|Home | About | Journals | Submit | Contact Us | Français|
The introduction of diagnostic clinical genome and exome sequencing (CGES) is changing the scope of practice for clinical geneticists. Many large institutions are making a significant investment in infrastructure and technology, allowing clinicians to access CGES especially as health care coverage begins to extend to clinically indicated genomic sequencing-based tests. Translating and realizing the comprehensive clinical benefits of genomic medicine remains a key challenge for the current and future care of patients. With the increasing application of CGES, it is necessary for geneticists and other health care providers to understand its benefits and limitations, in order to interpret the clinical relevance of genomic variants identified in the context of health and disease. Establishing new, collaborative working relationships with specialists across diverse disciplines (e.g., clinicians, laboratorians, bioinformaticians) will undoubtedly be key attributes of the future practice of clinical genetics and may serve as an example for other specialties in medicine. These new skills and relationships will also inform the development of the future model of clinical genetics training curricula.
To address the evolving role of the clinical geneticist in the rapidly changing climate of genomic medicine, two Clinical Genetics Think Tank meetings were held which brought together physicians, laboratorians, scientists, genetic counselors, trainees and patients with experience in clinical genetics, genetic diagnostics, and genetics education. This paper provides recommendations that will guide the integration of genomics into clinical practice.
Genetic diagnostics have evolved rapidly over the past six decades, changing the practice of clinical genetics and helping all medical specialties to achieve improved clinical outcomes. With techniques to see the correct complement of human chromosomes,1 genomic diagnostics through light microscopy was initiated. Three years later, the first genomic disorder, Down syndrome, was found to be associated with trisomy 21.2 The application of this “genomic” technology was limited for the next 30 years to diagnosing various germ line and somatic aneuploidies and translocations, with increasing resolution and the use of fluorescence in situ hybridization (FISH) beginning in the late 1980s.3–5 Comparative genomic hybridization (CGH) and single nucleotide polymorphism (SNP) array-based methodologies allowed for the first genome-wide high-resolution analysis for small deletions and duplications,6 and replaced the G-banded karyotype as a first-tier diagnostic test.7 Ten years after this, the first applications of genome sequencing8 and exome sequencing9–12 to clinical medicine occurred and have since become widely used in diagnostics.13
With the expanded comprehensiveness of genetic testing, the interpretation of results has become more complex. Clinical genome and exome sequencing (CGES) promises to improve identification of the molecular determinants for many disease categories, but carries a degree of uncertainty given the large number of variants of uncertain significance (VUS) identified in any individual, as well as secondary findings unrelated to the primary indication for testing. Secondary findings are identified in 0.5–3.5% of individuals undergoing CGES,14–18 but will increase as our knowledge about genetic determinants of disease advances. Indeed, as CGES evolves toward screening healthy populations, what we now consider secondary findings will become the primary findings, which will form the foundation for the practice of genomic precision medicine.19
The increasing complexity of CGES coupled with the broad scale application to a wide range of medical conditions requires a transformative approach to the ways we currently practice clinical genetics and medicine in general. New clinical models are needed to enable the integration of clinically relevant genomic information into specialized and routine healthcare.
The concept of a Clinical Genetics Think Tank (CGTT) evolved as it became evident that several clinical genetic centers in the United States and Canada were independently developing research and clinical programs to respond to demands for genomic services and were facing common challenges.20 The goals of the CGTT were to identify the major challenges to integrating CGES into clinical practice and to provide practical recommendations based on published evidence and professional experience.
Here, we outline the recommendations agreed upon by the CGTT, which are presented as a framework to influence policy discussions, provide guidance in the creation of new individualized care models incorporating family and patient preferences, and direction regarding how to integrate new genomic knowledge across organizations to ensure optimal management of patients undergoing CGES.
Five critical areas of focus in the clinical workflow to effectively integrate CGES were identified including; the pre-test process, pre-test education for patients and providers, phenotyping, sequence data interpretation and post-test patient care. For each area, we present challenges and recommendations followed by key questions, summarized in Table 1.
Many of the challenges of integrating CGES into clinical care are directly related to its huge diagnostic potential and its cost.20,21 As the cost/benefit ratio of CGES improves, some barriers to ordering this test are likely to fade.22 The diagnostic value will increase as genomic and phenotypic data are shared and interpreted, making it imperative that a robust pre-test process is developed to allow clinicians to utilize this test in the care of patients who are most likely to benefit.
The American College of Medical Genetics and Genomics (ACMG) recommended that CGES should be accompanied by consultation with a genetics professional and adequate genetic counseling.23 Given the relatively small number of genetics professionals in the workforce and the projected increase in demand for CGES, the CGTT recommends that a clinician should be able to order CGES if he or she has a minimum knowledge base, with an opportunity to subsequently request additional expertise from a genetics professional (e.g., for pre-test evaluation or interpretation).
This minimum knowledge is defined by:
There are limited publications guiding the decision to order CGES.13,24,25 Figure 1 exhibits an algorithm following modification of a prior guideline25 and may require further modifications as evidence is published regarding the sensitivity of CGES in specific patient populations.
The general principles guiding this algorithm are:
Some institutions have established a gatekeeper role for access to CGES to promote appropriate patient selection.21 Boston Children’s Hospital BEST (Bringing Excellence to Selecting Tests) committee consists of members with sub-specialty expertise, who review all requests for CGES and provide educational feedback to the ordering clinician if a request is declined. Applications are initially reviewed by the content expert most closely related to the phenotype, who may approve the application or send it for full committee review. Given the potential benefits of education, quality and cost control associated with a gatekeeping role, institutions should give consideration to developing such a process, with diverse representation from institutional stakeholders. The precise model for each institution will vary, but may include the provision of guidelines (either institution-specific or peer-reviewed publications) that are a pre-requisite to accessing CGES, in addition to (or instead of) an individual or committee review of each application. The skill set required to review CGES applications is likely to be the same as that required to order and interpret the test, as outlined previously in the section entitled ‘Who should order CGES?’.
Insurance companies and the Canadian Provincial Ministries of Health differ in their policies towards covering the cost of CGES. Several insurance providers have specific policies limiting and, in some cases denying, access based on the perception of CGES as an experimental or investigational technology.29 In the absence of access to CGES, less appropriate tests, often of equivalent or increased cost, may be ordered because those tests are covered by insurance.30
A recent study looking at insurance coverage for exome sequencing found extreme variability both among and within insurance providers which was heavily dependent on the indication for testing, and often involved large costs to the patients due to high deductibles and co-pays even when approved.22 A registry of next generation sequencing tests and reimbursement policies has been set up at UCSF’s Center for Translational and Policy Research on Personalized Medicine (https://pharm.ucsf.edu/transpers), in order to track and inform decisions made by insurance representatives. Genetics professionals spend significant time writing letters of medical necessity to justify testing, and patients are referred to clinical genetics services for assistance in this process. Increased demand for this type of assistance would limit patient access to CGES since clinical genetics services are not available in the majority of healthcare settings. In the short term, sample letters of medical necessity should be made publically available in order to provide ordering clinicians the opportunity to adapt appropriately worded letters for their specific patient request. The letters should include references to regularly updated, peer-reviewed literature on clinical validity and utility, and patient satisfaction associated with CGES.
In the longer term, enhanced communication between professional bodies (advocating for CGES to be used on the basis of published evidence for improved diagnostic, management and patient satisfaction outcomes) and payers (articulating the industry’s pre-requisites to providing coverage for CGES) is necessary to increase accessibility to CGES. The American Society of Human Genetics (ASHG), ACMG and Canadian College of Medical Geneticists (CCMG) groups will need to play an active role and continue to work with payers to substantiate the clinical validity and utility of CGES. The periodic publication of practice guidelines by these professional bodies should be part of this effort.24
Multiple stakeholders require education about CGES, including physicians, genetic counselors, trainees, insurance and government stakeholders, laboratorians, and patients and their families. Education and informed consent have been thoroughly discussed elsewhere.31–37 However, in order to facilitate broad implementation of CGES into clinical care, some areas require refinement in the context of increasingly advanced and complex genomic testing.
Patient education regarding CGES should be designed specifically for the target audience and extend beyond written information in the informed consent document. Several published articles and guidelines have outlined the topics considered essential to the informed consent process.38,39 The CGTT, primed by its patient and parental participants, concluded that it is most important to focus education on the types of results that are possible. A summary of the key patient education topics identified by the CGTT can be found in Table 2.
Provider education should be designed to enable assessment by clinicians of their ability and educational needs prior to ordering CGES. Topics should include: what a basic clinical genetics evaluation and family history entails, review of different types of genetic tests to allow for a critical summary of prior evaluations, overview of CGES (indications, strengths and limitations) to allow clinicians to determine whether this is the best test for the specific clinical indication, what is entailed in pre-test counseling, interpretation of results, post-test counseling, and when to seek support from a genetics professional.
Insurance providers and government bodies require an understanding of CGES in order to make informed decisions about test coverage and reimbursement levels. Key education topics are the clinical validity as it relates to the diagnosis and management of patients and their families as well as the utility of CGES for specific patient populations, which will be informed by data from research that is largely in its infancy.
Clinician-patient interaction is key for adequate patient education; however patient educational materials should be provided as a supplement to the patient – clinician interaction.40 There are numerous online resources appropriate for patient education in CGES available at no cost (see Appendix 1). Their formats include video, interactive modules and expert-authored test descriptions designed to meet different educational requirements. Some have been translated into multiple languages to minimize disparities in access to CGES.
Clinical geneticists are already highly competent in many of the areas required to order and interpret CGES; however obtaining the specific skills required for clinical interpretation of CGES is a critical element of translating sequence into patient care. Development of specialty-specific educational opportunities should be promoted, such as the genomics case conferences sponsored by ACMG which focus on variant interpretation as it relates to CGES. Other short courses and continuing medical education (CME) modules are available across North America and worldwide, and details can be found on the websites of the professional societies including ASHG (http://www.ashg.org/education/Health_Professionals.shtml), ACMG (www.acmg.net/ACMG/Education/) and the European Society of Human Genetics (ESHG) (www.eshg.org/courses.0.html). Clinician education methods may also include “on demand” learning, for example the electronic health record may contain prompts to training material when CGES is being electronically ordered.
The integration of clinical data, particularly a patient’s “phenotype”, is of central importance in determining what variants are most relevant during CGES interpretation. Accurate and consistent phenotype measures are needed to fully realize the utility of CGES; however, phenotype data are obtained by a wide range of medical specialties across many health care systems, and current methods for determining and recording phenotype information are not standardized. Recognition of the need to use standardized terms and measures for optimal phenotyping has stimulated worthy educational efforts to define phenotypic terms.41–49 One powerful approach to standardize clinical data is the Human Phenotyping Ontology project (HPO) (http://www.human-phenotype-ontology.org). The HPO was designed to reflect normal clinical phenotyping processes, and is “a computational representation of a domain of knowledge (phenotype) based upon a controlled, standardized vocabulary for describing entities and the semantic relationships between them.”50–52 Collecting clinical data for standardized phenotyping requires clinical databases to be either created or modified from pre-existing tools such as Phenotips53 (http://phenotips.org). This resource supports the use of HPO terms and allows for data entry functionality for PhenomeCentral, or PhenoDB (http://phenodb.net), permitting clinicians secure access to genotype-phenotype information and to collaborate on diagnoses. A full summary of online phenotype-genotype resources can be found in Table 3.
One approach to maximize the utility of standardized phenotypic data for analyzing CGES data would be to grant access to phenotypic data to the laboratory interpreting staff.
The phenotype may be focused for a specific clinical condition, suggesting a modular approach to phenotyping with the data content of individual modules defined for specific organ systems or diseases. In addition, a universal phenotypic data set could be established to assist in interpretation of as many primary and secondary findings on genomic testing as possible. Phenotypic measures developed for disease-specific research purposes (e.g., obesity) will help define which data should be captured for optimal CGES interpretation. Electronic health records (EHRs) need to be adapted to encompass universally consistent ontologic terms allowing for a common phenotyping language, and would play an essential role towards complete annotation of the variants in the human exome and genome from a clinical perspective.
Once phenotypic fields are delineated, it will be critical to get buy-in across organizations, nationally and internationally, to help disseminate robust information related to the significance of variants and mutations and their impact on phenotype, management, and clinical outcomes. A centralized and well-curated database containing both clinical and research genomic results linked to standardized phenotypic data would rapidly increase the accuracy of variant interpretation. Large-scale data collection will require broad scale efforts and institutional support, but the benefits in terms of standardized care, portability of information, and improved metrics to inform cost/benefit analyses would be potentially great.
The ability to update phenotypic fields will be essential to understand the significance of variants identified on genomic testing across an individual’s lifespan. There is also a need to reassess variants of uncertain significance that may have been implicated in human disease in the interval since testing was performed. A post-CGES follow-up visit with a genetics professional would provide an opportunity to proactively manage patients in relation to their identified disease and risk variants, as well as update patients with new information related to the variants identified on initial testing. These visits would also provide an opportune time to update and establish comprehensive phenotypes on individuals and to institute the application of correct terminologies into the EHR. Establishment of automated care assistant models that could link genomic variants to populations at risk and provide recommended management options would streamline the process of reviewing an individual’s genome. Initial work in this domain is actively being pursued by research consortia supported through the NIH including the Electronic Medical Records and Genomics (eMERGE) Network (https://emerge.mc.vanderbilt.edu), the Clinical Sequencing Research Consortium (CSER) (https://cser-consortium.org) and the Implementing Genomics in Practice (IGNITE) consortium (http://www.ignite-genomics.org/ignite_about.html). Recent work in applying such models to pharmacogenetic variants has shown the potential viability of such an approach.54,55
Much of the workflow (sequencing, alignment, variant calling) is standardized in diagnostic genomics laboratories. In comparison with gene panel testing, CGES interpretation requires closer collaboration between laboratorians and clinicians in order to reduce the number of genomic variants to be analyzed to a reasonable number. Fostering existing and establishing new relationships and processes to support these important interactions will ultimately ensure that CGES interpretation is accurate and efficient, and that the relevant information is communicated in a comprehensive clinical report.
Determination of the correct diagnostic test for an individual patient should be the decision of the physician caring for the patient; however the diagnostic laboratory directors and genetic counselors should be available to provide feedback regarding the suitability of CGES relative to the clinical indication. Factors such as how well relevant genes are covered by CGES, as well as what types of variation are predominant in those genes are important for the ordering clinician to understand (e.g., triplet repeat expansions, copy number variants and other structural variants may not be well-covered by CGES).
Identification and annotation of variants that are sufficiently relevant to the patient’s phenotype to be included in a focused clinical report are complex steps in the process of CGES interpretation.54–57 Maintaining high stringency with respect to proof of pathogenicity at both gene and variant level will help to reduce the potentially severe consequences of misdiagnosis.58 The following recommendations aim to optimize this process:
Variability in depth of coverage across the exome and genome leads to the possibility of gaps in sequencing and missing or uninterpretable data due to regions with high homology. The laboratory should make clear whether it will ensure completion or “fill-in” of any missing sequences or classes of variation that are relevant to the patient’s indication but technically difficult to access by CGES.55 Currently, laboratories are not routinely “filling in” gaps with Sanger sequencing. Diagnostic laboratories should establish mechanisms, by which clinicians can be informed of technical test limitations on a patient and indication-specific basis.
The common practice of confirming all pathogenic variants included in a clinical report by Sanger sequencing will become less warranted as validation efforts increase over time, and as laboratories ensure integrity of sample identity across complex genomic sequencing workflows.
A critical issue is the decision regarding which variants are included in a formal clinical report.60 Results should be reported that definitively or likely explain the indication for CGES. If the laboratory decides to report variants in genes of uncertain significance (GUS) such as those with limited, no or conflicting evidence according to ClinGen’s framework but which represent strong candidates to explain the clinical findings, efforts should be made to refer to research-based evaluation for further clarification; it should be made explicit when evidence is lacking. Clinicians and CGES laboratory staff should be cognizant of the distinction between research and clinical care. Deposition of cases with candidate genes into Matchmaker Exchange is recommended (http://www.matchmakerexchange.org/). For secondary findings, the threshold for inclusion in a clinical report should be more stringent than for the primary indication, and a minimal standard set of genes for inclusion of secondary variants (e.g., the ACMG recommendations)61 should be considered. The types of variants reported should be limited to likely pathogenic or pathogenic.54 In addition, reports should make it clear that complete detection and interpretation of all disease-associated variants is not possible.
The post-test patient care phase begins once a CGES report has been returned to the ordering physician. In theory, this phase could last a lifetime, for the aspirational goal is to use DNA sequencing results to make a primary genomic diagnosis and to maintain health for a generation and even beyond. Additional factors that support the notion of a continuum of post-test patient care are the ever-changing methods of DNA analysis, the growing body of evidence and rules for interpreting the pathogenic potential of a variant, the anticipated inclusion of ‘omics’ methods, and animal models to further delineate clinical phenotypes. The initial appointment to discuss CGES results is likely to be the first of several post-test contacts between the patient and the medical team.
The process of returning CGES results to the patient and other members of the medical team must be tailored to the diagnostic issue, the patient and family needs, the clinical setting, and the available workforce. The urgency and complexity of information to be returned to the patient should be balanced according to the patient’s stated wishes obtained during the pre-test consent process. Patients should have the choice of one or more results disclosure sessions to improve understanding and avoid fatigue. While we understand that it may be difficult to accommodate two sessions, and that in many circumstances one session only may be preferable for families, the feedback received from families and patient advocates should challenge the current health care model to allow for more than one session to discuss results of CGES. The first session addresses the principal diagnostic result. If the diagnostic or management issues relating to the primary diagnosis are outside the expertise of the ordering clinician, additional medical specialists may be invited to attend. Alternatively, the patient should be made aware of the need for obtaining specialist opinion(s), and the ordering clinician would complete a comprehensive referral including interpretation of the CGES results.
The second session should address secondary findings that are medically actionable, and be offered within a short time period after the first session if such findings exist. Pharmacogenomics variants would be addressed at this time, in consultation with a pharmacist or pharmacologist trained in interpreting pharmacogenetic data. This process is summarized in Figure 2.
A significant number of institutions have implemented large-scale EHR systems that are uniformly aimed at improving patient care, safety and also address billing related issues. As CGES moves into daily clinical care, several providers of EHRs have begun developing modules that allow full integration of CGES results into the EHR of each patient. There are a number of aspects that require consideration for this to occur, which are beyond the scope of this manuscript. However, it is important to emphasize that detailed phenotypic analyses accompanied by primary, secondary and pharmacogenetic findings resulting from CGES should form part of the EHR in the most transparent way, in order to ensure patient safety and appropriate management.
Periodic re-evaluation of an individual’s genome with the aim of refining diagnostic interpretation is desirable, but requires significant time and financial commitments from laboratories, clinicians, and patients. Re-evaluation would apply both to those individuals whose CGES was diagnostic for their primary disorder and to those who did not receive a genomic diagnosis. A major issue concerning re-evaluation of CGES clinically is how to bill and be reimbursed for these services. The value of re-evaluation has been demonstrated in the research realm and several initiatives have demonstrated that automated re-querying of CGES data to capture novel disease associated variants is computationally possible. These initiatives are also able to compare variants amongst related cohorts to identify novel pathogenic variants associated with specific phenotypes. An excellent example of this is the Epilepsy Genetics Initiative (EGI) (http://www.cureepilepsy.org/egi/index.asp) that is warehousing phenotype and genotype data (primarily exomes) contributed by clinicians and investigators and offering a re-analysis of the entire cohort every 6 months. While several diagnostic laboratories are offering free reanalysis within a specified time frame, guidelines need to be set for reimbursement for the diagnostic laboratories for these types of labor-intensive services. With 50–80% of clinical CGES tests having no identifiable definitive pathogenic variant on initial analysis, subsequent re-analysis as new information and pathogenic variants are identified will be critically important. Ideally at routine follow-up appointments, the family and personal health history would be updated, original sequencing results and interpretation reviewed, and DNA variant databases re-queried. If indicated, the physical examination would be repeated and follow up tests ordered. As functional testing becomes more accessible, VUS may be re-classified to allow diagnostic certainty. Supplementary ‘omics’ profiles, including metabolomics, transcriptomics, epigenomics and metagenomics, as well as animal models, already widely used in research may enable functional correlation with important genomic variants. These innovations should be the subject of early research on costs and effectiveness.
A number of papers have been written outlining issues related to implementation of genomic medicine into the clinic. Some focus on specific issues such as the evolving role of the medical geneticist,62 optimization and integration of genomic information in the EHR,63 research needs,64 ethical and legal aspects of returning results and distinguishing research from clinical care,65 and descriptions of institutional approaches to integration of genomic information into clinical care.20,66,67 Few have dealt with the practical aspects of implementation of genomic medicine into the clinic. The review of Biesecker and Green13 provided an overview of CGES, but did not focus on the obstacles to implementation. At a 2011 Colloquium of Genomic Medicine, sponsored by the US National Human Genome Research Institute (NHGRI), 20 groups of clinicians and investigators working on projects to implement genomic medicine summarized early findings.21 Some common challenges were seen: lack of consensus on the pathogenicity and clinical relevance of identified variants, reimbursement for genomic tests and interventions, the burden on clinicians and patients in managing the information generated from genomic testing, as well as infrastructural needs, such as open access knowledge base for cataloging variants and phenotypes. The Colloquium recognized that much of the work in clinical genomic medicine was being done in isolation and would benefit from active collaborative efforts and sharing of best practice, a catalyst for the CGTT workshops and this paper.
We hope that this outline will serve as the basis for continued interactions of various professional groups and will facilitate the establishment of collaborative infrastructures that are essential to implement best practices of clinical genomic medicine. It will only be through collaborative efforts and the application of consensus recommendations from different institutions and diverse populations (ethnically and racially, economically, prenatal, pediatric and adult for a wide spectrum of disease states and in healthy populations) that the robust data and metrics needed to evaluate the impact of genomic medicine on the health of individuals and populations will be able to be evaluated and best practice guidelines established for implementation of genomic medicine.
Funding to cover attendance and accommodation for all participants at both meetings was provided by the Centre for Genetic Medicine, Hospital for Sick Children, Toronto, Canada and The Departments of Pediatrics and Pathology at The Children’s Hospital of Philadelphia, Philadelphia, Pennsylvania, USA.
The Clinical Genetics Think Tank was funded by the Centre for Genetic Medicine, Hospital for Sick Children and the Departments of Pediatrics and Pathology at The Children’s Hospital of Philadelphia, with additional funding from the National Human Genome Research Institute (5UO1HG006546) (IDK, NBS). The authors acknowledge the parent and patient members of the CGTT – M. Hardy, B. Kovalski, H. Kovalski, D. Siciliano, J. Strautnieks, T. Stoppa and the support of Dr. Joe St. Geme and Dr. Bob Doms at the Children’s Hospital of Philadelphia. LGB was supported by the Intramural Research Program of the National Human Genome Research Institute.
Conflict of Interest
The authors of this manuscript declare no conflict of interest.
World Wide Web:
American College of Medical Genetics and Genomics (ACMG). Available at: http://www.acmg.net
Matchmaker Exchange. Available at: http://www.matchmakerexchange.org/
ClinGen. Available at: http://www.clinicalgenome.org/knowledge-curation/gene-curation/
US Food and Drug Administration. Available at: http://www.fda.gov/AboutFDA/CentersOffices/OfficeofMedicalProductsandTobacco/CDER/ucm221248.htm
Gene Dx Exome Slice. Available at: http://www.genedx.com/test-catalog/xomedxslice/
UCSF Center for Translational and Policy Research on Personalized Medicine. Available at: https://pharm.ucsf.edu/transpers
Human Phenotyping Ontology project (HPO). Available at: http://www.human-phenotype-ontology.org
PhenoTips. Available at: http://phenotips.org
PhenoDb. Available at: http://phenodb.net
Electronic Medical Records and Genomics (eMERGE) Network https://emerge.mc.vanderbilt.edu
Clinical Sequencing Research Consortium (CSER) https://cser-consortium.org
Implementing Genomics in Practice (IGNITE) consortium http://www.ignite-genomics.org/ignite_about.html
Epilepsy Genetics Initiative (EGI) http://www.cureepilepsy.org/egi/index.asp