martin C.L.
        • Goehringer S.R.
        • Freimuth R.R.
        • Williams M.S.
        Integrating genomic resources with electronic health records using the HL7 infobutton standard.
        Appl Clin Inform. 2016; 7: 817-831
        • Overby C.L.
        • Heale B.
        • Aronson S.
        • Cherry J.M.
        • Dwight S.
        • Milosavljevic A.
        • Nelson T.
        • Niehaus A.
        • Weaver M.A.
        • Ramos E.M.
        • Williams M.S.
        Providing access to genomic variant knowledge in a healthcare setting: a vision for the ClinGen Electronic Health Records Workgroup.
        Clin Pharmacol Ther. 2016; 99: 157-160
        • Riley J.D.
        • Procop G.W.
        • Kottke-Marchant K.
        • Wyllie R.
        • Lacbawan F.L.
        Improving molecular genetic test utilization through order restriction, test review, and guidance.
        J Mol Diagn. 2015; 17: 225-229
        • Desai K.
        • Hooker G.
        • Adeboyeje G.
        • Kachroo S.
        • Sen S.S.
        HSR20-083: real-world utilization and coding variability in medical claims for next-generation sequencing (NGS)-based diagnostic tests among cancer patients in the U.S..
        JNCCN. 2020; 18: HSR20-083
        • Hsiao S.J.
        • Sireci A.
        • Pendrick D.
        • Freeman C.
        • Yang J.
        • Schwartz G.K.
        • Mansukhani M.M.
        • Carvajal R.D.
        • Oberg J.A.
        Clinical utility and reimbursement for expanded genomic panel testing in adult oncology.
        J Clin Oncol. 2019; 37: 6593
        • Mandel J.C.
        • Kreda D.A.
        • Mandl K.D.
        • Kohane I.S.
        • Ramoni R.B.
        SMART on FHIR: a standards-based, interoperable apps platform for electronic health records.
        J Am Med Inform Assoc. 2016; 23: 899-908
        • Alterovitz G.
        • Warner J.
        • Zhang P.
        • Chen Y.
        • Ullman-Cullere M.
        • Kreda D.
        • Kohane I.S.
        SMART on FHIR genomics: facilitating standardized clinico-genomic apps.
        J Am Med Inform Assoc. 2015; 22: 1173-1178
        • Alterovitz G.
        • Heale B.
        • Jones J.
        • Kreda D.
        • Lin F.
        • Liu L.
        • Liu X.
        • Mandl K.D.
        • Poloway D.W.
        • Ramoni R.
        • Wagner A.
        • Warner J.L.
        FHIR genomics: enabling standardization for precision medicine use cases.
        NPJ Genomic Med. 2020; 5: 13
        • Lubin I.M.
        • Aziz N.
        • Babb L.J.
        • Ballinger D.
        • Bisht H.
        • Church D.M.
        • Cordes S.
        • Eilbeck K.
        • Hyland F.
        • Kalman L.
        • Landrum M.
        • Lockhart E.R.
        • Maglott D.
        • Marth G.
        • Pfeifer J.D.
        • Rehm H.L.
        • Roy S.
        • Tezak Z.
        • Truty R.
        • Ullman-Cullere M.
        • Voelkerding K.V.
        • Worthey E.A.
        • Zaranek A.W.
        • Zook J.M.
        Principles and recommendations for standardizing the use of the next-generation sequencing variant file in clinical settings.
        J Mol Diagn. 2017; 19: 417-426
        • den Dunnen J.T.
        • Dalgleish R.
        • Maglott D.R.
        • Hart R.K.
        • Greenblatt M.S.
        • Mcgowan-Jordan J.
        • Roux A.F.
        • Smith T.
        • Antonarakis S.E.
        • Taschner P.E.M.
        HGVS recommendations for the description of sequence variants: 2016 update.
        Hum Mutat. 2016; 37: 564-569
        • Stram M.
        • Seheult J.
        • Sinard J.H.
        • Campbell W.S.
        • Carter A.B.
        • De Baca M.E.
        • Quinn A.M.
        • Luu H.S.
        A survey of LOINC code selection practices among participants of the College of American Pathologists Coagulation (CGL) and Cardiac Markers (CRT) proficiency testing programs.
        Arch Pathol Lab Med. 2020; 144: 586-596
        • Carter A.B.
        • de Baca M.E.
        • Luu H.S.
        • Campbell W.S.
        • Stram M.N.
        Use of LOINC for interoperability between organisations poses a risk to safety.
        Lancet Digit Health. 2020; 2: e569
        • Stram M.
        • Gigliotti T.
        • Hartman D.
        • Pitkus A.
        • Huff S.M.
        • Riben M.
        • Henricks W.H.
        • Farahani N.
        • Pantanowitz L.
        Logical observation identifiers names and codes for laboratorians.
        Arch Pathol Lab Med. 2020; 144: 229-239
        • MacConaill L.E.
        • Burns R.T.
        • Nag A.
        • Coleman H.A.
        • Slevin M.K.
        • Giorda K.
        • Light M.
        • Lai K.
        • Jarosz M.
        • McNeill M.S.
        • Ducar M.D.
        • Meyerson M.
        • Thorner A.R.
        Unique, dual-indexed sequencing adapters with UMIs effectively eliminate index cross-talk and significantly improve sensitivity of massively parallel sequencing.
        BMC Genomics. 2018; 19: 30
        • MacArthur J.A.L.
        • Morales J.
        • Tully R.E.
        • Astashyn A.
        • Gil L.
        • Bruford E.A.
        • Larsson P.
        • Flicek P.
        • Dalgleish R.
        • Maglott D.R.
        • Cunningham F.
        Locus Reference Genomic: reference sequences for the reporting of clinically relevant sequence variants.
        Nucleic Acids Res. 2014; 42: D873-D878
        • Dalgleish R.
        • Flicek P.
        • Cunningham F.
        • Astashyn A.
        • Tully R.E.
        • Proctor G.
        • Chen Y.
        • McLaren W.M.
        • Larsson P.
        • Vaughan B.W.
        • Béroud C.
        • Dobson G.
        • Lehväslaiho H.
        • Taschner P.E.M.
        • den Dunnen J.T.
        • Devereau A.
        • Birney E.
        • Brookes A.J.
        • Maglott D.R.
        Locus Reference Genomic sequences: an improved basis for describing human DNA variants.
        Genome Med. 2010; 2: 24
        • Richards S.
        • Aziz N.
        • Bale S.
        • Bick D.
        • Das S.
        • Gastier-Foster J.
        • Grody W.W.
        • Hegde M.
        • Lyon E.
        • Spector E.
        • Voelkerding K.
        • Rehm H.L.
        Standards and guidelines for the interpretation of sequence variants: a joint consensus recommendation of the American College of Medical Genetics and Genomics and the Association for Molecular Pathology.
        Genet Med. 2015; 17: 405-424
        • Li M.M.
        • Datto M.
        • Duncavage E.J.
        • Kulkarni S.
        • Lindeman N.I.
        • Roy S.
        • Tsimberidou A.M.
        • Vnencak-Jones C.L.
        • Wolff D.J.
        • Younes A.
        • Nikiforova M.N.
        Standards and guidelines for the interpretation and reporting of sequence variants in cancer: a joint consensus recommendation of the Association for Molecular Pathology, American Society of Clinical Oncology, and College of American Pathologists.
        J Mol Diagn. 2017; 19: 4-23
        • Gulley M.L.
        • Braziel R.M.
        • Halling K.C.
        • Hsi E.D.
        • Kant J.A.
        • Nikiforova M.N.
        • Nowak J.A.
        • Ogino S.
        • Oliveira A.
        • Polesky H.F.
        • Silverman L.
        • Tubbs R.R.
        • Van Deerlin V.M.
        • Vance G.H.
        • Versalovic J.
        Clinical laboratory reports in molecular pathology.
        Arch Pathol Lab Med. 2007; 131: 852-863
        • Ogino S.
        • Gulley M.L.
        • den Dunnen J.T.
        • Wilson R.B.
        • Payne D.
        • Lowery Nordberg M.C.
        • Gong J.Z.
        • Krafft A.E.
        • Uphoff T.S.
        • Donahue P.
        • Hunt J.
        • Garrison G.
        Standard mutation nomeclature in molecular diagnostics: practical and educational challenges.
        J Mol Diagn. 2007; 9: 1-6
        • Haga S.B.
        • Kim E.
        • Myers R.A.
        • Ginsburg G.S.
        Primary care physicians’ knowledge, attitudes, and experience with personal genetic testing.
        J Pers Med. 2019; 9: 29
        • Nightingale B.M.
        • Hovick S.R.
        • Brock P.
        • Callahan E.
        • Jordan E.
        • Roggenbuck J.
        • Sturm A.C.
        • Morales A.
        Hypertrophic cardiomyopathy genetic test reports: a qualitative study of patient understanding of uninformative genetic test results.
        J Genet Couns. 2019; 28: 1087-1097
        • Macklin S.K.
        • Jackson J.L.
        • Atwal P.S.
        • Hines S.L.
        Physician interpretation of variants of uncertain significance.
        Fam Cancer. 2019; 18: 121-126
        • Recchia G.
        • Chiappi A.
        • Chandratillake G.
        • Raymond L.
        • Freeman A.L.J.
        Creating genetic reports that are understood by nonspecialists: a case study.
        Genet Med. 2020; 22: 353-361
        • Farmer G.D.
        • Gray H.
        • Chandratillake G.
        • Raymond F.L.
        • Freeman A.L.J.
        Recommendations for designing genetic test reports to be understood by patients and non-specialists.
        Eur J Hum Genet. 2020; 28: 885-895
        • Zahabi M.
        • Kaber D.B.
        • Swangnetr M.
        Usability and safety in electronic medical records interface design: a review of recent literature and guideline formulation.
        Hum Factors. 2015; 57: 805-834
        • Solomon I.B.
        • McGraw S.
        • Shen J.
        • Albayrak A.
        • Alterovitz G.
        • Davies M.
        • Del Vecchio Fitz C.
        • Freedman R.A.
        • Lopez L.N.
        • Sholl L.M.
        • Van Allen E.
        • Mortimer J.
        • Fakih M.
        • Pal S.
        • Reckamp K.L.
        • Yuan Y.
        • Gray S.W.
        Engaging patients in precision oncology: development and usability of a web-based patient-facing genomic sequencing report.
        JCO Precis Oncol. 2020; 14 (PO.19.00195)
        • Gray S.W.
        • Gagan J.
        • Cerami E.
        • Cronin A.M.
        • Uno H.
        • Oliver N.
        • Lowenstein C.
        • Lederman R.
        • Revette A.
        • Suarez A.
        • Lee C.
        • Bryan J.
        • Sholl L.
        • Van Allen E.M.
        Interactive or static reports to guide clinical interpretation of cancer genomics.
        J Am Med Inform Assoc. 2018; 25: 458-464
        • Westendorf L.
        • Shaer O.
        • Pollalis C.
        • Verish C.
        • Nov O.
        • Ball M.P.
        Exploring genetic data across individuals: design and evaluation of a novel comparative report tool.
        J Med Internet Res. 2018; 20: e10297
        • Marsolo K.
        • Spooner S.A.
        Clinical genomics in the world of the electronic health record.
        Genet Med. 2013; 15: 786-791
        • Jensen K.
        • Konnick E.Q.
        • Schweizer M.T.
        • Sokolova A.O.
        • Grivas P.
        • Cheng H.H.
        • Klemfuss N.M.
        • Beightol M.
        • Yu E.Y.
        • Nelson P.S.
        • Montgomery B.
        • Pritchard C.C.
        Association of clonal hematopoiesis in DNA repair genes with prostate cancer plasma cell-free DNA testing interference.
        JAMA Oncol. 2021; 7: 107-110
        • Fujiyoshi K.
        • Bruford E.A.
        • Mroz P.
        • Sims C.L.
        • O’Leary T.J.
        • Lo A.W.I.
        • Chen N.
        • Patel N.R.
        • Patel K.P.
        • Seliger B.
        • Song M.
        • Monzon F.A.
        • Carter A.B.
        • Gulley M.L.
        • Mockus S.M.
        • Phung T.L.
        • Feilotter H.
        • Williams H.E.
        • Ogino S.
        Standardizing gene product nomenclature-a call to action.
        Proc Natl Acad Sci U S A. 2021; 11 (8:e2025207118)
        • Schinkel M.
        • Paranjape K.
        • Nannan Panday R.S.
        • Skyttberg N.
        • Nanayakkara P.W.B.
        Clinical applications of artificial intelligence in sepsis: a narrative review.
        Comput Biol Med. 2019; 115: 103488
        • Richards C.S.
        • Bale S.
        • Bellissimo D.B.
        • Das S.
        • Grody W.W.
        • Hegde M.R.
        • Lyon E.
        • Ward B.E.
        ACMG recommendations for standards for interpretation and reporting of sequence variations: revisions 2007.
        Genet Med. 2008; 10: 294-300
        • Riggs E.R.
        • Andersen E.F.
        • Cherry A.M.
        • Kantarci S.
        • Kearney H.
        • Patel A.
        • Raca G.
        • Ritter D.I.
        • South S.T.
        • Thorland E.C.
        • Pineda-Alvarez D.
        • Aradhya S.
        • Martin C.L.
        • on behalf of the ACMG
        Technical standards for the interpretation and reporting of constitutional copy-number variants: a joint consensus recommendation of the American College of Medical Genetics and Genomics (ACMG) and the Clinical Genome Resource (ClinGen).
        Genet Med. 2020; 22: 245-257