563643
research-article2015
DSTXXX10.1177/1932296814563643Journal of Diabetes Science and TechnologyKlonoff
Editorial
Precision Medicine for Managing Diabetes
Journal of Diabetes Science and Technology 2015, Vol. 9(1) 3–7 © 2015 Diabetes Technology Society Reprints and permissions: sagepub.com/journalsPermissions.nav DOI: 10.1177/1932296814563643 dst.sagepub.com
David C. Klonoff, MD, FACP, FRCP (Edin), Fellow AIMBE1 Keywords big data, diabetes, genomics, personalized medicine, pharmacogenomics, precision medicine
Precision medicine is a modern concept that has been used since 2011 to describe tailored accurate medical treatments selected according to individual characteristics of each patient. Each patient’s disease is analyzed according to molecular data, genomics, and systems biology in this model to establish the patient’s disease process at the molecular level and select appropriate treatments. The patient’s response is then closely monitored with direct or surrogate measures such as biomarkers, and the treatments can then be adapted according to the patient’s response. The combination of traditional gross and microscopic metrics combined with molecular profiling is precision medicine.1
Personalized Medicine vs Precision Medicine Precision medicine is a similar concept to personalized medicine, which is a term that has been used for the past decade. Personalized medicine refers to determining specific information about a patient and then prescribing a treatment that is specific for that patient. Personalized medicine involves defining disease subtypes and defining biomarkers that can identify (1) which patients who are most likely to benefit from a specific treatment and (2) which patients are most unlikely to respond or likely to experience side effects.2 Although precision medicine resembles personalized medicine, the term “precision medicine” is now the preferred term among most scientists who are applying genomics to classify disease for individualized treatments. We are now hearing about many new precision medicine research programs and companies. The main reason for the shift to precision medicine, in my opinion, is that personalized medicine as a concept has not delivered as many specific gains from genomic analyses as proponents of that field were hoping. Therefore, the genomics community has embraced a new term, “precision medicine,” as a refresh term to get people thinking again about individualizing therapy based on molecular information as well as other new types of genotypic and phenotypic information.3
introduction of big data into the process of individually profiling diseases and patients. Currently, over 2 billion people worldwide are connected to the Internet, and over 5 billion individuals own mobile phones. Furthermore, remote sensors are continuously producing much heterogeneous data that are either structured or unstructured. By the year 2020, 50 billion devices are expected to be connected to the Internet. At this point, predicted data production will be 44 times greater than that in 2009.4 Big data refers to the use of multiple unrelated complex databases analyzed together with nontraditional data processing applications to identify novel relationships. Many new types of databases can be linked with molecular databases to power a precision medicine concept. Most of these data sets were not readily available or usable just a few years ago. Thanks to emerging technologies, patient-specific data have recently become available from previously unmined sources.5 Modern tools for collecting large amounts of new types of digital data are listed in Table 1. These new sources of data have created new interest in personalized medicine treatments based on these new phenotypic databases combined with increasingly available molecular information, such as the genome, the epigenome, and the proteome. Greater individualization of medical care will become possible when traditional sources medical information (the history, physical examination, and laboratory panel) can be augmented by 2 new powerful sources of information. These sources are (1) data mining for multidimensional phenotypic data6 and (2) improved genomic and omics analyses for structural and functional genotypic data, including pharmacogenomic data.7 This is now the right time to reclassify the use of genomic information in medicine because of newly available sources of phenotypic and genotypic big data. The addition of data mining to improved genomic analyses has led to the new term “precision medicine” being applied to what was previously a more limited concept of classifying and specifically treating diseases known as “personalized medicine.” 1
Mills-Peninsula Health Services, San Mateo, CA, USA
Big Data An important reason for the migration from “personalized medicine” to the new term of “precision medicine” is the
Corresponding Author: David C. Klonoff, MD, FACP, FRCP (Edin), Fellow AIMBE, Mills-Peninsula Health Services, 100 S San Mateo Dr, Rm 5147, San Mateo, CA 94401, USA. Email: [email protected]
Downloaded from dst.sagepub.com at KAI NAN UNIV on February 18, 2015
4
Journal of Diabetes Science and Technology 9(1)
Table 1. Modern Tools for Collecting Data That Can Be Incorporated Into Precision Medicine. 1 2 3 4 5 6 7 8 9 10 11 12 13 14
Gene sequencers Omics testing Lab-on-a-chip for biomarkers Electronic nose Microbiome analysis 3-D medical imaging Accelerometers Global positioning systems Wearable physiologic sensors Implanted physiologic sensors Ubiquitous video cameras Medication compliance systems Electronic medical record Social media
Table 2. Recommendations of the National Research Council Committee on a Framework for Development of a New Taxonomy of Disease.9 A new taxonomy will lead to better health care The time is right to modernize disease taxonomy A new taxonomy should be developed A knowledge network of disease would enable a new taxonomy New models for population-based research will enable development of the knowledge network and the new taxonomy Redirection of resources could facilitate development of the knowledge network of disease
Role of Analyses in Precision Medicine
A greater understanding of the pathogenesis of a disease will lead to identification of clinically relevant biomarkers, possibly actionable genetic mutations, and ideally druggable (or therapeutic) targets. If a potential target is identified, then drug target validation must be conducted to ensure that the target has a causal role in the disease.8
The New Taxonomy Proponents of using the new term “precision medicine” instead of “personalized medicine” point out that some people had used the term “personalized medicine” incorrectly to mean that unique treatments can be designed for each patient or that research could legitimately include anecdotal success stories or “n of 1” studies. Precision medicine on the other hand implies that through analysis of 1 or more databases (probably including a genomic analysis), it is possible to identify a patient’s precise disease or subpopulation of that disease and prescribe treatments that can be predicted to be highly likely (but not certain) to be effective, based on the outcomes from the multiple databases that powered the analysis.3 The term “precision medicine” was first widely used in an important 143-page National Academy of Sciences report published in 2011 from the National Research Council Committee on a Framework for Development of a New Taxonomy of Disease, titled Toward Precision Medicine: Building a Knowledge Network for Biomedical Research and a New Taxonomy of Disease.9 The report presented the conclusions of a NRC committee that met to explore the feasibility and need for developing a new taxonomy of human diseases based on molecular biology and a framework for creating one. The committee reached made 6 recommendations on this topic (Table 2). The committee concluded that a major beneficiary of the proposed knowledge network of diseases and new taxonomy would be precision medicine, whose ultimate end point is the selection of a subset of patients who are most likely to benefit from a drug or other treatment (Figure 1).
A human genome contains more than three billion base pairs.10 The cost of obtaining a single human genome sequence has come down rapidly over the past decade and a half so that the type of genomic analysis required for precision medicine is much more readily available (ie, affordable) than ever. The National Human Genome Research Institute estimated that the total cost of obtaining a single human genome sequence in 2001 was $95 million. By the time of the taxonomy report in 2011 the cost was $21 000.6 In 2014 Illumina launched its HiSeq X Ten Sequencer, which delivers the first $1000 genome.11 The greater availability of such genetic information combined with new databases have both led to renewed hopes that specific descriptions of patients at the molecular level and individualized predictions of responses to treatments will soon become powerful tools for precision medicine. Thanks to new diagnostic tools and genomic analysis, predictive biomarkers and prognostic biomarkers have been identified for many diseases that can guide which treatment to be prescribed and at what point in the course of a disease. Specific therapies can be developed that are targeted against cells with specific features or genetic mutations. This precision medicine approach was first used on cancer and can also be used with other diseases. As drugs become developed targeted at specific genetic mutations, it is likely that accompanying diagnostic tests for biomarkers will also become available to confirm whether the target biomarker is present.
Genomic Analyses for Diabetes Risk Genes signifying increased risk for both type 1 and type 2 diabetes have been identified. Genomewide association studies have identified over 50 loci associated with an increased genetic risk of type 1 diabetes. Several T1D candidate genes for increased risk of developing type 1 diabetes have been suggested or identified within these regions, but the molecular basis by which they contribute to islet cell inflammation and beta cell destruction is not fully understood.12 Also, several candidate genes for increased risk of developing type 2 diabetes have been identified, including peroxisome proliferatoractivated receptor gamma (PPARγ2), angiotensin converting
Downloaded from dst.sagepub.com at KAI NAN UNIV on February 18, 2015
5
Klonoff
Figure 1. How the new taxonomy of disease will support precision medicine. An individual’s information commons contains all available biological knowledge to inform a knowledge network. Validated relationships within the knowledge network that define new diseases or disease subtypes will be incorporated into the new taxonomy of diseases. This taxonomy will lead to more accurate diagnoses, better targeted treatments, and improved outcomes, the 3 of which are the hallmark of precision medicine. Source: Modified from National Research Council Committee on a Framework for Developing a New Taxonomy of Disease.9
enzyme (ACE), methylene tetrahydrofolate reductase (MTHR), fatty acid binding protein-2 (FABP2), and fat mass and obesity associated gene (FTO).13 The conclusions of a “Workshop on Metformin Pharmacogenomics,” sponsored by the National Institute of Diabetes and Digestive and Kidney Diseases, were published in 2014.14 The meeting was intended to review metformin pharmacogenomics and identify both novel targets and more effective agents for diabetes. The idea behind the meeting was that understanding the genes and pathways that determine the response to metformin has the potential to reveal new drug targets for the treatment of diabetes. The group noted that there have been few genes associated with glycemic control by metformin, and the most reproducible associations have been in metformin transporter genes. They acknowledged that nongenetic factors also contribute to response to metformin and that broader system biology approaches will be required to model the combined effects of multiple gene variants and their interaction with nongenetic factors. They concluded that the overall challenge to the field of precision medicine as it relates to antidiabetes treatment is to identify the individualized factors that can lead to improved glycemic control. Omics is the characterization of various types of biomarkers that compose the structure and function of an organism, known as the metabolome. Omics reflects how information encoded at the genomic level is implemented at the transcriptomic, proteomic, and glycomic levels.15,16 In recent years it has become apparent that understanding the metabolomic pattern is a critically important supplement to genetic analysis and using metabolomic biomarker data will be important for precision medicine.17
Precision Diabetes Clinics I predict that soon we will be seeing the appearance of precision diabetes clinics. These clinics will be multidisciplinary, with emphasis on prevention for patients with prediabetes and adherence with therapy for patients with diabetes. Such clinics would use diabetes technology devices and mHealth tools. The clinic would be equipped to interpret wirelessly transmitted data streams of glycemic patterns delivered by self-monitoring of blood glucose and continuous glucose monitoring systems, as well as insulin doses and timing delivered by pumps18 and pens.19 All the data would be wirelessly transmitted to the cloud for subsequent review by patients, relatives, and health care professionals. A clinic would also work with emerging wearable devices that make physiologic measurements, track exercise, track sleep, or monitor food intake. The clinic would provide new technologies for monitoring adherence to treatment so physicians would know when poor outcomes should lead to dose escalation and when they should lead to new (and sometimes more costly or more risky) therapies. The clinic patients would be offered tests of the latest biomarkers, including genomic sequences, circulating biomarkers, and other types of biomarker metrics. These tests would be for (1) classification of disease and its complications, (2) assistance with selecting treatments, and (3) prognosis of complications to assist with setting treatment goals. A precision diabetes clinic will provide (1) genetic and omics information and counseling; (2) a digital phenotypic assessment and counseling based on an extensive set of laboratory tests, sensor information, digital images, and personalized data analysis; (3) sensor-based behavior assessment and lifestyle counseling; and (4) genomic-based pharmacotherapy where appropriate (Figure 2).
Downloaded from dst.sagepub.com at KAI NAN UNIV on February 18, 2015
6
Journal of Diabetes Science and Technology 9(1) precision medicine tests must be well validated for them to be useful, but for many genes associated with type 2 diabetes, the increased risk for each candidate gene is small and less than the risk of widely practiced unhealthy behaviors. Serial measurements of biomarkers may prove to be more useful than static measurements for predicting individual risk.24 Furthermore, there is too little information known at this time about what findings in one ethnic group might mean for a different group.
Conclusions
Figure 2. Services provided by a precision medicine diabetes clinic.
Barriers Several potential clinical and technical barriers will have to be overcome for precision medicine clinics to be successful. Four technical barriers to the success of precision medicine clinics include analysis, usability, privacy, and security. First, big databases are often siloed and can be distributed across multiple big data repositories, spreadsheets, mobile applications, random access memory caches, external clouds, videos, and unstructured social media websites.20 New analytics tools are needed to analyze and interpret the various types of data to reveal meaningful insights. Second, it is important for a medical clinic using big data to have a user gateway that can efficiently decompose queries and then quickly and accurately compose results.21 The gateway interface must contain a good user interface to facilitate effective data mining. Third, the privacy of sources of data composing the big database must be assured. Fourth, the security of data regarding clinic patients must be assured for clinic patients.22 Four clinical barriers to the success of precision medicine clinics include specificity of therapy, cost, reimbursement, and accuracy. First, phenotypic and behavioral data provided by sensors and other digital measurement tools can lead to effective structured treatment plans, but for many genetic and genomic variants, there is no clear consensus on how to treat patients with these findings.23 Second, the cost of genomic analysis and interpretation can still be a problem, even though the cost of such testing has fallen considerably in recent years. Third, there is no clear policy for reimbursement for this new approach. Fourth, the specificity of the
Precision medicine promises to combine individual data about genetic predispositions to diseases, biomarker information about disease risks and responses, and physiologic and behavioral data from new sensors and databases to create rich patterns that can predict risks and responses in a very precise way. The new taxonomy of diseases that is now being developed will aid the pursuit of precision medicine by defining at the molecular level new diseases and subtypes that can be treated in a specific way. Using genetic data and other databases from a big data perspective, diabetes and other diseases will become classified into disease subsets, all with their own best treatments. Precision medicine clinics will channel new sources of information assembled into clinically focused data sets and will use the emerging data streams to deliver individualized treatments for patients with phenotypically similar but genotypically and molecularly dissimilar diseases. Abbreviations ACE, angiotensin converting enzyme; FABP2, fatty acid binding protein-2; FTO, fat mass and obesity associated gene; MTHR, methylene tetrahydrofolate reductase; PPARγ2, peroxisome proliferator-activated receptor gamma.
Declaration of Conflicting Interests The author declared the following potential conflicts of interest with respect to the research, authorship, and/or publication of this article: consultant for Google, Insuline, Roche, Sanofi, TempraMed, Voluntis; stockholder in TempraMed.
Funding The author received no financial support for the research, authorship, and/or publication of this article.
References 1. Mirnezami R, Nicholson J, Darzi A. Preparing for precision medicine. N Engl J Med. 2012;366(6):489-491. 2. Klonoff DC. Personalized medicine for diabetes. J Diabetes Sci Technol. 2008;2(3):335-341. 3. Timmerman L. What’s in a name? A lot, when it comes to “precision medicine.” February 4, 2013. Available at: http://
Downloaded from dst.sagepub.com at KAI NAN UNIV on February 18, 2015
7
Klonoff www.xconomy.com/national/2013/02/04/whats-in-a-name-alot-when-it-comes-to-precision-medicine/?single_page=true. Accessed September 9, 2014. 4. Khan N, Yaqoob I, Hashem IA, et al. Big data: survey, technologies, opportunities, and challenges. Scientific World Journal. 2014;2014:712826. 5. Klonoff DC. Twelve modern digital technologies that are transforming decision making for diabetes and all areas of health care. J Diabetes Sci Technol. 2013;7:291-295. 6. Munson ME, Wrobel JS, Holmes CM1, Hanauer DA. Data mining for identifying novel associations and temporal relationships with Charcot foot. J Diabetes Res. 2014;2014:214353. doi:10.1155/2014/214353. 7. Heinzel A, Perco P, Mayer G, Oberbauer R, Lukas A, Mayer B. From molecular signatures to predictive biomarkers: modeling disease pathophysiology and drug mechanism of action. Front Cell Dev Biol. 2014;2:37. doi:10.3389/fcell.2014.00037. 8. Lemoine C. Precision medicine for nurses: 101. Seminar Oncol Nurs. 2014;30(2):84-89. 9. National Research Council 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: National Academies Press; 2011. 10. National Human Genome Research Institute. The Human Genome Project completion: frequently asked questions. October 30, 2010. Available at: http://www.genome. gov/11006943. Accessed September 9, 2014. 11. Illumina. Systems/HiSeq X Ten. Available at: http://sys tems.illumina.com/systems/hiseq-x-sequencing-system.ilmn. Accessed September 9, 2014. 12. Santin I, Eizirik DL. Candidate genes for type 1 diabetes modulate pancreatic islet inflammation and β-cell apoptosis. Diabetes Obes Metab. 2013;15(suppl 3):71-81. 13. Abbas S, Raza ST, Ahmed F, et al. Association of genetic polymorphism of PPARγ-2, ACE, MTHFR, FABP-2 and FTO genes in risk prediction of type 2 diabetes mellitus. J Biomed Sci. 2013;20(1):80.
14. Pawlyk AC, Giacomini KM, McKeon C, et al. Metformin pharmacogenomics: current status and future directions. Diabetes. 2014;63(8):2590-2599. 15. Trifonova O, Lokhov P, Archakov A. Postgenomics diag nostics: metabolomics approaches to human blood profiling. OMICS. 2013;17(11):550-559. doi:10.1089/omi.2012. 0121. 16. Frey LJ. Informatics tools to advance the biology of gly cosaminoglycans and proteoglycans. Methods Mol Biol. 2015;1229:271-287. doi:10.1007/978-1-4939-1714-3_23. 17. Peer D. Precision medicine—delivering the goods? Cancer Lett. 2014;352(1):2-3. doi:10.1016/j.canlet.2014.04.011. 18. Driscoll KA, Johnson SB, Wang Y, et al. Importance of manually entering blood glucose readings when wireless-compatible meters are not being used with an insulin pump. J Diabetes Sci Technol. 2013;7(4):898-903. 19. New KiCoPen developed by Cambridge Consultants could make managing diabetes treatment easier. October 6, 2014. Available at: http://www.cambridge-news.co.uk/NewKiCoPen-developed-Cambridge-Consultants-make/story23050566-detail/story.htm. Accessed November 19, 2014. 20. Merelli I, Pérez-Sánchez H, Gesing S, D’Agostino D. Managing, analysing, and integrating big data in medical bioinformatics: open problems and future perspectives. Biomed Res Int. 2014;2014:134023. doi:10.1155/2014/134023. 21. Taha K, Elmasri R. GMB: an efficient query processor for biological data. J Integr Bioinform. 2011;8(2):165. doi:10.2390/ biecoll-jib-2011-165. 22. Khan N, Yaqoob I, Hashem IA, et al. Big data: survey, technologies, opportunities, and challenges. Scientific World Journal. 2014;2014:712826. doi:10.1155/2014/712826. 23. Shah S, Horne A, Capellá J. Good data won’t guarantee good decisions. Harvard Business Review. April 2012. Available at: https://hbr.org/2012/04/good-data-wont-guarantee-good-decisions/ar/1. Accessed November 19, 2014. 24. Herder C, Kowall B, Tabak AG, Rathmann W. The potential of novel biomarkers to improve risk prediction of type 2 diabetes. Diabetologia. 2014;57(1):16-29.
Downloaded from dst.sagepub.com at KAI NAN UNIV on February 18, 2015
research-article2015
DSTXXX10.1177/1932296814563643Journal of Diabetes Science and TechnologyKlonoff
Editorial
Precision Medicine for Managing Diabetes
Journal of Diabetes Science and Technology 2015, Vol. 9(1) 3–7 © 2015 Diabetes Technology Society Reprints and permissions: sagepub.com/journalsPermissions.nav DOI: 10.1177/1932296814563643 dst.sagepub.com
David C. Klonoff, MD, FACP, FRCP (Edin), Fellow AIMBE1 Keywords big data, diabetes, genomics, personalized medicine, pharmacogenomics, precision medicine
Precision medicine is a modern concept that has been used since 2011 to describe tailored accurate medical treatments selected according to individual characteristics of each patient. Each patient’s disease is analyzed according to molecular data, genomics, and systems biology in this model to establish the patient’s disease process at the molecular level and select appropriate treatments. The patient’s response is then closely monitored with direct or surrogate measures such as biomarkers, and the treatments can then be adapted according to the patient’s response. The combination of traditional gross and microscopic metrics combined with molecular profiling is precision medicine.1
Personalized Medicine vs Precision Medicine Precision medicine is a similar concept to personalized medicine, which is a term that has been used for the past decade. Personalized medicine refers to determining specific information about a patient and then prescribing a treatment that is specific for that patient. Personalized medicine involves defining disease subtypes and defining biomarkers that can identify (1) which patients who are most likely to benefit from a specific treatment and (2) which patients are most unlikely to respond or likely to experience side effects.2 Although precision medicine resembles personalized medicine, the term “precision medicine” is now the preferred term among most scientists who are applying genomics to classify disease for individualized treatments. We are now hearing about many new precision medicine research programs and companies. The main reason for the shift to precision medicine, in my opinion, is that personalized medicine as a concept has not delivered as many specific gains from genomic analyses as proponents of that field were hoping. Therefore, the genomics community has embraced a new term, “precision medicine,” as a refresh term to get people thinking again about individualizing therapy based on molecular information as well as other new types of genotypic and phenotypic information.3
introduction of big data into the process of individually profiling diseases and patients. Currently, over 2 billion people worldwide are connected to the Internet, and over 5 billion individuals own mobile phones. Furthermore, remote sensors are continuously producing much heterogeneous data that are either structured or unstructured. By the year 2020, 50 billion devices are expected to be connected to the Internet. At this point, predicted data production will be 44 times greater than that in 2009.4 Big data refers to the use of multiple unrelated complex databases analyzed together with nontraditional data processing applications to identify novel relationships. Many new types of databases can be linked with molecular databases to power a precision medicine concept. Most of these data sets were not readily available or usable just a few years ago. Thanks to emerging technologies, patient-specific data have recently become available from previously unmined sources.5 Modern tools for collecting large amounts of new types of digital data are listed in Table 1. These new sources of data have created new interest in personalized medicine treatments based on these new phenotypic databases combined with increasingly available molecular information, such as the genome, the epigenome, and the proteome. Greater individualization of medical care will become possible when traditional sources medical information (the history, physical examination, and laboratory panel) can be augmented by 2 new powerful sources of information. These sources are (1) data mining for multidimensional phenotypic data6 and (2) improved genomic and omics analyses for structural and functional genotypic data, including pharmacogenomic data.7 This is now the right time to reclassify the use of genomic information in medicine because of newly available sources of phenotypic and genotypic big data. The addition of data mining to improved genomic analyses has led to the new term “precision medicine” being applied to what was previously a more limited concept of classifying and specifically treating diseases known as “personalized medicine.” 1
Mills-Peninsula Health Services, San Mateo, CA, USA
Big Data An important reason for the migration from “personalized medicine” to the new term of “precision medicine” is the
Corresponding Author: David C. Klonoff, MD, FACP, FRCP (Edin), Fellow AIMBE, Mills-Peninsula Health Services, 100 S San Mateo Dr, Rm 5147, San Mateo, CA 94401, USA. Email: [email protected]
Downloaded from dst.sagepub.com at KAI NAN UNIV on February 18, 2015
4
Journal of Diabetes Science and Technology 9(1)
Table 1. Modern Tools for Collecting Data That Can Be Incorporated Into Precision Medicine. 1 2 3 4 5 6 7 8 9 10 11 12 13 14
Gene sequencers Omics testing Lab-on-a-chip for biomarkers Electronic nose Microbiome analysis 3-D medical imaging Accelerometers Global positioning systems Wearable physiologic sensors Implanted physiologic sensors Ubiquitous video cameras Medication compliance systems Electronic medical record Social media
Table 2. Recommendations of the National Research Council Committee on a Framework for Development of a New Taxonomy of Disease.9 A new taxonomy will lead to better health care The time is right to modernize disease taxonomy A new taxonomy should be developed A knowledge network of disease would enable a new taxonomy New models for population-based research will enable development of the knowledge network and the new taxonomy Redirection of resources could facilitate development of the knowledge network of disease
Role of Analyses in Precision Medicine
A greater understanding of the pathogenesis of a disease will lead to identification of clinically relevant biomarkers, possibly actionable genetic mutations, and ideally druggable (or therapeutic) targets. If a potential target is identified, then drug target validation must be conducted to ensure that the target has a causal role in the disease.8
The New Taxonomy Proponents of using the new term “precision medicine” instead of “personalized medicine” point out that some people had used the term “personalized medicine” incorrectly to mean that unique treatments can be designed for each patient or that research could legitimately include anecdotal success stories or “n of 1” studies. Precision medicine on the other hand implies that through analysis of 1 or more databases (probably including a genomic analysis), it is possible to identify a patient’s precise disease or subpopulation of that disease and prescribe treatments that can be predicted to be highly likely (but not certain) to be effective, based on the outcomes from the multiple databases that powered the analysis.3 The term “precision medicine” was first widely used in an important 143-page National Academy of Sciences report published in 2011 from the National Research Council Committee on a Framework for Development of a New Taxonomy of Disease, titled Toward Precision Medicine: Building a Knowledge Network for Biomedical Research and a New Taxonomy of Disease.9 The report presented the conclusions of a NRC committee that met to explore the feasibility and need for developing a new taxonomy of human diseases based on molecular biology and a framework for creating one. The committee reached made 6 recommendations on this topic (Table 2). The committee concluded that a major beneficiary of the proposed knowledge network of diseases and new taxonomy would be precision medicine, whose ultimate end point is the selection of a subset of patients who are most likely to benefit from a drug or other treatment (Figure 1).
A human genome contains more than three billion base pairs.10 The cost of obtaining a single human genome sequence has come down rapidly over the past decade and a half so that the type of genomic analysis required for precision medicine is much more readily available (ie, affordable) than ever. The National Human Genome Research Institute estimated that the total cost of obtaining a single human genome sequence in 2001 was $95 million. By the time of the taxonomy report in 2011 the cost was $21 000.6 In 2014 Illumina launched its HiSeq X Ten Sequencer, which delivers the first $1000 genome.11 The greater availability of such genetic information combined with new databases have both led to renewed hopes that specific descriptions of patients at the molecular level and individualized predictions of responses to treatments will soon become powerful tools for precision medicine. Thanks to new diagnostic tools and genomic analysis, predictive biomarkers and prognostic biomarkers have been identified for many diseases that can guide which treatment to be prescribed and at what point in the course of a disease. Specific therapies can be developed that are targeted against cells with specific features or genetic mutations. This precision medicine approach was first used on cancer and can also be used with other diseases. As drugs become developed targeted at specific genetic mutations, it is likely that accompanying diagnostic tests for biomarkers will also become available to confirm whether the target biomarker is present.
Genomic Analyses for Diabetes Risk Genes signifying increased risk for both type 1 and type 2 diabetes have been identified. Genomewide association studies have identified over 50 loci associated with an increased genetic risk of type 1 diabetes. Several T1D candidate genes for increased risk of developing type 1 diabetes have been suggested or identified within these regions, but the molecular basis by which they contribute to islet cell inflammation and beta cell destruction is not fully understood.12 Also, several candidate genes for increased risk of developing type 2 diabetes have been identified, including peroxisome proliferatoractivated receptor gamma (PPARγ2), angiotensin converting
Downloaded from dst.sagepub.com at KAI NAN UNIV on February 18, 2015
5
Klonoff
Figure 1. How the new taxonomy of disease will support precision medicine. An individual’s information commons contains all available biological knowledge to inform a knowledge network. Validated relationships within the knowledge network that define new diseases or disease subtypes will be incorporated into the new taxonomy of diseases. This taxonomy will lead to more accurate diagnoses, better targeted treatments, and improved outcomes, the 3 of which are the hallmark of precision medicine. Source: Modified from National Research Council Committee on a Framework for Developing a New Taxonomy of Disease.9
enzyme (ACE), methylene tetrahydrofolate reductase (MTHR), fatty acid binding protein-2 (FABP2), and fat mass and obesity associated gene (FTO).13 The conclusions of a “Workshop on Metformin Pharmacogenomics,” sponsored by the National Institute of Diabetes and Digestive and Kidney Diseases, were published in 2014.14 The meeting was intended to review metformin pharmacogenomics and identify both novel targets and more effective agents for diabetes. The idea behind the meeting was that understanding the genes and pathways that determine the response to metformin has the potential to reveal new drug targets for the treatment of diabetes. The group noted that there have been few genes associated with glycemic control by metformin, and the most reproducible associations have been in metformin transporter genes. They acknowledged that nongenetic factors also contribute to response to metformin and that broader system biology approaches will be required to model the combined effects of multiple gene variants and their interaction with nongenetic factors. They concluded that the overall challenge to the field of precision medicine as it relates to antidiabetes treatment is to identify the individualized factors that can lead to improved glycemic control. Omics is the characterization of various types of biomarkers that compose the structure and function of an organism, known as the metabolome. Omics reflects how information encoded at the genomic level is implemented at the transcriptomic, proteomic, and glycomic levels.15,16 In recent years it has become apparent that understanding the metabolomic pattern is a critically important supplement to genetic analysis and using metabolomic biomarker data will be important for precision medicine.17
Precision Diabetes Clinics I predict that soon we will be seeing the appearance of precision diabetes clinics. These clinics will be multidisciplinary, with emphasis on prevention for patients with prediabetes and adherence with therapy for patients with diabetes. Such clinics would use diabetes technology devices and mHealth tools. The clinic would be equipped to interpret wirelessly transmitted data streams of glycemic patterns delivered by self-monitoring of blood glucose and continuous glucose monitoring systems, as well as insulin doses and timing delivered by pumps18 and pens.19 All the data would be wirelessly transmitted to the cloud for subsequent review by patients, relatives, and health care professionals. A clinic would also work with emerging wearable devices that make physiologic measurements, track exercise, track sleep, or monitor food intake. The clinic would provide new technologies for monitoring adherence to treatment so physicians would know when poor outcomes should lead to dose escalation and when they should lead to new (and sometimes more costly or more risky) therapies. The clinic patients would be offered tests of the latest biomarkers, including genomic sequences, circulating biomarkers, and other types of biomarker metrics. These tests would be for (1) classification of disease and its complications, (2) assistance with selecting treatments, and (3) prognosis of complications to assist with setting treatment goals. A precision diabetes clinic will provide (1) genetic and omics information and counseling; (2) a digital phenotypic assessment and counseling based on an extensive set of laboratory tests, sensor information, digital images, and personalized data analysis; (3) sensor-based behavior assessment and lifestyle counseling; and (4) genomic-based pharmacotherapy where appropriate (Figure 2).
Downloaded from dst.sagepub.com at KAI NAN UNIV on February 18, 2015
6
Journal of Diabetes Science and Technology 9(1) precision medicine tests must be well validated for them to be useful, but for many genes associated with type 2 diabetes, the increased risk for each candidate gene is small and less than the risk of widely practiced unhealthy behaviors. Serial measurements of biomarkers may prove to be more useful than static measurements for predicting individual risk.24 Furthermore, there is too little information known at this time about what findings in one ethnic group might mean for a different group.
Conclusions
Figure 2. Services provided by a precision medicine diabetes clinic.
Barriers Several potential clinical and technical barriers will have to be overcome for precision medicine clinics to be successful. Four technical barriers to the success of precision medicine clinics include analysis, usability, privacy, and security. First, big databases are often siloed and can be distributed across multiple big data repositories, spreadsheets, mobile applications, random access memory caches, external clouds, videos, and unstructured social media websites.20 New analytics tools are needed to analyze and interpret the various types of data to reveal meaningful insights. Second, it is important for a medical clinic using big data to have a user gateway that can efficiently decompose queries and then quickly and accurately compose results.21 The gateway interface must contain a good user interface to facilitate effective data mining. Third, the privacy of sources of data composing the big database must be assured. Fourth, the security of data regarding clinic patients must be assured for clinic patients.22 Four clinical barriers to the success of precision medicine clinics include specificity of therapy, cost, reimbursement, and accuracy. First, phenotypic and behavioral data provided by sensors and other digital measurement tools can lead to effective structured treatment plans, but for many genetic and genomic variants, there is no clear consensus on how to treat patients with these findings.23 Second, the cost of genomic analysis and interpretation can still be a problem, even though the cost of such testing has fallen considerably in recent years. Third, there is no clear policy for reimbursement for this new approach. Fourth, the specificity of the
Precision medicine promises to combine individual data about genetic predispositions to diseases, biomarker information about disease risks and responses, and physiologic and behavioral data from new sensors and databases to create rich patterns that can predict risks and responses in a very precise way. The new taxonomy of diseases that is now being developed will aid the pursuit of precision medicine by defining at the molecular level new diseases and subtypes that can be treated in a specific way. Using genetic data and other databases from a big data perspective, diabetes and other diseases will become classified into disease subsets, all with their own best treatments. Precision medicine clinics will channel new sources of information assembled into clinically focused data sets and will use the emerging data streams to deliver individualized treatments for patients with phenotypically similar but genotypically and molecularly dissimilar diseases. Abbreviations ACE, angiotensin converting enzyme; FABP2, fatty acid binding protein-2; FTO, fat mass and obesity associated gene; MTHR, methylene tetrahydrofolate reductase; PPARγ2, peroxisome proliferator-activated receptor gamma.
Declaration of Conflicting Interests The author declared the following potential conflicts of interest with respect to the research, authorship, and/or publication of this article: consultant for Google, Insuline, Roche, Sanofi, TempraMed, Voluntis; stockholder in TempraMed.
Funding The author received no financial support for the research, authorship, and/or publication of this article.
References 1. Mirnezami R, Nicholson J, Darzi A. Preparing for precision medicine. N Engl J Med. 2012;366(6):489-491. 2. Klonoff DC. Personalized medicine for diabetes. J Diabetes Sci Technol. 2008;2(3):335-341. 3. Timmerman L. What’s in a name? A lot, when it comes to “precision medicine.” February 4, 2013. Available at: http://
Downloaded from dst.sagepub.com at KAI NAN UNIV on February 18, 2015
7
Klonoff www.xconomy.com/national/2013/02/04/whats-in-a-name-alot-when-it-comes-to-precision-medicine/?single_page=true. Accessed September 9, 2014. 4. Khan N, Yaqoob I, Hashem IA, et al. Big data: survey, technologies, opportunities, and challenges. Scientific World Journal. 2014;2014:712826. 5. Klonoff DC. Twelve modern digital technologies that are transforming decision making for diabetes and all areas of health care. J Diabetes Sci Technol. 2013;7:291-295. 6. Munson ME, Wrobel JS, Holmes CM1, Hanauer DA. Data mining for identifying novel associations and temporal relationships with Charcot foot. J Diabetes Res. 2014;2014:214353. doi:10.1155/2014/214353. 7. Heinzel A, Perco P, Mayer G, Oberbauer R, Lukas A, Mayer B. From molecular signatures to predictive biomarkers: modeling disease pathophysiology and drug mechanism of action. Front Cell Dev Biol. 2014;2:37. doi:10.3389/fcell.2014.00037. 8. Lemoine C. Precision medicine for nurses: 101. Seminar Oncol Nurs. 2014;30(2):84-89. 9. National Research Council 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: National Academies Press; 2011. 10. National Human Genome Research Institute. The Human Genome Project completion: frequently asked questions. October 30, 2010. Available at: http://www.genome. gov/11006943. Accessed September 9, 2014. 11. Illumina. Systems/HiSeq X Ten. Available at: http://sys tems.illumina.com/systems/hiseq-x-sequencing-system.ilmn. Accessed September 9, 2014. 12. Santin I, Eizirik DL. Candidate genes for type 1 diabetes modulate pancreatic islet inflammation and β-cell apoptosis. Diabetes Obes Metab. 2013;15(suppl 3):71-81. 13. Abbas S, Raza ST, Ahmed F, et al. Association of genetic polymorphism of PPARγ-2, ACE, MTHFR, FABP-2 and FTO genes in risk prediction of type 2 diabetes mellitus. J Biomed Sci. 2013;20(1):80.
14. Pawlyk AC, Giacomini KM, McKeon C, et al. Metformin pharmacogenomics: current status and future directions. Diabetes. 2014;63(8):2590-2599. 15. Trifonova O, Lokhov P, Archakov A. Postgenomics diag nostics: metabolomics approaches to human blood profiling. OMICS. 2013;17(11):550-559. doi:10.1089/omi.2012. 0121. 16. Frey LJ. Informatics tools to advance the biology of gly cosaminoglycans and proteoglycans. Methods Mol Biol. 2015;1229:271-287. doi:10.1007/978-1-4939-1714-3_23. 17. Peer D. Precision medicine—delivering the goods? Cancer Lett. 2014;352(1):2-3. doi:10.1016/j.canlet.2014.04.011. 18. Driscoll KA, Johnson SB, Wang Y, et al. Importance of manually entering blood glucose readings when wireless-compatible meters are not being used with an insulin pump. J Diabetes Sci Technol. 2013;7(4):898-903. 19. New KiCoPen developed by Cambridge Consultants could make managing diabetes treatment easier. October 6, 2014. Available at: http://www.cambridge-news.co.uk/NewKiCoPen-developed-Cambridge-Consultants-make/story23050566-detail/story.htm. Accessed November 19, 2014. 20. Merelli I, Pérez-Sánchez H, Gesing S, D’Agostino D. Managing, analysing, and integrating big data in medical bioinformatics: open problems and future perspectives. Biomed Res Int. 2014;2014:134023. doi:10.1155/2014/134023. 21. Taha K, Elmasri R. GMB: an efficient query processor for biological data. J Integr Bioinform. 2011;8(2):165. doi:10.2390/ biecoll-jib-2011-165. 22. Khan N, Yaqoob I, Hashem IA, et al. Big data: survey, technologies, opportunities, and challenges. Scientific World Journal. 2014;2014:712826. doi:10.1155/2014/712826. 23. Shah S, Horne A, Capellá J. Good data won’t guarantee good decisions. Harvard Business Review. April 2012. Available at: https://hbr.org/2012/04/good-data-wont-guarantee-good-decisions/ar/1. Accessed November 19, 2014. 24. Herder C, Kowall B, Tabak AG, Rathmann W. The potential of novel biomarkers to improve risk prediction of type 2 diabetes. Diabetologia. 2014;57(1):16-29.
Downloaded from dst.sagepub.com at KAI NAN UNIV on February 18, 2015
