IT & informatics in translational research: a case study

Abstract

Organizational Information Technology (IT) is only just beginning to comprehend and support the unique requirements and objectives of translational research. Methods, tools and technologies are emerging organically and evolving dynamically from other clinical and research applications. In late 2008, BloodCenter of Wisconsin (BCW) embarked on a 5-year strategy to integrate our IT services with the strategic direction of our research organizations – Blood Research Institute (BRI) and Medical Sciences Institute (MSI). The following represents a case study of our journey in the employment of bioinformatics in support of translational research initiatives. In his 2003 address to Congress, Dr Elias Zerhouni, the Director of the National Institutes of Health (NIH) and architect of the transformational NIH Roadmap warned, ‘Major trends in science, society, and economic forces will usher in an era of exceptionally rapid change in the way biomedical research is conceived and performed’ [1]. These trends exist on both sides of the capacity and demand chart. Economic dimensions such as increasing competition, return on investment and an aging demographic are heralding a sea change in the way biomedical research is conducted. New technologies have released a deluge of data that has accelerated the pace of this change. While technology has indeed accelerated the pace of biomedical discovery, conversion of these discoveries to clinical application has waned. Within the past decade, genomic and molecular research has led to tremendous discovery in basic science; however, these discoveries have been slow to yield clinical advances. As a sponsor of this biomedical research, the NIH has transformed its programs to also reward researchers with novel research ideas that possess tangible translational plans, provide for the sharing of data, and leverage interdisciplinary, inter-institutional collaboration that collectively depends on the use of information technology. Whereas basic science is concerned with knowledge discovery, translational research is focused on validating this knowledge and ultimately translating discoveries to patient care. To be successful across the clinical research continuum, research organizations must successfully negotiate translational hurdles, one concerned with translation from basic to clinical investigation and another from clinical investigation to better practices and decisions that improve outcomes [2]. Cursory review of the literature will turn up a diverse set of descriptions and definitions for translational research. Variations on these definitions suggest that while translational research is viewed differently depending on the particular industry, organization, or institution implementing its practice, its descriptions embody fairly consistent themes that favour bidirectional collaboration along the clinical research continuum in an evidence-based and patient-centric manner. Indeed, the declared objective of translational research aims to break down classical boundaries between research disciplines that have historically represented hurdles to innovation. For researchers and institutions that either depend on Federal funding to conduct research or depend on innovation to drive sustainable competitive advantage, the implication is obvious: successfully competing for scarce resources will be a function of how well they have adapted to accommodate this transformation. In Blood and Transfusion Medicine, translational research is still very much in its infancy. Yet, our investigators are increasingly proposing new research that seeks to better understand the molecular mechanisms of diseases in haemostasis and thrombosis, benign haematology, immunology and transfusion medicine. An overarching goal of these projects is to not only create new knowledge, but to appropriate this knowledge for practical clinical use in advancing patient care by developing and commercializing new diagnostic tests, new methods of care and new therapies. As an enabler of research, IT must also adapt to help facilitate this transformation, evolving beyond merely aligning with Research to actively partnering with our investigators as a facilitator of translational research. Understanding the role of IT and informatics in enabling biomedical research is a first step towards developing a strategy for how it must change to enable research within BloodCenter of Wisconsin, its BRI and its MSI. To appreciate the evolution of IT in translational research compels an awareness and acknowledgement of the distinction between informatics and information technology (IT). Bernstam et al. [3] distinguished between IT and informatics by describing the differential contribution of each to support translational research. Whereas the former generally leverages technology to provide both administrative and operational support to research, the later leverages domain relevant data, information and knowledge to support research. The authors further emphasize that failure to recognize this distinction or improper deployment of resources within each can propagate confusion and lead to project delay or failure. While translational investigations are often characterized by relatively small populations, less than 500 subjects, they often involve highly specialized data collection and large numbers of data points. Often these investigations are conducted in a facility that is equivalent to a hospital’s intensive care unit in terms of personnel and specialized equipment. As a result, they require a high level of information technology support and resources with diverse capabilities and skills. Operational IT functions to keep the lights on, ensure high availability, network uptime and support for email and desktop applications. Moreover, as the future of biomedicine is interdisciplinary, operational IT increasingly finds itself contemplating the use of Web 2·0 technologies and methods to promote collaboration, community interaction and sharing among organizations and consortia with common interests [3]. Research IT and even operational IT support may be instrumental in managing the vast quantities of complex and diverse data, generated by automated laboratory equipment and next-generation high-throughput instrumentation. Organizing and increasing data interoperability to promote interdisciplinary data sharing depends on yet a different orientation that embraces knowledge management, including cross-domain ontology, controlled vocabularies and data standards [4]. Further, eliciting meaning and even knowledge from these data may depend on specialized computer science resources and capabilities to make data more readily consumable and visible. Skilled informatics personnel, novel methods, and new technologies must enable researchers to more efficiently ask complex questions across disparate data types (that are rarely integrated). Increasingly, convergence of these roles, talent and capabilities to address the unique needs of translational studies has been embodied within the context of a new discipline – translational informatics [4]. Translational informatics is borrowing from relevant aspects of conventional disciplines, including bioinformatics, computational biology, clinical informatics and even public health, to bridge the clinical research continuum. The commercially available market for technologies serving translational research is still dynamically evolving. As a result, no single technology or vendor seems to possess market dominance within this vertical. Many integrated platforms – those providing an ability to manage study data, clinical data, sample data and scientific data – have historically positioned products to biobanks and pharmaceutical firms. With the launch of the Clinical Translational Science Award (CTSA), several technology vendors repositioned their products to CTSA awarded sites. During our market evaluation, we discovered that many of these established solutions evolved out of National Cancer Institute (NCI)-sponsored research institutions. Still, other technologies were, and still are, emerging organically from the ground up to shake free of the constraints imposed by traditional research models they supported. Many Academic Medical Centers (AMC) – particularly those with a deep bench of skilled IT and informatics resources – have developed their own solutions using open source models as lower cost alternative to licensing. A more comprehensive review of these solutions is, however, beyond the scope of this topic. BloodCenter of Wisconsin has long held biomedical research to advance patient care as core to our mission, values and business. A large proportion of this research is conducted at our dedicated research facility, Blood Research Institute (BRI), located on the Regional Medical Center grounds in close proximity to Medical College of Wisconsin, Children’s Hospital of Wisconsin and Froedtert Hospital (the region’s only level 1 trauma centre). BRI has also become synonymous with the nearly 150 dedicated faculty and staff conducting primarily basic science research and limited clinical research in immunology, transfusion medicine, vascular biology and haemostasis and thrombosis. Since the lion’s share of our research has historically been funded by federal extramural grants and contracts, each investigator-led laboratory had procured and maintained its own information technology, while corporate IS mainly supported operational IT; in recent years, however, several driving influences have prompted a shift in this thinking that has shaped the existing enterprise strategy for IT and informatics. Paramount among these drivers was the recognition that we could deploy our research resources and capabilities to focus on opportunities in clinical and translational research, propelled in part by our application to become a CTSA site in collaboration with Medical College of Wisconsin. Beyond CTSA, we also had significant experience in commercializing research discoveries – which has led to new products, creation of new clinical laboratories and even the foundation of a company that BCW sold in 2008 – but we lacked a dedicated structure to repeatedly manage this process. This changed in 2009 with the conception and creation of the Medical Sciences Institute (MSI). Comprised of physician researchers, MSI provided the infrastructure to integrate clinical care and research with a focus in benign haematology, haemostasis and thrombosis, transplant immunology and transfusion medicine. MSI provided a venue to carry the discoveries of our basic science researchers into the clinic, and to contribute our specialized knowledge. We also recognized that we were suboptimally utilizing one of our most precious assets, the combination of human specimen and clinical data that existed in different silos throughout BCW. When we first embarked on developing a strategy, we inventoried nearly 100 databases created by our researchers – primarily in desktop database technologies and spreadsheets – to manage specimen, study – and even clinical data. These IT systems that have supported research grew by accretion and not by design, emerging in isolation from one another to collectively create a formidable obstacle to realizing the value of the underlying data. Data within these databases, together with the potential research value of the health data on a vast population of participating, volunteer donors, became collectively known as the treasury. This treasury holds great potential as control data in research studies and further as a means of stratifying patient populations for correlative and prospective studies. In addition to these opportunities, greater potential efficiencies and economies of scale could be realized merely by centralizing informatics resources, expertise and capabilities, rather than keeping them distributed among individual laboratories. As our research organization moved towards clinical and translational research, IS endeavoured to move out ahead of it. We recruited informatics and IT resources with cross-domain knowledge in biomedical research and with an ability to communicate with researchers in their own language. We reorganized IS to better align our resources and capabilities with business units. We researched the market to understand what technologies were deployed, where they were deployed and how they were being used. Finally, and perhaps most importantly, we interviewed our investigators to identify shared vision and common requirements. From our analysis of this vision, these requirements and the marketplace, we devised a strategy which included deployment of an integrated, enterprise research platform intended to break down the boundaries around data silos to not only support basic discovery, but to also facilitate and foster translational research and innovation from bench to bedside and back, closing the loop between discovery and commercialization and providing tangible value to our donors and patients. Such a platform would marry features, including – biospecimen management, clinical data management, study management, scientific data management and knowledge management – together within a unified platform. Biospecimen management implies a robust ability to annotate specimens by multiple dimensions including – phenotype, genotype, environmental conditions and experimental findings; moreover, it includes the ability to segment, query and mine the data by these dimensions. Beyond enabling correlative banking studies, biospecimen management introduces the potential to conduct prospective banking, where stratification of the donor or patient for a specific study is not determined in advance of the study. Clinical data – medications, history, procedures and laboratory results – fulfils a vital component of the bidirectional translational loop, connecting bedside to the bench. Effectively managing clinical data and merging it with specimen data introduces the ability to segment the data by clinical dimensions, increasing the capacity for correlative analysis and potentially, the ability to identify relationships between phenotype and the molecular basis for disease. These characteristics demanded a solution sufficiently open and capable of integration with clinical information systems. Activities associated with day-to-day management of a research study span a broad continuum, ranging from protocol design and Institutional Review Board (IRB) submission, to site co-ordination, recruitment, consenting, enrollment, visit management, budgeting and more. Historically, our research teams managed these activities either on paper or by utilizing desktop databases or even documents, most of which were neither cross-referenced, shared, or indexed for searching. Study management functionality to empower researchers and their teams to rapidly launch and administer their research studies was, therefore, one of the critical features of an enterprise research solution. Because each study may not always require all of these dimensions, researchers valued the ability to configure on an ala cart basis, specific functionality and reusable data capture templates most relevant to the characteristics of the study. Experimentation is the medium of scientific exploration and holds the potential to reveal new discoveries, particularly when traceability is preserved between the experiment data, the specimens on which experiments were conducted, and the activities conducted to obtain the data. Experiments further bridge the translational gap between basic science and the clinic. Most laboratories maintain experiments on paper or in Word or Excel, limiting utility for correlative analysis with clinical variables. Integration of specific experiment management features – sometimes also called LIMS (Laboratory Information Management Systems) – within clinical research systems is a relatively novel concept intended to close the bench to bedside gap in translational research. Few commercially available clinical solutions positioned in the translational research market possess this functionality. Beyond just the annotation of specimens with experimental results, experiment management encompasses use of workflow technologies and reusable templates to standardize experiment design and data capture, as well as enable importation of scientific data generated either by instruments or by primary and secondary tertiary data analysis tools. Beyond organizing data and information, research information systems must also be capable of supporting knowledge management (KM). KM is incredibly broad, but in this context, KM refers to technologies and semantic methods that promote data interoperability and intelligent search through use of controlled vocabularies and ontology, and data modelling structures that support inference, meaning, knowledge representation and distribution. Systems with integrated services that permit crosstalk between standard and local taxonomies and that rigidly enforce hierarchical relationships in process ontology begin to tackle the formidable problem of merging disparate information and data together before knowledge can be gleaned. Moreover, research data can be made inestimably more valuable if researchers are permitted to ask questions in a robust analytical environment where they can explore the possible correlations between the data. Classical data warehousing may not always lend itself well to translational research partly because research evolves rapidly and by its very nature, requires a highly adaptive method for interacting with data to investigate these data from different perspectives. Semantic web methods together with intelligent search may enable researchers to combine often heterogeneous data from disparate sources to answer a question or solve a problem. Mindful of these core criteria and our guiding principle to focus our efforts on the novel and avoid duplicating what already exists, we completed a 6-month-long comprehensive vendor evaluation and package selection of commercially available clinical and translational research solutions that possessed these requirements. From this process, we selected a vendor on the basis of optimal fit between product features and requirements, price to budget comparison, and overall vendor relationship, as well as their willingness to indulge our creativity. In the months that followed, we partnered closely with a visionary investigator to pilot the deployment and collaborated with our partner to customize experiment management features that did not previously exist. We have therefore launched an additional project to deploy this solution for another researcher and have plans for additional deployments into 2011. Collectively, the knowledge and experience we gained throughout this journey has taught us that translational research will not deliver on its full potential unless supported both by investment in informatics and IT and by a collaborative partnership between Research and IT. Additionally, these investments must balance the right people, process and technology and this partnership must be characterized by open communication and realistic expectations. From these efforts will come the potential to leverage IT and informatics to create a dynamic environment of iterative study, experimentation, adaptation and innovation. None.