Community Case Studies

Return to Table of Contents

Baylor College of Medicine

Houston, Texas

Peter Traber, MD

President and CEO, Baylor College of Medicine.

OVERVIEW

• Baylor College of Medicine (BCM) seeks to establish an institution-wide approach to personalized medicine with the three core values of quality, service, and individualized care that ensures that each patient receives the proper intervention or treatment at the right time based on his or her unique biology.
• The Personalized Medicine Alliance of BCM seeks to align the missions of the College and members of the community into a single vision of individualized patient care in the genomic age.
• BCM is actively working towards a future in which literacy about genomics and personalized medicine is broad-based in society, by adding innovative resources and programs across the educational spectrum.
• The Baylor Clinic and Hospital, with a planned opening in 2011, is designed de novo to combine the best in science, information technology and compassionate treatment for the kind of personalized medicine now possible and in readiness to implement future phases of genomic medicine.
• BCM has the breadth and depth in human disease genetics − from medical testing through basic genomics research − to fulfill the scientific promise of personalized medicine. The BCM chip, for example, is being developed as an important new tool for diagnosis and treatment.
• A key component of BCM’s comprehensive personalized medicine program will be an information system that integrates electronic health records with other scientific data on a standardized platform to facilitate inter-institutional collaboration.

PHILOSOPHY

Personalized medicine involves the tailoring of prediction, prognostication, diagnostics and therapeutics to the individual, based on that person’s particular biological makeup, to ensure that the right thing is done for the right person at the right time. This requires not only advances in medical technology, but also the coincident development of better information infrastructure, better integration of clinical and research efforts, continuing innovations in medical education, and finally, a deep relationship with the patient that makes that person a partner in the healthcare he or she receives.

Academic medical centers, such as Baylor College of Medicine (BCM), are not only especially qualified to forge a path towards this future, but they have a social obligation to do so.  Academic medical centers bring vast experience in the rapid incorporation of research and clinical innovation into multiple levels of education; and, for personalized medicine to reach its potential, educating health care professionals and the general public to engage in a common dialogue will be as important as training the next generation of physicians.   Furthermore, personalized medicine has the potential to form an attractive unifying vision for academic medical centers and for healthcare solutions more generally, by focusing the efforts of scientists, educations, and clinicians directly on the patient. Thus, Baylor College of Medicine’s leadership seeks to establish an institution-wide approach to personalized medicine with three core values:

• Quality: Providing evidence-based care with the highest priority to patient safety, proven best practices and clinical outcomes.
• Service: Using processes and information that are convenient, easy, understandable and focused on the individual patient to provide an integrated healthcare experience.
• Individualized Care: Ensuring that each patient receives the proper intervention or treatment at the right time by understanding and taking into account the biology of that person.

Making significant progress towards the goal of personalized medicine will require the cross-institutional involvement of many individuals active in all three of BCM’s missions of education, research and patient care. This poses an organizational challenge for institutions as complex as academic medical centers. To bridge gaps in philosophy and execution, BCM will establish a Personalized Medicine Alliance to include all members of the BCM community – physicians, students, residents, bench researchers, administrators, its eight affiliated hospitals and institutions as well as alumni and members of the community – in an evolving organization that can combine the best of genomic medicine with patient care.  BCM views such an alliance, rather than a new department or center, as the most effective way to affect the culture across multiple activities in the institution.

Establishing the Personalized Medicine Alliance as a virtual presence at Baylor College of Medicine will provide an important nexus for future innovation and planning. As a measure of institutional commitment, The Alliance will be chaired by BCM’s President and CEO and include experts and advisors from across the College. We plan to announce the Alliance as an entity during fall 2008 and begin actively recruiting partners from within the College and its affiliated institutions.

Using technology that will also evolve with the science, BCM seeks through its Alliance to make delivery of healthcare a personal experience from the moment a patient enters its clinic and/or hospital. The genetic information that defines an individual will be only part of the picture. The Alliance also seeks to find ways to use that information that takes into account the needs and preferences of each patient, giving that person the best in personalized medicine.

We are designing the Baylor Clinic and Hospital, now under construction, from a blank slate to augment the goals of the Alliance, BCM and the patients who seek care from its physicians and institutions. In doing so, we hope to educate a new generation of physicians and allied health professionals. Achieving this goal at BCM involves more than developing a gene chip, a new hospital and clinic and an electronic medical record – all projects underway. It means bridging gaps within and outside of the institution itself in order to make medicine personal in an entirely new way, using the patient’s individual genetic code to define diagnosis, prevention and care for a lifetime.

HISTORY & RESOURCES

Founded in Dallas in 1900 as a proprietary medical school, BCM affiliated with Baylor University in 1903. In 1943, it moved from Dallas to Houston and in 1947, occupied the first building of the actual Texas Medical Center.  For many of the healthcare institutions that subsequently arose in the Texas Medical Center, an affiliation with BCM was a key driver of success. BCM became independent of Baylor University in 1969, leading to State support of medical education, enhanced federal funding, and a strong self-sustaining philanthropic Board of Trustees. This sparked a remarkable growth, particularly in research programs.

Today, BCM is affiliated with eight hospital or healthcare institutions, including Texas Children’s Hospital, the nation’s largest children’s hospital; the Harris County Hospital District and Ben Taub Hospital, where the indigent in the county receive care; the Michael E. DeBakey Veterans Affairs Medical Center, one of the best veterans institutions in the nation; The Institute for Rehabilitation and Research, a prominent site of rehabilitation medicine; The Menninger Clinic, a nationally ranked mental health facility; The Methodist Health System and St. Luke’s Episcopal Hospital, two of Houston’s largest private hospitals. It has academic affiliates across the state and works closely with NASA through the National Space Biomedical Research Institute.

BCM has a broad range of education, research, and patient care programs including 25 clinical and basic departments, more than 90 research and patient-care centers, over 2,000 faculty members, 650 medical school students, 600 graduate school students, 125 allied health students, 700 postdoctoral researchers, and over 1,000 resident physicians. The College receives more than $230 million in research dollars from the National Institutes of Health and a total of approximately $310 million in research funding.

Of particular relevance to the Personalized Medicine Alliance, is BCM’s proven expertise and experience in a continuum of basic biological, genetic, and genomic research and translation into clinical genetic testing. BCM’s Department of Molecular and Human Genetics is ranked first among similar departments in funding from the National Institutes of Health and is responsible for multiple groundbreaking contributions to both understanding the genetics of disease and translating genetic discoveries into the clinic. The Medical Genetics Laboratories at BCM have been dedicated to providing high quality comprehensive diagnostic services for over 30 years. Finally, BCM’s Human Genome Sequence Center (HGSC) was one of three teams leading the final sprint to complete the initial publically funded sequencing of the human genome, announced in June 2000.  Last year, the Center, in collaboration with 454 Life Sciences, completed the annotated DNA sequence of Nobel laureate and DNA pioneer, Dr. James Watson. Since that time, the Center has continued to work towards increasing the speed and decreasing the cost of sequencing an individual genome.

CURRENT EFFORTS, CHALLENGES AND PLANS

Education

While academic medical centers advance patient care and research, the primary mission of medical schools is to educate the next generation leaders in healthcare and biomedical research systems. Academic medical centers are responsible for providing budding physicians with the tools necessary to navigate and lead a future of dramatic change.

Postgraduate medical education in specialties and subspecialties has been done essentially the same way for decades. To advance the training of individuals who will champion personalized medicine, BCM is exploring a Genomic Leadership Residency Program to enable young physicians to make use of the new technology and provide a basis for translating new information from bench to bedside. Training these young physicians in the Baylor Clinic and Hospital, which is expressly designed to foster this kind of work, will speed the integration of gene-based research into direct patient care in a compassionate and intelligent manner. The period of graduate medical education, when the direction of a young physician’s practice patterns are set, provides the best possible “teaching moment” to move the way medicine is practiced into the next generation.

The vision is to develop a creative new approach to graduate medical education that focuses on preparing the next generation of clinician-scientists and academic leaders, who will lead the transformation of medicine in the 21^st century. The BCM Genomic Leadership Residency Program would be open to any physician in its ACGME-specialties, functioning as a multi-year academic track graduate medical education experience with a fully integrated research component. Residents would enter the program through the clinical specialty of their choice and at the end of the residency, be board-eligible in the specialty they choose. While the new Baylor Clinic and Hospital – designed to foster personalized medicine – will serve as the residency’s “home,” residents will also rotate through affiliated institutions for a well-rounded experience.

Key characteristics of the Genomic Leadership Residency Program will include:

• A core curriculum in the science of genetics/genomics, molecular and cell biology, etc., with an emphasis on the translation of science to clinical practice;
• A core curriculum in leadership that emphasizes inter-personal and communication skills, ethics and professionalism, system-based practice, and approaches to quality assessment and clinical outcomes, with each curricular element addressed in the context of the unique issues and challenges presented by genomic medicine;
• A strong multi-disciplinary emphasis, with most of the above core curricular elements delivered across traditional specialty lines;
• A required one or two year research experience, inter-disciplinary in nature where appropriate, which will be fully integrated into the residency program at intervals that assure that graduating residents are fully prepared as both clinicians and scholars; and
• A core set of multi-disciplinary clinical rotations that provide opportunities to work across specialties in applying genomic diagnostic tools and therapeutic interventions to the care of patients.

Educating the next generation of leaders, however, is not enough. BCM views its educational mission as spanning the entire life cycle of education, from the earliest years in grade school though continuing professional education programs. BCM sees a future in which literacy about genomics and personalized medicine is broad-based in society as critical to the success of personalized medicine.  However, informing academic peers, the general medical community and the community at large about the reality of personalized medicine today and its promise in the future looms as a major challenge for BCM and the personalized medicine community itself.

The majority of the healthcare and academic work force trained before the genomic era and lack knowledge of the building blocks of personalized medicine.  Furthermore, there is skepticism and an incomplete understanding of the role genomics along with an improved information technology infrastructure will play in personalizing medicine, which threatens to block acceptance and hinder efforts to go forward. Educational programs at academic medical centers must establish a realistic perspective on the opportunities provided by personalized medicine and a general timeline for reaching goals. Moreover, education cannot be focused at only a single level, but rather must be distributed across the spectrum of education, including training of teachers.  Finally, there is the challenge of organizing the ever advancing information in this field so that it can reach the intended audience and be authoritative and realistic in its assertions.

BCM is addressing these challenges by incorporating the building blocks of personalized medicine across a wide range of educational programs (see Table 1).   This includes revising both formalized curricula – from undergraduate genetics courses to continuing medical education accreditation – and more informal opportunities in the form of web-based seminars and novel educational tools for younger students.  We are also exploring ways to structure information and educational materials so that learners can tailor their experience by choosing from a broad range of educational opportunities.

Through these kinds of educational and community outreach initiatives and indeed, through the Personalized Medicine Alliance, BCM hopes to do its part in creating a culture that not only accepts personalized medicine but also anticipates its forward progress.

Baylor Clinic and Hospital

To realize the goals of personalized medicine, healthcare delivery systems and facilities must be effectively linked to innovative physicians and healthcare workers, scientists and translational research, and education programs. However, the healthcare system in the U.S. suffers from many challenges including poor patient access, service, and education; ineffective integration of outpatient, diagnostic and inpatient care; poor information transfer and communication; and a general lack of good outcomes information.   Although BCM has multiple outstanding affiliated hospitals, we decided that to deliver innovative personalized medicine to private adult patients, a new facility for faculty physicians was required.

The Baylor Clinic and Hospital, now well under construction with plans to open in 2011, promises to provide the kind of personalized medicine now possible and to be ready to implement new phases of genomic medicine as it matures. The new construction provides the opportunity to create a platform for personalized medicine without preconceptions or designs that impede the implementation of medical progress. In its first phase, Baylor Clinic and Hospital will have more than 1 million square feet with 252 inpatients beds (each of them private and 60 in intensive care units). There will be 15 operating room, 272 exam grooms and 270 faculty offices with specialties in cancer, cardiovascular disease, neurological disorders, transplantation and general surgery and medicine. It will encompass both outpatient and inpatient care in a coordinated fashion that links physicians and patients more closely than institutions in the past.

With today’s technology, hospitals can become interactive entities. Technologies can see farther and deeper, hear more, and sense the environment at great distances while managing workflow and decision-making. Whether implementing robotic surgery technology, enabling sensor networks for tracking equipment, or providing physician-patient videoconferencing, the Baylor Clinic and Hospital is committed to incorporating the best of technologies available today and expanding to those that come on-line in the future.

Of course, all the science and technology, bricks and mortar are a means to an end. That end is providing patients with care that is the best available, the most personal in understanding of their individual biology and understanding of their needs. In the final analysis, patients judge a healthcare facility not only on the quality of its technology, science and research, but also on the ability of its physicians, nurses and other staff to take care of them with humanity and compassion. Personalized medicine means more than a computer, a gene chip or a new test. It means a receptionist who answers the telephone and attempts to meet the patient needs. It is the physician who listens, and the nurse who remembers the salient points of the patient’s problems.

A key ingredient of the new Baylor Clinic and Hospital will be state-of-the-art quality improvement, with constant data collection and measurement. Personalized medicine is not static, and patients will benefit only if we maintain efforts across the board – research, clinical and educational. First, however, we must define the most important outcomes to measure. Then we must put into place the means for gathering the data for those measurements in both a quantitative and qualitative sense. Most important, those measures must become transparent, and reported to the public on a regular basis. Where we fall short, we will change our procedures and activities.

Many challenges face a new hospital that seeks to deliver innovative services in addition to the difficulty of opening and operating a clinic and hospital in today’s financial healthcare environment. New approaches, promoted by personalized medicine, will take time to implement and longer for reimbursement to come from third party payers. The current system rewards what is currently being done.  However, we are confident that our approach will lead to irrefutable advances in healthcare.  And as these outcomes become more apparent and the value of various undertakings are proven, BCM will move personalized medicine -- the technology, diagnosis and treatment – into its affiliate hospitals, making this care widely available across the community.

Human Disease Genomics and Diagnostics

Advancement of genomic medicine is dependent on implementing diagnostic testing in patient care situations. There are many hurdles to achieve this including the need for extensive test validation and government regulatory approvals, clinical trials to assess efficacy, patient information and education, physician education, and strategies to receive reimbursement for offered tests.  BCM has as a major strategic goal the marriage of its strengths in genomics and clinical medicine to advance the field of genomic medicine. The two general areas of discovery genomics and clinical genomic diagnostics will be instrumental in providing a scientific platform for clinical programs in personalized medicine in the Baylor Clinic and Hospital.

BCM has proven capabilities in discovering new links between genetics, treatment and disease. The BCM Human Genome Sequencing Center (HGSC) in collaboration with faculty from Genetics and multiple BCM departments has several ongoing projects to apply sequencing information directly to human disease and many other investigators are pushing forward the frontiers of genomic discovery. BCM and the Personalized Medicine Alliance will continue to support and catalyze research in this area.

Capitalizing on what we know already about the links between genetics, disease, and treatment, BCM is currently implementing several advances in diagnostics. One project in general genomic testing that is of particular note is the BCM Chip.

The BCM Chip grew out of work begun with the BCM HGSC and the HapMap project, the first international effort to study human genetic variation. The HapMap project made use of technology developed by ParAllele, since acquired by chip developer Affymetrix.  This chip platform technology was inspired by BCM scientists, John Belmont, MD/PhD, and Richard Gibbs, PhD, through brainstorming with ParAllele ways of simultaneously testing for thousands of gene variants that were relevant to adult disease.

A pilot project in the HGSC involving approximately 160 people last year enabled the team to test and refine a prototype chip. The latest version of the chip uses a platform from Illumina Corporation, and has migrated from HGSC into the clinical laboratories of the BCM Department of Molecular and Human Genetics.

In its current form, the BCM Chip contains a select, but not exhaustive, set of diagnostic aids:

• 2,300 assays devoted to pharmacogenetics relating to 32 gene variations that can affect the way individuals respond to different drugs. Understanding those genetic variations can help doctors determine optimal drug choice and dosing for individual patients.
• 800 tests for single nucleotide polymorphisms (SNPs) that increase the risk of common diseases such as breast and prostate cancer, coronary artery disease and type 2 diabetes. Information from the chip allows physicians to create a disease risk profile for individual patients.
• 400 assays that look at different SNPs used for human leukocyte antigen (HLA) typing. HLA typing is most important currently in transplantation as well as identifying the risk of some autoimmune disorders. However, some HLA types can also affect response to drugs, such as abacavir, a protease inhibitor used to treat HIV/AIDS.
• 3,3000 areas on the chip will be used for the diagnosis of Mendelian disorders as well as cancer susceptibility, neurological disease (including Alzheimer’s and Parkinson’s disease) and cardiovascular disorders.

This potentially rich source of diagnostic information challenges bioinformaticians to render some 6,000 test results in a reportable digital form that can be used now as well as in light of future advances. Bioinformaticians are developing a system that can do almost all the interpretation of results automatically, filtering the relevant information into the patient’s electronic medical record. Later, as information evolves, the database can be configured to allow new information to “pop up” automatically as it proves important to the patient’s care.

In keeping with our ethic of personalized medicine centered around the patient, a collaborative effort involving the Cleveland Clinic and BCM is exploring patient understanding of the potential value and pitfalls associated with the kind of information produced by the BCM chip. In addition, this project seeks to quantify for the patients how the chip information fits into their medical care, avoiding the risk of over expectation or exposing information that could affect the patient’s peace of mind.

Plans are now underway to validate the chip and put it into the current Baylor Clinic for adult outpatient care before the end of the year. By 2010, we hope to have the BCM Chip in a stable configuration that meets the standard of the College of American Pathology. Its clinical role will become more apparent during that period.

Health Information Technology

Underlying the promise of personalized medicine is a new information infrastructure that supersedes the current system of fragmented medical records, most not even in digital format.

Currently, a trip to the physician’s office might or might not be coordinated with reference to an available patient record. In the future, even before entering the hospital, a patient should be able to register via a Web portal, perhaps even interacting with established personal health record systems that are currently in development.

Personalized medicine will absolutely require a new infrastructure and methodologies for recording and tracking health information. Information systems must not only make patient data immediately available to the healthcare team, but also provide an effective interface between scientific and clinical information and analysis tools for exploring hypotheses. High throughput genetic sequencing, high throughput microscopy, high throughput functional magnetic resonance imaging scanning – all these techniques hold promise for the future of healthcare, but they carry a heavy burden of data.  The system must manage information overload while linking all the necessary components to support scientific and clinical needs, including bio-banks, genomic and expression data, patient data and more. It must also make that information available to physicians across a host of platforms and in varying locales and situations.

Partnering with the Epic Systems Corporation, BCM is developing an interactive, responsive electronic health record that will not only meet current needs but also enable use of information in the future. For example, as new information about disease states and risks linked to genetic information becomes available, the Baylor College of Medicine EHR will alert physicians to new facts that may influence the way in which they provide care for a particular patient, perhaps influencing the choice of drugs or lifestyle advice that person receives.

Accomplishing this prototype system within the confines of an existing, antiquated structure with longstanding investment in old, tired systems would be almost impossible. However, the information structure of the Baylor Clinic and Hospital is a blank slate awaiting a new form of writing that will enable personalized medicine at its most efficient. Technology will be omnipresent but not overriding in the new hospital, with computer terminals available but not always visible as physicians, nurses and technicians provide the care that patients need and want in a compassionate manner with the best biological information available.

Once the system is established at the Baylor Clinic and Hospital, the College will collaborate with its affiliates, which are already developing or honing their own electronic medical records. Texas Children’s Hospital and the Harris County Hospital District, which operates Ben Taub Hospital, have already selected Epic as their vendor, increasing the interoperability of those systems as they develop. Because electronic data is a moving target, establishing seamless systems will become easier as institutions seek to work together and regulation pushes most institutions toward similar solutions.

The EHR will need to feed into a larger system that brings multiple types of data together for analysis of healthcare quality and research efforts. Our goal is that the platform will support and use those systems that are being developed for standardization of data and inter-institutional collaboration, such as caBIG.

Bridging the Gaps

The Personalized Medicine Alliance at BCM will face enormous challenges in providing guidance and leadership across the range of projects and plans that comprise our goal, which is nothing short of transforming the future of healthcare.   We have provided a snapshot of the work we are doing to educate all constituencies, build a brand new hospital, maintain the pace of diagnostic and treatment progress, and incorporate that progress with patient data into a seamless electronic environment.  To bring rapid advances on all fronts that still remain integrated in a larger picture of excellent patient care would be a impossible without an Alliance to bridge the gaps within and outside the institution and provide a continuity of vision for the future of personalized medicine.

Return to Table of Contents

National Cancer Institute

Bethesda, Maryland

The Biomedical Informatics Grid (BIG):
A Platform for 21^st Century Biomedicine

Kenneth H. Buetow, Ph.D.
Associate Director, Bioinformatics and Information Technology
Director, Center for Biomedical Informatics and Information Technology
National Cancer Institute

I.  Personalized Medicine Requires a 21^st Century Systems Approach

Personalized Medicine as a New Biomedical Paradigm

Personalized Medicine is a new paradigm in biomedicine.  Its successful implementation requires integration of unprecedented amounts of information and diverse communities. The ability to collect, analyze, share, and integrate massive quantities of biological and clinical data in real time is a prerequisite for Personalized Medicine.

Biomedicine is a complex system.  There are key interdependencies between the sectors that compose this complex system. Personalized Medicine’s goal is to transform this system and must therefore recognize and embrace its complexity.  Key opportunities to create a self-sustaining Personalized Medicine ecosystem come from understanding resource and information flows within the larger system.

Strategies for Addressing Complexity.    

Industry provides best practices for active design of complex systems.  First, best practice requires one to recognize the system as a whole.  Next, it identifies the interfaces between the components.  Within the boundaries of the interfaces, individual components are developed and manipulated iteratively and incrementally.  It is also important that initial development occur in a limited context, but one with sufficient complexity that it faithfully captures the complexity of the system component.  Finally, additional complexity is also added incrementally with the controlled expansion of scope.  This approach permits rapid incremental success without being stymied by the complexity of having to “boil the ocean.” 

The Essential Role of Information Technology.

The daunting complexity of the personalized medicine ecosystem makes the use of information technology critical.  But information technology within the biomedical enterprise has been slow to develop and is rarely connected between laboratories even within a single institution, much less between different institutions. In contrast with other national efforts, such as in defense or federally-funded physics research, the U.S. biomedical research enterprise has never had any such information technology system.

Thus, to address the complexities of cancer and these discontinuities of the research process, a 21st century biomedical enterprise requires interoperability; that is, access to integrated tools to collect, analyze, and share data in standardized formats. This interoperability is a means to link together all the scientists, clinicians, patients, and other participants so that they can share such standardized information rapidly.

The current generation of internet and world wide web technologies makes information technology approachable to biomedicine.  Information technology is critical to the interface connecting the components of the biomedical ecosystem.  It enables efficient operations within components.

A Systems View of Personalized Medicine
Multidimensional Stakeholder Ecosystem

The full ecosystem of Personalized Medicine encompasses members of the axes of biomedicine.  It includes researchers, physicians, and consumers as participants.  The researcher category includes discovery, translational, and clinical arenas.  In an alternative axis, the ecosystem includes the academic, not-for-profit, commercial, and government sectors.  A complete survey of the ecosystem also contains gatekeepers, such as regulators and payers.

Connectivity Through Information Technology

The needs of Personalized Medicine for information-sharing are accommodated by best practices in information technology.  Applications of information technology are arbitrarily segmented between approaches used to connect information and approaches to connect people.

Best practices to connect information call for the use of a services-oriented architecture.    Services should interoperate through well-defined interfaces. The architecture defining the interface should include Enterprise, Information, Computational, and Engineering viewpoints, and be technology platform neutral.  The information should be represented utilizing internationally accepted standards where available.

Communications using information technology is rapidly evolving.  Tremendous opportunities exist in utilizing web technologies, especially the emerging Web 2.0 approaches to community organization and business. 

Personalized Medicine Ecosystem as a Learning System. 

A key benefit of conceptualizing the complete Personalized Medicine ecosystem is the capacity to convert biomedicine into a learning system. More specifically, by capturing the entire biomedical life cycle, it is possible to synergistically combine research, care delivery, effectiveness measurement, quality assessment, and safety.   

Cancer as the Pioneering Field in Personalized Medicine

Cancer researchers have been at the leading edge of the Personalized Medicine revolution, and many of the first-generation personalized medicine products have been developed for cancer indications.  There are three obvious reasons for this early focus:

• First, cancer is a complex set of diseases, for which molecular medicine approaches predate even the Human Genome Project.  It has been known for decades that cancers are caused by genetic changes – either inherited or acquired – that result in abnormal cell proliferation, cell division or cell death.
• Second, cancer is a serious, often deadly condition, for which the efficacy rates of therapeutics have traditionally been extremely low. Since selection of the most efficacious treatment for the patient can be an urgent life-or-death decision, personalized medicine approaches vs. time-consuming “trial and error” are compelling. 
• Third, the adverse effects of cancer therapeutics are extremely unpleasant, disfiguring and potentially fatal, thereby making it even more important to select the optimum therapeutic choice the first time, to avoid the doubly-negative impact of adverse effects from futile treatment. 

The National Cancer Institute’s 21st Century Biomedical Test-bed
The NCI’s Unique Research and Care Delivery Platforms.

The NCI has a unique collection of administrative platforms that capture the entire lifecycle of biomedicine development, and hence it supports a unique environment in which the Personalized Medicine paradigm can be prototyped. For over 30 years, NCI has supported Comprehensive Cancer Centers, which blend research, care delivery, and prevention.  There are more than 60 of these centers, distributed nationally and housed at the most prestigious research and care delivery institutions throughout the United States.  More specialized programs include more than 50 NCI Specialized Programs of Research Excellence (SPOREs) that support translational research, and 10 NCI Cooperative Group programs that conduct multi-institutional clinical trials.  Most recently in the care delivery area, the NCI has launched a Community Cancer Center Program (NCCCP) with 16 sites that cover 20 million lives.

II.   The Cancer Biomedical Informatics Grid (caBIG®):  Proof of Concept Platform for Personalized Medicine

Origins and Development of caBIG®

The National Cancer Institute (NCI) identified the need in 2003 for an informatics initiative of unprecedented scope for the biomedical community in recognition of three factors:  the growing clinical and economic burden of cancer; the transformation of research catalyzed by the molecular revolution and multiple genomics technologies that were generating massive amounts of data; and the recognition that the “essential unity” of research and clinical care had powerful potential to improve the outcomes of all cancers, as it had done in the field of pediatric oncology.

As a first step in building an informatics infrastructure that would enable Personalized Medicine, the NCI officially launched the caBIG® (cancer Biomedical Informatics Grid) initiative in 2004 as a pilot program.  Its initial objective was to develop capabilities that would meet the self-identified needs of the NCI Cancer Center community.  (For more information on the history of caBIG^® , see the caBIG® Pilot Phase Report at

http://cabig.cancer.gov/resources/report.asp)

caBIG® Strategic Principles.

Four fundamental principles underlie the activities of caBIG® and guide all of its operations:

• Open Access: Participation in caBIG^® and the products delivered by caBIG^® are open to all, enabling access to tools, data, and infrastructure by the cancer and greater biomedical research communities.
• Open Development: Software development projects are assigned to particular participants, but are informed iteratively with multiple opportunities for review, comment, further modification, and development by the caBIG^® community.
• Open Source: The software code underlying caBIG^® tools developed with the support of the NCI is available to software developers for use and modification. This software is licensed as open source to promote the reuse of existing code, hence optimizing the full benefit of the research dollars spent. Nonetheless, caBIG^® recognizes the need for and importance of commercial software to the biomedical enterprise, and accommodates it through caBIG^® interfaces.  The open source license is industry-friendly, allowing commercialization of derivative products and fostering industry interest and innovation, while still adhering to the principle of open source for caBIG®-funded activities.
• Federation: caBIG^® software and standards enable local organizations, such as Cancer Centers, to share data resources with the larger cancer care and research community and to use resources contributed by others. On the grid, these resources can be aggregated from multiple sites to appear as an integrated research dataset, while the individual resources remain under the control of the local organizations.

caBIG® as the World Wide Web of Cancer Research.

caBIG^®  provides infrastructure for creating, communicating, and sharing bioinformatics tools, data, and research results, while using shared applications, shared data standards, and shared data models, all operating on a cancer community network (caGrid).

caGrid is underlying service oriented architecture that provides universal mechanisms for enabling interoperable programmatic access to data and analytics in caBIG®.  caGrid also creates a self-described infrastructure wherein the structure and semantics of data can be programmatically determined, and provide a means by which services available in caBIG® can be programmatically discovered and leveraged. 

There are to date over 100 grid nodes currently online at a variety of U.S. government, academic and commercial organizations, enabling those entities to share data. 

Use-driven Capabilities – Real Solutions to Real Problems

caBIG® provides more than 40 software tools, as well as the connecting network called caGrid, by which every function required in the molecular-based discovery and clinical research continuum can be performed and linked together.  

The extensive and continually evolving portfolio of caBIG^® capabilities can be reviewed at the website (www.cabig.nci.nih.gov ) and freely downloaded for use. 

III.  caBIG® Enterprise:  Platform for Networking the Global Biomedical Community

Unifying Research and Care

Beyond providing the informatics needed for molecular based research, there is a need in Personalized Medicine to link the research endeavor back to health care delivery.  Specifically, caBIG® is providing the ability to integrate molecular profiling, family history and molecular diagnostics into the Electronic Health Record, as well as to share back clinical outcomes data and clinical trial results into the discovery enterprise to achieve a ‘rapid learning’ system.

Following the completion of the pilot phase of the caBIG^® initiative, the NCI took the next step towards an infrastructure for Personalized Medicine by extending caBIG^® to an “enterprise phase”, with expanded capabilities to network the larger cancer community and beyond.

Today, caBIG® is a network of interconnected data, individuals, and organizations, designed to share data and knowledge, simplify collaboration, speed research to move new diagnostics and therapeutics from bench to bedside faster and more cost effectively, and ultimately to realize the potential of Personalized Medicine to improve patient outcomes.

A total of 56 NCI-designated Cancer Centers across the nation are working to connect their research and clinical care capabilities into a caBIG®-enabled information network.  Through the NCI’s Community Cancer Centers Program (NCCCP), 16 Community Cancer Centers that in the aggregate touch 20 million lives are also becoming a part of this network.  caBIG®-enabled connectivity enables these Centers to participate in clinical research studies and to bring the benefits of Personalized Medicine to their patient population in real time. 

More than 1,000 individuals from over 200 organizations have participated in caBIG® activities since the initiative’s inception.  Moving forward, however, it will be difficult to count the participants, since research users are increasingly applying caBIG® tools automatically as part of their studies without even noticing that they are “powered” by caBIG^® infrastructure.  In addition, as more and more software becomes caBIG^®-compatible, countless users will benefit from its interoperability features without awareness of its presence.

caBIG^® in Action

In the “enterprise” phase, caBIG^® infrastructure and tools are becoming ubiquitous among NCI intramural and extramural programs, as it enables and accelerates basic and clinical research.  Representative examples of such caBIG^®-enabled activities are:

Inter-SPORE Prostate Biomarker Study (IPBS).

The SPORES (Specialized Programs of Research Excellence) are NCI-sponsored clinical research groups each specializing in a particular type of cancer.  While each SPORE conducts its own trials, when biomarkers have been compared between centers, there has been a high degree of variability in the clinical significance of biomarkers screened from one center to another.  The IPBS study was designed to assess ways to unify the data collection and analysis of samples, improving consistency of results.  The IPBS study leverages caGrid to connect all participating centers, and applies caTissue to track the samples and manage the analysis results.

caBIG™ and Mutational Analysis.

The International HapMap project is a continuing effort to compare the genetic sequences of groups of different individuals to identify chromosomal regions where genetic variants are shared. The first two phases were completed in 2007 and opened the door to wider use of Genome Wide Association Studies (GWAS), where DNA markers are scanned across the genomes of many individuals to find genetic variations associated with a particular disease.  In the past year, GWAS studies have found genetic associations for coronary heart disease, Type I diabetes, and breast cancer, among others. 

However, researchers need sophisticated tools in order to make sense of the potentially millions of data points generated in a single GWAS study. To make these studies both simpler to interpret and more productive to find disease associations, caBIG^® has created several tools to analyze data from GWAS and other mutational studies.  The cancer Genome-Wide Association Studies (caGWAS) model allows researchers to integrate, query, report, and analyze significant associations between genetic variations and disease, drug response or other clinical outcomes, helping researchers to find the “needle in a haystack”.  Originally developed for use in cancer research, the caGWAS model was extended to accommodate the specific study needs of the cardiovascular research community as well.

In addition, the Cancer Genetic Markers for Susceptibility (CGEMS) project represents the first public release of a GWAS study for cancer.  Accessible by the CGEMS data portal (http://cgems.cancer.gov), over 500,000 SNPs have been analyzed so far, facilitated by caGWAS to produce and upload pre-computed results tables rapidly. 

The data generated as part of the CGEMS program has already helped identify variations in FGFR2, associated with increased risk for breast cancer, and multiple loci associated with increased risk for prostate cancer.

The Cancer Genome Atlas and the Cancer Molecular Analysis Portal.

One of the biggest challenges to researchers of high throughput genomics technologies is how to collect and work with the large quantities of diverse experimental data. The caBIG^®–enabled Cancer Molecular Analysis (CMA) Portal (http://cma.nci.nih.gov) provides powerful tools and resources that enable cancer researchers across the world to explore, visualize, and integrate genomic characterization, sequencing, and clinical data from a variety of data sets.  

The Portal exemplifies the caBIG® core principles of open development and federation. The CMA Portal allows researchers to use analysis programs developed at three different organizations, and to access data produced by more than 10 different institutions, all by a unified web interface.  The tools available on CMA Portal allow researchers to access clinical characteristics such as survival data and tumor staging, and correlate those with mutation and other genomic data. This capability enables researchers to conduct cross-platform queries, helping them to find correlations between research and clinical data that would be difficult, if not impossible, to find using conventional means.

The first data set accessible from the CMA Portal is from The Cancer Genome Atlas (TCGA).  TCGA is a comprehensive and coordinated effort to improve understanding of the molecular basis of cancer through the application of genome analysis technologies, including large-scale genome sequencing.  TCGA is an integrative, multidisciplinary effort to develop and assess a framework for systematically identifying and characterizing the genomic changes associated with three cancer types: glioblastoma multiforme, squamous cell carcinoma of the lung, and serous cystadenocarcinoma of the ovary. Together, TCGA and CMA advance the opportunities for scientists and clinicians to analyze and employ TCGA data, to develop a new generation of targeted diagnostics, therapeutics, and preventives for cancer, and pave the way for more personalized cancer medicine. 

FIREBIRD.

  To participate in FDA-sanctioned clinical trials, all investigators must fill out a variety of certification documents; key among them is the FDA registration Form 1572.  Until recently these were paper-based forms, but the Federal Investigator Registry of Biomedical Informatics Research Data (FIREBIRD) application is changing that process. FIREBIRD is the first module implemented toward the vision for a Clinical Research Information Exchange (CRIX) infrastructure. FIREBIRD will leverage legally enforceable digital signatures compliant with Title 21 Regulations using an Identity Assurance infrastructure, Secure Access for Everyone (SAFE).

FIREBIRD enables investigators to register online with the National Cancer Institute and other sponsors, including medical product companies. Through a single web-based interface to a secure central repository, investigators will be able to maintain their profile containing the accreditation information required for their participation in biologic, drug, or medical device trials. Investigators electing to participate in government, academic, or industry trials can access and apply their profile information to regulatory submission documents automatically, thus removing paper-based latencies and infrastructure costs. FIREBIRD is already in wide use across the clinical research community.

National Lung Screening Trial.

Medical images play a critical role in cancer diagnosis and treatment, and the DICOM (Digital Imaging and Communications in Medicine) image standards allow technical interoperability between various medical imaging hardware and software systems.  These standards, however, do not address workflow issues or how to integrate medical images with other types of biomedical information, such as genomic data, or clinical outcomes information. In addition, a standard part of the DICOM format includes the patient’s name within the structure of the image file, complicating de-identification of the images for later population studies.

The caBIG® Imaging program has several collaborations underway:

• The National Lung Screening Trial uses caBIG® imaging tools to integrate radiology and pathology data.
• The Grid-enabled MAX project involves integration of caBIG® tools with all the current cooperative group quality assurance activities for imaging and radiation therapy from the Quality Assurance Review Center (QARC) and the University of Massachusetts Medical School and NCI Advanced Technology Consortium (ATC).
• An additional project expands the application of caBIG® imaging tools to optical images generated by digital histology imaging tools.

I-SPY.

Unlike the treatments provided for most other diseases, cancer therapeutics are virtually all toxic compounds. To minimize the side effects and improve efficacy of these treatment with these agents, it is vital to identify biomarkers to predict which agents will be most effective for a particular cancer.

The I-SPY 1 (Investigation of Serial Studies to Predict Your Therapeutic Response with Imaging And moLecular analysis) trial is a national study to identify these biomarkers that may be predictive of response to therapy for women with late stage breast cancer.

Informatics support for the I-SPY trial includes integrating and analyzing clinical, MRI imaging, gene expression, CGH, Immunohistochemistry, and other data types. By correlating MRI image data with this collection of molecular characterization data from the tumors, researchers hope to identify biomarkers predicative for outcomes, ultimately resulting in more effective patient treatments.  The integration for I-SPY comes from caIntegrator, providing data warehousing and data mining access to researchers via a web portal, and provides an excellent example of cross-study integration and analysis in support of translational research.  Over 300 women with state II and III breast cancer have been enrolled to date. The study has also established standards for MR imaging and developed novel tools for data sharing, tissue tracking, common information repositories and clinical trial automation.

The TRANSCEND project (TRANslational Informatics System to Coordinate Emerging Biomarkers, Novel Agents, and Clinical Data) is a follow-on to the I-SPY 1 trial.  The goal of TRANSCEND is to develop the next generation of clinical trials data collection by the use of web-based case-report forms (CRFs) to simplify data collection, improve collection of clinical data in support of the CRF forms at 2 I-SPY trial sites, demonstrate integration with an electronic health record system (Tolven eCHR) with the bioinformatics infrastructure in place for the I-SPY 1 trial, and develop common data elements (CDEs) for breast cancer. In addition to the caBIG® tools used in I-SPY 1, caTissue and NCIA are part of the informatics infrastructure being developed for TRANSCEND.

Clinical Data Management System (CDMS).

An overarching goal of caBIG^® is to increase collaboration between basic and clinical researchers by encouraging the adoption of standards-based tools and data collection.  One area where the lack of standards seriously inhibits large-scale data comparison is in multisite clinical trials. This issue was recognized by the Clinical Trials Working Group of the National Cancer Advisory Board report “Restructuring the National Cancer Clinical Trials Enterprise”, which recommended creating an interoperable information technology platform for clinical trials. Broad use of standards-based electronic data capture systems improves the quality and comparability of data obtained at the different sites, facilitates multicenter trials, reduces trial administration overhead, and provides significant cost and time savings when compared with paper-based systems.

The NCI recently announced that it had acquired licensing rights from Medidata to distribute the Rave^® Clinical Data Management System (CDMS) software package, with related installation, support, and maintenance services free to any interested NCI-funded organizations conducting oncology clinical trials. The new software will interoperate with other caBIG^®-compatible software tools, and will enable data sharing and collaboration within each research organization, between diverse research organizations, and with NCI itself. 

Beyond Cancer

The tools and infrastructure of caBIG^® can be generalized and applied in a variety of biomedical settings beyond the initial cancer community, as follows:

• Beyond cancer, the tools and infrastructure of caBIG^® are being used to enable Personalized Medicine approaches in other therapeutic areas, such as in cardiovascular disease at the National Heart Lung and Blood Institute (NHLBI).
• Beyond research, caBIG^® is linking discovery, clinical research and care delivery, in order to achieve the essential unity of research and care.
• Beyond the National Institutes of Health, caBIG^® is being integrated into the federal health architecture to connect the Nationwide Health Information Network.
• Beyond U.S. borders, caBIG^® tools and infrastructure are being adopted to enable biomedical enterprises in the United Kingdom, India, Singapore, China, and some countries in Latin America to achieve Personalized Medicine.
• Beyond the “silos” of the traditional  health care enterprise, the caBIG^® infrastructure is being applied to link a complex ecosystem of constituencies in the BIG Health Consortium (see Section IV below), to demonstrate Personalized Medicine in real settings, in real time.

caBIG^® Architecture and Health

Effective communication and collaboration between the clinical research and clinical care communities requires the use of common standards-based systems for data collection and management. Unfortunately, it is often competing standards rather than a lack of standards that inhibits interoperability between these communities.

To address this problem, the stakeholders from the Clinical Data Interchange Standards Consortium (CDISC), the HL7 Regulated Clinical Research Information Management Technical Committee (RCRIM TC), the National Cancer Institute (NCI), and the US Food and Drug Administration (FDA) have worked together to produce a shared view of the dynamic and static semantics that collectively define the domain of clinical and preclinical protocol-driven research, and the associated regulatory “metadata” that describes the clinical trial.

The caBIG® policies and tools that specify controlled terminologies, data element structure, data models, and computable metadata about those data elements are all openly developed, made freely available, and provide a pre-made framework for an effort like BRIDG.  The caBIG® program has been a key partner and supporter of BRIDG and was instrumental in bringing the interested parties together at the outset.  caBIG® continues to play a critical role in future plans to produce a similar data standards model for the non-clinical research space.

IV. The BIG Health Consortium:  21^st Century Model for Biomedicine

As the next step in its strategy to achieve Personalized Medicine, the NCI is pro-actively working to break down the traditional silos of the biomedical enterprise and work collaboratively with all the key stakeholders that must be empowered in this new paradigm. 

On September 10, 2008, the NCI convened 25+ leaders from academe, government, advocacy, policy, and commerce, to grapple with the daunting challenge of transforming the biomedical enterprise to achieve the benefits of Personalized Medicine and demonstrate that the “disconnected islands” of the 20^th century can be reconfigured to improve health care.  A new group, known as the BIG Health Consortium, was formally launched that day.

Mission and Goals

The BIG Health consortium is a partnership comprised of all the key stakeholders in health care:  patients, providers, payers, product innovators, advocates, investors, and information technologists.  Conceived by the National Cancer Institute (NCI), its mission is to show – in real settings, in real time – how and why personalized medicine works.  Through a series of demonstration projects, BIG Health is modeling a new approach in which clinical care, clinical research, and scientific discovery are linked. 

The goals of BIG Health are to:

• Demonstrate feasibility of implementing a new model of translational medicine
• Create an “ecosystem” of participants that seamlessly integrate research, care delivery and consumer health information
• Break down traditional silos that are barriers to rapid discovery and learnings
• Accelerate and enhance research productivity and improve clinical outcomes

Assembling a New, Integrated Ecosystem

The BIG Health Consortium™ is designed to foster an integrated and interactive ecosystem (or “mega-community”) of previously unlinked sectors within life sciences and health care, who gather to conduct demonstration projects to make Personalized Medicine a reality.  Each participating organization is expected to share its capabilities, as well as to derive benefit, in order to have a self-sustaining endeavor.

Among the organizations that are participating in the BIG Health Consortium™ are cancer centers; integrated healthcare providers; academic centers; medical schools; diagnostic laboratories and product developers; personal genomics firms; patient advocacy and action-tank organizations; venture capitalists; biopharmaceutical companies; and government programs.

The informatics infrastructure of caBIG® will be generalized to “BIG” (Biomedical Informatics Grid) and applied as the underlying connectivity or “electronic glue” for BIG Health. 

Return to Table of Contents

Coriell Institute for Medical Research

Camden, New Jersey

The Coriell Personalized Medicine Collaborative
Examining the Utility of Genome-Informed Medicine

Michael F. Christman, Ph.D.
President and Chief Executive Officer, Coriell Institute for Medical Research

Executive Summary

The Coriell Personalized Medicine Collaborative (CPMC) is a research study that employs an evidence-based approach to determine the utility of using personal genome information in health management and clinical decision-making. The CPMC also aims to build a cohort with rich genotypic and phenotypic data with which to discover genetic variants that affect drug toxicity and efficacy, as well as to discover presently unknown gene variants that elevate a person’s risk of cancer and other complex diseases.

This forward-looking, collaborative effort involves physicians, scientists, ethicists, genetic counselors, volunteer study participants, and information technology experts. Its goal is to better understand the impact of personalized, or genome-informed, medicine and guide its ethical, legal and responsible implementation. The study will enroll 10,000 individuals by the end of 2009 with an ultimate goal of 100,000 participants. As of October 2008, there were 3,000 participants enrolled in the study. There is no charge to study participants.

Challenges of Implementing Personalized Medicine

Genome-informed medicine is the use of an individual’s genetic information to predict disease, avoid adverse drug reactions and tailor treatment ^1-3. The successful implementation of personalized medicine is dependent upon several factors. First, there is a critical need to educate health professionals ^4-7. The amount of genetics traditionally taught in medical schools is limited and typically focused on single-gene disorders and chromosome abnormalities, with little exposure of students to complex genetics. Second, the implementation of personalized medicine requires government support and regulatory oversight ^8-10, as well as public vetting of ethical issues ^{11, 12}. Third, medical records systems must be structured to accept genetic data and integrate them with the patient’s existing health record in a way that facilitates its use in clinical decision-making.

Additional challenges to evidence-based research into the effectiveness of personalized medicine include the need for large cohorts and longitudinal data collection to generate sufficient data to compute the treatment effect and gauge the potential costs and benefits. Cohort size must be large enough to address 1) genetic variants of low frequency (~1 to 2 percent), 2) gene-environment effects, 3) gene-gene interactions, and 4) loss of participants to follow-up. There are also consent and privacy issues that come into play in large cohort studies ¹³. In addition, genetic studies of large cohorts require significant biobanking, genotyping and information technology infrastructure ¹⁴.

The Importance of Biobanking

The mission of the Coriell Institute includes the collection, characterization, storage, and distribution of valuable biomaterials and associated data for scientific research. Coriell has more than forty years experience in developing and maintaining biorepositories as national and international resources for the study of human diseases and aging. The Institute continues to expand its information management systems to meet evolving business and scientific requirements. Coriell has a state-of-the-art laboratory and data management system and a web-based catalog for biomaterials and associated data.

Since the inception of the Coriell Cell Repositories, more than 150,000 cell cultures have been distributed to investigators in laboratories in the United States and sixty-two foreign countries. More than 500,000 aliquots of DNA have been shipped from all Coriell-managed repositories to investigators throughout the world. Coriell’s Repository Information Management System was designed to facilitate and streamline high-volume biomaterials and data distribution management. Coriell has been managing web-based access to genome-wide genotype data on hundreds of samples in its collections during the past several years. Its biorepository capabilities include significant phenotypic data management, with use of standardized phenotypic language and collection of longitudinal data for its disease collections ^{15, 16}. Additionally, Coriell has partnered with several regional healthcare systems that are rapidly moving toward comprehensive electronic medical record systems. These assets position Coriell and its partners to meet the challenges of translating genomics into clinical practice.

Need for Evidence-Based Research Studies

The Human Genome Project ¹⁷, the SNP Consortium ¹⁸ and the HapMap Project ¹⁹ have served to lay the foundation for the next generation of efforts to map complex disease genes and the quantitative trait loci (QTLs) ²⁰ that may be preclinical indicators of pending disease. To make this information useful in improving health and the quality of life, the mechanism for sharing genetic variation information associated with complex diseases with individuals and healthcare providers must be constructed, and evidence-based studies must be performed to assess the outcomes from receipt and utilization of this information. These are the major goals of Coriell’s research study.

The importance and need for an evidence-based initiative has not gone unrecognized by others in the scientific community. Dr. Francis Collins, former National Human Genome Research Institute director and human genome project pioneer, stated in a June 6, 2008 interview with Science magazine, “We desperately need, in this country, a large-scale, prospective, population-based cohort study. And we need to enroll at a minimum half a million people. We would need to have their environmental exposures carefully monitored and recorded, their DNA information recorded, their electronic medical records included, and have them consented for all sorts of other follow-ups.” The cost to perform such a study has been estimated at 300 to 400 million dollars per year. Coriell’s CPMC study has been constructed such that participants may opt to share their anonymized genotypic and phenotypic data with the scientific community, where it can be combined with other datasets in large genome-wide association studies.

The Coriell Approach

The CPMC aims to be a model for the ethical, legal and responsible implementation of genome-informed personalized medicine. The CPMC study is structured to allow dynamic communications between Coriell and study participants using a secure web portal. Web-based surveys will be used to assess health and behavioral outcomes related to the personal genetic variant information released by the study. Additionally, this portal will allow participants to share their data with healthcare professionals. Currently, the CPMC is funded through philanthropic donors and institutional support, with no cost to individual study participants. An outline of the CPMC research study is shown below (Figure 1).

After participants have given their informed consent, they are asked to donate two milliliters of saliva for genome profiling using a microarray platform (Affymetrix 6.0 Genechip, Affymetrix, Santa Clara, CA) and targeted SNP profiling using a bead-based platform (Illumina BeadXpress, Illumina, San Diego, CA). An outside panel termed the “Informed Cohort Oversight Board” (ICOB) meets at least twice per year to review genetic variants, submitted by Coriell, as risk variants for health conditions. Only genetic variants associated with health conditions considered to be potentially medically actionable (i.e., where there is the potential to mitigate risk, and those variants for which a significant association has been replicated) are then returned to participants via a secure web portal. Participants are able to grant access to their physician(s) to view the results and may request to discuss their results with a CPMC genetic counselor at no cost. A variety of outcome measures are assessed via web-based surveys completed by participants regarding their actions, physician actions, attitudes and, ultimately, health outcomes. Participants are asked to update their medical, family and lifestyle information annually such that longitudinal datasets are generated. Thus, there are several dynamic aspects of the CPMC, including ongoing review of association studies to identify variants for submission to the ICOB, continual outcomes research and the longitudinal collection of participant medical records on an annual basis.

Figure 1. Outline of the CPMC Research Study

The CPMC research study involves (1) informed consent and saliva collection; (2) genotyping; (3) viewing of genetic results; (4) optional sharing of genetic results; and (5) outcomes research.

Engagement of Hospital Partners and Medical Professionals

With respect to the challenge of integrating genomic information into the practice of medicine, the education of medical professionals, particularly doctors and nurses, is likely to be a rate-limiting step. Coriell understands the importance of engaging clinicians and other medical professionals to develop successful strategies for integrating complex genetic information into the current medical paradigm, and does so by engaging these individuals in the CPMC both as collaborators and participants. In addition, Coriell appreciates the commonality of cancer in society and the enormous potential for cancer research and cancer care to be impacted by personalized medicine. Thus, Coriell has established collaborations with neighboring healthcare partners for the CPMC study.

Coriell established a partnership with next-door neighbor and tertiary teaching hospital, Cooper University Hospital, in March 2008. Cooper University Hospital is the clinical campus for the Robert Wood Johnson Medical School of the University of Medicine and Dentistry of New Jersey and has more than 550 physicians in more than seventy-five subspecialties. In July 2008, Coriell announced its collaboration with community-based Virtua Health. The collaboration with Virtua was born out of the understanding that most of the population is treated in community health centers, as opposed to academic medical centers, which are often located in urban areas. Virtua is a community health system with four hospitals, numerous outpatient centers and more than 1,800 physicians in its network. Coriell also formed a collaborative relationship with Fox Chase Cancer Center, one of thirty-nine National Cancer Institute-designated comprehensive cancer centers and a center with a long tradition of excellence in combining state-of-the-art patient care with cutting-edge genetic research. In addition, a number of other partnerships with the CPMC are being discussed. Coriell encourages the enrollment of medical professionals and health center employees into the research study. These ties energize the study and open the door to educate medical professionals about genomics.

One of the strategies to educate medical professionals will involve seminars given by Coriell scientists and hospital partner physicians. Coriell is developing a seminar series on genomic medicine in collaboration with partner hospitals. Seminars will focus on diseases included in the CPMC and will meet the requirements of Continuing Medical Education (CME) such that attendees may gain CME credits. In an attempt to make education as accessible to healthcare providers as possible, Coriell may post the genomic medicine seminars online as webcasts.

Coriell will also look to medical professionals for input to ensure effective mechanisms are developed for using genomic data in the clinical setting. Questions to be addressed include:

• How is genome information best conveyed in the typical twelve-minute office visit?
• What type of information do healthcare providers want to see in a genome-wide genetic test report and in what context?
• What resources and tools are needed by healthcare providers to appropriately use genome information and educate their patients?

Realization of genomic medicine will require a two-way exchange in which scientists educate medical professionals and vice versa. This exchange will involve traditional communication in addition to that of medical and genetic datasets (in the form of electronic medical records and large numbers of genetic test results, respectively). Coriell expects that the deep engagement of several hospitals partners in the CPMC will catalyze this dialogue. Moreover, it is anticipated that as CPMC participants invite their healthcare providers to view their personal genetic results, Coriell will have an engaged and accessible population of healthcare providers to whom targeted surveys may be directed regarding use of genome information in medical care.

Recruitment of CPMC Study Participants

Recruitment of individuals into the CPMC is primarily conducted during informed consent sessions held at the Coriell Institute, partner hospitals or other community locations. The principal investigator of the CPMC, or a CPMC scientist, discusses the details of the study, possible risks, the content of the Informed Consent document, and provides attendees with the opportunity to ask questions. Upon signing of the Informed Consent document, newly enrolled individuals are invited to submit a small saliva sample.

Eligibility requirements are limited to requiring that participants are eighteen years old and older, have a valid email address and are willing to complete web-based surveys throughout the course of several years. Participants may opt (at the time of enrollment or any time thereafter via the secure web portal) to release their anonymized genome-wide variant data and medical history data to the scientific community for association studies. There is no charge to participants in the CPMC study.

CPMC’s Cancer Arm

Coriell’s partnership with healthcare centers including Fox Chase Cancer Center enables the study to have a cancer arm in addition to the wellness arm described above. Among the first 10,000 participants, the goal is to enroll 2,500 patients with breast cancer and 2,500 patients with prostate cancer. There is evidence that the baseline risk to develop cancer is strongly influenced by genetic variation and that in cancer patients, the response to chemotherapeutic agents, adverse events from medication and clinical outcomes are influenced by a patient’s genetic makeup. Thus, the creation of a large cohort of breast and prostate cancer patients with rich phenotypic datasets from the national cancer registries, as well as genome-wide genetic information, will allow researchers to examine the role of genetic variants in pharmacogenomic and clinical endpoints. For those participants who agree to allow the CPMC to share their anonymized data, such data will be made available to the larger scientific community through the National Center for Biotechnology Information (NCBI) database of Genotype and Phenotype (dbGaP) resource.

Clinical Laboratory Improvement Act (CLIA) Compliance

The CPMC’s goal to examine the potential use of genome information in clinical practice requires that the testing be performed in a Clinical Laboratory Improvement Act (CLIA)-approved laboratory. Therefore, the Coriell Genotyping and Microarray Center applied for and obtained CLIA certification to perform genotyping assays using the Affymetrix GeneChip platform. Soon, Coriell will expand its initial application to include CLIA certification for genotyping using the Illumina BeadExpress platform.

Powerful Analysis: Coriell Genotyping and Microarray Center

The Coriell Genotyping and Microarray Center uses the Affymetrix Genome-Wide Human SNP Array 6.0. The Affymetrix array was designed to provide broad coverage of SNPs across the entire genome through genotyping at more than 900,000 SNPs. Due to this design, SNPs known to have an association with a particular phenotype may not be present on the chip or represented through a perfect proxy SNP. To compensate, Coriell will use custom-designed SNP panels to include the disease-relevant SNPs absent from the Affymetrix platform. These panels will be analyzed on the Illumina BeadExpress platform.

Regulatory Body: Informed Cohort Oversight Board

The purpose of the Informed Cohort Oversight Board (ICOB) is to evaluate the medical actionability of health conditions and the evidence of a genetic risk variant’s potential medical “actionability” with regard to this health condition. A major prerequisite for consideration of genetic variants is the validity of association studies in the published literature that suggest a significant association between genetic variants and specific medical conditions. The ICOB thereby determines what personal genetic variant information will be returned to study participants. Approval is given when knowledge of a participant’s status for a particular genetic variant has the potential to affect a healthcare provider’s treatment course or permit the provider to offer advice about the participant’s health or lifestyle that has the potential to mitigate risk. Using prospective, web-based outcomes surveys, the CPMC study will determine whether or not the use of variant information does indeed mitigate risk.

This external advisory board comprises highly esteemed scientists, healthcare professionals, an ethicist, and a community pastor. The concept of such a board was proposed by Dr. Kohane and colleagues ²¹. This approach provides a model for a national system for evaluation of genome-informed medicine.

CPMC scientists review medical and scientific literature to identify candidate gene variants and provide summary reports to the ICOB. The ICOB reviews each report and votes to approve, disapprove or to request more information on each variant and condition. Factors to be considered include:

• Recommendations by the US Food and Drug Administration, Centers for Disease Control and Prevention, National Institutes of Health, National Associations for Medical Subspecialties, or other governmental advisory bodies.
• Seriousness of the disease, condition or potential adverse drug response.
• Number, size and quality of studies demonstrating a statistically significant association of a gene variant with the condition. Meta- analyses, when available, are reviewed.
• Magnitude of the effect of the particular genetic variant.
• The risks and benefits of clinical or lifestyle intervention(s) to minimize or reduce the risk.
• Data elements to measure outcomes.

Approval by the ICOB means that the association between the genetic variant and the condition has been validated and that the condition is considered to be potentially medically actionable. Approval does not require that there be clear evidence that the variant has utility in affecting health outcomes. The goal of the CPMC is to provide the outcomes data to determine the utility of each genetic variant.

The ICOB meets at least twice per year. This frequency allows the study to integrate findings from peer-reviewed association studies for new associations and validations of prior findings. It is likely that over time, the CPMC will request the ICOB to re-review both previously rejected variants for which there is new scientific evidence and previously rejected health conditions for which prevention or treatment options have changed the potential actionability. ICOB decisions are determined by a majority vote. The group deliberations are conducted in private, assuring that scientific issues are debated in an objective, critical and unencumbered environment. However, the outcome of all deliberations is publicly disclosed through the web portal.

Dynamic Participant Engagement: Results Viewed Through Secure Web Portal

The CPMC web portal is a website with several functions. It allows for 1) data collection through online surveys, 2) genetic variant results reporting, 3) education of participants and medical professionals, 4) secure sharing of personal genetic variant information with healthcare professionals, 5) web-based requests for access to data from scientists, and 6) web-based requests for genetic counseling from participants. It is a public site with a portal for participants to log in to a secure server. In the secure portion of the site, participants may set up their CPMC account with a password, change their contact information (email address), update their consent options (e.g., opt to release their anonymized data for genome-wide association studies (GWAS)), and view their personal genetic variant information as it is released.

Additionally, the CPMC web portal has a significant amount of genetic education material. This material is written for two distinct audiences, the lay participant and the medical professional, although any individual may access the more advanced educational material if desired. The educational pages include information on basic genetics and scientific milestones such as the Human Genomic Project and HapMap project. Educational material is also provided on inheritance, cancer, the multifactorial nature of complex disease, the meaning of “risk” and how to interpret disease risk assessments, and reasons why this type of study is only possible today.

With each visit to the web portal, participants are re-engaged. Participants must elect to view each genetic variant result independently, assuring that control over the results lies with the participant and that participants are not informed of results that they are not actively seeking. Individuals who choose to view CPMC results will watch a short educational video of a genetic counselor giving anticipatory guidance for that specific variant prior to viewing their personal genetic variant information. The CPMC encourages study participants to invite their healthcare providers to view their results. Participants may authorize access to their results directly from their CPMC web portal account.

In addition, the site has current information about opportunities available to participants such as no-cost genetic counseling, educational forums and additional surveys related to the study. There will be the potential for the CPMC to post information about other studies for which participants may be eligible. Figure 2 provides a diagram of the information system architecture for the study.

Figure 2. CPMC Study Information Architecture

Important to the maintenance of participant privacy is the fact that all personally identifying information is both encrypted and stored separately from genotype and medical information. Two-factor security is used to dynamically build the web pages as participants view their personal data.

Realistic Risks: Explanations of the Magnitude of Risk Elevation

The CPMC is committed to reporting realistic risks associated with genetic associations in a format that is understandable by the lay population. All results presented will illustrate the known population disease risk (specific to racial/sex/age groups, if known) and the adjusted risk based on the genetic variant genotype. Although in some cases a particular genotype may increase the risk significantly, it is expected that most genetic variants associated with complex (multifactorial) diseases will increase the risk only modestly. Until validated algorithms are available to combine risks associated with more than one genetic variant, each will be reported individually. References to the primary literature are included on all result reports.

To ensure that participants and healthcare providers understand the risks conferred by the genetic variants included in the CPMC results, an educational section of the web portal called “Understanding the Odds” has been created. This section, written for both lay and medical professional audiences, describes the concept that the risk of complex diseases is dynamic and involves the interaction of genes with the environment. Additionally, the genetic contribution toward a complex disease is discussed, addressing the likelihood that tens of individual genes, not a single variant whose current results are being reported and viewed, influence the genetic risk of complex disease. It is also explained that given the current state of knowledge, family history is likely to be a larger risk factor for most complex diseases than any one genetic variant.

Coriell is using an additional tool to educate CPMC participants. Study participants will be invited to attend educational forums hosted by CPMC genetic counselors and clinicians from hospital partners. Upcoming events are announced to study participants through the CPMC web portal. The purpose of the forums is to educate participants about health conditions for which genetic variant information has been released as part of the CPMC study. At these sessions, the clinician will discuss the health condition, its causes (genetic and non-genetic), screening, treatment, and prevention strategies. The CPMC genetic counselor will discuss the genetic variants that are part of the CPMC study and their association with the condition, as well as the risk assessment supplied with the genetic information.

Understanding Results: Genetic Counseling

Genetic counseling in the era of genomics and personalized medicine will require a new approach from traditional counseling for single-gene disorders ²². Coriell employs full-time, board-certified genetic counselors who are dedicated to the CPMC study and available to provide genetic counseling to participants via email, phone and face-to-face office consultation, as well as through educational forums open to CPMC participants. Medical professionals whose patients are participating in the study may also request access to CPMC genetic counselors to discuss the study and the reported genetic variant information.

The genetic counselors will record all encounters with CPMC participants in a secure, password-protected tracking database that is only accessible to the CPMC genetic counselors. This database will allow the genetic counselors to have easy access to the history of contact between themselves and a participant. It will allow genetic counselors to track the amount of time and type of consults being made and to gather statistics on the types of diseases and variants for which the consults are being requested. This tracking system will also allow the genetic counselors to identify common areas of confusion around which future educational sessions for both the lay public and medical professionals can be tailored.

Medical History, Family History, and Lifestyle Questionnaires

Participants are required to complete extensive medical history, family history and lifestyle questionnaires online after establishing their personal, online CPMC account. These surveys must be completed prior to viewing genetic results. Participants will be asked to update their medical history, family history and lifestyle information one year after the information is entered and every twelve months thereafter. These data will be used for two purposes: 1) they will be used in combination with genotype data to calculate personalized risk, whenever possible, and 2) they will be used in combination with genotype data in GWAS studies to identify additional genetic variants which contribute to complex disease and/or drug metabolism (for those participants who opted to allow their anonymized data to be used for association studies).

Coriell recognizes the importance of CPMC data in GWAS studies and has created a mechanism (via the participant consent form) for participants to indicate their willingness for their anonymized data to be shared with researchers (both from non-profit and for-profit organizations). As such, anonymized data from the CPMC will be available to all qualified researchers through the NCBI dbGaP web portal. The model is to perform surveys through the web portal, allowing cross-validation of data across questionnaires. The longitudinal nature of this project, the on-going release of genetic variant results, and the request for annual updates of survey information will allow for the collection of data that are traditionally hard to acquire, such as diet and exercise patterns over time and environmental exposures as they happen.

Longitudinal Data Collection: Electronic Medical Records

Participants may opt to release recent medical records from their primary care healthcare provider via hard copy, or in electronic form if they are in a hospital partner’s Electronic Medical Records (EMR) system. Updated medical records will be requested annually to ensure longitudinal data collection. These datasets will be monitored for changes in health outcomes relevant to conditions for which the CPMC has released genetic variant information. Medical records will be compared to self-reported patient medical history reports.

CPMC staff will transcribe a subset of the information in the medical record into a Personally Controlled Health Record for each participant. All Coriell information technology systems will allow compliance with established standards for interoperability (HL7) and medical data definitions such as SNOMED and LOINC.

Participant Privacy and Security

Coriell has several provisions in place to maintain integrity, confidentiality and security of its data and information systems. Coriell has security policies in place to assure that all data are protected from unauthorized access, and maintains audit trails, backup procedures and error checking to assure accuracy and protection of CPMC data. Data security is a balanced combination of management and staff actions, operational activities and technological control measures. The CPMC information technology infrastructure requires three highly integrated technology layers: 1) web portal, 2) laboratory information management system for inventory management, phenotypic data management and process management, and 3) secure hardware infrastructure that contains web application servers, database servers, a storage array network, and network security appliances. Personally identifying information is encrypted and stored in a separate database from the genotype and medical data. Participants will also be required to log in to the secure web portal using their barcode identifier, username and a strong password.

Outreach to Minority Populations

As the population of participant volunteers in the CPMC grows, Coriell is dedicated to ensuring that the genetic data collected are representative of the ethnic composition of the region. Camden, NJ, the community in which the Coriell Institute is located, is one of the poorest urban communities in the country, primarily made up of African-American and Hispanic residents. Coriell’s aim is to develop mechanisms to reach these historically underserved communities.

Coriell has enlisted the support of several groups to aid in minority recruitment. First, Coriell approached the religious community in Camden County, NJ. Second, prominent leaders are taking part in the study and offering their infrastructure to the project. Within the Hispanic and Latino community, Coriell has engaged local Hispanic political leaders including United States Senator Robert Menendez (D-NJ), Co-sponsor of S.976, “Genomics and Personalized Medicine Act of 2007.” Finally, Coriell hosts enrollment events in Spanish and offers a Spanish version of the CPMC Informed Consent document.

Availability of CPMC Data to Researchers Worldwide

The CPMC team has discussed with National Human Genome Research Institute a strategy for hosting anonymized data from CPMC participants that opt to share their data with scientists for research through the dbGaP web portal. Coriell is committed to ensuring widespread access to this valuable dataset. The Institute has a history of posting data with dbGaP for use by qualified scientists and has been involved in the return of genotypic data generated from samples in the Framingham Heart Study, as well as in the National Institutes of Neurologic Diseases and Stroke and the National Institute of General Medical Sciences repositories at Coriell.

Outcomes Research

Follow-up studies of the actions of CPMC participants and healthcare providers, as well as participant health outcomes, are at the heart of this evidence-based study. A thorough assessment of medical history, family history and lifestyle at baseline is made prior to the release of personal genetic variant results. In addition, participants will be able to take part in other assessments, such as an examination of baseline knowledge of genetics.

When scaled appropriately, the data collected from the CPMC will be used to assess whether healthcare costs increase as a result of genome-informed medicine using objective criteria such as number of physician visits, tests ordered, data related to hospital admission, and drug prescriptions. Measures of physician practice based on surveys of physician beliefs and recommended practices will be balanced by examining choices made by participants in selection of healthcare options. Coriell will work with hospital partners to develop such metrics and with organizations such as the Technology Evaluation Center to ensure appropriate clinical data elements are monitored.

Summary

The CPMC is an evidence-based research study designed to determine which elements of personal genetic data are valuable in clinical decision-making and healthcare outcomes. Medical records and genomic data will be updated dynamically. There is no charge to CPMC participants and, for participants who choose to release their data, anonymized genotypic and phenotypic data will be made available to qualified scientists. The CPMC will enroll 10,000 participants by the end of 2009 into wellness and cancer arms. Close partnerships with area hospitals are designed to catalyze physician engagement in personalized medicine.

References

1. Burke W, Psaty BM. JAMA. Oct 10 2007;298(14):1682-1684.
2. Hoffman EP. N Engl J Med. Dec 27 2007;357(26):2719-2722.
3. Katsanis SH, Javitt G, Hudson K. Science. Apr 4 2008;320(5872):53-54.
4. Frueh FW, Gurwitz D. Pharmacogenomics. Jul 2004;5(5):571-579.
5. Levy H, Young JH. Clin Pharmacol Ther. Mar 2008;83(3):492-493.
6. Patel KK, Babyatsky MW. Gastroenterology. Mar 2008;134(3):656-661.
7. Morrow T. Manag Care. Feb 2008;17(2):59-60.
8. Obama B. Clin Adv Hematol Oncol. Jan 2007;5(1):39-40.
9. Fox JL. Nat Biotechnol. May 2007;25(5):489-490.
10. Raich T, Irwin J. MLO Med Lab Obs. Aug 2007;39(8):48.
11. Abrahams E, Ginsburg GS, Silver M. Am J Pharmacogenomics. 2005;5(6):345-355.
12. Landon MR. J Biolaw Bus. 2005;8(3):28-36.
13. McCarty CA, Nair A, Austin DM, Giampietro PF. Community Genet. 2007;10(1):2-9.
14. Roden DM, Pulley JM, Basford MA, et al. Clin Pharmacol Ther. Sep 2008;84(3):362-369.
15. Gwinn K, Corriveau RA, Mitsumoto H, et al. PLoS ONE. 2007;2(12):e1254.
16. Keller M, Gwinn, K, et al. Neuropsychiatric Disease and Treatment. 2007;4:613-618.
17. Venter JC, Adams MD, Myers EW, et al. Science. Feb 16 2001;291(5507):1304-1351.
18. Miller RD, Phillips MS, Jo I, et al. Genomics. Aug 2005;86(2):117-126.
19. Manolio TA, Brooks LD, Collins FS. J Clin Invest. May 2008;118(5):1590-1605.
20. Fan R, Xiong M. Eur J Hum Genet. Oct 2002;10(10):607-615.
21. Kohane IS, Mandl KD, Taylor PL, et al. Science. May 11 2007;316(5826):836-837.
22. Weil J. EMBO Rep. Jul 2002;3(7):590-593.

Return to Table of Contents

Harvard-Partners

Boston, Massachusetts

AN INFORMATATION TECHNOLOGY INFRASTRUCTURE
FOR BRINGING GENETICS TO MEDICINE

Samuel Aronson and Raju Kucherlapati
Harvard-Partners Center for Genetics and Genomics

Abstract

The Harvard Medical School-Partners HealthCare Center for Genetics and Genomics (HPCGG) was founded in 2001 to develop and implement strategies for the incorporation of genetic/genomic information and knowledge into clinical medicine (“personalized medicine”) with the belief that such incorporation has the potential to change clinical outcomes and radically improve medical practice.  As a part of this effort HPCGG was charged with enhancing the Partners HealthCare Systems (PHS) clinical enterprise infrastructure in a manner that would both speed the adoption and improve the quality of personalized medicine.   To this end, HPCGG created facilities capable of incorporating new genetic and genomic instruments as they are developed.  In addition to supporting research activities, these facilities support the HPCGG’s CLIA certified molecular diagnostic laboratory that is called the Laboratory for Molecular Medicine (LMM).  The LMM offers gene based tests to clinicians for use in routine clinical care.   Together with the PHS hospitals, these facilities and the LMM form an integrated healthcare delivery network capable of developing and offering molecular diagnostic tests and leveraging such knowledge to improve healthcare. 

Early in the HPCGG’s development it became clear that substantial information technology (IT) investments would be required to enable personalized medicine to reach its potential.  Five years ago a partnership between the HPCGG, the Partners HealthCare Information Systems Department and Hewlett Packard Corporation was formed to begin building the required infrastructure.  This paper will describe the IT functionality that has been deployed to support and link together the HPCGG’s facilities, the LMM and the Partners HealthCare Electronic Health Record (EHR).   We will also describe projects underway to further enhance our genetics/genomics based IT capabilities.  In particular, we have identified two areas where new inter-institutional networks will be needed to prepare for wider adoption of genetic and genomic techniques in medicine.  We will describe infrastructure we have begun to construct that might enable the establishment of these networks.

The Nature of Genetic Based Diagnostic Tests

Molecular diagnostics, also known as genetic or genomic based diagnostics, provide the bridge that enables physicians to bring genetic knowledge to routine patient care.  Physicians use the information these tests generate for assessing disease risk, for diagnosis, for prognosis and for making specific treatment decisions.  The revolutionary effect these technologies will have on the health care system is already being felt: genetic tests are being used to guide treatment in clinical domains as diverse as heart disease, cancer, infectious disease, and many other common illnesses.  At our Center we offer molecular diagnostics in different medical areas.  These include cancer, cardiovascular disease, pharmacogenetics, pharmacogenomics, several childhood disorders and sensory neural disorders.  Some tests are relatively simple in which the laboratory tests for a few genetic variants and others very complex where tens of kilobases of DNA are examined for variants and mutants.  Specific examples include:

• Hypertrophic Cardiomyopathy that may result from mutations in any one of eleven different genes and is severe enough to result in sudden death if left untreated, and to assess the risk of their relatives for these disorders.
• Identify whether a patient’s hearing loss is caused by genetic variants that are also correlated with other serious medical conditions.
• Determine whether non-small cell lung cancers have genetic variants that correlate with either Tyrosine Kinase Inhibitor (TKI) efficacy or resistance.
• Identify whether a patient has genetic variations that will cause him or her to metabolize Warfarin abnormally, either quickly (risking stroke) or slowly (risking brain hemorrhage).

The field of clinical molecular diagnostic testing is evolving quickly. A few years ago, nearly all genetic tests were gene-based tests that involved examining a small number of specific base pairs in a patient’s DNA to determine whether particular mutations or variants were present. Today, we commonly run sequencing tests that read long segments of patient DNA in one gene or many genes and determine all the variations present in those sequences. There are many different technologies for DNA sequencing including Affymetrix resequencing Chip-based technologies that we have implemented in our Center.  This technology has made it possible to survey increasingly large segments of DNA in a cost-effective manner.  Newer sequencing technologies that promise even higher throughput and lower cost are ready to be implemented.  Several national governments are funding research and many commercial entities are investing significant capital to reduce the cost of sequencing a person’s entire genome to approximately $1,000. 

While we are at least a few years away from reaching this goal, new technologies will continue to drive down the cost of sequencing to costs that may be even less than $1,000/genome. 

The continuous reduction in DNA sequencing costs will significantly affect how genetics is leveraged in the clinic.  At present the cost of DNA sequencing is a barrier to increased use of molecular diagnostic testing.  As this barrier is reduced, we believe the amount of patient DNA being sequenced will increase.  We expect the number of variants identified in the patient population to grow continuously until it becomes feasible to cost effectively sequence all of the nearly 3 billion base pairs of a patient’s genome.  Current estimates place the number of variants that such a test would yield at 4-5 million per person.  (Levy et al. 2007)   The data generated by such a whole genome sequencing test would be good for a person’s lifetime.  Sequencing would only need to be redone in the case of cancer or other disorders where somatic changes are important and for infectious agents.

We are rapidly approaching the day where we will be able to determine the precise DNA variations present in each patient.  However, the process of determining the implications of each of these variants, let alone the implications of each combination of variants, will take longer.  Our knowledge of the impact of genetic variation is constantly expanding and this knowledge expansion is likely to continue for many years to come.   Ideally, clinicians would take into account the most up to date discoveries on every variant discovered in each patient as they prescribe care, but doing so is clearly beyond the capacity of the human mind.  To reach this goal, clinicians will require far more extensive IT infrastructure than that exists today. The rate of progress in technologies to accomplish the goal of sequencing the entire human genome is much greater than the rate at which progress is being made to provide IT support for such efforts.  Clinicians will likely encounter substantial challenges well before whole genome sequencing becomes routine.  Present day chip based genotyping technologies are already capable of generating datasets that would overwhelm existing clinical knowledge management systems.  For the past five years the Harvard Partners Center for Genetics and Genomics, the Partners HealthCare Information Systems department, and Hewlett Packard have been developing components of the IT infrastructure needed to address these issues.  The applications we have built are supporting the use of genetics in the clinical environment; however, much work remains to be done to provide the depth of support clinicians will ultimately need.

The Partners HealthCare Genetics IT Infrastructure

To be truly effective, IT infrastructure that helps manage genetic information must integrate laboratories, genetic professionals, the Electronic Health Record (EHR) and automated clinical decision support engines.  Infrastructure of this scope must be built incrementally.  In our institution, we began by constructing a platform to support the laboratories that generate genetic and genomic data.  Next we built infrastructure that supports professional genetic experts and other healthcare professionals including genetic counselors.  Then we integrated this infrastructure with the Partners HealthCare EHR.  Finally we began the work of creating genetics based clinical decision support (CDS) functionality.  We are now deepening our support for genetics in the EHR to enable broader genetics based clinical decision support. (Figure 1)  As we do this, we are encountering challenges that cannot be solved by an individual organization acting independently.  Therefore, we are working to establish networked infrastructure that will be needed to fully support personalized medicine.

Supporting the Laboratory:  The Gateway for Genomics-Proteomics Applications and Data (GIGPAD)

Almost all genetic and genomic data are generated in laboratories by complex machinery.  Laboratory Information Management Systems (LIMS) are critical to an overall genomics IT infrastructure because they can help with workflow issues as well as capture genetic and genomic data in structured form when it is initially generated. Downstream bioinformatics, report generation, and clinical decision support systems depend on this structured genetic and genomic data. LIMS are also important for ensuring data integrity across the inter-organizational process flows associated with genetic testing. For these reasons, integrated LIMS support is an essential part of a genomic IT enterprise architecture. In addition, LIMS can help reduce costs and increase quality through process automation, reducing errors, facilitating communication and reducing the need for manual entry of information.

A large number of genetic and genomic technologies are used in research and they will migrate to clinical use.  Maintaining multiple LIMS within an enterprise is both challenging and expensive.  We have found that creating an Enterprise LIMS Superstructure can help address these problems.   We created a system called the Gateway for Integrated-Genomic Proteomic Applications and Data (GIGPAD) to serve this function.  GIGPAD serves as an umbrella over the individual LIMS in the environment and integrates them together. The system exposes unified user and system interfaces to the rest of the enterprise.

GIGPAD’s umbrella style architecture (Figure 2) serves two purposes.  First, it enables us to develop common functionality that can be shared across facilities.  Order entry, accessioning, results return, the financial interface, user authentication and authorization functionality are all shared across facilities.  Low volume facilities can leverage the umbrella layer to provide the required IT support.  Higher volume facilities tend to need specialized LIMS functionality to assist in their workflow and integrate their instruments.  We make individual laboratory build versus buy decisions for this type of specialized LIMS functionality.  When we choose to buy, we integrate the purchased LIMS under the GIGPAD umbrella.  The scope of these build or buy projects is smaller because of the common functionality contained in the umbrella layer.

Figure 2:  GIGPAD is architected to enable reuse of functionality across the clinical and research environments.

BSF, Biosample services facility; BWH GCRC, Brigham and Women’s Hospital General Clinical Research Center

Second, we have enhanced the GIGPAD umbrella to operate in both the research and clinical context.  We strive to isolate the differences between research and clinical process flows in the umbrella layer.  As a result, the laboratory LIMS become relatively agnostic as to whether they are servicing a research or clinical process.  This enables GIGPAD to provide an important translational medicine function.  When geneticists identify a clinical use for a research technology, we can quickly enable well validated clinical IT support for that technology’s workflow – in the case of resequencing microarrays, one person was able to affect this transition in less than a week.    While workflow support can be established very quickly, creating the necessary quality assurance / quality control (QA/QC) functionality can take longer.  GIGPAD contains a case management system (CMS) that is responsible for:  (1) supporting the wet bench work that is required to break a sample into the required constituent assays, (2) managing the QA/QC functionality which often involves performing specific follow-on assays on a second technology to validate results and (3) managing the laboratory signoff process that occurs in advance of results being sent to geneticists.  We have found it worthwhile to continuously focus a significant amount of our development resources on improving and building new forms of automation into the CMS.  The processes that the CMS supports cover a significant percentage of the cost of genetic testing.  They are also important in ensuring the test quality.   

GIGPAD has been operational in our environment since April of 2004.  As of September 10, 2008 there were 1,007 registered users of the system and 1,076,379 data files under management.  GIGPAD currently provides support for the initial phase of the Molecular Diagnostic testing process.  This includes all steps up until the point that we determine what genetic variations are present in the stretches of DNA that are sequenced.  At this point GIGPAD forwards this information to clinicians for interpretation (Figure 3).  GIGPAD is designed to handle DNA based, RNA based or protein based testing efforts.

Supporting the Geneticists:  GeneInsight and the Genomic Variant Interpretation Engine (GVIE)

Most clinicians have neither the training nor the time to assess the clinical significance of variants that have been identified in their patients.  For this reason, molecular diagnostic laboratories typically employ genetic professionals who interpret test results and produce a text report describing the significance of any genetic variants identified. The process of generating this report can be time-consuming and expensive, so streamlining and automating portions of the process through IT can be valuable. IT can also help standardize result reporting by reducing variability between the ways different geneticists might interpret the same result. When test results are sent to the EMR, it is useful to capture interpretations in structured form in addition to the genetic variants themselves. Capturing structured interpretations requires IT support during report generation.

We have constructed two tools to support the report generation process in our environment: GeneInsight and the Genomic Variant Interpretation Engine (GVIE). Because our understanding of the clinical implications of particular variants can change over time, it is important to have a database that tracks current knowledge relative to individual variants. We use GeneInsight to perform this function. Keeping this type of database current is extremely challenging.  There are numerous heterogeneous research databases that contain information about genetic variants but very few clinically validated data sources.  Genetic professionals must review these research sources to formulate clinical interpretations.  GeneInsight has data structures that associate information with diseases, genes, tests, and genetic variations. When a new test is brought on line, developers load data from existing data sources. GeneInsight is then integrated into the geneticists’ reporting processes so that it is maintained as a by-product of the process of signing out reports. This is made possible through integration with GVIE.  GVIE is a reporting tool that is interfaced to GIGPAD.  As variants are identified in patients, they are passed to GVIE which then looks up the information stored in GeneInsight on those particular variants.  GVIE then produces a draft interpretive report which a geneticist and/or genetic counselor reviews.  During this review process, they are shown statistics related to the variant’s frequency and given the ability to review previous cases where the variant was identified.  Geneticists and genetic councilors have the option of modifying these reports.  We track which reports are modified.  This provides us with a metric for assessing the maturity of each part of GeneInsight.

As a result of this process, geneticists can maintain the data in GeneInsight for the diseases they report on without a significant incremental time investment when they encounter a new variant. As a benefit, the time required to report on previously identified variants is significantly reduced. Overall, the combined GVIE/GeneInsight system saves geneticists time, which promotes systems utilization.  The amount of data contained in GeneInsight has grown over time and we are now evaluating additional uses for this information in the clinical environment.  However, as we will describe later, we need to find ways to dramatically increase the depth and breadth of the data in GeneInsight if it is to solve our core genetics related knowledge management needs.

Supporting Front Line Clinicians:  Electronic Health Record (EHR) and Clinical Decision Support (CDS) System Integration

Molecular diagnostic reports are ideally delivered to the clinician through an EHR.  Doing so ensures that genetic test results are stored in an organized manner and are consistently accessible to authorized clinicians.  It also opens up the possibility of leveraging automated CDS systems to proactively assist clinicians in the use of this information.  We created a specially secured area in our EHR where we maintain patient genetic profiles and a custom screen to organize the genetics results.  GVIE is interfaced, through our hospital pathology LIMS, to this part of our EHR.  This interface allows us to transfer genetic laboratory test results in both human readable and highly structured electronic formats.  The structured genetic test result format is designed to be read by CDS algorithms.      

As the number of variants stored in patient genetic profiles increases, it will become increasingly difficult for clinicians to review these profiles during the care delivery process.  Properly applying the information in these profiles will be even more challenging.  Clinicians will need to rely on CDS functionality to surface relevant genetic information at the appropriate times.  This functionality is required for gene based personalized medicine to reach its potential, but it will be very difficult to build.  We have taken an initial step within Partners by establishing a genetics aware clinical decision support rule that alerts physicians if they order a particular class of tyrosine kinase inhibitors for a patient who has a genetic mutation associated with resistance to these drugs.  The process of establishing this rule helped us understand the modifications to our EHR infrastructure we need to make to support broader based genetic aware clinical decision support. 

The enhanced genetics IT architecture we have begun constructing is shown in Figure 4.  We chose to employ a service oriented architectural (SOA).  Patient genetic data will be stored in a specially secured Genetic Marker Repository (GMR).  Test definitions will be stored in a Genetic Test Definition Catalog (GTDC).  GeneInsight will serve as the EHR’s genomics knowledgebase.  A service layer will be constructed on top of these repositories.  Our general CDS infrastructure will leverage these services as will our front end EHR displays.   An additional display in the form of a Patient Genome Explorer (PGE) will be constructed to provide clinicians with an additional specialized view into patient genomic profiles.  We are focused on constructing the GMR, GTDC, PGE and GeneInsight wrappers in a modularized fashion.  Our goal is to ultimately package the GMR, GTDC, PGE and GVIE/GeneInsight components together to form a Genetics Enabler Kit (GEK) that could be used to genetics enable other EHRs, PHRs or Pharmacy systems.    

The test results stored in the GMR and the knowledge in GeneInsight are the heart of this architecture.  While the information stored in these repositories is critical, the inter-institutional interfaces required to populate them do not currently exist.  The next sections describe the network infrastructure we are constructing to help address this problem.

Figure 4:  PHS Genetic Enabled Architecture Currently Under Construction.  LMR, Longitudinal Medical Record; CPOE, Computerized Physician Order Entry; PEPR, Patient Enterprise Problem Repository;  CDR, Clinical Data Repository; GMR, Genomic Marker Repository; PGP, Patient Genetic Profile; GVIE, Genomic Variant Interpretation Engine.

Linking it all Together:  Establishing the Data and Knowledge Flows Needed to Drive Genetics Aware CDS

We have established a flow that links together GIGPAD, GVIE, GeneInsight and our EHR.  When a Partners HealthCare patient is tested in our Laboratory for Molecular Medicine (LMM), the results flow into our EHR in structured form.  Up to date knowledge about the implications of any variations found by the LMM is maintained in GeneInsight.  When we test our own patients, we have both the knowledge and data resources required to construct genetics based CDS. 

Genetic tests are performed by many different laboratories throughout the world. Many of the genetic tests performed on our patients are performed by external laboratories.  Similarly, our diagnostics laboratory often tests patients for other providers.  In both of these cases, interfaces do not exist to transfer the variants identified in electronic form.  Therefore, neither the structured genetic data nor the structured knowledge is ultimately represented in an EHR.  Without this information, CDS is impossible. 

This problem must be addressed by both establishing appropriate standards and creating appropriate data and knowledge networks. 

Establishing the Standards for Genetic Data Exchange

A member of the HPCGG IT team serves as one of the co-chairs of the HL7 Clinical Genomics Workgroup.  We have developed and contributed internal message formats to HL7 and worked with them to develop a standard model to transfer genetic laboratory test results.  We have also worked extensively with the leadership of LOINC to establish appropriate coding schemes for genetic results and their associated clinical implications.  We also interact with government institutions focused on supporting the development of standards for personalized medicine data and health record functionality including the Department of Health and Human Services, the National Library of Medicine's Lister Hill Center for Biomedical Communication, and the Evaluation of Genomic Applications in Practice and Prevention (EGAPP) initiative.

In addition to developing standards, we are also investing in two infrastructure projects designed to facilitate the exchange of genetic data and knowledge. 

VariantWire:  A Network for Moving the Data

Interfaces between laboratories and provider EHRs are expansive to build and maintain.  In the case of a reference laboratory serving the majority of a provider’s clinical testing needs, this cost can often be justified based on the volume of tests that pass through the interface.  However, the genetic testing market is dispersed.   In the majority of cases, a single interface between a provider and a laboratory will not carry enough volume to justify the cost.   It would be economically infeasible to establish point to point interfaces linking every provider to every genetics laboratory that tests their patients.  Standardization efforts can substantially reduce the costs of establishing interfaces, but these costs will always remain significant.  Unless this issue can be addressed, it will be impossible for providers to maintain patient genetic profiles that contain all variants identified through laboratory tests. 

We are constructing a system, called VariantWire, which is designed to serve as a hub that will enable the secure transfer of genetic test results.  Any institution that connects to VariantWire will be able to communicate with all other connected institutions through a single interface.  In this way, VariantWire can help address the “many-to-many problem.”  The economics of creating interfaces to VariantWire should continuously improve as additional providers and laboratories connect.  Each additional node should increase volume through existing interfaces and thereby reducing the transaction costs of maintaining those interfaces.  VariantWire is constructed on top of Intersystems’ Ensemble product.   We are building validation functionality into the hub that will enable us to enforce standards by rejecting any non-conforming messages.  We are currently in the process of linking together the HPCGG LMM, the Intermountain HealthCare EHR and will link the PHS PGE when it is completed.

We are currently designing an interface between the GEK and VariantWire.   Once this interface is built, any EHR that leverages the GEK will be able receive structured genetics results from any laboratory on the VariantWire network without any additional software development.   We hope the GEK will be leveraged in this way to dramatically reduce the costs associated with joining the VariantWire network and thereby increase network participation.

Leveraging GeneInsight to Create a Knowledge Network

VariantWire is designed to enable EHRs to gather complete patient genetic profiles where every variant identified in a patient is stored in a GMR.  However, just knowing patient specific variants is not sufficient to fully enable EHR’s to handle genetic information.  Structured knowledge detailing the significance of variants is also required.  A laboratory can report its assessment of the significance of a variant at the time a test is run.  However, over time additional information may be learned about the significance of a variant.  Mechanisms must be built into the EHR to make this knowledge accessible.  Some laboratories make an effort to update their historical reports when new knowledge is discovered but this solution is not scalable for tactical and strategic reasons.  From a tactical standpoint, laboratories can go out of business, laboratories usually cannot “follow” a patient when they change providers, and providers may or may not agree with the knowledge sources leveraged by the laboratory.   From a strategic standpoint, as the breadth of DNA covered in each genetic test increases, it will become infeasible for individual laboratories to curate knowledge for all of the variants their tests can identify.  Providers face a similar issue.  In our environment GeneInsight serves as our genomic knowledgebase.  Our geneticists continuously update GeneInsight as they learn more about individual variants.  However, they only maintain information on genes covered by our LMM tests.  No single institution could ever hope to employ enough geneticists to maintain up to date information on all of the variations that could be identified in its patient population. 

It is impossible to genetics enable an EHR in a scalable manner without access to comprehensive, continuously updated, genomics knowledge.  This information will clearly be needed to maintain broad spectrum genetics aware CDS but it will also be needed to perform more basic functions.  For example, as the amount of data in the GMR grows, we will need to provide filtered displays of the data through our PGE.  We will need to rely on the knowledge in GeneInsight to keep these filters up to date over time.  We are also working to build a flag into our PGE that alerts clinicians if new information has been learned about a variant since it was reported.  This flag will be driven off of the information in GeneInsight.     

Genetic knowledge will continue to be curated by many different institutions for the foreseeable future.  A mechanism is needed to assemble the knowledge curated by many disparate groups and make it accessible to EHRs, PHRs and pharmacy systems.  We are exploring whether GeneInsight could be enhanced to enable this knowledge network.  The concept we are investigating involves building a central GeneInsight hub and then distributing instances of GeneInsight that can communicate with this hub.  Each participating organization would gain the ability to enter three levels of data into GeneInsight: 

1. Public Data:  information that can be shared with any organization that has access to the GeneInsight network. 
2. Private Data: information that is never transmitted to the centralized hub and therefore remains proprietary to the organization that enters it.
3. Protected Data:  information that an organization is willing to disclose under certain conditions.  Potential conditions could include a fee per use, a subscription fee or through a collaborative agreement. 

Our goal is to support the business models required to incent organizations to share protected data.

EHRs, PHRs and pharmacy systems could implement an instance of GeneInsight.  Because GeneInsight is a component of the GEK, any organization implementing the GEK would have access to the GeneInsight network.  Connected organizations would gain access to public data and have the ability to negotiate access to protected data if they desire.  Each organization will presumably also specify the degree to which they trust the different data sources that contribute to GeneInsight.

We are seeking partnerships with commercial genetic testing laboratories, healthcare providers as well as pharmaceutical and biotechnology companies to assess the feasibility of implementing the tools that we have developed within their environments and in creating the networks described above.

Figure 5:  VariantWire and GeneInsight Areas of Focus

Summary

We have implemented and are pursuing several genomics related IT projects because we believe that in the long term they will enable significantly improved patient care.  Physicians have a limited amount of time to spend with each patient.  Genomic technologies are capable of generating an overwhelming amount of data and our knowledge of the implications of these data constantly expands.   A robust inter-institutional IT infrastructure must be established to enable clinicians to harness the increasing power of genetics for the benefit of their patients.

Reference

Levy S, Sutton G, Ng PC, Feuk L, Halpern AL, et al. (2007) The diploid genome sequence of an individual human. PLoS Biol 5(10): e254. doi:10.1371/journal.pbio.0050254

Return to Table of Contents

HealthMapRx

Washington, D.C.

COMMUNITY-BASED APPROACH TO PERSONALIZED HEALTH CARE: HEALTHMAPRx^TM

A Patient Self-Management Program Utilizing Community-Based Pharmacists

Summary Document

Authors:

• William M. Ellis, RPh, MS, Executive Director and CEO, The American Pharmacists Association Foundation
• Toni Fera, Pharm.D., Senior Director, Patient Self-Management Programs, The American Pharmacists Association Foundation
• Jamie Kirkwood, BS Marketing and Sales Representative
• Benjamin Bluml, RPh, Vice President, Research, The American Pharmacists Association Foundation

Program Overview

HealthMapRx^TM is a service of the American Pharmacists Association (APhA) Foundation.  The service has evolved from the previous decade of research by the Foundation, including the “Asheville Project^®," a community pharmacy-based program that began in 1996 and continues today. (1) The success of this model has helped business leaders to recognize that health care can be an investment in well-being rather than an expense for sickness.

HealthMapRx^TM is a patient-focused collaboration between employers, their covered health plan beneficiaries, and specially trained community pharmacists who provide, face-to-face counseling sessions where participants learn how to better manage their chronic conditions (such as diabetes, high blood pressure, hyperlipidemia) and reduce associated health risks.

Local networks of pharmacists are established to provide the self-management services to the patients. The program is collaborative and designed to complement and reinforce existing health care team provider roles, including the patient’s primary care physician. In addition, the program establishes a benefit model that aligns incentives for employers, patients, and providers.

What the Program Does: The Value Proposition

• The HealthMapRx™ program creates a collaborative team of employers,   employees, pharmacists, physicians and diabetes educators ─ and aligns incentives ─ to focus on wellness, patient self-management and workplace cost savings.
• Educates and supports employees with information and guidance to become active participants in managing chronic diseases, such as diabetes, based on a proven model and demonstrated research outcomes.
• Employer waives co-pays on medications or provides other incentives to encourage active engagement in self-care.
• Employee (or dependent beneficiary) meets regularly with pharmacist to discuss their care and learn new ways to monitor and control their disease.
• Centers care around the patient and positions pharmacists as accessible, valuable resources in helping patients understand and control chronic disease.
• Reduces unscheduled absenteeism in the workplace and associated costs.
• Improves health outcomes as measured by key indicators.
• Saves health care dollars by investing in patient well-being ─ keeping people healthy rather than paying for care when they become seriously ill.

How the Program Works

• Specially trained community pharmacists “coach” participants on how to manage their chronic disease, including setting goals, using medications properly, and tracking their condition consistently with recognized clinical indicators such as cholesterol tests, blood pressure, foot exams and eye exams.
• Collaborative care teams ─ including pharmacists, diabetes educators and physicians ─ are assembled in the community, educated about the program and are compensated for their involvement.  Team members communicate regularly to optimize patient care.
• Employees choose to participate through a voluntary benefit offered by their employer that aligns employee benefit incentives to encourage success.
• Success is measured with the following indicators:
  • improvement in A1C concentrations (blood sugar control)
  • body mass index
  • blood pressure control
  • lipid control
  • increased patient satisfaction with pharmacy services
  • decreased costs of medical care

The Model

Employers/Payers

The practice model implemented for HealthMapRx^TM is designed as a collaborative care model that emphasizes the roles of the employer, physician, pharmacist and patient. The employer/health plan agrees to invest in incentives for patients and pharmacist providers. At a minimum, these incentives include waived co-pays for medications and certain supplies. Some employers add other incentives as a way to integrate the program into their existing plan offerings. Other incentives have included counting participation toward wellness points, waiving co-pays for education classes and/or laboratory test co-pays. Most employers participating in the program are self-insured employers.

Employers work closely with their Third Party Administrators (TPA) and Prescription Benefit Managers (PBM) in order to establish a process to implement incentives (such as waived co-pays) and to provide basic claims data information on an annual basis to allow for program economic performance review. In some situations, the TPA or PBM can assist the employer with other aspects of program implementation, such as sending announcement letters to potential participants or managing enrollments.

Participants/Patients

Enrollment is voluntary; the employer educates eligible beneficiaries about the program through various announcement methods, including direct mailings, e-mail, newsletters and live orientation sessions. All participants are required to complete enrollment materials and a participant agreement. Enrolled participants are matched with a pharmacist “coach” and/or location from a local pharmacy network directory.

Pharmacists

Patient assignments are coordinated by a local pharmacy network coordinator. Services may be provided in a local pharmacy or at the participant’s workplace. During regularly scheduled visits, pharmacists apply a prescribed process of care that focuses on clinical assessments and progress toward clinical goals, establish self-management goals specific to each patient, and work with other health care providers and may recommend adjustments in the patients’ treatment plans. Pharmacists who participate in the program are required to complete an ACPE-accredited training program in the relevant clinical area (such as diabetes or hypertension), or are otherwise certified. They generally follow national treatment guidelines unless otherwise specified by the physician.  Pharmacists collect subjective and objective assessment information and enter it into a web-based documentation system for outcomes reporting.  Pharmacists are reimbursed by employers for patient visits according to fee schedules negotiated by the local pharmacy network.

Physicians and other Providers

Physicians are informed of participant enrollment and are encouraged to share their care plan with the pharmacists, who reinforce that plan with the participants.  Pharmacists communicate with physicians after every visit, as necessary, and refer patients as needed to their physician (for follow-up visits, laboratory tests or resolution of medication-related problems), or other providers, such as a dietician (for intensive nutrition education) or diabetes education centers (for additional education support).

Program Experience

The  HealthMapRx^TM Program evolved from early published works in Asheville, North Carolina and the APhA Foundation’s Project ImPACT Hyperlipidemia (1, 2, 3, 4).  Since that time, the APhA Foundation has conducted projects in a variety of sites throughout the country to assess the replicability of the model in diverse settings. Results from the initial pilot site replications were published in 2005 (See Appendix A) (5). These results demonstrated positive clinical, economic and patient satisfaction improvements for participants enrolled in the program. In order to test the scalability of the program, the APhA Foundation launched the Diabetes Ten City Challenge at the end of 2005. The interim clinical results were published in March, 2008 (6). Currently, the program is implemented under the brand name, HealthMapRx^TM.  The program has now been implemented by more than 80 employers in 20 states, with more than 3,000 active participants. Several employers are continuing the program into multiple years. 

The majority of employers implementing the program have been self-insured and include private companies, school districts, city and county municipalities, and health systems. Program design has core elements that are required to ensure the integrity of the model, but there is significant opportunity to tailor the program and its implementation at the local level. The HealthMapRx^TM team provides implementation consultation for employers, as well as templates for announcing the program and for managing enrollment.

Success Factors

There are key qualities that seem to drive successful program implementations:

• An Employer/Payer that will invest in incentives for patients and providers to improve health and lower costs
• Employers who are more involved in the program implementation, and have an open culture with their employees tend to have faster and higher percentage of enrollments of eligible beneficiaries
• Receptiveness of health care providers who support community-based collaborative care
• A local network of pharmacists that have the motivation, training and time to help patients manage their care
• Accessibility to pharmacist services
• Following the HealthMapRx established process for employer implementation, patient care, and documentation
• Willingness of TPA/PBM to provide claims data for analysis

Challenges

The program is implemented at the local level and developed to address needs, resources, cultural and political issues within the employer’s community. Thus, challenges may occur at the local level. Although some employers have unique challenges, there are some challenges that appear to occur more frequently. For example, an employer “champion” usually drives the initial approval and implementation of the program. If there is a change in staff or lack of a true “champion,” this is challenging and may even jeopardize continuation of the program.  Without strong employer support and a plan for consistent and clear communication about the program benefit design for participants, the full enrollment potential (and therefore, results) may not be realized.

On the pharmacist network side, since this is a relatively new practice model for community-based pharmacists, it is important to balance participant access with network capacity. In addition, there needs to be adequate resources to support network services, coordination, and management. The pharmacist provider shortage, particularly in rural areas can also be a challenge.

Employer Profile

In order to implement the program, employers should have the following characteristics:

• Willingness to invest in employees’ health to enhance quality of life, reduce sick days and lower hospitalization costs;
• Willingness to promote the program, orient and enroll patients;
• Capability to (or use a PBM) provide reduced/waived co-pay prescription cards or other incentives;
• Ability to provide access to data from TPA to track total health care costs for enrollees; and
• Willingness to provide payment to pharmacist providers/the provider network.

Patient Profile

When employees enter the program, they are asked to sign a participant agreement, which outlines consistent requirements for their patients who participate. Generally, the program is introduced as a voluntary benefit for employees and/or dependents who agree to meet with a qualified pharmacist on an ongoing basis for education, monitoring and set personal goals for diabetes self-management. The patient agrees to work with pharmacist coaches to set goals and monitor their progress. Participants must agree to meet at least quarterly with a qualified pharmacist to set self-management goals, have scheduled assessments and procedures to monitor performance.

Pharmacist Provider Profile

Specially trained pharmacists or those willing to complete the required training are recruited to pharmacy networks as providers for the program. Providing medication therapy management services, including identifying and preventing drug-related problems is a key component of the pharmacist’s role. In addition, pharmacist providers have received additional training in chronic care and the program processes of care. Examples of requirements include:

• Pharmacists must have designated certification or completed a comprehensive ACPE-accredited program in diabetes or other disease state as specified (such as a CDE, BCPS certified or APhA Diabetes program certification).
• A private consultation area must be available for patient education.
• Self-management coaching to patients in relevant lifestyle areas, such as smoking cessation, diet, exercise and nutrition must be provided.
• There must be collaboration with local health care providers, including primary care physicians and refer, or recommend for referral, participants to existing resources.
• Outcomes documentation must be maintained.

Other Health Care Providers

It is important to stress that, in this program, physicians will remain responsible for overall care of patient and changes in therapy. Physicians will receive summary reports after each patient’s session with the pharmacist as applicable, and will be notified about the program when patients enroll. Physicians are still responsible as required to make therapy changes or referrals as required. Data from the early projects indicate that physician outpatient and diabetes education center visits increase.

Summary

A collaborative practice model utilizing community-based pharmacists to provide coaching and self-management education to patients, that aligns incentives for participants, sponsoring employers and health care providers has been successfully implemented in a variety of settings. An investment in “well care” has led to lower costs, improved employee satisfaction, and better outcomes for patients with chronic disease.

HISTORY OF HEALTHMAPRx™(Appendix A)

The following milestones and research have paved the way for HealthMapRx™

• 1996          The APhA Foundation creates Project ImPACT: Hyperlipidemia™, the first collaborative care program designed to show how pharmacists, physicians and patients with high cholesterol can work together to make lifestyle changes and improve medication adherence to achieve cardiovascular goals. (4) Over a three-year period, nearly 400 people with high cholesterol in 12 states, working together with 26 pharmacies, participated in this landmark program.
  The results, published in the Journal of the American Pharmacists Association in 2000, showed that more than 90 percent of patients stayed on their medications and 67.5 percent reached the National Cholesterol Education Program (NCEP) treatment goals.
• 1997          The diabetes management program, the Asheville Project, is first offered to employees, dependents and retirees in the City of Asheville, North Carolina, in partnership with the North Carolina Center for Pharmaceutical Care.  The program starts with 47 initial participants.
• 1998          Mission-St. Joseph’s Hospitals and the Blue Ridge Paper Company add the diabetes management program for beneficiaries in their health plans.  It grows to more than 300 people with diabetes over the next three years.
• 2003          Long-term results of the Asheville Project, published in the Journal of the American Pharmacists Association, showed that patients improved A1C levels (key diabetes indicator), employers had lower total health care costs, employees had fewer sick days and increased satisfaction with pharmacist services, and pharmacists developed thriving patient care services. (1)  Asheville Project results also appeared in Business Insurance and The Washington Post.
• 2003          The APhA Foundation begins follow-up research based on the Asheville Project and Project ImPACT: Hyperlipidemia™ to assess the feasibility of expanding the model to multiple employer types and geographic locations.  A pilot program, “Patient Self-Management Program™ for Diabetes,” is initiated in four states at five employer sites with more than 300 patients. 
• 2004          The APhA Foundation completes development of “The Patient Self-Management Program™ Diabetes Credential,” the first and only credential for education in diabetes that can be awarded to individual patients for completing study in diabetes and its management as part of the Patient Self-Management: Diabetes™ program research. 
• 2005          Patient Self Management Program™ for Diabetes program results are published with compelling findings that indicate the ability to replicate and expand the scale of the Asheville model in diverse settings (5):
  • Participants dramatically improved in key indicators of diabetes control, including reducing average A1C values from 7.9 to 7.1 percent, using the goal set by the American Diabetes Association
  • More patients kept up to date with key indicators of diabetes care, including influenza vaccinations, foot and eye exams, recorded blood pressure, and lipid profiles (average increase of more than 40 percent)
  • Employers realized a $918 net cost savings per employee
• 2005          Diabetes Ten City Challenge™ is announced in October, inviting participation from employer groups that want to seize the opportunities for improved patient health and cost savings demonstrated in the Asheville Project and Patient Self-Management Program™ for  Diabetes.  The Pittsburgh Business Group on Health and the Northwest Georgia Healthcare Partnership are the first employer groups selected to participate.
• 2006          HealthMapRx^TM is established.
• 2008          The Interim Results of the Diabetes Ten City Challenge are published in JAPhA (6) 
  The report released analyzed aggregate data on 914 DTCC participants who were in the program at least three months as of September 30, 2007.  It documented clinical improvements in all the recognized standards for diabetes care, including:
  • Decreases in laboratory measures (mean) for hemoglobin A1C (a laboratory test showing the patient’s average blood sugar control over the previous two to three months), LDL cholesterol and blood pressure over the initial year of the program
  • Increases in the number of participants with current influenza vaccinations, foot examinations and eye examinations
  • 21% increase in the number of participants achieving the American Diabetes Association goal of A1c level <7.0
  • Increase from 43.8% to 57.7% in participants achieving nationally recognized National Cholesterol Education Program goals for LDL cholesterol
  • 15.7% increase in the number of people achieving recognized goals for systolic blood pressure
  • The number of DTCC participants who felt their overall diabetes care was “very good to excellent” increased from 39% to 87%
  • More than 97% of participants reported being “very satisfied” or “satisfied” with diabetes care provided by DTCC pharmacists
  • The number of participants setting self-management goals to control their diabetes also increased significantly:  those with nutrition goals increased from 22% to 66%; those with weight goals increased from 23% to 64%; and the number of participants setting exercise goals increased from 24% to 72%

HealthMapRx™ Testimonials

Pharmacist Testimonial:

Society and even families don’t realize how bad this disease is. They often don’t know that uncontrolled diabetes can lead to blindness, amputation, end-stage kidney disease, and cardiovascular complications such as stroke or heart attack. It’s important that patients can access a coach on a regular basis to help them through the ups and downs and help them control their diabetes as best they can. You can’t just show someone how to use a glucometer and send them on their way – it takes constant education, encouragement, and support to empower the patient to self manage. Our participation in this pharmacist directed wellness program gives us a chance to give back to the community by providing much-needed diabetes patient education and make a difference in the health outcomes of people with diabetes. Pharmacists have the ability to apply their scientific knowledge in making therapeutic decisions that will affect health outcomes.

In the first three months of the program we ideally like to see the patient once a month to understand their health history, set goals, and go over basics like nutrition, exercise, and how to use a glucometer. We see what patients need in terms of education and make sure they understand what each medication does and how to take it.

Employer Testimonial:

This program enables people to understand what they need to do in order to become healthy or stay healthy. The more that people take advantage of it, the healthier our employees will be, which can be a win for everybody. Improving health means improving energy and attitude, and there is less down time from lost workdays.

Everyone I have talked with who is involved in the program has been pleased with their results, whose personal encouragement helped recruit employees to the program. They tell me they have learned more about diabetes than ever before.

Patient Testimonial:

The program is a good support to help you stay on track, and an excellent resource for information. Having an hour set aside allows me to sit down, focus and ask questions without feeling rushed. I take three different medications, and the pharmacist explained what each one does in my body. I also learned that my medications might not have been working right because of how I was taking them. I probably wouldn’t have asked my doctor about that.

My morning readings were very high, Working with (my pharmacist), my doctors increased the dosage of medicine I take at night, had me take it with my meals and have a snack before I go to bed. I’ve been able to bring down my numbers and work on losing some weight, which has been a major factor. I feel much better.

When I found out I had diabetes, I was devastated. Since enrolling in this program, I’ve made major changes in my life, including losing weight and exercising every day. My pharmacist coach has become one of my closest friends and she continues to inspire me at every visit. This program has taken away so much of my fear and truly saved my life.

Physician Testimonial:

The key to success of the program is to make sure that additional burden isn’t placed on the physician for managing these patients. Physician engagement is driven by the patients, and they will respond best when they hear from their patients why this is a huge benefit.

How the Program Works

• APhA Foundation contracts with employers to implement its HealthMapRx program which include program guidelines, templates and software for documentation. The Foundation also provides staff support and assists with identifying a local network of pharmacist providers to establish the program in the selected community.
  
• Employers offer a voluntary employee benefit with incentives to encourage success (typically waiving participants’ co-payments for diabetes medications and supplies) and compensate pharmacists for the care provided.
• Participants meet regularly with a specially trained pharmacist “coach,” learn how to self-manage their diabetes and track key indicators with medical tests, foot exams and eye exams.
• Pharmacists are specially trained and use the Patient Self-Management Program to educate patients and record their clinical progress on key diabetes quality-of care indicators.
• Collaborative care teams including pharmacists, diabetes educators and physicians communicate regularly to optimize care.
• Success is measured by evaluating:
  • improvement in clinical outcome indicators, such as A1C concentrations (blood sugar control)
  • increased patient satisfaction with pharmacy and diabetes care services
  • decreased costs of medical care.

REFERENCES

(1)  Cranor CW, Bunting BA, Christensen DB. The Asheville Project: Longterm clinical and economic outcomes in a community pharmacy diabetes care program. J Am Pharm Assoc. 2003;43:173–84

(2)  Bunting BA, Cranor CW. The Asheville Project: Long-Term Clinical, Humanistic, and Economic Outcomes of a Community-Based Medication Therapy Management Program for Asthma. J Am Pharm Assoc. 2006;46:133–147.

(3)  Bunting BA, Smith BH, Sutherland SE. The Asheville Project: Clinical and economic outcomes of a community-based long-term medication therapy management program for hypertension and dyslipidemia. J Am Pharm Assoc. 2008;48:23–31.

(4)   Bluml BM, McKenney JM, Cziraky MJ.  Pharmaceutical Care Services and Results in Project ImPACT: Hyperlipidemia.  J Am Pharm Assoc. 2000;40:157-65(5)  Bluml BM, Garrett DG. Patient Self-Management Program^SM for Diabetes: First Year Cost Savings J Am Pharm Assoc. 2005; 45: 130-137

(6)  Fera TF, Bluml BM, Ellis WM.  The Diabetes Ten City Challenge : Interim Clinical and Humanistic Outcomes of a Multisite Community Pharmacy Diabetes Care Program.  J Am Pharm Assoc. 2008 Mar-Apr;48:181–90

Return to Table of Contents

Marshfield Clinic

Marshfield, Wisconsin

Community Based Approaches to Personalized Health Care: Marshfield Clinic

Authors

Stephen Wesbrook, PhD; Philip F. Giampietro, MD, PhD; Ingrid Glurich, PhD;
Catherine A. McCarty, PhD; Peggy Peissig, MBA; Justin B. Starren, MD, PhD;
Timothy S. Uphoff, PhD; Christina Zaleski, MS; Humberto Vidaillet, MD

MarshfieldClinic Research Foundation, Marshfield, WI 54449

This is an abbreviated version of the paper prepared for the Department of Health and Human Services Summit on Personalized Health Care, October 5–7, 2008. The full paper can be requested from the Director, Marshfield Clinic Research Foundation, 1000 North Oak Avenue, Marshfield, Wisconsin 54449.

I.    PIONEERING

On September 19, 2002 Wisconsin Governor Scott McCallum joined the leadership of Marshfield Clinic and its Research Foundation (MCRF) in announcing the enrollment of the first of what would be 20,000 research subjects into the Clinic’s Personalized Medicine Research Project (PMRP). Governor McCallum stated in his remarks:

“Marshfield Clinic and its research division are dedicated to the public good, using science, scientific research, and scientific discovery to improve the quality of life in Wisconsin, throughout the nation, and really, throughout the world. I congratulate the people of north central Wisconsin for their understanding of the importance of research to our health and well-being, for their commitment to participation in research, and for the community spirit shown in efforts to promote health and the health of future generations.”

The President of Marshfield Clinic, Frederic Wesbrook, MD, summarized for the audience the objectives of the research project, which was supported by $2 million from the State government, $800,000 from the Federal government, and $1 million from Marshfield Clinic.

“This project seeks to accelerate the reality of personalized medicine, a concept that envisions an individually tailored approach to detecting, preventing, and treating disease based on a person’s specific genetic profile. Some day your doctor will have a set of genetic tests that will tell you personally what diseases you are at risk of getting, what you should do to prevent or delay those diseases, and what medicines you should take or not take.”

The start of PMRP enrollment was preceded by almost 2 years of planning. The Clinic’s potential to contribute to personalized medicine was first realized and the vision created by Michael Caldwell, MD, PhD, then the Director of MCRF and PMRP’s first principal investigator (PI). The grant proposals were prepared and the project organized by MCRF’s Associate Director, Steve Wesbrook, PhD. PMRP was executed beginning with enrollment by Catherine McCarty, PhD, MPH, then Director of the newly created Personalized Medicine Research Center and currently the project’s PI. The project’s inception, design, and implementation were guided by a steering committee of co-investigators that included, in addition to those mentioned above, the Clinic’s then Director of Clinical Research, Kurt Reed, MD; Director of Laboratory Medicine, Robert Carlson, MD; Director of Medical Genetics, David Schowalter, MD, PhD; Director of Corporate Communications, Donna Chapman-Stone (responsible for community and population information and education); and Clinic’s Chief Information Officer, Carl Christensen. The team received advice and counsel from an Ethics and Security Advisory Board (ESAB), which was led by Norman Fost, MD from the University of Wisconsin, and a Scientific Advisory Board (SAB), which was lead by David Altshuler, MD, PhD from Massachusetts’s Institute of Technology and Jurg Ott, PhD from the Rockefeller University.

Between a meeting of the ESAB in August 2001 and the first meeting of the SAB scheduled for September 13 and 14, the PMRP team was wrestling with a fundamental question raised by the ESAB. “Is the Marshfield Clinic Personalized Medicine Program [as it was then titled] a service program or a research program?” At the recommendation of the ESAB, the name would be changed and the focus narrowed, in large part out of the need to be clear to research study participants that they would not be receiving any personalized medicine services. Below is an unedited slide (Figure 1) that was sketched by the Associate Director of MCRF to guide the discussion of this question at a meeting of the steering group on Monday, September 10, 2001.

Figure 1. Outline of Personalized Medicine

This working slide still remains a reasonable outline of the dimensions of ersonalized medicine. But what was perhaps most prescient about it was the recognition that even though the boundaries of PMRP would be narrowed to research, 1) the research project would serve as a catalyst for implementation of personalized medicine throughout the Clinic and 2) would create new linkages between scientific investigators and clinicians. What no one could predict, of course, was how much the United States would be changed by the events of the next day.

For the October 5-7, 2008 personalized health care (PHC) Summit in Utah, Marshfield Clinic was asked to focus on “how a community-based healthcare system has brought the key elements of PHC together to deliver more effective health care.” In doing so, we will address three elements of PHC: biomedical informatics; clinical care, including medical genetics and laboratory medicine; and medical research. The final section will provide some insights from our limited perspective that may have general import on future change and also address four of the major initiatives that are defining the PHC way ahead at Marshfield Clinic.

II.  BIOMEDICAL INFORMATICS

In 1964 a group of physicians at Marshfield Clinic determined that the future of high quality health care would depend on computers. The Clinic has held to that vision for the past four decades and, as a result, developed one of the largest regional integrated health care information systems in the nation. The system spans most of the northern half of Wisconsin and is used by over 12,000 users, not only Marshfield Clinic employees, but also affiliated hospitals and treatment centers, and even competing physician group practices. The information network maintained by Marshfield Clinic has been structured in parallel to the health care delivery process and is of strategic importance in an effort to provide consistent, quality health care to a large geographical area. Marshfield’s ability to develop information systems has provided the needed flexibility to react to evolving clinical needs in a rapid manner and has assisted in point-of-care decision support for PHC.

Where We Are Today. Effectively delivering PHC requires many different systems working in concert at Marshfield Clinic. These include:

• Regional, integrated electronic health record (EHR)
• Semantic interoperability
• Clinical data warehouse
• Decision support
• An Internet-based portal that enables patients to directly interact with the Clinic’s information systems
• Tablet computers
• Population-based tools

Leveraging Clinical Information Systems to Support PHC Research.  Marshfield Clinic believes that PHC is a process of continually improving knowledge and care, not a single endpoint. To that end, Marshfield has a history of integrating clinical computing with research computing.

• Population-based research. Marshfield Epidemiological Study Area (MESA) is a geographic population cohort in a 24 ZIP code area. The MESA database tracks a subject’s geographical location since 1991 and has the ability to link subjects to data stored within the Clinic’s data warehouse.
• Development of the Personalized Medicine Research Database (PMRD). PMRP leveraged existing Marshfield Clinic practice management and laboratory systems to recruit and collect genetic specimens. A cryptographic key system was developed to allow genotypic and clinical data to be combined for research studies, while protecting the privacy of research subjects. PMRD was created to store specimen sample identification numbers and the corresponding subject identification information. Later, PMRD was modified to accept validated genotype and phenotype data from the data warehouse.
• Phenotyping efforts. One essential requirement in assessing genetic impact on health and disease is the ability to characterize reliable phenotypes. Strong informatics and data management techniques, clinical guidance, statistical expertise, and clear communication with the disease experts enhance the ability to generate thoughtful and accurate phenotypes.
• Data mining. MCRF has entered into several collaborative data mining ventures with scientists from the University of Wisconsin-Madison to analyze large complicated genetic and phenotypic databases and develop algorithms that can predict patient reactions and outcomes to treatment.
• Episode-of-care. System monitors events in the EHR for patients who require special handling.

Challenges and Future Directions in Biomedical Informatics. The most important lesson we have learned is that systems to support PHC are not something that can be purchased “out-of-the-box” or simply “bolted-on” to existing systems and processes. They require commitment that spans years or decades. Achieving PHC requires a commitment to change, not only computer systems, but also health care processes. This implies that practicing clinicians must be involved at all stages of the development and implementation lifecycle. Another lesson is that the necessary integration cannot be achieved by silos, each focusing only on its own needs. Managers of clinical systems must believe that research is of value to the entire organization. Similarly, researchers must take the time to understand the ever-increasing demands on health care providers. It is acutely obvious to everyone at Marshfield Clinic that converting health care records to electronic form and eliminating paper charts (something that took 40 years to achieve) is only the first step toward a health care computational infrastructure that truly enables the vision of PHC. Marshfield Clinic is actively engaged in many projects to keep working toward the vision.

• Anonymized research data warehouse. The objective of this project is to develop a data warehouse that contains genetic, environmental, and clinical data.
• Natural Language Processing (NLP) of clinical documents. With over 55 million electronic documents containing health habits, family history, symptoms, environmental, and social factors, Marshfield Clinic is actively advanced in NLP to extract additional information for phenotyping and decision support.
• Phenotyping advances. Tools and techniques to improve the efficiency and interoperability of the phenotyping process are being developed.
• Pedigree mapping. Pedigrees add power to the genetic studies and allow rare disease studies to be conducted with limited cases.
• PHC reference library. The reference library will provide information on clinical, environmental, and genomic data and validated phenotypes.
• Research web portal. An Internet-based application (portal) that enables researchers to access genotype, clinical, and environmental information will be developed.
• Optimizing care through integration. Systematic workflow analysis and process mapping techniques are needed to seamlessly integrate not only the EHR, but also research innovations and discoveries into a busy practice setting.

III. CLINICAL CARE

Clinical Medical Genetics

Medical Genetic Services at Marshfield Clinic. The Medical Genetic Services Department at Marshfield Clinic provides clinical genetic consultation, diagnostic testing, and genetic counseling for patients and their families with genetic concerns. The greatest demand for clinical genetic services in adults is for single gene disorders including inherited cancers (BRCA1, BRCA2, HNPCC, FAP); connective tissue disorders such as Marfan syndrome, hemochromatosis, cardiomyopathy, Brugada, hemoglobinopathies, Huntington’s disease; and genetic susceptibility to adverse drug reactions. Annual unique patient referrals to Medical Genetic Services have grown from <30 in 1999 to over 300 in 2007.

The clinical genetics team realizes how important it is to raise awareness among patients regarding seeking genetic services, and has received grant support to increase the patient’s understanding about the value of genomic medicine. The initiative, led by Christina Zaleski, MS earned her the 2007 Leadership in Excellence Award for Community Service. The project involved creating multilingual (English, Spanish, Hmong) brochures and posters that discuss when and how to access genetic counseling for families with high-risk newborns or those who have experienced miscarriages, a stillbirth or other infant death. These materials were distributed to all birth centers in Wisconsin.

Utilization of genomic medicine to provide optimal clinic care requires that practitioners feel comfortable with ordering and interpreting genetic test results, as well as discussing these results with patients and their families. To help increase the awareness about medical genetics among primary care providers in Wisconsin and encourage practitioners to utilize clinical genetic services, Marshfield Clinic offers an annual state-wide conference entitled “Practical Genetics for Health Care Providers.”

The future of clinical medical genetics. With PHC emerging as an important contribution to clinical care, it is important to sustain and grow clinical genetics as a state-of-the-art service to both patients and healthcare professionals. The American Board of Medical Genetics is in the process of expanding the role of clinical geneticists and suggesting that geneticists broaden their services to liaison with other departments and be viewed as a resource for primary care patient management. It is also critical to attract students to consider careers in genetic counseling.

While genetic medicine has largely centered on provision of diagnoses and treatment for individuals with well-defined single gene disorders, genomic medicine when fully realized will decipher genetic information derived from a person’s genome into predictors of disease susceptibility. A personalized medicine approach can be implemented for a particular individual and may consist, for example, of avoidance of certain disease risk factors or implementation of various screening modalities. Pharmacogenomic advances will facilitate testing for multiple genetically-mediated drug sensitivities, and genetic counseling will be needed for patients and their family members to understand the relationship between drug metabolism capacity and genetics that underlie them. Prospective genetic testing will be invaluable to the primary care provider in planning appropriate treatment.

To realize the promise of genomic medicine, health insurance barriers need to be overcome. Genetic referral and testing represents an exclusion in many insurance policies. Insurance denials entail additional workloads to genetics professionals, and in some instances appeals need to be made by the patient and not the healthcare provider. In order for patient care to be optimized there needs to be a three-way transfer of information between clinicians, researchers, and community members. PMRP has influenced interconnectedness between clinical care and research through its Community Advisory Group (CAG) and quarterly Personalized Medicine News (Figure 2). Accurate family histories from patients is challenging for many reasons. Bioinformatics approaches appear to have great promise.

Figure 2. Personalized Medicine News

Laboratory Medicine

Where we are today. Advances in genomics and related technology in the past decade have resulted in significant growth in molecular diagnostic testing and services, which has impacted almost all areas of laboratory medicine. Marshfield Laboratories was an early adopter of molecular diagnostics and has been performing such testing for over 12 years. As molecular testing increased in breadth and crossed into more traditional lab sections, it became apparent that for a regional laboratory, such as ours, to rapidly adopt this technology, a core laboratory with expertise in molecular testing was necessary. In 2005, Marshfield Laboratories formed its molecular pathology section to expand, coordinate, and standardize this growing area of testing services utilized for bacteriology, virology, coagulation, hematology, genetics, histology, and pathology. Prioritization of resources for development, validation, and implementation of new molecular testing is determined with input from clinicians and laboratory personnel representing all these areas. Test volumes in molecular pathology have grown faster than any other areas of the clinical lab during this time period (Figure 3).

The Molecular Pathology Laboratory is continuing to develop improved automation and expanding test menus. PHC tests in the area of oncology are increasingly prevalent, as these treatments are very expensive and carry high risks. Identification of KRAS gene mutations in codons 12 and 13 can predict whether or not an epidermal growth factor receptor (EGFR) inhibitor will be useful to treat colorectal cancer. A challenge for our laboratory to offer this testing is that the only commercially available testing product in the U.S. is labeled RUO (“for research use only"). Regardless of how well the test is validated, the use of RUO reagents in clinical testing significantly reduces or eliminates most forms of third party reimbursement. In the overall scheme of patient health care, it is clear that identifying patients who will not respond to costly treatments is prudent. However, KRAS mutation test reagents alone cost approximately $100/test. With little hope of reimbursement, the laboratory’s financial prospects of this testing are bleak. Our implementation of JAK2 testing to identify myeloid proliferative disorders, such as polycythemia vera, has been similarly hampered because the company holding exclusive intellectual property rights for clinical testing offers only RUO reagents for sale. These are not isolated instances.

Figure 3. Test Volume Growth in Molecular Pathology

Challenges and future directions in laboratory medicine. Biomarker discovery has proven to be much more difficult than initially envisioned. In addition, many new markers are part of a complex interaction with other genes and environmental influences making clinical utility difficult to ascertain. Also, while great strides have been made in the technology involved in DNA sequencing and genotyping, the availability of accurate phenotypes is lagging far behind. Another significant hindrance to bringing new molecular testing into the clinical laboratory is affordability, which is often linked to gene patents. Also, the complexity of patent and intellectual property regulations limits availability. The costs associated with advanced medical technology for PHC are disproportionately higher than traditional diagnostic services. If PHC is to grow, healthcare institutions, Centers for Medicare and Medicaid Services, and insurance providers must recognize the overall healthcare savings of PHC and support testing through appropriate reimbursement.

IV. GENETIC RESEARCH

Personalized Medicine Research Project. PMRP was designed to support genomics research in three areas: pharmacogenetics, genetic basis of disease, and population genetics. The project required a concerted effort to develop not only genotyping and phenotyping capability and informatics infrastructure, but also needed to address issues such as logistics of population-based enrollment, bioethics, and stewardship of the biobank. From its inception, the project was intended to serve as a national resource for hypothesis generation and testing.

Nearly 20,000 adults have enrolled as of August 2008. Over 99% have consented to be re-contacted. In addition to the extensive EHR, the temporal span in years of clinical data available for PMRP subjects sets the cohort apart from other similar projects. The average span of clinical history for PMRP participants is 29+ years.

Pharmacogenetics. The study of the genetic impact on drug metabolism and disposition, and how this translates into drug efficacy or contributes to adverse drug events, has been a research priority at Marshfield Clinic. Below are ongoing extramurally-funded pharmacogenetics research studies.

• Efficacy and safety of statins
• Genetic impact on metformin metabolism and management of patients with type II diabetes
• Pharmacogenetics of tamoxifen response in treatment of breast cancer
• Pharmacogenetics underlying response to beta blockers in patients with glaucoma
• Pharmacogenetics of warfarin metabolism

Genetic Basis of Disease. Outlined below are ongoing extramurally-funded studies being conducted at Marshfield Clinic on the genetic basis of disease.

• Cataracts
• Scoliosis and other vertebral malformation
• Genetic and environmental interaction and risk factors contributing to multiple sclerosis
• Myocardial infarction risk and influence of genetic variation on chromosome 9p21

In addition, Marshfield Clinic investigators are conducting internally funded studies using the PMRP research infrastructure in hypertensive heart disease, Alzheimer’s disease, fibromyalgia syndrome, and osteoporosis.

V.  THE WAY AHEAD

Marshfield Clinic is a robust, comprehensive and highly integrated health care system. It has over 750 physicians and 6,500 additional staff at 41 centers in a primary service area that includes 60% of Wisconsin geographically. Annually, the Clinic sees approximately 370,000 unique patients. In support of its mission to serve patients through accessible, high quality health care, research, and education, Marshfield Clinic supports strong programs in research and graduate education, maintains its own 147,000 member HMO, and provides through its Community Health Center and other programs health care for people irrespective of their ability to pay. But the capability of even a large and progressive health care system to deliver PHC depends largely on external factors, including the general advancement of science, speed with which industry commercializes discoveries, intellectual property law, government and private medical insurance payment schedules, and many others.

To the degree to which Marshfield Clinic can control its own PHC future, it will continue to do what it has done well over the past decade.  This includes striving for even better integration of medical research and clinical practice. Marshfield Clinic and MCRF are either leading or substantially contributing to a number of new initiatives that will also shape its way ahead.

Clinical and Translational Science Awards (CTSA)

University of Wisconsin School of Medicine and Public Health and MCRF partnered to receive a NIH Clinical and Translational Science Award in September 2007. Currently supporting a national consortium of 34 academic medical research institutions, plus partnering institutions, CTSA is scheduled to link 60 such institutions by the year 2012. The consortium has been designed to ensure broad access to CTSA resources, enhance communication, and encourage information sharing.

Marshfield Clinic also joined in the fall of 2007 with five academic health science schools/colleges at the UW-Madison to create the UW Institute for Clinical and Translational Research (ICTR). ICTR was established to “create an environment that transforms research into a continuum from investigation through discovery and to translation into real-life community practice, thereby linking even the most basic research to practical improvements in human health.”

Wisconsin eHealth Care Quality and Patient Safety Initiative

On November 2, 2005, Wisconsin Governor Jim Doyle came to Marshfield Clinic to sign Executive Order 129 creating the Wisconsin eHealth Care Quality and Patient Safety Board. Governor Doyle charged the Board with establishing an action plan for the statewide adoption of EHRs and the exchange of health care information by the year 2011 (http://ehealthboard.dhfs.wisconsin.gov/). Referring to the Clinic’s leading role in developing and using EHRs, its successful quality initiatives, and its history of championing for health care reform, he stated that “Marshfield Clinic is truly the place to make this announcement.”

Whereas progressive and committed organizations can do much on their own, no single institution has enough information on all their patients to provide optimal health care, and health information exchange cannot be done by a single institution. A statewide eHealth information infrastructure will improve the quality and cost of health care in Wisconsin by 1) ensuring health information is available at the point of care for all patients, 2) reducing medical errors and avoiding duplicative medical procedures, 3) improving coordination of care between hospitals, physicians, and other health professionals, 4) furthering health care research, and 5) providing consumers with their health information to encourage greater participation in their health care decisions. The goal is to achieve 100% electronic health data exchange between payers, health care providers, consumers of health care, researchers, and government agencies as appropriate.

Personalized Health Care Testbed

Achieving the computational infrastructure for PHC will require the integration of many different components. It will also require two very different types of research endeavors. The first type of endeavor will involve research groups that have deep expertise in one, or a few, of the components. These groups will develop new theories and approaches to specific problems in PHC. For example, one of these groups may develop improved knowledge discovery tools and a health care workflow engine. Essentially, these groups focus on one piece of the puzzle. The second type of endeavor will involve the development of a PHC Testbed (Figure 4). The number of potential PHC interventions is likely to increase exponentially in the coming years. The role of a PHC Testbed is to evaluate the impact of implementation of possible PHC interventions in real clinical practice, at a speed that is more rapid and at a cost that is much below what would be required by a de novo conventional clinical trial. Testbed institutions will need to have expertise across a broad range of domains.

Figure 4. Personalized Health Care Testbed Architecture

There are three foundation components in a PHC Testbed: large scale genotyping, population coverage, and longitudinal clinical data. Potential PHC interventions that will be evaluated by a PHC Testbed can be divided into major groups. The first are those that involve associations between genetic or metabolic markers that are measured by existing broad screening tools, such as large-scale SNP chips. The second base-level attribute of a PHC Testbed is population coverage. Having a large and stable population base allows prior information to be applied to future clinical care in a large percentage of cases. The third base level competency for a Testbed site is an extensive repository of longitudinal clinical data. Since the future of PHC is certain to include lifetime EHRs, attempting to evaluate PHC interventions without many years of prior clinical data can yield spurious results.

The next level in the PHC Testbed is semantic interoperability. Without a consistent framework for what individual clinical terms and concepts mean, it will be impossible to reliably identify patients or evaluate outcomes. Above the semantic layer is the knowledge layer, which includes two components, analytics and knowledge discovery and knowledge assimilation. The process of defining clinical phenotypes and clinical outcomes involves the use of knowledge discovery techniques. Any successful PHC Testbed must have an active knowledge discovery group that can rapidly address new questions. Knowledge assimilation is the process of incorporating structured knowledge from outside an organization into the computational knowledge framework of the institution. Any PHC Testbed will need a structured approach to knowledge assimilation so that new PHC intervention can be incorporated efficiently into the institution’s knowledge base.

Clinical care involves complex processes with multiple steps. Simple rule-based systems are inadequate to capture these clinical workflows. The implementation of PHC will require the implementation of electronic workflows that support the complex, multi-actor nature of clinical care. The ultimate goal of PHC-driven electronic workflows is to improve clinical outcomes. PHC Testbeds will need both experience evaluating clinical outcomes and access to comprehensive data in order to determine true outcomes. This is much easier in sites with stable patient populations and broad population coverage. Two other components of the PHC Testbed span all levels: ethics and security and standards.

For PHC interventions to become mainstream therapy, they will need to be evaluated not only in highly controlled studies but also in a real world practice setting, like those represented by a PHC Testbed. A successful PHC Testbed will require expertise in such a broad range of domains. Many institutions have strengths in one, or a few areas, but very few have strength across the entire spectrum required for a PHC Testbed. However, the success of PHC will be markedly delayed if such PHC Testbeds are not available. Marshfield Clinic represents a unique combination of capabilities across this spectrum.

Wisconsin Genomics Initiative (WGI)

On October 10, 2008 Wisconsin Governor Jim Doyle announced the Wisconsin Genomics Initiative, which is a collaborative research effort among Marshfield Clinic, Medical College of Wisconsin, University of Wisconsin School of Medicine and Public Health, and University of Wisconsin–Milwaukee. He stated that “By capitalizing on the unique strengths of each institution, we have a rare opportunity to meet an important scientific and public health need that could otherwise not be met.” The vision of WGI is to be able to predict for individual patients in a clinical setting the risks of disease susceptibility and treatment response using the combined power of cutting edge genetic, phenotypic, and environmental analyses, thereby making the promise of personalized medicine a reality (Figure 5).

Figure 5. Predictive Personalized Medicine

The key elements of its phase I WGI strategy are to 1) genotype up to 20,000 PMRP participants for 1,000,000 genetic markers, 2) validate selected target phenotypes and multiple clinical attributes from the Marshfield Clinic EHR for the PMRP cohort, 3) integrate genetic, phenotypic, and environmental information databases and develop the search engines to use data efficiently for scientific discovery, and 4) to build predictive computational models using machine learning and super-computer capability, for the key equation, Genetic + (Environment and Clinical) = Phenotype. It will then conduct initial predictive studies (diabetes, obesity, coronary artery disease, and atrial fibrillation) to test and improve the scientific platform, as well as a genome-wide association study (GWAS). WGI institutions anticipate making the WGI scientific platform, information, and methods available to scientists across the country. In phase II, WGI plans to add a 20,000-person urban cohort, a pediatric cohort, and to expand substantially predictive studies.

Return to Table of Contents

Moffitt Cancer Center

Tampa, Florida

Community-Based Personalized Health Care

William S. Dalton, PhD, MD; David Fenstermacher, PhD; Paul Jacobsen, PhD;
L. David de la Parte, JD; Timothy Yeatman, MD

Personalized Cancer Care Defined

Cancer represents the second leading cause of death as of 2005. The sequencing of the human genome has unraveled many mysteries as to how a normal cell can go awry and become cancerous. Further understanding of not only the genetics of cancer but the biology and metabolism of cancer has increased our knowledge of biologic systems that support cancer growth, and this new knowledge has been translated into novel strategies for preventing and treating cancer. And yet, these new  discoveries which have heightened expectations of success, have in large part, fallen short in delivering dramatic cures anticipated by society. The reality is that we have learned that cancer is actually an array of many diseases masquerading under the single title or name of "cancer." We need to embrace the complexity of this disease we call cancer, and stop focusing on treating the cancer, and instead, focus more on caring for the patient. National policy must promote the search for solutions, not just cures. These solutions will reduce, and ultimately eliminate death and suffering due to cancer. Solutions for reducing and eliminating suffering due to cancer will be accomplished by individualizing and personalizing cancer care with the following goals:

1. Identification of the needs of the individual patient
2. Identification of markers that will predict needs and risks so that interventions can be applied earlier
3. Development of methods for early detection of cancer
4. Identification of signatures predicting which patients will not respond to standard of care therapies
5. Utilization of clinical characteristics and molecular profiling, matching the right treatment for the right patient
6. Improvement in the performance of clinical trials by patient matching
7. Raising the standard of care for all patients by integrating new technologies in an evidenced based approach to maximize benefits and reduce costs

At the Moffitt Cancer Center in Florida we are developing a personalized approach to cancer care we call Total Cancer Care (TCC). TCC represents a holistic approach to cancer that places the patient at the center of their life journey, as illustrated in the diagram at left.

Accomplishing these goals will require the following:

1. Identification of high risk populations
   1. genetic markers
   2. environmental factors of influence
   3. molecular epidemiology
   4. mathematical modeling to predict risk
2. Improved technology for early detection of cancer
   1. identification of biomarkers
   2. development of new imaging techniques
   3. study of metabolomics
3. Improved therapies by utilizing multi-modality approaches designed for individuals based on molecular profiling of tumors and analysis of treatment tolerance
   
4. New therapies for patients who do not benefit from standard therapy
   
5. Improved performance of clinical trials by reducing time and number of patients on trial by trial matching
   
6. Better methods/models of drug discovery
   
7. Reduced suffering by improving psychosocial and palliative care for patients and their families
   
8. Development of factors that predict patients at risk and providing early intervention
   
9. Creation of evidenced-based guidelines that define when to use certain technologies, and improve access for all patients

Development of large regional cancer biorepositories in parallel with the development of a related information system containing the patient’s clinical outcomes data holds the greatest promise for achieving the goal of personalized cancer care. Such an endeavor will facilitate discovery of biomarkers for the identification of high risk populations, early detection, prognosis, predictors of response to therapy, new drug targets, predictors of toxicity and late effects, and clinical trial matching. Such an effort requires prospective patient consent to participate in a trial that requests the following of patients:

1. Permission to follow the patient clinically throughout their lifetime
2. Permission to store tumor specimens for molecular analysis
3. Permission to collect patient clinical data to integrate with scientific data using secure data management systems

Once created, computational approaches can be developed to compare and analyze data from all patients so that relationships can be discovered and evidence generated to develop best practices. From the evidence generated, knowledge can be derived so that effective technologies will be utilized in appropriate circumstances for individual patients to promote best solutions for their cancer disease.

The Road to Development of Total Cancer Care at Moffitt

At the heart of Moffitt’s Total Cancer Care^TM Program is an obligation to serve as a resource to Florida communities, and the nation, in both cancer prevention and treatment. TCC seeks to overcome barriers to personalized cancer care and provide far-reaching access to the latest discoveries in lifesaving research.

The vision of TCC required three significant building blocks, among others: 1) establishing a network of partners to provide access to an NCI Comprehensive Cancer Center’s expertise; 2) improving patient participation in clinical trials by matching the right patient to a trial; and 3) developing a database as a source for the collection, storage, integration and management of clinical data and scientific findings.

Our initial quest began in 1999 with the creation of a network of affiliate hospitals and physicians across the State of Florida. This network was designed to provide access to expert cancer care for patients located hundreds of miles from a comprehensive cancer center at locations closest to the patient’s home. Goals of the affiliate network partnership are to improve access to technological advances, improve quality of care, and increase participation in clinical trials. Following passage of the NIH Revitalization Act of 1993, the NIH established guidelines for inclusion of women and minorities in clinical research. This led to cancer centers making concerted efforts to expand access to clinical trials throughout their diverse communities. Establishment of the affiliate network provides the infrastructure to expand the participation of patients in clinical trials within their own communities. Building this partnership with community centers is a critical building block to our development of personalized cancer care.

Because TCC is a partnership with the patient and involves tissue acquisition and collection of clinical data, a formal protocol and patient consent process was developed. In addition, opinions of patients, patient advocates as well as bioethicists were sought. The TCC protocol was developed as a comprehensive, prospective research study to acquire tumor tissue and data from cancer patients across time and was approved by the Institutional Review Board (IRB) in 2006. The information technology platform to support this massive effort was developed to provide for a highly robust “warehouse” of clinical and molecular profiling data for patients participating in the TCC project. Upon consent to participate, patient clinical data is stored over time, including results of laboratory test results, radiographic images, treatment response data and tumor molecular profiling signatures.

The information system has to integrate electronic data from a multitude of data sources, in multiple formats and ultimately be available for research and clinical uses in a highly secure environment.  

The TCC data warehouse is designed to provide a disease management system that integrates clinical data into all aspects of cancer diagnosis, treatment, and care. The integration of these data, over time, will allow the ability to identify populations of patients that may be eligible for current and future clinical trials. Consequently, patients throughout the state of Florida and beyond benefit from the knowledge gained from previous patients through maintenance of an ongoing data repository, resulting in an improved quality of cancer diagnosis, care and prevention.  

Taking TCC from “vision to reality” was estimated to cost over $100M in the first five years. It became abundantly clear that innovative approaches to funding these efforts were needed. Several funding sources were considered and pursued: philanthropy, government sponsorship and private sector partnerships. Discussions began with a pharma partner who shared the same vision of developing personalized drugs for the “right patient at the right time.” These discussions led to a successful collaboration with Merck & Co. in 2006. 

Capitalizing on the State of Florida’s focus on economic development and the growth of biotech in the state, Moffitt created a wholly-owned subsidiary, Moffitt Genetics Corporation (M2Gen), to focus on administering the collaboration with our pharma partner and meeting the contractual obligations of our newly developed personalized medicine venture. The investment by the State of Florida and local (county & city) governments provided additional capital and land for the construction of a state-of-the art biorepository able to house tumor, blood and urine specimens critical to the project and provide for additional expansion of the sites participating in the project. In return for the state’s investment, new jobs are being created in Tampa Bay and Florida.

Current State

Since protocol approval in 2006, the TCC Personalized Medicine research project has enrolled over 20,000 patients across 16 sites; eight (8) in Florida including the Moffitt Cancer Center, and sites in seven (7) other states: Louisiana, Connecticut, North & South Carolina, Kentucky, Nebraska and Indiana. These sites were identified based on their volume of cancer patients, and shared interest and commitment to the development of personalized medicine, the willingness to offer clinical trials to patients and their infrastructure to support clinical trials operations and bio-specimen collection. Three of the consortium sites were identified because of the efforts of the NCI National Community Cancer Centers Program (NCCCP) which provided the infrastructure for biological specimen collection and clinical trial performance. 

The central data warehouse is in place, receiving interfaced data from a number of systems
at Moffitt including the tumor bank, cancer registry, and the clinical information system. Portals to access the database are now being constructed for researchers and patients, and in the future clinicians, using the database as a decision tool.

The ability to capture data in a discreet format is extremely valuable, yet a challenging effort. As part of the project, a standardized template has been created to capture response to therapy data. This data will be critical for future research efforts in developing evidence-based, personalized treatment protocols.

Digital imaging technology and the implementation of the Emageon™ product as the DICOM-compliant imaging database for all imaging files at Moffitt provides the ability to link diagnostic images into the data warehouse. This provides the platform to implement a means to quantify response to therapy with a consistently applied methodology such as RECIST (Response Evaluation Criteria in Solid Tumors) or other modes to measure progression of disease.

Compliant and appropriate access to the data is under development that provides a single point of access and presentation to a variety of data elements. Viewing permissions and corresponding review policies and processes are essential to ensure HIPAA and human subject research compliance, while not undermining the pursuit of statistically valid research hypotheses. 

An underlying premise to the information technology aspect of the Total Cancer Care™ project is meeting the data needs of various stakeholders. Although there are many others, Moffitt has initially defined three key data stakeholders in this effort, both at Moffitt and the participating consortium sites: researchers, clinicians and patients. In an effort to meet the needs of these stakeholders three portals to the data warehouse are under development. As illustrated below, these portals will allow views into the data to meet the needs of each constituent group.

The Patient Portal, currently under development, will initially focus on providing cancer survivors access to a summary of their treatment and a personalized survivorship care plan. It will also have transactional features, such as appointment scheduling and bill paying. In addition, the portal will use demographic, disease, and treatment data in the data warehouse to generate educational information tailored to the individual patient. This goal will be accomplished using a sophisticated search engine that retrieves only highly specific and relevant sets of information. The tailored educational information is expected to help patients/survivors better understand and address their ongoing psychosocial and physical needs as they move from active treatment into follow up. 

The Research Portal has been constructed and is being used by basic scientists, clinical scientists, and cancer control scientists.  We anticipate that the database will support research for new drug discovery targets, molecular signatures to predict therapy response and resistance, risk for relapse and new primary cancers, as well as, clinical studies for outcomes analysis.

The Clinician Portal will be developed over the next several years as a means to provide clinicians with evidence-based treatment guidelines, generated through the outcomes research conducted through TCC as well as consensus-based guidelines through NCCN or other generally accepted national standards.

An “honest-broker” system has been developed for access to data, as well as, for access to the bio-specimen repository by creating an institutional Tissue & Data Release Committee. This multidisciplinary group is comprised of faculty, Tissue Core leadership, Technology Transfer and regulatory/legal experts. A coordinator for the committee provides investigators consulting and facilitation in ensuring the research projects are appropriately defined and approved through the scientific and IRB reviews.

The bio-specimen repository supporting the TCC™ effort provides a critical resource for the development of personalized cancer care. Through the molecular profiling of thousands of tumor samples, along with blood, urine samples and corresponding clinical history, over time we will be able to identify genetic targets that can identify various cancers, develop diagnostic and prognostic tools, perform Phase-2 clinical trial enrichment and ultimately, develop a personalized approach to each patient’s cancer treatment. As of August 2008, the TCC™ project has over 8,000 tumor specimens, of which approximately 1,900 have been profiled. 

Partnerships with state and local government have provided the funding for building a state-of-the-art, custom designed, fully automated freezer system with robotic capability for storage and retrieval. By 2011, it is estimated that the bio-repository will house over 3 million sample tubes: tumor tissue, normal tissue and pre/post operative liquids.

Challenges in the Development of Personalized Cancer Care

Taking the development of personalized cancer care from concept to patient care is not without significant challenges, some of which are addressed briefly below.

Funding

TCC requires the development of an integrated data management platform that interfaces with a broad range of information sources: electronic and paper, both at Moffitt and at the consortium sites participating in the project. As such, this effort has required a significant commitment of capital towards technology and personnel. These commitments currently exceed $100M for the initial five years of the project. Significant additional capital, however, will be needed to sustain and advance the project. 

Potential funding sources were considered and pursued, including: state, federal, philanthropy, and strategic partnerships. Ultimately, a non-traditional approach permitted us to initiate the development of TCC by creating a “win-win” scenario for the partners involved.

The state, county and city benefit through expansion of a knowledge-based economy, creation of new jobs and impact on the economy of nearly $211M in direct and indirect income and more than $56M in direct and induced capital investment. 

Private industry benefits by access to a unique resource of human tumor samples and associated de-identified clinical data, leading to molecular and genetic signatures, novel drug targets and clinical trial enrichment.

Ultimately, the patient benefits from these financial investments through the development of personalized treatments tailored to their tumor’s genetic profile, thus increasing safety and efficacy of treatments, while decreasing toxicities.

Regulatory Compliance

A complex and sometimes conflicting framework of federal and state regulations governing human subject research, patient privacy (HIPAA) and ownership of tissue and rights to intellectual property requirements govern the collection, storage, dissemination and use of human biological specimens and the corresponding patient data. Ethical considerations both medical and legal must also be applied to the use of patient tissue and clinical data for research.

This patchwork of regulation and agency guidance is a matter of concern as research is conducted using the tissue and corresponding data that has the capability of generating intellectual property and commercialization opportunities for the investigators and institutions involved. Also challenging is the management of any contractual obligations established through partners invested in the endeavor.

Recognizing the challenges faced by cancer centers in developing this research resource, in June 2007 the National Cancer Institute published the National Cancer Institute Best Practices for Biospecimen Resources.

The development of a clear, but broad protocol for the collection of tissue and data, along with a correspondingly clear and understandable informed consent and authorization for use of protected health information in research is a critical first step in addressing this challenge. 

All patients at Moffitt are approached with an invitation to participate in TCC, regardless of whether their care involves the collection of tumor.  Patients are asked to be a partner in Total Cancer Care for life.

This vast resource and investment by the patients themselves carries with it significant moral, ethical and legal obligations to protect the participants from harm (i.e. breaches of patient privacy) and provide a benefit to these individuals who have placed their trust in us.  The return for these valuable stakeholders is: a contribution to the development of advances in personalized cancer treatment that may ultimately lead to clinical trials and new drugs that will treat their specific cancer; development of evidence-based guidelines to improve the standard of cancer care and an integrated information system that will allow them access to their health information.

With the goal of ensuring the protection of the patients/research participants in TCC, and ensuring the highest quality biospecimens and their use in scientifically sound research, Moffitt adopted an “honest broker” system for their data and tissue repository.

A multidisciplinary steering committee comprised of faculty, Tissue Core staff, regulatory and legal staff functions as a means to ensure that requests for access to tissue or data in TCC are
properly vetted to ensure coordination of IRB, Privacy Board and scientific reviews. Links to patient identifiers are retained only through the honest broker. Identifiable data, linked with specimens is only provided upon approval from the IRB for a specific use protocol from the tissue bank/data warehouse. 

Standard Operating Procedures for the optimal collection, processing and storage have been developed to ensure the highest quality of the specimens.  This is of critical importance due to the participation of sites across the country, all with varying degrees of expertise in best practices.  Protocols are also necessary to standardize the procedures for the shipment of these specimens
to Moffitt for molecular profiling and storage.

Patient Concerns

Patients are a critical partner in the success of Total Cancer Care. We have an obligation to provide the educational tools to help patients overcome their concerns about participating in a life-long research study. Patient advocacy organizations play a vital role in helping cancer patients understand the implications of their participation and the value to them in contributing to the future of cancer research. 

Equipping patients with the knowledge they need to participate in their health care decisions, including an understanding of their contribution to the development of personalized medicine requires a concerted effort. 

First, patients must overcome concerns regarding the creation of molecular data (e.g. gene expression, sequencing, etc) from their tumors. Patients may fear this research identifies them as having predictive markers for developing cancer, such as BRCA1gene mutations. It is imperative that patients, and their families, be considered major partners in the development of these resources. Ultimately, these challenges are only overcome with extensive education, communication, proper informed consent and the involvement of patient advocacy groups as well as patients themselves.

Changes in Physician Practice

Health care providers across the country understand the need for highly efficient processes and practices in order to reduce costs.  The development of evidence-based personalized cancer treatments requires the gathering of significant amounts of discreet data on a patient’s staging, treatments and response to therapy. 

This requires the creation of new data collection tools (i.e. CRFs) for the research that gathers the needed data while not interfering with the clinical care process of health practitioners. 

Information Technology

Although the collection of this data electronically is ideal, there is no uniformity of an electronic medical record across health care organizations and information technology capabilities vary from site to site. 

The capital investment required to develop an integrated, electronic health record is astronomical and well out of reach of many community cancer centers.

The development of personalized cancer care requires an information platform that can sustain a substantial amount of data, not only from one site but from several, while providing secure and appropriate access to the stakeholders to conduct the research leading to the discoveries that will translate to patient care. This is not only a challenge to address in terms of financial sustainability, but is a challenge to ensure the most visionary architecture is adopted to provide for data quality, security and integrity. NCI initiatives such as caBIG (Cancer Biomedical Informatics Grid) will hopefully address needs and supply solutions for these major challenges. Given the complexity of the challenge, it is also likely that solutions will emerge from private/public partnerships that address the needs of multiple stakeholders with the ultimate beneficiary being the patient and families.

The Future of Total Cancer Care

In the future, we hope that the TCC™ database will be robust enough so that it can be used as a decision tool for clinicians caring for cancer patients. The figure below illustrates this concept by considering a patient diagnosed with breast cancer in, for example, Pensacola, Florida. 

A newly diagnosed cancer patient would be enrolled in the TCC™ protocol at their community medical center and surgery is performed. The tumor specimen is sent to Moffitt for profiling and entry of results into the database. As the patient undergoes treatment within the community, the database is electronically (via web) updated with the patient’s discreet clinical information over time, including diagnostic images, and response data.  As the number of patients in the database grows, the ability to match patients to effective clinical trials increases. In addition, as outcomes data increases, so does the ability to create evidence-based, in lieu of consensus-based, guidelines for each tumor type. Ultimately, the database would provide the stakeholders the ability to query: physicians for the most effective treatment guidelines for patients they are seeing with a particular tumor profile; scientists to develop new biomarkers and health outcomes research; and patients to have access to their own information more effectively and help them better understand their disease and treatments.

It is imperative that as many patients as possible with diverse backgrounds be entered into the database so that the variables of genetics and environment can be considered. The best approach to ensuring that the database represents the community being served is to, in fact, make this protocol and approach available to as many community medical centers as possible, and not limit participation to tertiary research centers.

Impact of Personalized Medicine to the Cancer Patient

The Moffitt Total Cancer Care™ project is an ambitious approach to cancer care and research by identifying patient needs and developing solutions by integrating new technologies into the standard of care, improving performance of clinical trials, and generating evidenced-based cancer care that will increase response rates, reduce toxicity for patients, and increase access to state of the art cancer treatments within the patient’s own community.

Once realized, patients participating in TCC will provide a tumor sample for profiling, their physician will be able to query the database to match the patient to optimal evidence-based guidelines, personalized for that patient.  Patients will be able to query the data warehouse through a patient portal and receive their survivorship care plans and personally relevant information regarding their cancer through a highly sophisticated search engine. 

Achievement of this vision will take years and a continued investment by all the stakeholders
in the effort.

¹ National Vital Statistics Reports, Volume 56, Number 10, April 24, 2008. (http://www.cdc.gov/nchs/data/nvsr/nvsr56/nvsr56_10.pdf)

Return to Table of Contents

Partnership for Personalized Medicine

Arizona and Washington

Biodesign Institute at Arizona State University, Tempe, AZ
Translational Genomics Research Institute, Phoenix, AZ
Fred Hutchinson Cancer Research Center, Seattle, WA

Science and technology are beginning to provide revolutionary insights into medicine through a comprehensive molecular understanding of human health and disease. However, the promise of better health for all is undermined by the growing cost of medical treatments, which threatens the very viability of health care systems around the world. By 2015, health care spending in the U.S. is expected to reach $4 trillion, or 20% of GDP, and by 2020, spending will double in OECD countries. The challenge we face is to use our new knowledge to improve patient outcomes while stabilizing or reducing the costs of health care. We believe that this is possible by realigning our science to meet the needs of health care.

Current economic incentives assure that companies will develop the most expensive new therapeutics and devices while neglecting the power of new diagnostics to improve health at reduced cost. The promise of personalized medicine to improve health care outcomes and reduce health care costs will not be manifest by the marketplace where incentives align with expensive therapeutics for late-stage disease. It is paramount that health care providers and insurers play a new role in medical research, becoming the vehicle for the discovery, validation and implementation of new diagnostic platforms that can achieve the goals desired by patients and providers—prevention, early detection and effective intervention at reasonable cost.

Critical opportunities exist in all diseases for better molecular diagnostics to improve patient outcomes while reducing health care costs:

• Risk assessment: Identifying individuals at greater risk of developing specific diseases will enable the implementation of preventive measures that could eliminate both the suffering from disease and the costs associated with treatment.
• Early detection: For many diseases, diagnosis at earlier stages of disease progression allows intervention when there is a greater likelihood of effective treatment and cure. For example, in nearly all forms of cancer, early diagnosis can lead to a cure at a fraction of the cost of ineffective treatments for late-stage disease.
• Definitive diagnosis: The diagnosis of many diseases is challenging due to a lack of distinctive symptoms. Improved diagnostics will allow more rapid and effective implementation of appropriate treatments for those who will benefit while preventing adverse side effects and the costs of treatment for those who will not.

Archimedes said, “Give me a lever long enough and a fulcrum on which to place it and I shall move the world.” The lever that will help health systems move health care in a revolutionary new direction is the Partnership for Personalized Medicine (PPM). PPM is a nonprofit initiative with substantial foundation support whose goal is the development, validation and clinical application of new molecular diagnostics, designed to improve health outcomes and, importantly, reduce health care costs.

The Partnership for Personalized Medicine is led by Dr. Lee Hartwell, President and Director of the Fred Hutchinson Cancer Research Center and 2001 Nobel laureate; Dr. Jeffrey Trent, President and Scientific Director of the Translational Genomics Research Institute (TGen); and Dr. George Poste, Director of the Biodesign Institute at Arizona State University.

How Does the Partnership for Personalized Medicine Work?

The PPM model is based on the formation of collaborative partnerships that leverage a full suite of genomic and proteomic capabilities provided by PPM with dedicated health care systems to complete demonstration projects that integrate four key elements:

• A cohesive and interactive partnership between health insurers, providers, clinicians, and researchers;
• Epidemiologic, clinical and economic analysis to identify critical intervention points in disease management;
• Systematic and empirically based discovery, development and validation of new diagnostic tests to improve patient outcomes and reduce system costs; and
• Collaborative, prospective and evidence-based evaluation of the test within health systems to validate and implement the new test in patient management.

An Evidence-Based Approach

PPM features the following approaches:

• Health care economics: Economic analysis will identify major disease costs and opportunities for interventions to reduce costs; examples include earlier disease detection to enable preventive measures, and testing to avoid unnecessary therapy for patients who will not respond. In the current health care paradigm, the cost-effectiveness of a diagnostic test is generally not evaluated until after implementation, if at all. Thus PPM introduces a new approach whereby economic models drive diagnostic development.
  
• Clinical management: Following consultation with clinical experts, PPM will construct decision trees to outline current treatment management. A decision tree will enable PPM to identify steps in disease management that would benefit from improved diagnostics. The value of a new diagnostic will lie in its ability to better facilitate clinical decisions and prompt and appropriate intervention to improve patient outcomes. Based on models utilizing clinical, epidemiologic and economic data, the performance criteria needed to both improve outcomes and reduce costs will be decided by all partners, including insurers, providers, clinicians, and scientists.
  
• Biomarker discovery: Research clinicians in the health care system will identify appropriate patients, obtain tissue or blood samples and record clinical outcomes. PPM will use these samples to identify hundreds of biomarkers that distinguish diseased individuals from healthy individuals. An iterative process between clinicians, patients and PPM will locate biomarkers that are sensitive and specific for the desired point of disease intervention. Markers that meet agreed upon performance criteria will move forward in the development pathway into clinical testing.
  
• Implementation: After pre-specified performance criteria have been demonstrated by prospective analysis of patient and economic outcomes, the new test will be introduced into clinical care. The insurer will then reimburse for the test. Patient outcomes will continue to be tracked, providing opportunities to further enhance test performance.

WHY NOW?  THE SCIENCE OF MOLECULAR DIAGNOSTICS

With the completion of the human genome came great expectations for personalized medicine. It was thought that discovery of genetic variations that confer significant risk for major diseases would permit the widespread adoption of preventive measures and focused screening for early disease detection. The promise has not materialized. In fact, except for rare mutations, most common genetic variations associated with prevalent diseases confer very small risk for disease. Transcriptomics, the analysis of the activity of genes in different tissues, has shown improved diagnostic capability but is complex, and clinical correlations have been difficult to reproduce. Both genomics and transcriptomics will continue to inform medical science and occasionally provide useful clinical information, but their ultimate role in personalized medicine remains uncertain.

The Promise of Proteins

Recent advances in proteomics and improvements in mass spectrometry now make it possible to identify and quantify proteins at previously undetectable levels. This opens new opportunities for the development and application of protein biomarkers across a broad range of disease areas. It is these advances that lead us to believe that dramatic new opportunities in molecular diagnostics are at hand. Proteins are more informative than DNA or RNA as diagnostics DNA mRNA protein and can be applied to a broader spectrum of diseases for a number of reasons:

• DNA reveals only hereditary predisposition, whereas proteins change dynamically in response to physiological conditions and can reveal disease onset and progression as well as lifestyle and environmental risk exposures.
• A single gene can produce a family of 10 to 100 variant proteins. This variation adds to the amount of information available from the spectrum of proteins.
• Proteins from diseased tissue are found in the bloodstream, whereas DNA and RNA are generally obtained by biopsy of the disease tissue itself. Therefore, clinicians can measure protein biomarkers by a simple blood test that is much less invasive than tissue biopsy.

Protein biomarkers will also be useful in the further development of medical imaging tools such as X-rays, magnetic resonance imaging (MRI), ultrasound, and positron emission topography (PET). Combining protein biomarkers with imaging technology will enable the precise identification of disease activity within the body. However, imaging tests are expensive. Therefore, using less costly blood-based protein diagnostics as an initial step to identify which patients require imaging tests will also contribute to the reduction of health care costs.

Technology and Proteomics Production Facility

To effectively facilitate the development of diagnostic tests, PPM draws upon the strengths of two of Arizona’s leading bioscience entities, the Biodesign Institute at Arizona State University and the Translational Genomics Research Institute (TGen). PPM will integrate the shared expertise of these entities in proteomics, biomarker discovery, cell biology, and bioinformatics, with each bringing specific capabilities and facility resources to the collaboration. An industrial-scale, high-throughput proteomics facility will be uniquely positioned to serve as a hub for biomarker discovery. PPM will employ state-of-the-art technology platforms and research in supercomputing, nanotechnology and health economics, as well as genomics, transcriptomics and tissue sampling.

Technology Sharing

The initial discovery and development work for demonstration projects will take place at the Biodesign Institute, TGen, Fred Hutchinson Cancer Research Center in Seattle and other collaborating institutions. However, technological innovations and knowledge may also be transferred to partners. Should a partner wish to establish a facility in their own country, PPM would provide support for such a venture through training and advice. This arrangement will allow partners to leverage the initial project into new disease areas, with the potential for further improvements in health care and cost savings.

THE BENEFITS OF PARTNERSHIP

Partnering with PPM will offer a number of valuable benefits and opportunities:

• PPM partners will be participants in applying health care economics in their solutions, combined with the application of information technologies to track patient outcomes that are correlated with molecular diagnoses, will be integrated into the cycle of creativity.
• PPM partners will be collaborators in the effective use of genomics and proteomics to identify those at risk for disease, detect the presence of early-stage disease, match the needs of individual patients to effective therapy, and monitor for disease recurrence.
• PPM partners will join an expanding network of health care systems and laboratories dedicated to transforming the practice of medicine through the application of molecular knowledge to patient care. Partnership will provide a unique opportunity for learning, innovation and solution sharing.
• PPM partners will be able to establish their own diagnostic technology centers, enabling them to stay at the forefront of the health care revolution. PPM is dedicated to helping its partners establish their own technology capability by providing advice, best practices and training.
• As knowledge and improved methodologies become available to medicine, there will be an increasing need for governments, insurers and health care providers to develop robust policy for implementing change. Through the auspices of the Pacific Health Summit (www.pacifichealthsummit.org), PPM partners will have a forum for ongoing policy development.

Return to Table of Contents

University of Utah / Intermountain Healthcare

Salt Lake City, Utah

Pioneering a Shared Genetic Legacy

For more than six decades, Utah researchers, institutions and citizens have collaborated to extend the boundaries of genetic understanding and its application in clinical practice. The fruits yielded by this collaboration have included the creation of the first artificial limb (the Utah Arm), the first successful artificial heart transplant, the seeds of the Human Genome project, one of the world’s largest and most comprehensive population and health databases, the first real-time cardiovascular monitoring system and, most recently, the 2007 Nobel Prize in Physiology or Medicine. 

Currently, two leading Utah organizations – its flagship institution of higher learning, the University of Utah; and its largest indigenous healthcare system, Intermountain Healthcare – are building on the state’s storied genetics legacy. This Community Report explores the progress of key initiatives at each of these pioneering institutions. 

The University of Utah: Moving FURTHeR Toward Personalized Healthcare

With its long-standing strengths in genetics, informatics and model systems, the University of Utah (U of U) brings unique assets to the personalized healthcare enterprise. Credit for this belongs in large part to the Utah citizens, leaders and community partners who have supported genetic research for more than 60 years, when U of U physicians first recognized the power of the state’s large families and meticulous genealogical records in studying inherited traits. In 1946, the U of U was awarded the first-ever extramural research grant by what is now the National Institutes of Health (NIH), for “The Study of Metabolic and Hereditary Disorders” in Utah families. Today, Utahns’ continued support translates into a greater-than-95-percent participation rate in biomedical research studies – the nation’s highest.

One of Utah’s leading contributions to modern genetics is the Utah Population Database (UPDB). Established in the mid-1970s through a collaboration between the U of U, the State of Utah and the Church of Jesus Christ of Latter-day Saints (LDS or Mormon Church), the UPDB began as a computerized genealogy file containing demographic and kinship information for more than one million individuals in about 170,000 Utah families. Today’s UPDB continues to grow, with data for approximately six million individuals in families that span up to seven generations. Individual records are now linked to other data sets, including Utah birth and death certificates, electronic medical records from the state’s major healthcare provider networks, all Utah hospital discharge summaries, and cancer registries from Utah and Idaho.

Analysis of UPDB data led to some of the earliest descriptions of the heritable component of cancer and the identification of major cancer predisposition genes, including BRCA1, BRCA2, p16 and HPC2.^{1, 2} Current areas of research involving the UPDB include high-risk pedigree studies of melanoma, prostate cancer, breast cancer, colon cancer, major depression, asthma, influenza, intracranial aneurysms, pelvic organ prolapse and incontinence syndromes, autism and chronic fatigue syndrome.

Concurrent with UPDB work in the 1980s and 1990s, U of U researchers Drs. Mark Skolnick, Ray White, Mark Leppert and Jean-Marc Lalouel were leading efforts to develop molecular markers to map inherited diseases in Utah pedigrees. Within a few years, the power of chromosome mapping methods was demonstrated with the localization of the genes for conditions such as retinoblastoma, cystic fibrosis, neurofibromatosis type 1, several forms of neonatal epilepsy, the inherited colon cancer syndrome and the familial adenomatous polyposis coli syndrome. U of U research into diseases caused by inherited traits continues to the present, with the identification of genes responsible for a variety of neurodegenerative disorders, cardiac ion channel disorders and developmental disorders.

In 1984, Utah played host to a now-legendary event known as “The Alta Summit,” which took place during a blizzard at the Alta ski resort in Utah’s Wasatch Mountains. Sponsored by the U.S. Department of Energy and the International Commission for Protection Against Environmental Mutagens and Carcinogens, the gathering of scientists, including U of U professors Drs. Ray Gesteland and Ray White, focused on measuring heritable mutations in atomic bomb survivors from Hiroshima and Nagasaki. Their conversations generated a flood of ideas and plans that ultimately influenced the design of large-scale genome mapping and sequencing projects – and are now widely credited as the genesis of the Human Genome Project.³

Complementing the U of U’s strength in human genetics res­earch are its resources for animal models of human development and disease, including the products of Dr. Mario Capecchi’s pioneering work in mouse gene targeting. This method allows scientists to alter any gene of interest in a single mouse, and from that mouse produce a lineage of animals that pass the mutation – and its effects – from generation to generation. Using these lineages, scientists can assess the effects of disease-causing mutations, determine biological mechanisms and test new therapies. It is fair to say that mouse gene targeting technology changed the way the world does biomedical research. This work earned Dr. Capecchi the 2007 Nobel Prize in Physiology or Medicine and provided the foundation for a host of U of U research programs that use genetically malleable model systems such as nematodes, planaria, zebrafish, fruit flies and newts.

The U of U’s legacy in genetics and model systems, together with its one-of-a-kind population database, position it to make unique and valuable contributions to the discovery science facet of personalized healthcare. In addition, Utah has led key developments in clinical informatics, an area that is crucial to the delivery of personalized healthcare. In the early 1970s, Salt Lake City-based Intermountain Healthcare, with the guidance of U of U medical informatics professor Dr. Homer Warner, implemented one of the nation’s first electronic medical record (EMR) systems. The U of U also participated with the Veteran’s Affairs (VA) Salt Lake City Health Care System and other VA sites in the development of the award-winning VISTA/CPRS electronic health record software.⁴ Today the U of U’s Department of Biomedical Informatics (DBMI) fosters partnerships across Utah’s major health care systems to refine EMR and clinical decision support systems, develop new tools and demonstrate their impact on medical care and quality of life.

Like our peer institutions, the U of U is reviewing its portfolio of strengths relative to the numerous challenges in personalized healthcare, and determining next steps in building its programs. We recognize that many of these challenges – data management, regulatory, economic, policy, ethical, education and social issues – can be effectively addressed only through close collaboration with local and national partners. The remainder of this report will focus on how the U of U and its partners are addressing a critical challenge upon which many of our future efforts will rely: Developing a collaborative infrastructure for data integration and delivery across diverse research, clinical and community domains.

While traditional informatics allows us to create the clinical or bioinformatics data infrastructure for a single institution, personalized healthcare requires the integration of disparate research, clinical, and community data both within and across institutions. Above and beyond creating a secure yet accessible informatics infrastructure, this requires rethinking the tools used to process data and the training strategies needed for effective adoption. To be truly integrative, the infrastructure should be scalable – ultimately to a national network of biomedical research centers – while incorporating stringent regulatory and policy implementation strategies. Equally critical is the need for adopting strategies that ensure effective social and cultural change.

To address this challenge, the U of U is spearheading a statewide collaborative, named the Federated Utah Research Translational Health e-Repository (FURTHeR), which will provide the informatics infrastructure for integrative, collaborative, and transformative research. FURTHeR will link genotypic, phenotypic, genealogic, clinical, environmental and public health data from disparate sources statewide and present them through a Web-based portal to patients, community providers and researchers according to appropriate access policies and regulations. It will integrate the substantial research resources within the U of U as well as data from the State of Utah Department of Health and the state’s three major healthcare delivery networks (University of Utah Healthcare, Intermountain Healthcare and the VA Salt Lake City health care systems).

Covering more than 85 percent of the state’s clinical data and nearly 100 percent of statewide population-based public health data, this remarkable array of phenotypic data will be integrated with the UPDB and enhanced with genotype data as planned biobanks become a reality. The result will be an unparalleled platform for translational training, research, and innovation, one with a special strength in assessing multiple factors – genetic, genealogic, environmental and geographic – that contribute to complex diseases. Designed with scalability and extramural connectivity in mind, FURTHeR employs biomedical and industrial standards while enforcing patient and institutional privacy protections.

THe FURTHeR project has begun with available data resources, and it is growing incrementally as additional systems are incorporated. While it is currently focused on providing research data infrastructure, FURTHeR could eventually contribute to processes that inform clinical decision support. In a unique first step toward this end, we are working in concert with the National Cancer Institute (NCI) to use some of the basic Cancer Bioinformatics Grid (caBIG) ⁵ tools and methods in the development of the FURTHeR system.

The repository is “virtual” in the sense that each data source will continue to reside in its home organization, with its own architecture, access policies and procedures. A common road map (meta-model) will describe what is in each system and how to access it. FURTHeR will federate these data using a grid-computing framework^6,7 for resource sharing and a fair-broker data integration service⁸ to ensure that data-access policies are enforced among the partners. In Figure 1, the central node, shown in green, will initially provide three key services:

• Metadata integration service. Using metadata services, FURTHeR will classify and describe data from disparate sources⁹ and deliver a consistent data model to the end user.
• Fair broker data access service. The fair broker service enforces access policies unique to each data source, acting as the final gatekeeper to metadata and data. This is a critical social engineering and regulatory task. At first it will provide security and regulatory compliance during user-initiated data searches. Over time we expect that it will also assist in discovery and ontology management.¹⁰
• De-identified “data sandbox” service. In translational science, researchers, policymakers, administrators and other users often need aggregated views of data across institutions, or they need limited views of de-identified data for individuals (i.e., data conforming to the HIPAA Privacy Rule 45 CFR Parts 160, 162, and 164). The FURTHeR “data sandbox” will provide these views, promoting collaboration through comprehensive, statewide access to data. It also provides a powerful educational resource for a variety of courses and programs.

The strategic plan for FURTHeR is ambitious, but its potential for success is bolstered by numerous existing and planned institutional and cross-institutional activities that are being integrated to realize the overall vision. We have a running start with a 50-year history of informatics in Utah, with generous institutional support, and with an extensive set of collaborative projects that have already laid groundwork for the project. Table 1 contains a list of these projects.

In addition, we capitalize on the laws of the State of Utah, which allow for the aggregation of data from multiple sources (with adequate privacy protections) for the purpose of research to enhance the health of the citizens of the state (Utah Code Sections 26-3-7, 26-5-2, 26-25-2, and 63-2-206). The current linkage of data in the UPDB derives from multiple sources, including the Utah Department of Health, the genealogies of the State’s founders, and the electronic medical records of the U of U Health Care System and Intermountain Healthcare. Few states promote (or even allow) this sort of linkage; Utah both promotes and protects them with the power of the law. The U of U has worked within this system for 30 years and has developed a separate review board, the Resource for Genetic and Epidemiologic Research (RGE), to set and enforce access policies for these data.

The FURTHER project is supported in part by a Clinical and Translational Science Award (CTSA; Public Health Services research grant numbers UL1-RR025764 and C06-RR11234 from the National Center for Research Resources) as well as generous institutional support from the U of U Research Foundation.

Figure 1. FURTHEeR Informatics Architecture Overview

Table 1. Resources That Contribute to FURTHeR

Intermountain Healthcare and the Clinical Genetics Institute

Intermountain Healthcare is a not-for-profit health system based in Salt Lake City. Serving the health care needs of Utah and southeastern Idaho residents, Intermountain’s system includes 21 hospitals, numerous clinics, a health plan and a physician group of more than 500 practitioners. Intermountain has been named the nation's first- or second-place integrated healthcare system for the past nine years, by Modern Healthcare magazine and by the leading healthcare data analytics firm, Verispan. Verispan presents an annual study reporting on its examination of more than 500 health systems around the nation. The annual list rates local and regional healthcare systems on factors such as services and access, technology, hospital utilization and financial stability.

In 2007, the Dartmouth Medical School study found that Medicare spending could be reduced by a third – while maintaining or improving quality – if the nation provided healthcare the way it's provided in the greater Salt Lake City area. The study specifically cited Intermountain Healthcare as an organization that provides high-quality, highly-efficient care. These accolades reflect Intermountain Healthcare’s commitment to best practices that provide a backbone for developing, testing and implementing high-quality, evidence-based personalized medicine with optimal value. 

Electronic Medical Record (EMR) system

In partnership with Dr. Homer Warner and the University of Utah’s Department of Medical Informatics, Intermountain developed one of the nation’s first large EMR systems (see also above in the University of Utah section). The Intermountain electronic data warehouse contains longitudinal medical records on nearly six million individuals, some of which go back more than 30 years. In addition to the electronic storage of information, the informatics system assists physicians with point-of-care clinical decisions for treating diabetes, community-acquired pneumonia, ventilator management, and coronary heart disease. E-resources imbedded in the EMR allow providers access to a wide range of electronic context-specific information sources.

Intermountain Healthcare has been recognized by many outside groups for its innovation in medical informatics.  The American Hospital Association’s magazine, Hospitals & Health Networks, named Intermountain Healthcare in its list of the “Top 100 Most Wired” health care organizations in the country, the Health Information Resource Center awarded Intermountain Healthcare gold, silver, and bronze medals at its 2005 National Health Information Awards and the journal of the American Hospital Association named Intermountain Healthcare one of the nation's most technologically savvy hospital systems for the eighth time in nine years.

Quality improvement

Another critical and unique resource at Intermountain is the Institute for Healthcare Delivery Research, created in 1990 to assist Intermountain in the formal application of quality improvement techniques within the clinical care setting. The approach of Intermountain Healthcare to quality management has been accepted as a national model by the Hospital Research and Education Trust, the research affiliate of the American Hospital Association. Since 1992, the Institute has trained thousands of Intermountain employees and hundreds of physicians and other clinicians from across the United States and around the world in the principles of health care quality improvement.

Intermountain has received top national awards for providing quality health care, including the 1996 NCQHC Quality Health Care Award. U.S. News & World Report ranked Intermountain’s LDS Hospital as one of America’s Best Hospitals in 2006 for orthopedic care, treatment of respiratory disorders, pulmonary medicine, endocrinology or diabetes care, and urology.

Intermountain Healthcare’s Clinical Genetics Institute (CGI)

In 2005, after more than three years of strategic planning and in recognition of the increasingly important role that genetics and genomics will play in the future of medicine, Intermountain launched the CGI, which aims “To promote excellence in the quality and value of healthcare throughout our service area by implementing developments in genetics/genomics within Intermountain Healthcare.” CGI’s values include:

• The belief that genetics/genomics can be a catalyst for a revolution in disease prevention, management and treatment
• The belief that patients and families have the right to understand genetic implications across generations when making health care decisions (personalized medicine)
• The belief that Intermountain can contribute to regional, national, and international development and application of genetics/genomics information into clinical healthcare practices
• The belief that personalized medicine can be realized only through the use of Intermountain Healthcare’s strengths in informatics, clinical decision support, quality improvement and evidence-based model care processes.

The CGI will use this mission and these values to address the rapid and dynamic changes anticipated with the addition of genomic information to the existing repertoire of clinical data used in the practice of medicine. CGI staff members have a wide range of expertise, including clinical genetics, genetic counseling, health care delivery, payer issues, education, technology assessment, economic analysis, quality improvement and informatics.

As the largest integrated health system in the Intermountain West (and one of the largest in the country), Intermountain provides a wealth of opportunities to research the effectiveness and value of genetics and genomics in health and disease.  One of the most active areas of research is the use of genomic tests to guide the use of drugs that target the products of disease-causing genes.

A significant challenge facing personalized medicine in the future is the ability to put genetic and genomic information in the hands of providers in a useable form at the point of care. The CGI staff serve on several national boards and committees including the American College of Medical Genetics Board of Directors; the Secretary’s Advisory Committee for Genetics, Health and Society; the Personalized Health Workgroup of the American Health Information Community (DHHS); Ethics Advisory Group and Billing and Reimbursement Task Force of the National Society of Genetic Counselors; and the Clinical Genomics workgroup of HL7. 

Additionally, the CGI has worked closely with the CDC’s EGAPP (Evaluation of Genomic Applications in Practice and Prevention) and CETT (Collaboration, Evaluation and Test Translation) programs in order to promote evidence based evaluation of emerging genetic and genomic tests.  Dr. Williams is actively promoting incorporation of formal quality improvement methods into clinical genetics by founding the Quality Special Interest Group of the American College of Medical Genetics and spearheading a national quality project through the Mountain States Genetics Regional Collaborative Center (funded by the Health Resources Services Administration-HRSA).

Current CGI Projects:

Family History

With grant support from HRSA through the Genetic Alliance the CGI is exploring ways to increase the use and utility of the family history through integration into the EHR.  Specific projects include development of a patient-entered family history form in the patient portal (a web based interface into Intermountain’s EMR designed for use by patients).  Once created this tool will be studied to see how information could be sent to clinicians as well as providing information for clinical decision support and patient health messages.  In 2008 Intermountain Healthcare received a Microsoft HealthVault award to develop a tool for the HealthVault personal health record and explore how the information collected in that tool could be transmitted to EMRs.  The CGI has also contracted with the Department of Health and Human Services to provide technical expertise to the DHHS, the Veterans Administration, the Department of Defense and the Indian Health Service to develop a standardized family history collection tool.

Genetic Care Delivery

Comprehensive genetic cancer services are being developed throughout the Intermountain system, and links are being established with similar services provided at the Huntsman Cancer Institute.  The CGI is currently partnering with oncology, surgery and pathology to implement a system-wide program to identify patients and family members with Lynch syndrome and its associated increased risk for colorectal and other cancers.  

Genetics Resources in the EMR

In conjunction with the University of Utah Department of Biomedical Informatics and Intermountain informatics (see also above in the University of Utah section), genetic information resources (both general and disorder specific) are being integrated into the E-resources of the EMR and linked to infobuttons in the patient problem list that are capable of providing point of care “just-in-time” education for several hundred genetic conditions.  As genomic tests emerge into practice, educational resources to support appropriate use and interpretation of these tests will be deployed.

The Secretary of DHHS has championed an interoperable EMR in the United States by 2014.  The CGI in conjunction with Intermountain Healthcare’s informatics department is working with Harvard Partners Center for Genetics and Genomics to pilot transmission of genetic test results between a laboratory information system and EMR using the HL7 Genetic Variation messaging model, which is being offered as an international standard.  We hope to begin working with ARUP a Utah-based national reference laboratory to implement this on a larger scale in the near future.

Emerging Genetic Test Evaluation

CGI has initiated a multidisciplinary Genetic Testing Practice Council to evaluate new genetic tests for clinical utility and to develop methods to support appropriate use of new and existing tests found to have clinical validity and utility, while erecting barriers to discourage the use of tests without an adequate evidence base.

A unique aspect of the CGI is the presence of a full-time analyst to address issues of new technology assessment and cost effectiveness. Partnering with other groups such as the University of Washington School of Public Health Genomics’ pharmacoeconomic faculty, the Institute for Preventive Medicine, CDC, and the Economics of Genetics Technologies Seminars (organized by the Health Economics Research Centre [University of Oxford] and the North West Genetics Knowledge Park [University of Manchester]), the CGI is modifying and applying standard assessment tools within an integrated health care system.

Ultimately, Intermountain’s extensive experience in implementing evidence-based medicine in its hospitals and clinics will serve as a foundation in Utah for practicing genomic medicine in all levels of clinical practice. The CGI published a method for rapid assessment of genetic tests, a technique that has been taken up by a national specialty society for an evidence review to support development of a clinical guideline. In addition a paper on the economic implications of pharmacogenomics testing for Warfarin dosing presented at the International Society of Pharmaceutical Outcomes Research meeting in 2008 is currently under review for publication. 

These endeavors identify the CGI at Intermountain Healthcare as a national and international leader in evaluation and translation of genetic and genomic medicine which should bring the vision of personalized medicine to reality in clinical practice.

With committed support from the State of Utah – which has identified personalized healthcare as a core component of its Life Science Industry Clusters economic development program – the rigorous, groundbreaking research being performed at the University of Utah and the intensive, innovative clinical work being done at Intermountain Health Care will continue to redefine the parameters and reach of personalized health care far beyond the borders of Utah.

REFERENCES

Slattery ML, Kerber RA. A comprehensive evaluation of family history and breast cancer risk. The Utah Population Database. JAMA (1993) Oct 6;270(13):1563-8.

Kerber RA, O'Brien E. A cohort study of cancer risk in relation to family histories of cancer in the Utah population database. Cancer. (2005) May 1;103(9):1906-15.

Cook-Deegan RM. The Alta Summit, December 1984. Genomics 5 (1989) 661-663.

VISTA/CPRS electronic health record software: http://www1.va.gov/cprsdemo.

National Cancer Institute Cancer Bioinformatics Grid: https://cabig.nci.nih.gov.

Gannon D, Plale d, Christie M, et al. Service Oriented Architectures for Science Gateways on Grid Systems. https://cabig.nci.nih.gov/guidelines_documentation/caGRIDWhitepaper.pdf 17. Referenced on 09/10/2007.

Saltz J, Oster S, Hastings S, et al. caGrid: design and implementation of the core architecture of the cancer biomedical informatics grid. Bioinformatics. (2006) Aug 1;22(15):1910-6. Epub 2006 Jun 9.

Budgen D, Turner M, Kotsiopoulos I, et al. Managing healthcare information: the role of the broker. Studies Health Technol Inform. 2005;112:3-16.

Brandt CA, Gadagkar R, Rodriguez C, Nadkarni PM. Managing complex change in clinical study metadata. J Am Med Inform Assoc. (2004) Sep-Oct;11(5):380-91.

Budgen D, Rigby M, Brereton P, Turner M. A Data Integration Broker for Healthcare Systems. Computer. (2207) April; 40(4). p. 34-41.

CONTRIBUTORS TO THIS REPORT

Richard L. Bradshaw, MS                    Knowledge Management Team, Intermountain Healthcare

Julio C. Facelli, PhD                    & #160;        Biomedical Informatics, The University of Utah

Raymond F. Gesteland, PhD                 Human Genetics and Program in Genomic                      0;                    &# 160;                                   0;   Medicine, The University of Utah

John F. Hurdle, MD, PhD                     Biomedical Informatics, The University of Utah

Bernard A. LaSalle                    60;           General Clinical Research Center, The University of Utah

Jennifer Logan, PhD                    & #160;        Program in Genomic Medicine, The University of Utah

Susan A. Matney, MSN                    & #160;  Biomedical Informatics, The University of Utah

Geraldine P. Mineau, PhD                    Department of Oncological Sciences and Huntsman Cancer Institute The University of Utah

Joyce Mitchell, PhD                    & #160;         Biomedical Informatics and Health Sciences Information Technology, The University of Utah

Scott P. Narus, PhD                    & #160;        Biomedical Informatics, The University of Utah

Roberto A. Rocha, MD, PhD               Knowledge Management Team, Intermountain                      0;    Healthcare Biomedical Informatics, The University of Utah

Kimball Thomson                    60;             Co-Chair, National Summit on Personalized Health Care

Brent Wallace, MD                      0;        Chief Medical Officer, Intermountain Healthcare

Marc Williams, MD                    &# 160;         Director, Clinical Genetics Institute, Intermountain Healthcare

Return to Table of Contents

Vanderbilt University

Nashville, Tennessee

BioVU: The Vanderbilt DNA Repository

Dan M. Roden, M.D.,
Professor of Medicine and Pharmacology,
Assistant Vice-Chancellor for Personalized Medicine
Vanderbilt University School of Medicine, Nashville

 The Office of Research at Vanderbilt University Medical Center undertook a strategic planning exercise in 2005, to define priority areas for allocation and investment of resources.  The strategic plan identifies three priority areas: therapeutic discovery and translation; public health and healthcare; and personalized health and healthcare. The fundamental premise driving the personalized health and healthcare initiative is that we have reached a point in contemporary biology that a strategy of “one size fits all” fails to recognize individual variability in susceptibility to disease, expression of disease, and beneficial and adverse effects in response to therapies.  A major thrust of the program is discovery, with the long term goal of developing platforms for evaluation, delivery, and mining of high dimensional information (such as genomics and proteomics) to demonstrably improve delivery of healthcare (http://www.vanderbilt.edu/oor/about/strategic-plan12-20 05.pdf). One major component in implementation of this vision has been the creation of a DNA repository, whose development and implementation is described here.

History

The development of priorities within our strategic plan in general, and in the personalized health and healthcare initiative in particular, leverages a range of scientific disciplines in which Vanderbilt has had longstanding investments and recognized national and international expertise.  Vanderbilt’s Division of Clinical Pharmacology, established in 1963, is the largest and most successful division of its kind.  Science in the division, which was supported by over $20 million in direct funding from NIH in the last fiscal year, focuses on the very broad question of mechanisms underlying individual variability in response to drug therapy in human subjects.  Clinical Pharmacology is administratively and philosophically a Division of both the Departments of Medicine and of Pharmacology.  Pharmacology at Vanderbilt includes not only strengths in clinical pharmacology, but also in basic pharmacology and in pharmacoepidemiology.  The Division is home to the Vanderbilt’s site in the NIH’s Pharmacogenetics Research Network (PGRN). 

A second key resource that leverages our efforts in personalized medicine is a two decade-long investment in bioinformatics and information technology. The Vanderbilt Electronic Health Record (EHR), StarPanel, includes data on 1.7 million subjects over the last 10 years and the clinical environment is near-paperless.  StarPanel and the associated electronic order capabilities^1,2 are document-centered and all portions of the system are readily searchable for research and quality control purposes.  The EHR also includes extensive point of care ordering capabilities, which have now been licensed to and are being co-developed with McKesson as the Horizon Expert Order (HEO) system; HEO includes delivery of warnings that flag serious drug interactions, or potential dosage errors. Information technology support is provided by faculty and staff in the Bioinformatics Center and in the Department of Biomedical Informatics, which currently includes 57 faculty.  In addition to service components, faculty have extensive research activities in areas such as processing large datasets and natural language processing.

A strategic planning effort in the late 1990s identified genetics as a priority area of investment at Vanderbilt.  Robust research groups in genetics are now housed in the interdisciplinary Center for Human Genetics Research, as well as in divisions of the Departments of Medicine and Pediatrics.  Additional capabilities are key to executing our vision of Personalized Medicine and are in place: these include current and next generation sequencing, extensive core DNA storage and genotyping capacity, and informatics and analysis support. 

Current Efforts 

Specific programs: Individual investigators and groups have extensive and well-funded research programs directly relevant to personalized medicine areas such as HIV pharmacogenetics (Vanderbilt serves as the DNA repository for the AIDS Clinical Trial Group); the Ayers Institute, a philanthropically-supported effort to detect early circulating biomarkers in colon cancer; SPOREs in breast cancer, GI cancer, and lung cancer; and the Pharmacogenetics of Arrhythmia Therapy Program, the Vanderbilt node of the PGRN.  These efforts are supported by an extensive series of advanced core resources, notably the DNA storage and analysis capabilities described above as well as one of the largest mass spectrometry centers in the world, with a particular focus on proteomics and on molecular profiling for target discovery.  Parallel investments in the other arms of the strategic plan, such as a healthcare economics initiative in the outcomes sector and high throughput screening for new therapeutic in the drug discovery sector complement these capabilities.  Thus, the environment at Vanderbilt University Medical Center has been nurtured over decades to position the Medical Center to assume a leadership position in the area of personalized healthcare. 

BioVU: In 2004, institutional leadership committed to development of a DNA repository with the twin goals of accelerating biologic discovery as well as development and validation of methodologies to evaluate and deliver “omic” discovery to the bedside.  The model we developed, now termed BioVU, couples extraction of DNA from discarded blood samples with a de-identified “mirror” image of the Electronic Health Record (StarPanel and associated resources).  This mirror image, termed the “synthetic derivative” (SD), in essence allows the discarded samples to be used in a fashion designated as “non-human subjects” by the federal Office for Human Research Protections (OHRP) guideline of August 2004.  Implementation of this unique design required extensive preparatory work, including focus groups and community consultation, and evaluation by the IRB, multiple ethics boards, the Institution’s legal department, and OHRP.  The research is designated “non-human subjects”, although the IRB felt that, given the project’s unique scope and nature, continuing oversight was desirable. A key enabling step for the resource was a change in the “consent to treat” form that patients sign every year: this now includes a prominently positioned box, in bold, that allows the patient to “opt out” of DNA collection.  Only samples associated with a signed consent to treat form with an empty opt out box are included in the resource.  Details describing operation of the resource have been published.³

Challenges, Plans, Patient Impact

Advantages and disadvantages of an opt-out approach: There are major advantages to this method of sample accrual.  First, samples are acquired from individuals across the healthcare system and not selected for as in a clinical trial. Indeed, it is widely recognized that results from clinical trials may or may not be translatable to practice, given that some sets of patients, such as those with complicated medical histories (especially the elderly), are often not studied.  Further, multiple phenotypes are represented. Second, the resource has tremendous advantages of scale: sample accrual currently proceeds at 500-1,000 samples per week: sample accrual began in spring 2007 and the resource held 45,900 samples as of Aug. 25, 2008, making it the largest DNA repository in the country.  Third, in order to execute this design, the requirement for de-identification mandated an investment in the broad area of data privacy and security. In addition, the de-identification effort reduces re-identification potential, an increasing concern.

There are some disadvantages to this approach.  Because the individuals are de-identified, recontact is not possible.  Thus, any need for further information, such as environmental exposure or extensive family history, must be sought through other data collections. Focus groups suggested that the opt out rate would be 3-5% and this, indeed, has been our experience.  This resource’s unusual design required an extensive planning and implementation effort, as described above, and also including a number of important milestones, such as development and validation of sample handling and de-identification algorithms. 

Current work: A fundamental question that we are now addressing is whether healthcare information useful for research can, in fact, be extracted from an EHR. One validation study included genotyping the first 10,000 samples accrued into the resource at several dozen SNP sites, identified and validated in recent genomewide association studies as modulating susceptibility to common diseases.  The major challenge here has been development of natural language processing methods to identify cases and controls.  An initial evaluation of two dozen SNPs associated with Type II diabetes, atrial fibrillation, rheumatoid arthritis, Crohn’s disease, multiple sclerosis, demonstrates that for each disease, at least one previously validated SNP was replicated in our dataset.⁴ This is an extremely important milestone for the resource, since it strongly supports the notion that useful information can be extracted from such “real world” resources.  The National Human Genome Research Institute has launched an initiative to evaluate the utility of EHRs associated with DNA repositories, and Vanderbilt is one of five sites participating in this “eMERGE” Network; as well, Vanderbilt acts as the administrative coordinating center for the Network. The five sites will identify 15,000-18,000 subjects with extensive electronic health records and phenotypes of interest (or controls) for genomewide association.  This will not only propel the field forward, but will also provide a very rich dataset on which to explore genotype-phenotype associations within an EHR context (https://www.mc.vanderbilt.edu/victr/dcc/projects/acc/index.php/About).

Plans: The next step for BioVU is development of capabilities to make the resource available to the Vanderbilt investigative community. This involves the development of web-based interrogation tools, and roll out of data use agreements which include as one provision that all genetic information generated will be redeposited into the Synthetic Derivative.  Ultimately, when sufficiently large numbers of patients have been genotyped at large numbers of sites, it may be possible to examine genotype-phenotype associations in silico without the need for further genotyping.

Challenges: The long term goal of the BioVU project is to develop methods to identify, validate, and then implement on a clinical level new high dimensional information.  Our current vision suggests that point of care delivery systems such as next-generation HEO will include delivery of  increasingly patient-specific warnings and prescribing advice. There are multiple challenges that will have to be overcome for such a future tense vision is executed: What technology to use?  How to evaluate the added benefit of integration of patient specific information into healthcare? At what costs?  What sort of effect would be deemed “real”? We recognize these and multiple other challenges in the implementation of a next generation personalized healthcare delivery system, and consider BioVU as a key enabling step for studies to address these challenges. 

References

1. Stead WW, Borden R, Bourne J, Giuse D, Giuse N, Harris TR, Miller RA, Olsen AJ. The Vanderbilt University fast track to IAIMS: transition from planning to implementation. J Am Med Inform Assoc 1996;3:308-317.
2. Stead WW. Rethinking Electronic Health Records to Better Achieve Quality and Safety Goals. Annual Review of Medicine 2007;58:35-47.
3. Roden DM, Pulley JM, Basford MA, Bernard GR, Clayton EW, Balser JR, Masys DR. Development of a large-scale de-identified DNA biobank to enable personalized medicine. Clin Pharmacol Ther 2008;84:362-369.
4. Crawford DC, Ritchie MD, Denny JC, Havens A, Weiner J, Pulley J, Basford M, Masys DR, Roden DM, Haines JL. Electronic Medical Records Linked to DNA: A Valuable Resource for Large-Scale Genetic Association Studies. American Society for Human Genetics, accepted for presentation 2008.

Return to Table of Contents