The oral-systemic personalized medicine model at Marshfield Clinic.

Unique resources have aligned Marshfield Clinic to advance dental personalized medicine and set the stage for ushering in its integration at the clinical level. The key elements already in place that have positioned Marshfield Clinic to advance realization of dental personalized medicine include 1) the Personalized Medicine Research Project (PMRP) and its bio-repository; 2) Marshfield Clinic’s combined dental and medical clinic infrastructure; 3) an integrated medical-dental electronic health record (iEHR) supported by substantial bioinformatics expertise; and 4) a research infrastructure with a dual focus on genetics and metagenomics. How this infrastructure has been integrated to advance dental personalized medicine at Marshfield Clinic is the focus of the second part of this review. It is essential to provide an overview of one of Marshfield Clinic’s oldest and most valuable resources, its EHR, as well as characterising the central role of the EHR in realizing personalized medicine. Foremost, EHRs provide multiple benefits relative to tracking clinical care, patient outcomes, and maximizing efficiency. Further, the EHR is an invaluable resource for research aimed at assessing and improving patient care. In its landmark report, The Computer-Based Patient Record (CPR): An Essential Technology for Health Care ( Dick et al, 1997 ), the Institute of Medicine (IOM) eloquently formulated a vision for EHRs. The benefits of computer-based records (CPRs) articulated by the IOM include longitudinal and comprehensive patient documentation, improved data quality and validation, built-in decision support functions, universal and secure access, and availability. Since 1997, several other IOM reports have addressed how HIT can reduce medical error ( Kohn et al , 2000 ; Committee on Quality of Health Care in America, 2001 ) and increase quality in healthcare ( Thompson and Brailer, 2004 ). These benefits of CPRs provide the central rationale supporting the development of the National Health Information Infrastructure ( Osheroff et al , 2007 ).

While the more comprehensive understanding of infectious/inflammatory processes contributing to PD and DM has been facilitated in part by advances in biotechnology, rapid advances in informatics capabilities are also supporting acceleration of translational personalized medicine from bench to bedside. Many of these advances are being driven by an urgent need to manage the massive amounts of data being generated as a consequence of advances in genomic and metagenomic analysis that potentiate high throughput accumulation of both human and microbial data that must then be stored, organized, sorted, analysed, and interpreted in the context of other available data in order to decipher its significance. For example, the oral metagenome, currently under study as a component of the Human Microbiome Project ( NIH HMP Working Group et al , 2009 ) will generate vast amounts of genetic data whose interpretation will be aided by availability of the Human Oral Microbiome Database and, as a reference tool, its auxiliary website ( http://www.homd.org ) that has catalogued all microbial species presently found in the mouth ( Chen et al , 2010 ). Availability of these resources and analytical toolboxes to analyse their content in order to create accurate microbial phenotypes relative to pathogenic capacity in the context of microbiomes is critical to advancing applicability to personalized medicine. Similar advances in genomic sequencing capabilities are making sequencing technology less cost prohibitive and more accessible. Such efforts will undoubtedly provide increased understanding of the host genomic contribution to health and disease, especially in the context of shifts in profiles of constituent microbiomes of the systemic metagenome in the context of disease. However, these data will also need to be systematically catalogued in searchable, accessible databases for future analyses in the context of health and a myriad of disease entities. Central to the systematic evaluation of parallel data and distillation of its contribution to the global understanding of health maintenance and aberrations contributing to disease pathogenesis is the development and implementation of validated, standardized phenotypic definitions of the various host and microbial characteristics that contribute to oral (and systemic) health and disease. A phenotype is an observable trait that may define, for example, an organism’s appearance, behaviour, or signs and symptoms of disease among other features. Phenotypes also extend to referencing observable traits or attributes at a molecular level. Lack of uniformity in the clinical phenotypic definition of PD has created significant challenges in comparing and interpreting epidemiological data across studies ( Borrell and Papapanou, 2005 ). Advancing accurate clinical phenotyping capacity also requires advanced biomedical informatics capabilities to sift and analyse clinical and demographic data that will support accurate definition of attributes that allow affected individuals to be systematically distinguished and compared to unaffected individuals at molecular, clinical, and demographic levels in the context of environmental exposures that influence each individual. Ideally, to advance biomedical informatics capacity in this regard requires the availability of integrated dental and medical electronic health records that can be systematically interrogated to create phenotypes. Characterization of the host dental phenotype is facilitated by availability of an electronic dental record (EDR). There has been a slow increase in the use of computers to store not only administrative data but also clinical information at dental practices across the nation. An American Dental Association (ADA) survey in 2007 (American Dental Association Survey, 2007) showed that approximately 55% of all practitioners maintain their clinical information using an electronic (computer) format. A more recent survey of California dentists showed that about 23% of its dentists report having fully implemented an EDR in their practices ( California Healthcare Foundation 2010 ). The survey also reported that another one-third is either in the process of or planning to do so in the coming years. More recently there has been a strong push from the federal government for the adoption of the electronic health record (EHR) within all branches of healthcare. This will most likely support the adoption of EDR amongst dental practitioners who treat Medicaid patients. There has been a wide-spread adoption of EDRs in majority of the US dental schools. Additionally, US dental schools, as well as dental schools in Canada and Europe, are either using or adopting some aspects of a common EDR framework ( White et al , 2011 ). The adoption of EDR in dental schools and private practices throughout the nation is resulting in a wide array of clinical data being stored in the individual repositories (databases). This creates a potential opportunity to utilize these clinical data not only to support clinical practice but also to conduct cutting-edge research.

In 2000, Surgeon General David Satcher published a report entitled “Oral Health in America” drawing attention to the persistent ‘silent epidemic’ represented by the largely under-recognized national oral health crisis that has persisted for decades ( U.S. Department of Health and Human Services, 2000 ). Due to lack of a systematic definition and under-diagnosis of conditions affecting the gingiva, estimation of prevalence has been somewhat difficult to characterize ( Borrell and Papapanou, 2005 ; Page and Eke, 2007 ; Savage et al , 2009 ). However, existing attempts at epidemiological prevalence projections support Satcher’s characterization, with estimates of 75% of the US population affected by gingivitis, 35% of adults affected by periodontal disease (PD), and approximately 13% afflicted by severe periodontitis, with persons of low socioeconomic status disproportionately affected ( Albandar et al , 1999 ). Emerging evidence that PD impacts exacerbation of other pathological conditions systemically has heightened concern over the high prevalence of oral disease. Of highest concern was mounting evidence demonstrating that diabetes mellitus (DM), another escalating epidemic, and PD represented reciprocal risk and exacerbation factors for each other, presumably mediated by chronic inflammatory pathophysiological mechanisms ( Mealey and Oates, 2006 ). However, a critically important finding of some studies was that, with appropriate oral prophylaxis and intervention, exacerbation of both conditions could be dramatically curtailed. ( Khader et al , 2006 ; Darré et al , 2008 ; Janket et al , 2005 ; Simpson et al , 2010 ; Teeuw et al , 2010 ; Lalla and Papapanou, 2011 ). Shortly after the Surgeon General’s report was released ( U.S. Department of Health and Human Services, 2000 ), Harvey Schenkein (2002) prepared a prescient review predicting the advent a new era of personalized medicine relative to oral health and disease as a result of expanded understanding emanating from applied genomics studies. He proposed a scenario wherein personalized medicine approaches could be harnessed to: 1) facilitate identification and analysis of etiological factors contributing to oral disease; 2) permit assessment of environmental impact on oral health and disease; 3) evaluate host risk from a clinical and genetic perspective; 5) predict response to treatment; and 4) support development of diagnostic, prognostic and therapeutic approaches for application on a personalized basis. The realization of a growing dental crisis that has disproportionately affected low income and under- or non-insured populations, especially those populations residing in dental professional shortage areas, has been a strong conduit for advancement of personalized medicine at Marshfield Clinic. To put this in perspective, in 2002, the Marshfield Clinic’s Family Health Center (FHC) in Wisconsin found itself in the midst of a dental care crisis. Oral health care access had become a highly significant problem for many Wisconsin residents, with only 22.6% of Medicaid/BadgerCare recipients able to receive dental services due to a continuing decline in participation by most dentists in the Medicaid/BadgerCare insurance program. Of further concern was the prediction that continued attrition was expected over the next 10 years due to an aging dental workforce entering retirement. These statistics, coupled with an anticipated steady increase in demand attributable to expanding numbers of eligible individuals over time and the potential impact on their quality of life in a climate of escalating of healthcare cost, projected that future access to dental care would become increasingly problematic, necessitating urgent, immediate, proactive measures. Recognizing the impact of oral disease on both quality of life and systemic health, author (GN), Executive Director of the Family Health Center (FHC) of Marshfield Clinic in Wisconsin, a designated federally qualified health centre (FQHC), undertook an unprecedented grassroots effort, in collaboration with Wisconsin communities, to set up the first 8 of 15 projected dental clinics across their combined, largely rural, service area. Currently, these 8 dental centres are serving residents from all 72 counties in Wisconsin, with one additional site scheduled to open in 2012. Between November 2002 and January 2011, 58,718 unique patients were treated at these dental centres ( Figure 1 ), 85% of which were under 200% of the poverty level. Long-term plans call for establishment of dental facilities to support approximately 100 to 120 full-time equivalent (FTE) dentists and a slightly higher number of hygienists. To maintain a steady workforce to support the new dental clinics, a new dental school is being created as a pipeline to support provision of dental care in designated regional dental professional shortage areas. In addition to making huge inroads into providing urgently needed access to state-of-the-art dental care for its population, establishing these dental clinics has coincidentally created a natural laboratory in which to study the impact of improved dental care on a variety of health outcomes in low income populations with high prevalence rates of PD and systemic conditions such as DM. By aligning many unique resources at Marshfield Clinic, including its advanced clinical informatics and health information technology (HIT) as demonstrated by its interrogatable, highly sophisticated, dental-medical electronic health record, an established Personalized Medicine Research Project (PMRP) featuring a population-based biobank of >20,000 DNA samples, and metagenomics research capability, Marshfield Clinic finds itself positioned to accelerate realization of the personalized approach to oral health care envisioned by Schenkein (2002) . Evidence supporting the inter-relationship of PD and DM, two diseases that have achieved epidemic proportions, has been rapidly mounting over the past decade and the strength of the evidence has recently undergone extensive, systematic reviews and meta-analyses that are cited throughout this review. The present review will briefly frame the current understanding relative to the process driving these diseases as it relates to advancing oral-systemic personalized medicine. Further, an overview of Marshfield Clinic’s unique resources will be presented in the context of how these are advancing oral-systemic personalized medicine based on the current conceptual understanding of oral-systemic health and disease. Special emphasis is given to stemming the epidemic tide of PD in the context of DM through advancement of an interdisciplinary approach to disease management and how recognition of this double threat to population health has fuelled efforts to address the expanding dental crisis in Wisconsin.

Global epidemic prevalence of PD and DM and growing evidence supporting bidirectional exacerbation of these conditions on each other present an escalating health crisis requiring immediate intervention. In addition, evidence suggests that PD may be a risk factor for a broad spectrum of other chronic systemic conditions. Notably, cost effective interventions for PD are available, rendering it a modifiable risk factor for a large percentage of the affected population. At Marshfield Clinic, we are pursuing oral-systemic personalized medicine as an interventional approach to the growing problem. Advancement of personalized medicine is possible because we are positioned to align essential resources to advance the definition of phenotype, genotype, and environmental factors contributing to disease and health. These resources include the iEHR, iEDW, PMRP, a state of the art dental clinic infrastructure spanning much of the state and targeting service to high risk populations, and a research focus on oral microbiomes, metagenomics, and genetic and physiological factors regulating host response and interaction with infectious processes. This line of research will advance the understanding of pathogenic mechanisms contributing to PD and other systemic conditions such as DM and contribute further to developing personalized approaches to intervention and prevention. Harnessing these interdisciplinary resources into a multifaceted approach intended to move oral systemic personalized medicine further into the clinical arena will have more collateral benefits. In addition to advancing metagenomic and human genetic research relative to the oral cavity, advancing microbiomics and integrated dental HIT, we are also positioned to engage in clinical, translational, and comparative effectiveness research. Finally, because the dental infrastructural model in which we are advancing personalized medicine was put into place to address regional healthcare disparities, our efforts are ultimately establishing transformational care models that will inform and shape health policy for the future. In conclusion, we are envisioning a shift in paradigm from ‘personalized medicine’ to ‘personalized health’.

The term ‘clinical phenotype’ is broadly defined as a clinical manifestation or characteristics of a patient’s condition, where ‘condition’ may refer to such entities as disease (or lack of disease), physical characteristics such as systolic or diastolic blood pressure, gingival inflammation, tooth color, or definition of other factors that impact phenotypic expression. Phenotypic characterization involving infectious processes occur at two levels: 1) host phenotypic expression and response driven by infectious/inflammatory exposure; and 2) microbial phenotypic characterization of the microbial/inflammatory forces advancing the pathological manifestations. Phenotypic expression may vary among organisms with the same genotype due to the environmental influence. As an integral component of the human body, the microbiome imposes environmental signals on an individual. Dental phenotypes can be identified and defined from multiple sources. Dental radiographs (bite wings and panorex), dental casts, and dental photographs (intra-oral and extra-oral photographs) are some important sources for identifying dental phenotypes. Transforming EDR data into research-appropriate clinical phenotypes could be very beneficial to focusing on the personalized oral health and also translating the results of the research into practice. There has been little exploration focussed on the clinical phenotypes, genotype, and the oral microbial environment of the patients to understand the personal health concept in dental and oral health. Although EDR is a highly valuable source for this purpose, it is not a straight-forward process. EDR data is rarely collected for the purpose of research, and deriving valid phenotypic information from the secondary use of EDR data is complicated. Data collected in the course of routine patient care is not validated to the extent needed to guarantee the presence or absence of a phenotype, making clinical phenotyping an arduous process. Historically, the lack of precise and appropriate phenotyping has contributed to the often varying and non-reproducible results of some genetic and clinical studies due to variability in phenotypic definition ( Borrell and Papapanou, 2005 ). Care must be taken at multiple levels, with rigorous attention paid to data quality, completeness, comparability and recognition and reduction of clinical heterogeneity. These issues are particularly challenging to address when using data gathered with the EHR to identify phenotypic populations. The Marshfield Clinic’s approach involves development of EDR-based algorithms using a combination of structured data extraction, natural language processing, or other methods to identify patients based on clinical features. The development of a phenotype definition requires in-depth clinician review of records identified by the preliminary EDR-based algorithms. Although clinical phenotyping using the EDR is a time-consuming process involving a multi-disciplinary team to define the clinical attributes of a disease or condition, it is essential. As in all research that attempts to identify and quantify relationships between exposures and outcomes, rigorous characterization of study subjects, or phenotyping, is extremely important. Careful phenotyping is both critical to the eventual results discovered by the study and a source of great challenge due to the variety of phenotyping options and approaches that can be employed using the same data. Marshfield Clinic has had extensive experience in extracting phenotypes utilizing the combined resources of the EHR and the PMRP. Details about the PMRP can be found on the project website at http://www.marshfieldclinic.org/pmrp . The target cohort for PMRP is adults aged 18 years and older who reside in the Marshfield Epidemiologic Study Area (MESA). Established in 1991, MESA comprises a 19-Zip code region centered geographically around Marshfield, Wisconsin and a 9-Zip code area in northern Wisconsin (see Figure 1 ) and has a total population of approximately 90,000 residents. The majority of residents within MESA use the Marshfield Clinic for their health care, thus allowing for population-based epidemiologic research within the system. Except for the city of Marshfield (which has a population of 19,000), MESA residents reside rurally or in small towns or villages. The annual in- and out-migration is very low, making it ideal for prospective studies. By sharing a combined EHR with neighbouring hospitals, inpatient diagnoses, outpatient diagnoses and procedures are captured for MESA residents. Many MESA residents are also members of the Marshfield Clinic-sponsored health maintenance organization, Security Health Plan, allowing capture of health events that may occur outside the Marshfield Clinic system of care. Currently >20,000 MESA residents have participated in PMRP (approximately 45% of eligible residents that could be contacted). The demographics and health characteristics of PMRP subjects are similar to adult residents of Wood County in central Wisconsin (98.2% white Caucasian). Thus, the PMRP cohort is ideal for genetic studies, since population stratification is not likely to introduce bias to study results. PMRP was supported with Federal, State and Marshfield Clinic funds, and has a database containing genetic, phenotypic, and environmental information that can be used for large-scale association studies. With over 20,000 participants, the PMRP continues to be one of the largest population-based biobanks in the US. The PMRP cohort has been used in studies that aim to better understand the role of genetics in various diseases including cataracts ( Waudby et al , 2011 ), endometriosis ( McCarty, et al , 2012 ), Alzheimer disease ( Ghebranious et al , 2011 ), fibromyalgia ( Reeser et al , 2011 ), DM ( Wilke et al, 2007 ), congestive heart failure ( Wilke et al , 2009 ), and breast cancer ( Onitilo et al , 2009 ). Other studies within PMRP relative to personalized medicine include the pharmacogenetics of warfarin clearance ( Yazigi et al , 2010 ) and atorvastin efficacy ( Wilke et al , 2008 ). PMRP is one of several biobank assets that are part of the eMERGE consortium. ( McCarty et al , 2011 ). Analysis of genetic contribution requires phenotypic characterization of cases and controls to support investigation of genetic association. Marshfield Clinic has developed extensive expertise in developing and validating electronic algorithms for phenotypic definition of disease. Some approaches to defining phenotypes remain in progress while several have been published previously (Wauby et al , 2011; Conway et al, 2011 ; Wilke et al , 2010 ). To investigate the relationship between oral health and systemic health, Marshfield Clinic is currently recruiting subjects for an oral-systemic health research sub-cohort to PMRP. This cohort will also be used to better understand the genetic contributions to PD and caries. A cohort of 2000–3000 individuals is envisioned with iEHR data, DNA, plasma, serum, oral samples (saliva, subgingival plaque), and cross sectional longitudinal clinical tests (HbA1C, C-reactive protein, fasting glucose, urine microalbumin) collected at defined intervals. We aim to build on our past success in recruiting subjects for personalized medicine research, and indeed, these new subjects will become part of PMRP.

Marshfield Clinic, the largest, physician-owned, private group medical practice in Wisconsin and one of the largest in the US with 750 physicians representing 86 different medical specialties, is unique among organizations that are successfully harnessing the CPR to improve patient care. Marshfield Clinic physicians and surgeons provide care for approximately 400,000 unique patients residing in the mostly rural counties of central, northern, and western Wisconsin as well as Michigan’s Upper Peninsula through its network of 45 regional centres and in collaboration with local and regional hospitals, with roughly 1.8 million patient visits annually. Coordinating seamless care to this largely rural population across regional centres was a challenge that became greatly facilitated through the establishment of one of the oldest and most comprehensive EHRs in the US. Internally developed by the visionary efforts of its long-standing clinical informatics team, its proprietary EHR known as CattailsMD™, contains coded diagnostic data dating back to 1960, with other coded data and digital documents dating back to the mid-1980s. This EHR provides comprehensive electronic access to diagnoses, procedures, clinical notes, laboratory results, imaging results, pharmacy data, hospital progress notes and discharge summaries. This fully integrated and comprehensive EHR allows real-time electronic access to a patient’s medical record from any of its 45 regional centres and collaborating hospitals using its system. Additionally, it was the first physician practice-developed EHR to receive accreditation from the Certification Commission for Health Information Technology . Early development of the Medical Event Coding by Clinicians Application (MECCA) system has provided Marshfield Clinic with several decades of experience in the use of coded terminology in an EHR. In addition, the EHR contains roughly 70 million text documents. As such, Marshfield Clinic provides a unique environment for informatics research and the perfect laboratory to develop new tools for clinicians and staff. Marshfield Clinic physicians have been routinely using computers as an integral tool to capture and record relevant details of every patient encounter for over a decade. Marshfield Clinic continues to develop much of its own clinical software, thus allowing for rapid integration of new ideas and applications into clinical practice. With the introduction of hand-held wireless tablet computers, Marshfield Clinic is among the first health care provider in the country to put the electronic medical record truly at the source of health care decisions - in physicians’ hands. Marshfield Clinic was among the first institutions to respond to the challenge of the IOM, National Academy of Sciences report, “Dental Education at the Crossroads: Challenges and Change” ( Committee on the Future of Dental Education, 1995 ), which called for a greater collaboration and integration between medicine and dentistry at all levels of the health care: patient care, education and research. These developments also addressed the Surgeon General’s call to action to deal with the silent epidemic of oral disease disproportionately affecting the poor of all ages ( U.S. Department of Health and Human Services, 2000 ). As previously described, the partnership between Marshfield Clinic and its Family Health Center led to provision of dental services to rural communities throughout central and northwestern Wisconsin through a series of dental clinics that have become an integral component of Marshfield Clinic. With support from Delta Dental of Wisconsin, Marshfield Clinic was positioned to emerge as an early adopter of an integrated dental medical record, incorporating clinical and practice management functions into its comprehensive EHR. With medical/dental diagnoses, hospital data, imaging studies such as radiographs, and prescription information (including potential drug interactions) all available at the touch of a button, Marshfield Clinic clinicians have in their hands the information needed to make the best clinical decisions. The dental module, CattailsDental™, has been successfully implemented in all seven Marshfield Clinic dental centres. CattailsDental™ is based on an open-source dental software platform, Open Dental of Salem, Oregon. The integrated medical-dental electronic health record (iEHR) system is designed specifically for Marshfield Clinic physicians and dentists to access and share patient health information ( Figures 2 & 3 ). CattailsDental™ includes a dental dashboard; workflow management tools including tooth charting, periodontal charting, and treatment planning; access to centralized medications, allergies, special conditions, problems list, demographics, and Health Insurance Portability and Accountability Act (HIPAA) 1 forms; access to medical appointments to support better coordination of care; and highly secure remote access supporting devices such as laptops and computer tablets. Medication Manager is a recently introduced point-of-care integrated electronic prescribing and medication inventory system currently used by all physicians and dentists at Marshfield Clinic. Any changes in a patient’s medication history is available to physicians and dentists in real-time, irrespective of who made the change. Emerging approaches such as Clinical Decision Support Systems (CDDS) ( Sackett et al, 1996 ) and evidence-based practices ( American Recovery and Reinvestment Act, 2009 ) require availability of all data elements to support clinical decision making, with each element contributing a specific input needed by the system. In response to the American Recovery and Reinvestment Act of 2009 (Committee on Comparative Effectiveness Research, 2009), the IOM recommended comparative effectiveness research priorities as a focus of new national investment. Priority topics included healthcare delivery, health care disparities, oral health, and HIT ( Schmittdiel et al , 2004 ). Effectiveness of EHRs, eHealth, and other technology-based solutions to improve clinical care and self-management are urgently vital. EHRs increasingly have embedded evidence-based physician reminders to address clinical inertia ( O’Connor et al , 2009 ; Peterson et al , 2008 ; Chi et al, 2010 ). Historically, Marshfield Clinic has focused efforts on maximizing the capability of its EHR. With the iEHR almost complete, attention is now being focused on improving decision support at point-of-care. One of the key strategies to reverse the oral health access disparity is to enlist medical providers with appropriate decision support tools delivered through the iEHR system. This is an area targeted for improvement by the IOM, who recognises that true progress in this arena is possible through advances in health informatics and informatics technology. Currently, useful and readily available information to physicians and dentists to manage their patients in a comprehensive way is lacking. Apart from the systemic diseases that are interconnected with PD, there are several oral manifestations of systemic conditions ( Chi et al, 2010 . See Table 1 ). Careful examination of the oral cavity may reveal findings indicative of an underlying systemic condition and allow for early diagnosis and treatment. With the iEHR, this information can be used to develop the cross-disciplinary CDDS to improve quality of care, provider workflow, and decision making for physicians and dentists. This infrastructure further provides support for conducting translational oral-systemic disease research. The iEHR is supported by an integrated Enterprise Data Warehouse (iEDW) that houses patients’ medical and dental information. The intent of this sophisticated data warehouse is to be the sole reporting source for high quality integrated information related to patient care, research, and clinical operations. The iEDW is a high quality data repository that has been optimized for data analysis and reporting to support research activities, quality assurance, and decision support at Marshfield Clinic. The iEDW utilizes existing Marshfield databases and other data sources to store key information central to defining patient phenotypes. iEHR data are transferred to the iEDW, where they are linked accurately and completely as longitudinal continuums of care, ensuring that virtually no records are lost. As such, the iEDW serves as an essential resource for the development of research databases. As data is moved from clinical applications into the iEDW, data models and coding standards are normalized, algorithms are applied that ensure data quality, and data is transformed into a data model that supports population-based research. To give an example, the primary care departments at Marshfield Clinic currently use a CDSS (called iList) to create department/physician-based listings of patients whose quality care measures are not at goal for DM, hypertension, heart failure, and coronary artery disease. Patients who are diagnosed with one or more of these disease states and meet the quality care measures will not be displayed in a list. The patient listings allow staff to identify and intervene with patients not meeting the measures. The intent is to provide actionable information to improve patient outcomes and is used as a tracking tool to improve patient care. Any actions taken by physicians or staff members will help patients achieve a quality goal. Per protocol, support staff may call a patient to schedule an overdue fasting lipid panel or add an HgA1c test to a patient’s next scheduled appointment; or a provider may direct support staff to advise a patient to return to the clinic in one month for a blood pressure check. Currently, iList assists in ensuring that diabetic patients are scheduled for needed services such as lab orders and appointments with primary care physicians and non-physician team members by executing the rule-based algorithm. However, as part of this diabetes management protocol, there is no routine oral examination done for the diabetic patient. Additional rules are being developed for application to this diabetes management protocol along with the necessary triggers to filter out the diabetic patients who have not had an oral examination in the last 6 or 12 months for dentate or edentate patients, respectively, within the iList. The developed CDSS combines analytics with oral care referrals for diabetics in a real time clinical care application that provides actionable patient lists for diabetes disease management and prevention. On the other side, the dental CDSS at the dental site focuses on generating appropriate referral reminders based on previous diagnosis of diabetes and system alerts based on potential risk for diabetes/pre-diabetes. The dental CDSS then utilizes data already available within the integrated medical-dental EHR system or collected as part of patient history taking and examination to trigger the reminders and alerts. The dental CDSS algorithm also evaluates the potential risk for diabetes/pre-diabetes in the undiagnosed population. The dental CDSS utilizes patient’s age, height, and weight already available within the iEHR system or collected as part of history when logging in new patients. Based on specific criteria (e.g. patient’s age ≥45 years and body mass index ≥25) the dental CDSS will trigger an alert to perform a blood glucose test at the dental centre. Appropriate referrals to the medical centres will be made based on the test results.

The availability of saliva and subgingival plaque collected in the context of the oral-systemic sub-cohort of PMRP will provide an important resource for microbiome characterization in patients with and without PD or DM. Dental samples are being obtained from individuals with no PD as well as mild to severe PD at the time of their visit to a Marshfield Clinic’s dental centre. Accurate phenotyping of these patients is accomplished through conduct of a complete oral examination by trained, calibrated dental examiners. Periodontal status is assessed at six sites per tooth (mesiobuccal, midbuccal, distobuccal, mesiolingual, midlingual, and distolingual) for all teeth present. Probing depth (mm) and location of the gingival margin in relation to the cementoenamel junction is measured using a UNC-15 manual probe (HuFriedy, Chicago, Illinois, USA) ( Desvarieux et al, 2010 ). Dentists perform a full-mouth clinical examination including inspection of the teeth, oral mucosa, and periodontal tissues to characterize oral health status of all individuals. Inclusion criteria for subjects with untreated PD include no antibiotic usage in the past 3 months, no routine dental care in the past 12 months, at least eight sites with a pocket depth >4mm, and six sites with attachment loss levels >4mm. Inclusion criteria for subjects defined as having no PD include no periodontal pockets deeper than 3 mm, no antibiotic use in the past 3 months, and no clinical evidence of PD or other concurrent oral disease. In addition, patients are assessed for diabetic status by determining HbA1C levels. Collection of these specimens will allow characterization of microbiome in the presence of PD and following resolution of infectious processes. Further, comparisons of microbiomes in patients with and without DM are planned to gain insights into how the environment impacts the metagenomic footprint in the presence and absence of glycemic control.

Within the context of PD, another genetic component that needs to be considered is that of the oral and other microbiota in the body, especially those microorganisms associated with periodontal plaque that have been implicated in the pathogenesis of destructive inflammatory responses locally and systemically. Historically, early inroads into metagenomics were made with the analysis by Socransky et al (1998) using a whole genomic approach to characterize the microbial profile associated with PD and health. These investigators contributed recognition of the so-called ‘red complex’, that consists of three periopathogens – P. gingivalis, T. forsythia, and T. denticola – that tend to occur as a group in association with PD. Darveau (2010) postulated that this trio of organisms collectively has the capacity to disrupt innate immune responses. Evidence is mounting that health depends on the dynamic interplay between the host and the microbial community that harbours in different organs of the body. The microbiome represents the ecological community of commensals, symbiotic, and pathogenic microorganisms that share our body space, and any perturbation in microbiome is likely to have a serious role in health and disease ( Lederberg and McCray, 2001 ; Keijser et al, 2008 ). Human adults harbour ten times more microbial cells than human cells ( Gordon et al , 2005 ). Consequently, these microbial cells contribute more genomic material (genes, transcripts, proteins, metabolites) than the host; therefore, it is only logical to assume that the microbiota of an organ may play a huge role in the development, modulation, and/or maintenance of some chronic diseases. Advances made in characterizing oral microbiomes have been recently reviewed ( Zarco et al , 2012 ; Jakubovics and Kolenbrander, 2010 ; Avila et al , 2009 ; Xie et al , 2010 ; Dewhirst et al, 2010 ; Sakamoto et al , 2004 and 2005 ) and will be discussed here in the context of how advances in these fields have expanded the understanding of pathological mechanisms contributing to PD and how this impacts on advancing personalized medicine. Like other human organs such as the gut, the human oral cavity harbours a large community of bacteria. Any changes in the dynamics of the oral microflora could initiate the formation of a dental plaque. While the majority of these organisms could be commensals; some are opportunistic pathogens and play a role in oral conditions such as gingivitis and chronic periodontitis. Individuals with chronic periodontitis have an increased risk of developing coronary heart disease ( Beck et al, 2005 ; Kebschull et al , 2010 ), ischemic stroke ( Joshipura et al , 2003 ), hypertension ( Desvarieux et al , 2010 ), diabetes ( Genco et al , 2005 ), aspiration pneumonia ( Scannapieco 1999 ), bacterial endocarditis ( Berbari et al , 1997 ), renal disease ( Kshirsagar et al , 2005 ; Shultis et al, 2007 ) and preterm low birth-weight babies ( Offenbacher et al , 1998 ). Similarly, recognition of the threats posed by periodontal diseases to individuals with chronic diseases such as respiratory diseases and osteoporosis is relatively recent ( Seymour et al , 2007 ). Indeed, some findings from these studies relative to the constituents of the oral microbiome are challenging traditional paradigms that have been advanced to explain disease pathogenesis. As summarized by Zarco et al (2012) , the collective microbial genomic material, or metagenome that represents the genetic content of an estimated 700 species of microbes, expands the genomic content of any given human approximately 100 fold and adds systemic attributes that were not genetically endowed by the human genome. Collectively the microbiome systems throughout the body contribute to the global metagenome and establish a unique ecological system within each person that creates a symbiotic relationship with the host. Perturbations in the microbiome or changes in the host interaction with the microbial load may shift the balance within microbiomes and the global ecosystem from one supporting health to one supporting disease. This shift underscores the necessity for a personalized approach to care, since microbiomes will vary with each individual relative to microbial content and environmental characteristics contributed by the microbial milieu and metagenomic content and host responses. These authors emphasize the need to attempt definition of microbiome expression patterns in health and in the context of various diseases. An important insight afforded by advanced molecular methodological approaches into the composition of the new oral microbiome is its heterogeneity throughout the oral cavity and even within the layers of a biofilm, as dictated by the environment. Further, genotypic characterization of an isolated microbe alone is no longer sufficient to establish relative pathogenic status. As seen with the ‘red complex’ microbial trio, determining relative representation, analysis of other microbes present, and characterization of contextual interaction with the host is critical to establishing the relative contribution to health or disease of any constellation of organisms present in the microbiome. Broad heterogeneity in genomic content, even for any given species, has been noted to endow new virulence factors through horizontal transfer of genetic material and transfer of new functional capacity among species. The use of culture-independent techniques, such as 16S rDNA sequencing to identify uncultivable microorganisms has proven to be a powerful technique for identifying and characterizing novel pathogens found in clinical samples and the environment. With the advances in sequencing capability from Roche’s 454 DNA sequencer to Illumina’s sequencing by synthesis (SBS) technology, it is now possible to sequence millions of bases at a fraction of the cost compared to 10 years ago. This ability to identify previously uncultivable, therefore unsuspected, pathogens from complex infectious diseases has redefined the etiological agent(s) of a number of complex illnesses and metabolic diseases ( Eng et al, 2011 ). These approaches revealed that approximately 50% of organisms represent uncultivable bacteria and even include newly defined microbes ( Wade, 2011 ). The relative contribution of these organisms to health or disease is totally uncharted territory. Even the existent body of evidence limiting attribution of PD to the periodontal pathogens recognized to date has been challenged by data reporting low relative representation of the these organisms associated with the presence of other previously unidentified, undefined species in samples exhibiting disease ( Wade, 2011 ). Importantly, application of new technologies such as the Human Microbe Identification Microarray (HOMIM), including probes for 300 microbial species, is making it possible to discern a unique genetic profile in patients with refractory PD that distinguishes them from healthy subjects ( Colombo et al , 2009 ). These discoveries suggest a new paradigm wherein definable patterns of microbial profiles within microbiomes and shifts in these patterns in the context of disease may support more accurate characterization of pathogenic processes. Notably, a new hypothesis recently advanced by Amir et al (2011) proposes that hyperglycaemia associated with T2DM may be responsible for mediating translocation and relocation of gastrointestinal flora systemically. Translocation is postulated to impact the composition of the periodontal microbiome causing perturbation of the periodontal homeostasis and further up-regulation of acute phase response, cytokine signalling, and immune mediated response ( Amir et al, 2011 ). The impact of individual microbiomes on other microbiomes requires further examination. Increasingly, the importance of microbiome-based studies is underlined in identifying the potential role of etiological agents in complex chronic diseases such as DM and PD, where infectious/inflammatory processes are suspected contributors to pathogenesis ( Rajilic-Stojanovic et al, 2007 ).

At its most fundamental level, personalized medicine can be envisioned as making clinical and treatment decisions for an individual based on probabilities of events given the interactions between genetic, clinical and environmental factors affecting that individual ( Kinane and Heart, 2003 ). By contrast, traditional medicine bases treatment and clinical decisions for individuals on probabilities of population-based observations of outcomes and responses, which may or may not be appropriate at the level of the individual. Assuming the validity of genetic contribution, the following is a brief review of the current understanding regarding the contribution of genetics to PD. To date, genetic studies of disease association have estimated that there are approximately 307 human ‘periodontitis-associated’ genes. However, these studies are largely underpowered to find all but the largest of genetic effects, with >65% of reported associations based on examination of ≤ 100 patients, and approximately 40% based on ≤ 60 patients that require further validation. Approximately 14% of the 307 genes represent candidates based on biological significance or detection across multiple studies ( Diehl et al, 2005 ; Diehl et al, 2011 ; Nikolopoulos et al , 2008 ). Thus, more robust studies are required to test these candidates and explore further genetic associations in the light of advances in understanding of mechanisms underlying PD. Genetic analysis applying familial and twin studies support a heritable component for aggressive PD, whereas twin studies support a genetic basis for chronic PD. To date, no gene(s) have been clearly associated with causation, although autosomal dominant transmission, albeit with heterogeneity of loci for familial aggressive PD, appears to be operative following correction for ascertainment bias ( Kinane et al , 2008 ). An interesting study of 50 families with affected members presenting with aggressive PD by Diehl et al (2003) found that 38% of variance of immunoglobulin G allotype 2 (IgG2) levels could be explained by genetic contribution (p=0.0006). No correlation was noted following bi-trait variance component analysis of IgG2 levels and aggressive PD susceptibility, suggesting that different genes determine IgG2 levels and disease susceptibility. One meta-analysis of association studies reports association of single nucleotide polymorphisms (SNPs) occurring in chronic periodontitis in the IL1, IL6, VDR, and CD14 genes ( Laine et al, 2010 ), although further studies are required to validate these in larger, more stringently phenotyped populations. Other studies suggest that moderate and severe PD both share and have distinct genetic components ( Brett et al, 2005 ). The possibility that functional mutations, such as single-nucleotide polymorphisms (SNPs) occurring in pro-inflammatory pathways, may simultaneously impact on association with several conditions, is also of interest to explore. One such example can be found by SNPs identified in the pro-inflammatory cytokine genes for IL1 (including one each in IL1A and 1L1B) , that promote periodontal progression ( Huynh-Ba et al , 2007 ; Nikolopoulos et al, 2008 ). A further example includes SNPs occurring within the IL1 gene cluster that have recently been implicated both in PD and in hyperlipidemia. The impact IL1 SNPs on two distinct pathological conditions is simultaneously amplified by both increasing risk for myocardial infarction ( Laine et al , 2010 ; Keavney, 2010 ) and promoting diabetes pathogenesis ( Santos Tunes et al, 2010 ). Similarly, expression of IL6, a pro-inflammatory cytokine, increases with age and may contribute to promoting pathogenesis of conditions that are exacerbated by pro-inflammatory mechanisms. Further, IL6 impacts on bone resorption and may thus contribute to bone loss observed in PD. A recent study by Offenbacher et al (2009) investigated the role of the immune response and proinflammatory processes in gingival health from a host susceptibility perspective. These investigators explored changes in the patterns of whole-transcriptome that occur in humans during the induction and resolution of experimentally-induced gingivitis, the mildest form of PD, on serial biopsy specimens collected pre-induction, during active gingivitis, and after resolution. Whole-transcriptome gene-expression arrays revealed that 131 immune response genes were up- or down-regulated. Transient expression of inflammatory and oxidative stress mediators were predominantly expressed, including interleukin IL1A, IL1B, IL8 (CXCL8), Regulated on Activation, Normal T Expressed and Secreted, (RANTES), colony stimulating factor 3 (CSF3), and superoxide dismutase 2 (SOD2); while there was decreased expression of interferon gamma-induced protein 10 (IP10), interferon inducible T-cell alpha chemoattractant (I-TAC), matrix metalloproteinase 10 (MMP10), and beta 4 defensin (DEFB4). Reversal of these expression patterns was noted when inflammatory processes associated with gingivitis resolved. This study revealed that a subset representing approximately 10% of the immune response genes demonstrated specificity relative to the transcriptome-expression response and maintenance of tissue homeostasis at the biofilm-gingival junction ( Offenbacher et al, 2009 ). Another noteworthy study by Covani et al (2009) used a genomics approach to define genetic common denominators between DM and PD. The investigators set out to identify genes associated with the highest number of functionally interactive biological pathways associated with each condition. These genes, termed ‘leader genes’ by the authors, were defined independently for DM (n=12), PD (n=5), and sinusitis (n=6). Remarkably, comparison of the resulting ‘leader gene’ analysis for each condition demonstrated bidirectional overlap between four of five PD leader genes with DM leader genes. Overlapping genes included NFκB1, RELA, PIK3R1, and GRB2. Notably, such convergence was not present with any of the six leader genes associated with sinusitis, although infectious/inflammatory mechanisms also are implicated in establishment of pathogenesis and chronicity of this condition.