Cohort profile: Greifswald approach to individualized medicine (GANI_MED)

GANI_MED is conceptualized as cohort study with individual follow-ups. Starting with six main cohorts including patients with common diseases, three additional cohorts joined during the recruitment period. All subjects were regular patients of the university hospital except some of the patients with chronic renal failure who were also recruited from special treatment centers for dialysis. Study nurses or physicians from the participating departments enrolled the patients. All cohorts are thoroughly phenotyped using clinical methods, imaging technologies (ultrasonography, magnetic resonance imaging (MRI)), and OMICS approaches (Table 1). GANI_MED was approved by the ethics committee of the Medical Faculty of the Ernst-Moritz-Arndt University Greifswald. After complete description of the study to the patients, written informed consent is obtained. New infrastructures have been implemented (e.g. automated biobank, medical informatics) and additional research groups (e.g. research on ethical challenges) were founded. In the future, we plan to integrate all patients of the university hospital into the research data base, at a lower level of data assessment as compared to the present assessments in GANI_MED.

We are recruiting six main cohorts with common cardiovascular, cerebrovascular or metabolic conditions: heart failure (expected n = 1200), stroke (expected n = 600), periodontal disease (expected n = 800), renal insufficiency (expected n = 400), metabolic syndrome (expected n = 1600), and fatty liver disease (expected n = 400). The official start of the patient recruitment was July 7, 2011. Three further cohorts (patients with sepsis, pulmonary diseases, and adverse medication effects) have been launched in the meantime (for detailed description of the cohorts see below).

Besides the core examination program, which includes the computer-assisted basic medical history and a comprehensive medication assessment, blood pressure and anthropometric measurements, a basic dental examination, and the common laboratory measurements, each cohort is characterized by additional examination procedures that are detailed below. The clinical characterization of patients in the different cohorts is summarized in Table 1.

Individualized Medicine requires complex platforms to acquire and manage data sets for evidence-based research, including data transformation, normalization, and long term storage. A high degree of automation is essential to efficiently acquire, integrate and apply Use & Access rules to data from a variety of sources, including electronic Case Report Forms (eCRF), a patient’s health record, diagnostic devices such as MRI, sonography, and ECG, or sources for secondary data like a population register, health insurance or a patient’s family doctor [50]. Compliance with public and local regulations of data privacy as well as information security has to be ensured. Scalability and high availability of at least those tools that are used to record data is another key aspect to consider. Applicability of the Medical Device Directive (MDD) of the EU, together with country-specific regulations, must be carefully considered when designing and operating such platforms and tools. The informed consent is a basic requirement for the scientific use of patient data. GANI_MED adopts a modular consent design that allows patients to agree or disagree to particular modules in an opt-in model (e.g., use of data, biomaterials, information about incidental findings, contact in the future). An independent Trusted Third Party as institutional as well as technical instrument is implemented to centrally manage the informed consents of all GANI_MED participants. Furthermore, the Trusted Third Party ensures unique and platform-wide identification of persons, independent from data source and local identifiers. Another unique identifier (Master Person Index) is used as a link between all source-specific identifiers and identities. A probabilistic approach allows tolerating minor spelling mistakes or missing information; ambiguous matches are resolved manually at a later point of time while data can be recorded continuously. A conservative approach is taken in case of an ambiguous match: two identifiers are used, which can be combined once both identities have been approved as belonging to a single person. Furthermore the Trusted Third Party generates and manages all pseudonyms used throughout the platform. The tools used in GANI_MED can operate on almost arbitrary alphabets (numeric, characters, special characters, mixes). Parameters can be adjusted to optimize for error correction or error recognition. Use & Access of data is managed through a transfer unit. When transferring data to a research project, all pseudonyms are made project-specific pseudonyms to prevent aggregation of larger data sets across a number of projects. Figure 1 shows the high-level architecture of the GANI_MED research IT-platform.

Cohort management is done using Java-based offline-capable eCRF software on mobile clients, so-called Mobile Clinical Assistants [51]. These are used to record patients’ personal data, document the informed consent, and generate a cohort-specific number of dynamic eCRF-based data, including lab order IDs and similar links to external sources. All data are synchronized with a server application which functions as one of the data sources (left hand side of Figure 1). Currently, the cohort management is used for nine GANI_MED cohorts, implementing 225 electronic forms, more than 8,000 variables and 200,000 data sets. The Extraction Layer of the platform accounts for separating personal (i.e., identifying) data from medical data and performs source-specific transformations into a platform-internal unified data format for further processing. Presently (March 3, 2014), the platform operates on 18 million data points for 3,535 persons participating in GANI_MED. The three functional blocks of the Trusted Third Party have successfully been re-used and adapted for other projects such as the German Center for Cardiovascular Research (DZHK) or The National Cohort in Germany, a large representative population cohort that will start recruitment of 200,000 subjects in 2014.

An automated biobank (STC12k-ULT KiWi Store, FA Liconic, Liechtenstein) was installed to store and handle up to 500,000 cryo tubes of 1 ml volume at a constant temperature of -80°C. Nineteen cryo tubes with biomaterials including serum (N = 5), EDTA-plasma (N = 5), urine (N = 5), saliva (N = 2), and DNA (N = 2) are processed and stored for each patient.

Deep phenotyping of patients with genomics and functional genomics approaches provides molecular information at unprecedented detail that will allow much more in-depth description of the (patho)physiological status of a patient. The OMICs data generated at different levels (genomics, transcriptomics, proteomics, and metabolomics) will likely allow for better descriptions of health and disease states [52]. Within GANI_MED, standardized workflows for the OMICs characterization of biosamples will be developed. Currently, genomic variants are analyzed for 1,800 patients with the aid of the HumanCoreExome + v1.1-Psych Array. For a subset of 1,200 patients, whole blood expression data are generated using Illumina’s HT-12 bead chips. These data will be complemented with proteomics and metabolomics data using established workflows.

Metabolomics provide comprehensive snapshots of the metabolome of body fluids such as plasma, urine or saliva. Within GANI_MED, high-throughput metabolomics analyses mainly based on 1H nuclear magnetic resonance (NMR) spectroscopy, a non-destructive analysis with minimal preparation requirements, are performed. NMR spectroscopy provides robust and reproducible measurements. Mass spectrometry (MS) with high analytical sensitivity will be used for additional detailed studies.

Integrated analyses of these data for associations with clinical and subclinical phenotypes will be performed by the bioinformatics groups of the GANI_MED consortium.

The health economic analysis contributes to the comprehensive assessment of advantages and disadvantages of Individualized Medicine by comparing costs of innovative diagnostics with their effectiveness. One focus lies on the performance of an economic evaluation of the predictive value of genetic and non-genetic biomarkers to forecast health care costs. The analysis which is based on SHIP data, correlates genetic and non-genetic biomarkers with the health seeking behavior and/or health care expenditure [53]. For instance, we demonstrated that rs738409 (GG) in PNPLA3 is a significant predictor of hospital admissions for gallstone complaints. Secondly, we perform cost analyses of severe drug-related side effects as they are a major cause of hospital admissions and cause a large fraction of the global health care expenditure. Thus, it is crucial to understand the costs and benefits of specific treatment options or diagnostic tests such as genetic testing. We could show that the immense costs caused by intestinal bleeding as a side effect of phenprocoumon therapy are so high that a specific genetic test for every patient taking this drug would be cost-effective [54]. Thirdly, economic analyses evaluate the cost-effectiveness of specific therapeutic interventions of Individualized Medicine [55].

One general strategy in GANI_MED is the recruitment of patients during ongoing clinical routine care. This means that hospitalized patients are being asked to make their clinical data and biomaterial available for scientific research. Thus, one major challenge for the ethical working group was to develop an appropriate information concept to enable patients to give their informed consent. This informed consent procedure covers the multilevel assessment of personal, clinical and biological data as well as the storage and use of data and the biobanking. All informed consent procedures are carried out according to standards of law, data protection, and research ethics. This includes the preparation of the information and consent forms, an ongoing training for the GANI_MED staff as well as the design and implementation of workflows and standard operation procedures regarding all ethically sensitive processes in GANI_MED [56, 57].

Some GANI_MED cohorts also include non-treatment related study examinations. Ethical challenges emerge, if incidental findings, that might be relevant for the health of the participants, occur in these cases. Their management does not follow clinical routine care. Especially when the participant does not suffer from any symptoms of illness researchers need to weight potential harm against potential benefit of disclosure of medical findings for the participant. The benefit might be the early detection and treatment of pathological states whereas invasive diagnostic procedures as well as psychological distress associated with the disclosure of incidental findings represent relevant negative consequences.

The non-clinical use of imaging methods like the whole body MRI or genome-wide data represents an ethical and psychological challenge in case of incidental findings.

To address this issue, one of our ethical research projects specifically focuses on the disclosure of incidental findings. The project applies methods from empirical social research to examine how the disclosure of incidental findings will affect the participants within medical research projects. This is done within the framework of the population-based SHIP study [2] at the University Medicine Greifswald with a special focus on the whole body MRI. The view of the participants is inferred from their answers to questionnaires and interviews with SHIP participants who were informed about their incidental MRI findings [58‐60].

The University Medicine Greifswald has a long tradition in community-based research. SHIP is a large prospective cohort study that was implemented in 1997 (SHIP-0) to investigate a broad spectrum of health-related parameters in West Pomerania, a rural area in the north-east of Germany with a relatively short life expectancy as compared to that in the average German population [2]. Over the past 15 years, the population-based study SHIP has evolved as cohort study with one of the most comprehensive phenotype assessments worldwide covering the follow-up waves SHIP-1, SHIP-LEGEND and SHIP-2. From 2008 to 2012 a new, independent sample called SHIP-TREND was recruited (n = 4420). Both representative samples from the general population serve as reference sample for case–control comparisons with the GANI_MED patient cohorts. The university hospital Greifswald is the only major hospital in the area. The SHIP sample was derived from the same geographic area, ensuring that patient cohorts and control groups originate from the same population.

External collaborations are welcome. For the coordination of research projects based on the GANI_MED data, a request and transfer process has been established that largely follows the procedure established in SHIP. A web-based request form will be available soon. Meanwhile interested researchers may download a form at the data transfer unit (http://​www.​community-medicine.​de) and send the completed document to grabeh@uni-greifswald.de. Together with collaborators at the University Medicine Greifswald we will select the necessary variables and complete the application. Once the application has been approved, a contract is concluded between the external applicant and the Research Network of Community Medicine.

As data collection and OMICS analyses in GANI_MED are still ongoing, research is focusing on the analyses of SHIP and laboratory data to generate new hypotheses for translational research. Research in the field of hormones like testosterone, prolactin and IGF-1 as individual biomarkers and putative predictors of clinical outcomes (e.g., hypertension, depression, inflammation) has been especially successful [61‐66]. Novel biomarkers like angiopoietin-2, Tie-2 [67], homoarginine and asymmetric dimethylarginine (ADMA) [68] are under investigation. Metabolites have been analyzed by NMR and mass spectroscopy and are currently tested for their association and predictive properties in various diseases [69]. Strong associations between genetic variations and metabolites have been discovered [70]. Myocardial gene expression profiles have been used to successfully predict the treatment response to immunoadsorption therapy in patients with dilated cardiomyopathy [71]. The important interface between oral and systemic health is increasingly addressed in GANI_MED [72], especially the inflammatory effects of periodontitis [73]. Growth factor receptor-mediated signaling is meanwhile recognized as a complex signaling network. The interactome of the epidermal growth factor receptor (EGFR) was identified and quantified. The newly developed Cytoscape plugin ModuleGraph facilitated the extension and functional investigation of this network [74]. A short 5-item questionnaire for routine clinical assessment of childhood abuse and neglect has been developed and validated [3, 75]. Thus, the effects of gene-environment interactions in somatically burdened patients can be investigated [76, 77].

A standardized protocol for fat quantification in fatty liver disease on liver MRI has been developed and validated within this cohort [34, 78, 79].