Article Tools
Health Services Research, Clinical Informatics, and Quality of Care
Article Tools
Art and Challenges of Precision Medicine: Interpreting and Integrating Genomic Data Into Clinical Practice
Abstract
Precision medicine is at the forefront of innovation in cancer care. With the development of technologies to rapidly sequence DNA from tumors, cell-free DNA, proteins, and even metabolites coupled with the rapid decline in the cost of genomic sequencing, there has been an exponential increase in the amount of data generated for each patient diagnosed with cancer. The ability to harness this explosion of data will be critical to improving treatments for patients. Precision medicine lends itself to big data or “informatics” approaches and is focused on storing, accessing, sharing, and studying these data while taking necessary precautions to protect patients’ privacy. Major cancer care stakeholders have developed a variety of systems to incorporate precision medicine technologies into patient care as soon as possible and also to provide the ability to store and analyze the omics and clinical data aggregately in the future. Scaling these precision medicine programs within the confines of health care system silos is challenging, and research consortiums are being formed to overcome these limitations. Incorporating and interpreting the results of precision medicine sequencing is complex and rapidly changing, necessitating reliance on a group of experts. This is often performed at molecular tumor boards at large academic and research institutions with available in-house expertise, but alternative models clinical decision support software or of virtual tumor boards potentially expand these advances to almost any patient, regardless of site of care. The promises of precision medicine will be more quickly realized by expanding collaborations to rapidly process and interpret the growing volumes of omics data.
PRACTICAL APPLICATIONS
Learn about incorporating genomic data into EHRs.
Understand the role of data-sharing consortia in aggregating data from large numbers of patients.
Learn about molecular tumor boards, including virtual tumor boards, and other mechanisms for interpreting complex genomic data.
Learn about data safety and patient privacy as it applies to precision medicine.
Develop an appreciation for the necessity of incorporating precision medicine–based therapies into routine clinical practice.
The concept of cancer precision medicine, and that of precision medicine informatics, has always hinged on one key premise: that most cancer therapies are designed for the “average patient” as a “one-size-fits-all" approach. But there is no “average patient,” and thus most treatments will be successful for some patients but not for others. “Precision medicine” should not be conflated with “personalized medicine,” as both terms are sometimes used interchangeably. Precision medicine tailors therapies to classes of patients on the basis of the differences in people’s genes, environments, and lifestyles. Personalized medicine implies customization for an individual patient. Precision medicine simply expands this to the cohort level.1 Taken together, one can think of precision medicine as designed to target the right disease with the right treatment of the right set of patients at the right time.
For precision medicine informatics to be effective, precision medicine requires technology development that allows us to identify key altered pathways that are susceptible to molecularly targeted or immunologic therapies.2 The increasing customization of medical treatment to specific patient characteristics has been possible through continued advances in (1) our understanding of the physiologic mechanisms of disease, leading to the proliferation of omics data (e.g., proteomics, metabolomics), and (2) computing systems (e.g., patient and trial matching algorithms) that facilitate the development and application of targeted agents.3 These advancements allow improved outcomes and reduced exposure to the adverse effects of unnecessary treatment. They can help us better decipher the interpatient (between patients) and intrapatient (different tumors within the same patient) heterogeneity that is often a hurdle to treatment success and can contribute to both treatment failure and drug resistance.4
Precision medicine algorithms and strategies have already borne fruit. The introduction of U.S. Food and Drug Administration (FDA)–approved treatments that are tailored to specific characteristics of individuals, such as a person’s genetic makeup, or the genetic profile of an individual’s tumor have become more routine. Patients with a variety of cancers routinely undergo molecular testing as part of patient care, enabling physicians to select treatments that improve chances of survival and reduce exposure to adverse effects. Importantly, omics-based cancer medicine is here. In 2017, nearly 50% of the early-stage pipeline assets and 30% of late-stage molecular entities of pharmaceutical companies involved the use of biomarker tests.5 Furthermore, more than one-third of drug approvals have had DNA-based biomarkers included in their original FDA submissions.6
Although genomics data are commonly mentioned in the same breath as precision medicine, it is useful to point out that genomics is simply one type of precision medicine data. Other forms of precision data include, but are not limited to, such things as radiographic features (radiomics), patient-reported outcomes (personomics), and digital pathology. By “genomics data,” we are referring to the ability to interrogate the genome using next-generation sequencing techniques. It is equally important to note that omics-type data can be derived from (1) the tumor, (2) the patient, (3) tissue surrounding the tumor “stroma,” (4) circulating blood, and (5) other bodily fluids. Ongoing research has increased our understanding of the underlying pathophysiology of not only the tumor but also the patient-tumor interaction through these omics data. Acquisition of these omics data has required improvements in detection techniques and data analysis.
As an example, assaying proteins using immunohistochemistry, the use of singular antigens that bind to single proteins of interest in cancer tissue, is now being supplanted by mass spectrometry, which allows massively parallel identification of hundreds of proteins simultaneously. However, it has taken improved computer performance (and supercomputer clusters) to accurately identify this large number of proteins in a reasonable amount of time. This expanding field, proteomics, provides a far more accurate readout compared with immunohistochemistry, which is often subjective and difficult to parallelize. Advanced DNA sequencing, which ushered in the genomic revolution, has also improved greatly. Our ability to perform DNA sequencing with trace amounts of starting material (low-passage reads) with improved fidelity and detection is allowing us to detect circulating tumor DNA from the blood. Circulating tumor DNA is tumor-derived fragmented DNA circulating in blood along with cell-free DNA from other sources, measuring about 150 bp. Circulating tumor DNA provides an overview of the genomic reservoir of different tumor clones and genomic diversity. Circulating tumor DNA may finally provide a means of assaying intrapatient tumor heterogeneity, allowing us to get a sense of the relative abundance of genomic alterations across metastatic deposits within a patient. Other promising omics technologies include metabolite analysis. Metabolites have traditionally been singular molecules detected by immunoassay in the clinic. Metabolomics aims to measure abundances of all small molecules detectable in biospecimens, including blood, tissue, urine, and breath, among others. Typically, mass spectrometry and nuclear magnetic resonance techniques are applied to measure hundreds to thousands of metabolites in a given sample. The chemotherapeutic drug methotrexate, for example, has levels that are detected via immunoassay for quantification purposes.7 However, immunoassays measure only singular known metabolites, and it is well known that combinations of metabolites are more clinically relevant than singular metabolites.
To use these omics data meaningfully for clinical use requires systems that help clinicians sort through the omics and context and interpretation. Algorithmically, there has been a shift to using informatics methods such as gene signatures and nonlinear approaches such as neural networks and advanced aggregative techniques to model complex relationships among patients to facilitate this process.8 Importantly, these approaches are the root of cohort matching algorithms that aim to find “patients like my patient.”9 Results of these algorithms are simpler to understand and have propelled the growth of clinical trials matching algorithms. National trials such as NCI-MATCH10 that pair patient tumors with specific tumor alterations to targeted medications are a simplistic first step in this paradigm shift. The ability to perform complex matching, and matching rules, has relied on the growth of aggregated patient data sets and the ability to quickly assess tumor omics data. Although not all of these trials have been successful, there is evidence that as a general approach, patients treated with therapies that match the molecular profile of their tumors have better outcomes than those who are not.11 As more targeted agents become available, the number of laboratories offering molecular testing has increased, and large academic, tertiary care hospitals have begun conducting molecular tumor boards (MTBs)12,13 at which experts weigh in on the molecular profile as well as other relevant factors for specific patients to suggest matched therapies. However, the gold standard remains a genomics or domain expert to provide interpretation of the data. Thus, these interpretations are often facilitated by MTBs or by clinical decision support software. But given the shortage of subtype and pathway specific domain expertise, virtual tumor boards (VTBs) are often used to bring disease certain expertise in treatment planning. VTBs are particularly useful for rare tumors, in which domain expertise may be difficult to obtain locally, or in late-stage tumors, in which standard-of-care options have been exhausted and novel treatments are being explored.
The vast majority of patients with cancer in the United States are treated at community hospitals and practices. It is therefore paramount that precision medicine oncology technologies be available to these facilities. Not only is there a delivery issue, there is a need to encourage the widespread uptake of precision genomics by community oncology practices. This will not only facilitate potentially better patient care but also aid in accruing patients to omics-driven clinical trials. However, considerable challenges exist in implementing a precision medicine program, including administrative, logistic, and financial barriers. Arguably, the most imposing barrier may be the willingness and motivation of oncologists to incorporate routine genomic testing into their daily workflows when the clinical benefit of panel-based omics testing in all tumor types has not been definitively established. Overcoming these substantial obstacles by establishing a comprehensive platform to integrate omics data into clinical practice requires the support and engagement of the key stakeholders in each practice or health care system, which includes oncologists, pathologists, nursing and research staff members, administrators, electronic health records (EHRs), and information technology specialists and others.
As a case example, we use the Swedish Cancer Institute as a didactic means of uncovering and overcoming challenges in bringing a precision medicine program to the community. In September 2014, the Personalized Medicine Research Program was implemented at the Swedish Cancer Institute, a nonuniversity, community-based research practice in Seattle, Washington. The Swedish Cancer Institute is a component of Providence St. Joseph Health, a system comprising 50 hospitals across seven western states.14 In their model, patients are enrolled into an institutional review board–approved registration protocol, and tumors are profiled using a customized in-house gene alteration panel, originally composed of 68 gene alterations and recently expanded to 79 gene alterations, focused on solid tumors. Data are collected using a cloud-based integrated informatics platform to facilitate evidence-based analysis, clinical trials matching, and an MTB.
The underlying software platform to facilitate management of this clinical and research program has been developed by a third-party vendor, Syapse (San Francisco, CA). Provider acceptance of this technology was enhanced by integrating the platform into the Swedish Medical Center’s existing EHR. This reduced redundancy by pulling in patients’ clinical data directly from previously entered fields within the EHR and from the institution’s cancer registry. Similarly, results of the genomic analysis and recommendations of potentially suitable therapies are delivered to the provider within the EHR and did not require separate logins to an external web portal. Raw profiling data from the gene alteration panel are imported directly from an affiliated Clinical Laboratory Improvement Amendments–certified laboratory, with the additional capability of importing omics data from large commercial genomics test vendors. Subsequent iterations of the platform will integrate and present meaningful outcomes data. Data connections also exist, or are under development, with a clinical trials management system, institutional disease site registries, state and national registries, and the anatomic pathology laboratories. To optimize use of the emerging data set, projects are under way to apply machine learning to develop decision support tools.
Financial barriers continue to be a limitation to more widespread uptake of precision medicine testing, although an early study suggests that it may be a cost-neutral undertaking.15 Billing for the next-generation sequencing panel in the Personalized Medicine Research Program was submitted to patients’ insurance companies on the basis of medical necessity, and an interim analysis of reimbursement patterns has shown that approximately one-third of patients received reimbursement, with private and Medicare managed care plans reimbursing at the highest frequency and level.16 The recent FDA approvals of several commercially available NGS panel tests will likely result in greater accessibility to genomic testing.17 For biomarkers that strongly implicate a well-validated targeted therapy, obtaining the medication is difficult. These medications are designated as "off-label" as they generally do not have an FDA indication for the tumor type in question. Albeit time consuming, many insurance plans increasingly are willing to consider evidence from targeted therapies in the appeals process. Furthermore, many pharmaceutical companies provide copay or compassionate use programs for patients that demonstrate need.
The main goal for setting up the Personalized Medicine Research Program was to provide the ability to match genomic results with appropriate or promising therapies. Therapeutic suggestions, or the domain expertise that is clinically actionable, which may include on-label, off-label, and clinical trial options, are formulated by a molecular decision support service (N-of-One, Concord, MA). A molecular pathologist reviews these suggestions before being included together with the genomic results in the final report. A subset of these cases are discussed, by oncologist or pathologist request, at a biweekly MTB, to provide additional input on interpretation of molecular results. In the first 869 patients enrolled in this protocol, results of the next-generation sequencing testing was found to affect 105 patients (21%),18 consistent with the impact of molecular profiling in other published series.19,20 The MTB has proved to be a key venue for engaging and educating clinicians about precision medicine. Clinical trial enrollment was a key goal of establishing a precision medicine program. In addition to single tumor site–based molecular trials, the current framework also facilitates the recruitment of patients to large national basket trials, including NCI-MATCH (NCT02465060) and ASCO’s Targeted Agent and Profiling Utilization Registry (TAPUR; NCT02693535) trial.
Several initiatives are under way to further leverage genomic and clinical data from the growing percentage of cancer patients who are participating in precision medicine programs across the country. The American Association of Cancer Research initiated Project Genomics Evidence Neoplasia Information Exchange (GENIE), a multiphase, multiyear, international data-sharing project to assist in clinical decision-making in rare cancers, and in rare variants of common cancers. Research outcomes for Project GENIE are (1) identification of novel therapeutic targets, (2) aiding in the design of biomarker-driven clinical trials, and (3) identification of genomic determinants of response to therapy. In the first phase of the project, the eight founding international academic institutions released a 19,000-patient data set.21 Notable findings from the initial GENIE data demonstrate that almost all tumor types, especially carcinomas of unknown primary, have at least some samples with a high mutational burden, defined as mutation burden above the 90th percentile of all samples tested on the larger sequencing panels (12.3 mutations/Mb). This finding suggests that the strategy of checkpoint inhibition may be relevant to a proportion of patients in a wide variety of cancer types.22 As of early 2018, both the Swedish Cancer Institute and the Providence Portland Cancer Center’s Earle A. Chiles Research Institute, both institutes within Providence St. Joseph Health, have become participants in Project GENIE. This represents an important contribution to the GENIE data set from a large community-based health system.
Along similar lines, the Oncology Precision Network consortium was launched to share deidentified aggregated omics data among multiple community and academic institutions.23 Founding members of the consortium include Intermountain Healthcare (Salt Lake City, UT), the Swedish Cancer Institute, the Providence Portland Cancer Center, and the Stanford Cancer Institute (Stanford, CA), with the Syapse platform facilitating the data sharing. This consortium anticipates sharing genomic and outcomes data from more than 100,000 patients, representing more than 200 hospitals, annually by 2019, with mostly large community-based hospital systems, and some additional academic centers, having committed to joining the Oncology Precision Network. All contributing partner sites will be able to access the data, which are appropriately deidentified and Health Insurance Portability and Accountability Act and Health Information Technology for Economic and Clinical Health compliant, in real time and will allow treating physicians to view real-world outcomes for other patients with similar omics profiles.
As mentioned earlier, although molecular profiling and tumor board discussion are available to patients at many large academic hospitals, 95% of patients with cancer are treated at the community level. The Swedish experience provided one such example of using decision support software to provide a level of genomic interpretation. There are other growing approaches. One such effort involved was initiated in the pancreatic cancer domain and involves academic medical centers, patient advocacy groups, community hospitals, and a small company to develop a scalable VTB. A scalable VTB can take advantage of cloud-based computing, mobile device engagement, and collaborative platform software development. The rapid pace of publication of new results describing biomarker-drug interactions relevant to cancer makes it difficult for most oncologists to stay up to date. In addition, the presence of multiple actionable biomarkers in a single tumor can present challenges for physicians in that a strong informatics system becomes necessary when confronted with the huge number of potential combination therapies in clinical trials. A VTB can bring together all this information, along with patients’ clinical and molecular data, to the fingertips of clinicians using a single web-based portal. Other cancer domains and their respective oncologists who run MTBs at academic and commercial organizations are now adopting this VTB model.
The creation of the aforementioned pancreatic VTB example is described in better detail here. A multidisciplinary team of clinicians and informaticians from the Pancreatic Cancer Action Network, Perthera, and Georgetown University’s Lombardi Cancer Center orchestrated a precision medicine operation called Know Your Tumor in which 640 patients with pancreatic cancer referred from 287 community and academic centers in the United States were enrolled into the largest geographically distributed precision medicine program for pancreatic cancer. A VTB consisting of more than 10 oncologists reviewed each case and identified highly actionable findings in 27% of patients, including 15% of patients with homologous recombination-deficient tumors. An additional 27% of patient samples also carried molecular abnormalities, the targeting of which would alter therapeutic choices. In all, more than 50% of patient samples were actionable or highly actionable. These patients were followed longitudinally, and at the time of data cutoff, 24% of the patients (156 patients) had initiated new therapy, 81% (126 patients) of whom used therapies that matched their molecular profiles. These data and analysis of the VTB were quantitative in nature and required the informatics teams to collect and curate the data, which were received mostly in nonstandard formats. VTBs provide a unique opportunity to automate this process with a web-based, asynchronous, VTB with highly usable interfaces for clinicians, who are overburdened at most academic medical centers and worse so in community clinics.
VTBs have their own set of technical challenges aside from simply coordinating health care providers. The VTB process of delivering patient reports has guided its development. To date, the many features of this pancreatic VTB include the following:
An information technology infrastructure consisting of databases for multiple types of information to feed into the VTB: patient history, molecular profiling data, and external knowledge of biomarkers, drugs, and clinical trials.
A treatment scoring system (Fig. 1) used to rank single-agent and combination therapies on the baiss of the strength of molecular biomarkers, the activity of the drug(s) in the specific cancer type, and prior exposure of the patient to the drug under consideration (or drugs with similar mechanisms of action). In the VTB example above, the scoring model is based on best practices published by the ClinGen Somatic Working Group,24 Association of Molecular Pathologists,25 and OncoKB.26
FIGURE 1. Treatment Scoring Model Based on Best Practices Published by ClinGen Somatic Working Group, Association of Molecular Pathologists, and OncoKB
This scoring model was codeveloped with collaborators at Georgetown University’s Lombardi Cancer Center.
A user interface consisting of (1) a chat feature allowing asynchronous discussion of a case, thus avoiding scheduling conflicts for geographically dispersed oncologists; (2) a single-tab view allowing access to patient-related documents such as medical history, laboratory results, and reports all in one place without having to navigate away from the tab; (3) resources and references allowing access to external databases within the VTB (example databases include National Cancer Institute cancer dictionaries, clinical trials database, PubMed and Online Mendelian Inheritance in Man) with the intent that the user does not have to leave the VTB interface to access critical information from other references databases; and (4) a case-tracking feature allowing clinicians to know whether a case is ready for analysis or is awaiting laboratory results or medical history.
Data sources including Knowledgebase, Molecular DB, and Patient EHR DB, described below.
The selection of a precise therapy on the basis of a patient’s molecular profile requires computer-assisted analysis of enormous molecular, clinical, patient history, and pharmacological data sets that often come from disparate sources. For instance, arriving at an optimal decision may involve searching through tens of thousands of unique variants in ClinVar,27 more than 5 million somatic variants in the Catalogue of Somatic Mutations in Cancer,28 more than 25 million PubMed articles, more than 0.9 billion submissions to dbSNP,29 190 FDA–approved drugs with pharmacogenomic labeling and more than 300,000 globally registered clinical trials.30 Scientific publications remain a central source of information on the actionability of biomarkers. With advancement in tumor molecular profiling and cancer drug development, scientific evidence is witnessing a huge surge. For doctors commonly with time constraints, presenting just the therapeutically relevant information would ease the time pressure and expedite the biomarker-matched treatment selection process. To create a library of intelligently filtered oncologist-useful information such as drug, disease, biomarker alteration, biomarker-drug relations, manual curation from high-quality journals is often required on a day-to-day basis. In the pancreatic VTB example, biomarker-drug associations were scored on the basis of strength of evidence (preclinical or clinical) and grouped into implication levels as follows: 0 = pertinent negative, 1 = uncertain, 2 = evidence that could modify options, and 3 = biomarker alterations that are highly actionable.
After selecting biomarker-matched treatment, it is equally important for an oncologist to determine the accessibility of a recommended treatment. One recent study found that 19 of 95 patients (20%) were unable to enroll in a recommended study because of trial eligibility restrictions or inconvenient travel distances.13 Poor structuring of eligibility requirements at ClinicalTrials.gov is therefore an issue. Although each trial has inclusion and exclusion criteria, the actual enrollment information is spread throughout the sections in a listed clinical trial. For example, a trial might include “solid tumor” in its title but specify a particular tumor type in its inclusion criteria or specify a particular subtype in its exclusion criteria. Treatment setting (i.e., what line of treatment a patient eligible is for), although most often included in the inclusion and exclusion criteria, is sometimes included in the official title. A biomarker-based search might list biomarker-specific clinical trials, but it is important to know whether an alteration in a specific biomarker excludes or includes a patient. For example, EGFR-exon19 deletion might be an inclusion requirement, but EGFR-T790M might be present as an exclusion criteria. Therefore, to know the accurate eligibility criteria, manual population of clinical trial eligibility requirements is performed. This is continuously updated with any new information gathered either via our medical review panel or elsewhere (e.g., trial arm closure). There are many consortia and government-based efforts such as ClinGen31-34 that have disease-specific task forces that standardize such evidence from somatic testing laboratories.
The success of software applications, especially those built for clinicians and scientists, is dependent on understanding the complex cognitive processes of the intended users and their unique workflows. With this knowledge one can develop a user interface that is more intuitive, is easy to use, and, most important, meets the functional needs of the user. Human factors engineering and usability analysis facilitate users’ reasoning and support their decision making to arrive at the best treatment decisions for their patients and can be applied to VTBs.
Usability testing on VTB prototypes ensures that products meet the design objectives.35,36 Some common techniques for usability testing are think-aloud protocols and eye tracking. To better understand the connection between the design of the prototypes and user experience, one can track how a user visually processes the information on the interface. This attention to usability will help ensure that target oncology users will find their actions intuitive and easy to interpret.
With molecular testing, one possible risk is breach of confidentiality through the release of identifying information. Because it may deeply affect an individual’s sense of self, the privacy of patients with genetic disorders needs to be fiercely protected. All clinicians and researchers involved in this field are very sensitive to this and routinely enforce this in their daily practices. To minimize this risk, DNA samples and medical data are typically encoded and linked to a registry participant identification number. The link between contact information and personal identifying information is maintained at a single location, accessible only to honest brokers and principal investigators of the study or VTB. Any breach of confidentiality should be immediately reported to the institutional review board of record and the sponsor (e.g., National Institutes of Health).
Potential risks associated with clinical whole-exome sequencing are anticipated to be similar to risks associated with other forms of genetic testing, including anxiety and stress at learning the results, disrupted family relationships, possible disruption in social relationships, changes in reproductive choices, stigmatization, and potential loss of employability or insurability (although this potential is now lessened because of the Genetic Information Nondiscrimination Act). In addition, because whole-exome sequencing also produces incidental genetic information, there may be unforeseeable risks systems, and processes must be in place for appropriate return of results. Study enrollment and disclosure of genetic test results are typically performed in the context of genetic counseling by board-certified geneticists and genetic counselors to minimize these potential risks. During the informed consent process, individuals must be provided with alternatives for clinical genetic testing so that they can make informed decisions about whether to participate in the precision medicine research program. Participants may also change their minds and decline to learn their results prior to the test result disclosure session.
Precision medicine informatics platforms must therefore adhere to all federal regulations for handling sensitive data including the Health Insurance Portability and Accountability Act and the Federal Information Security Management Act. National Institutes of Health–funded programs are required to adopt and implement the policies, procedures, controls, and standards of the U.S. Department of Health and Human Services Information Security Program by aligning with data access policies and eRA Commons authentication framework. Large consortia sharing genomic data from thousands of participants use a secure virtual private cloud framework that provides the security controls necessary to meet Federal Information Security Management Act compliance requirements. Through this infrastructure, information protection is provided with security controls at the virtual network, server, and storage layer as well as the security controls offered by cloud providers such as Amazon Web Services. Teams must implement identity management, authentication and authorization services, storage security, and logging and event management to effectively secure patient derived precision medicine data sets.
Taken together, leveraging the rapid advances in precision medicine technologies to deliver the greatest benefits to patients is incredibly promising while at the same time constrained by considerable challenges. Using precision oncology requires innovative comprehensive programs to overcome barriers to implementation and scalable learning systems to keep pace with the huge and growing fund of knowledge and rapidly changing technological advances. Here we present two methods of precision medicine informatics, via clinical decision support software methodologies and a modern version of a classic tumor board. Either method will permit treatment of patients and also facilitate the collection, storage, and sharing of molecular and clinical data. This step is essential, as learning from the past is the only way to improve the collective care provided to patients in the future. Specific to bioinformatics, sophisticated, cloud-based platforms adhering to strict security standards are being developed to manage and analyze complex omics data. These rapidly evolving systems allow routine clinical care and can also facilitate the use of virtual expertise. Developing and strengthening collaborations (private, public, and government; laboratory and clinic) remain key in quickly achieving the promises of precision medicine.
S. M. acknowledges funding support from a National Cancer Institute Cancer Center Support Grant, National Human Genome Research Institute Big Data to Knowledge grants, and technical support and medical review panel from Perthera. S. S. and T. D .B. acknowledge the assistance of Mariko Tameishi in preparation of the manuscript. J. L. C. acknowledges funding support from the National Comprehensive Cancer Institute Young Investigator Awards.
The following represents disclosure information provided by authors of this manuscript. All relationships are considered compensated. Relationships are self-held unless noted. I = Immediate Family Member, Inst = My Institution. Relationships may not relate to the subject matter of this manuscript. For more information about ASCO's conflict of interest policy, please refer to www.asco.org/rwc.
Stock and Other Ownership Interests: GenomiCare Biotechnology
Consulting or Advisory Role: GenomiCare Biotechnology, Jiahui Health
Speakers' Bureau: Novartis Speakers Bureau, Precision Medicine, Syapse
Travel, Accommodations, Expenses: GenomiCare, Jiahui Health, Syapse
Consulting or Advisory Role: Immune Design, Novartis, Syapse
Speakers' Bureau: Foundation Medicine, Novartis
Research Funding: Eisai
Patents, Royalties, Other Intellectual Property: #MatchTX
Consulting or Advisory Role: Bayer, Exelixis, Lexicon, Takeda
Leadership: Perthera
Stock and Other Ownership Interests: Perthera
Consulting or Advisory Role: Perthera
Research Funding: Teewinot Life Sciences (Inst)
| 1. | National Research Council, Division on Earth and Life Studies, Board on Life Sciences, Committee on a Framework for Developing a New Taxonomy of Disease. Toward Precision Medicine. Building a Knowledge Network for Biomedical Research and a New Taxonomy of Disease. Washington, DC: The National Academies Press; 2011. Google Scholar |
| 2. | Goossens N, Nakagawa S, Sun X, et al. Cancer biomarker discovery and validation. Transl Cancer Res. 2015;4:256-269. Medline, Google Scholar |
| 3. | U.S. Food and Drug Administration. Paving the Way for Personalized Medicine. Washington, DC: U.S. Food and Drug Administration; 2013. Google Scholar |
| 4. | Catenacci DV. Next-generation clinical trials: Novel strategies to address the challenge of tumor molecular heterogeneity. Mol Oncol. 2015;9:967-996. Crossref, Medline, Google Scholar |
| 5. | Akhmetov I, Bubnov RV. Assessing value of innovative molecular diagnostic tests in the concept of predictive, preventive, and personalized medicine. EPMA J. 2015;6:19. Google Scholar |
| 6. | Koren G. Personalized medicine-disregarding the obvious: analysis of trends among articles published on “personalized medicine.” Ther Drug Monit. 2015;37:559. Google Scholar |
| 7. | Bouquié R, Grégoire M, Hernando H, et al. Evaluation of a methotrexate chemiluminescent microparticle immunoassay: comparison to fluorescence polarization immunoassay and liquid chromatography-tandem mass spectrometry. Am J Clin Pathol. 2016;146:119-124. Google Scholar |
| 8. | LeCun Y, Bengio Y, Hinton G. Deep learning. Nature. 2015;521:436-444. Crossref, Medline, Google Scholar |
| 9. | Johnson TLD, Chen JL. Opportunities for patient matching algorithms to improve patient care in oncology. Clin Cancer Inform. 2017;1. Google Scholar |
| 10. | Brower V. NCI-MATCH pairs tumor mutations with matching drugs. Nat Biotechnol. 2015;33:790-791. Crossref, Medline, Google Scholar |
| 11. | Schwaederle M, Zhao M, Lee JJ, et al. Impact of precision medicine in diverse cancers: a meta-analysis of phase II clinical trials. J Clin Oncol. 2015;33:3817-3825. Link, Google Scholar |
| 12. | Schwaederle M, Parker BA, Schwab RB, et al. Precision oncology: the UC San Diego Moores Cancer Center PREDICT experience. Mol Cancer Ther. 2016;15:743-752. Crossref, Medline, Google Scholar |
| 13. | Dalton WB, Forde PM, Kang H, et al. Personalized medicine in the oncology clinic: implementation and outcomes of the Johns Hopkins Molecular Tumor Board. Precision Oncol. 2017; (1):1-19. Google Scholar |
| 14. | Brown TD, Gold PJ, Drescher C, et al. Rapid implementation of a personalized medicine research program (PMRP), using a targeted next-generation sequencing (NGS) assay, in the community setting. J Clin Oncol. 2015;33(15_suppl):e22055. Abstract, Google Scholar |
| 15. | Haslem DS, Van Norman SB, Fulde G, et al. A retrospective analysis of precision medicine outcomes in patients with advanced cancer reveals improved progression-free survival without increased health care costs. J Oncol Pract. 2017;13:e108-e119. Link, Google Scholar |
| 16. | Brown TD, Tameishi M, Liu X, et al. Analysis of reimbursement (R) for next generation sequencing (NGS) on patients’ tumors in the context of a personalized medicine program. J Clin Oncol. 2017;35(15_suppl):6506. Abstract, Google Scholar |
| 17. | Allegretti M, Fabi A, Buglioni S, et al. Tearing down the walls: FDA approves next generation sequencing (NGS) assays for actionable cancer genomic aberrations. J Exp Clin Cancer Res. 2018;37:47. Google Scholar |
| 18. | Brown TD, Tittel PD, Gold PJ. Abstract 997: Impact of a personalized medicine research program (PMRP), using targeted tumor profiling and a cloud based clinical trials matching platform, on clinical decision-making. Cancer Res. 2017;77:997. Google Scholar |
| 19. | Le Tourneau C, Delord J-P, Gonçalves A, et al; SHIVA Investigators. Molecularly targeted therapy based on tumour molecular profiling versus conventional therapy for advanced cancer (SHIVA): a multicentre, open-label, proof-of-concept, randomised, controlled phase 2 trial. Lancet Oncol. 2015;16:1324-1334. Crossref, Medline, Google Scholar |
| 20. | Syn NL-X, Yong W-P, Goh B-C, et al. Evolving landscape of tumor molecular profiling for personalized cancer therapy: a comprehensive review. Expert Opinion on Drug Metabolism & Toxicology. 2016;12:911-922. Google Scholar |
| 21. | Litchfield K, Turajlic S, Swanton C. The GENIE is out of the bottle: landmark cancer genomics dataset released. Cancer Discov. 2017;7:796-798. Google Scholar |
| 22. | AACR Project GENIE Consortium. AACR project GENIE: powering precision medicine through an international consortium. Cancer Discov. 2017;7:818-831. Crossref, Medline, Google Scholar |
| 23. | Nadauld L, Haslem D, Tittel PD, et al. Abstract 998: OPeN: the oncology precision network data sharing consortium. Cancer Res. 2017;77(13 suppl):998. Google Scholar |
| 24. | Ritter DI, Roychowdhury S, Roy A, et al; ClinGen Somatic Cancer Working Group. Somatic cancer variant curation and harmonization through consensus minimum variant level data. Genome Med. 2016;8:117. Crossref, Medline, Google Scholar |
| 25. | Griffith M, Spies NC, Krysiak K, et al. CIViC is a community knowledgebase for expert crowdsourcing the clinical interpretation of variants in cancer. Nat Genet. 2017;49:170-174. Crossref, Medline, Google Scholar |
| 26. | Chakravarty D, Gao J, Phillips SM, et al. OncoKB: a precision oncology knowledge base. JCO Precis Oncol. 2017;2017:1-16. Google Scholar |
| 27. | Landrum MJ, Lee JM, Benson M, et al. ClinVar: public archive of interpretations of clinically relevant variants. Nucleic Acids Res. 2016;44(D1):D862-D868. Crossref, Medline, Google Scholar |
| 28. | Forbes SA, Beare D, Gunasekaran P, et al. COSMIC: exploring the world’s knowledge of somatic mutations in human cancer. Nucleic Acids Res. 2015;43(D1):D805-D811. Crossref, Medline, Google Scholar |
| 29. | Sherry ST, Ward MH, Kholodov M, et al. dbSNP: the NCBI database of genetic variation. Nucleic Acids Res. 2001;29:308-311. Crossref, Medline, Google Scholar |
| 30. | World Health Organization. International Clinical Trials Registry Platform. Geneva, Switzerland: World Health Organization; 2014. Google Scholar |
| 31. | Overby CL, Heale B, Aronson S, et al. 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. Google Scholar |
| 32. | Hunter JE, Irving SA, Biesecker LG, et al. A standardized, evidence-based protocol to assess clinical actionability of genetic disorders associated with genomic variation. Genet Med. 2016;18:1258-1268. Crossref, Medline, Google Scholar |
| 33. | Rehm HL, Berg JS, Brooks LD, et al; ClinGen. ClinGen—the Clinical Genome Resource. N Engl J Med. 2015;372:2235-2242. Crossref, Medline, Google Scholar |
| 34. | Kirkpatrick BE, Riggs ER, Azzariti DR, et al; ClinGen Resource. GenomeConnect: matchmaking between patients, clinical laboratories, and researchers to improve genomic knowledge. Hum Mutat. 2015;36:974-978. Google Scholar |
| 35. | Plaisant C, Wu J, Hettinger AZ, et al. Novel user interface design for medication reconciliation: an evaluation of Twinlist. J Am Med Inform Assoc. 2015;22:340-349. Google Scholar |
| 36. | Tarkan S, Plaisant C, Shneiderman B, et al. Reducing missed laboratory results: defining temporal responsibility, generating user interfaces for test process tracking, and retrospective analyses to identify problems. AMIA Annu Symp Proc. 2011;2011:1382-1391. Google Scholar |
