The UK Biobank

In this issue, Natalie Ryan, Martin Rossor and Nick Fox commemorate the centenary of the death of Alois Alzheimer by looking at how his eponymous disease has evolved from a rare, early-onset dementia to a major focus of global public health concern (Ryan et al., 2015). Central to our current understanding is that the disease processes are initiated—and may be amenable to therapeutic modification—decades before their clinical manifestations, a paradigm that is being extended across an ever-wider range of chronic brain disorders of late life. This has led to a new challenge for neurology: identifying people who are at risk early in order to allow interventions that delay or prevent later clinical disease. Valentina Escott-Price and colleagues describe an important step towards making this possible based on the discovery of polygenic risk markers meaningfully predictive of development of the disease (Escott-Price et al., 2015). This important development was made possible by the aggregation of very large genetic and clinical outcome datasets. The UK Biobank project has just taken another major step towards making it easier for researchers everywhere to make such discoveries in the future with the announcement of release of the uniquely large, openly accessible brain (and body) imaging dataset linked to extensive genetic, phenotypic, exposure and health outcomes data. UK Biobank is a prospective study of 500 000 UK participants recruited in middle-age during 2006–2010 (Sudlow et al., 2015). Extensive data were collected at baseline from all participants on their lifestyle, environment, personal and family medical history. Participants also underwent a wide range of physical and cognitive measures and provided samples of blood, urine and saliva for storage and future analyses. Samples from the participants have been genotyped on an Affymetrix chip including 800 000 variants (including those known to be relevant to neurodegenerative conditions), that already are being imputed to millions of further variants. Long-term follow-up of health outcomes is occurring through linkages to national health data and web-based questionnaires. An important aspect of the study is that the data acquired are progressively being enhanced in conjunction with repeats of their baseline assessments. The most recent major enhancement for UK Biobank promises to transform its potential for neurology with an ambitious plan for clinical imaging of 100 000 of its participants (see descriptions in UK Biobank, 2015). A pilot over 2014–15 has demonstrated the feasibility of achieving this goal. As this editorial goes to print, funders are considering approval for the full study. However, the first instalment of ∼4500 datasets including magnetic resonance images of the brain, heart, and body, DXA images of the bone and joints, and carotid ultrasound imaging has already been released for use by the research community. The data are of very high quality, have been obtained with rigorous attention to consistency of acquisition, and are linked to the extensive phenotypic, genetic and other data already gathered in the study. The UK Biobank is open to all scientists with research questions of public health interest to encourage independent new discovery. It promises a unique and welcome resource in a number of ways. The size of the first release and follow-on plans are unprecedented for research imaging data. It makes available data acquired from other imaging modalities addressing organ systems outside of the brain on the same individuals. All data were acquired at a single site using the same scanners and consistent imaging protocols for all subjects. Moreover, the imaging data in UK Biobank are linked to the extraordinary range of additional information described above. This population-based resource will complement more detailed disease-specific studies, such as the pioneering Alzheimers Disease Neuroimaging Initiative (ADNI), and the advanced high field diffusion MRI and structural neuroimaging on more than 1000 well-phenotyped younger people developed through the US Human Connectome Project (HCP). The UK Biobank evaluations and scanning are conducted by a dedicated team, currently operating out of a centre in Cheadle, near Manchester, England. However, design of the imaging protocols, developed over 5 years, technical advice and supportive oversight have been provided by an expert Imaging Working Group of UK imaging researchers, supported by international expert advisory groups involving >100 additional scientists. Input from professional bodies, including the Royal College of Radiologists and the Society and College of Radiographers, as well as from UK Biobank’s independent Ethics and Governance Council, has helped to guide a thoughtful approach to professional, societal and ethical issues raised by the project. The neuroimaging dataset, acquired during a ∼35-min long 3 T protocol, includes isotropic T1, T2 FLAIR, susceptibility-weighted T2*, diffusion, resting state and a task-based functional MRI testing brain responses to emotional faces. In addition, >700 image-derived phenotype measures describing, for example, brain and individual brain-atlas defined anatomical region volumes, will be available. As scientists throughout the world perform additional analyses on these data, their results will be added to create a growing database of quantitative image measures, along with descriptions of the methods used. This promises to create a large, curated dataset allowing direct comparisons of analytical outcomes from different methodological approaches for the first time. However, more important is that the release of the imaging data affords an opportunity to address a new range of questions. A recent report relating resting functional MRI differences between individuals to measures of trait and lifestyle conducted with a part of the HCP dataset illustrates how these new data should illuminate the brain basis of human personality, cognition, behaviours and diseases with a new level of precision (Smith et al., 2015). Immediately, it provides enough data to test—and replicate—genetic associations with imaging phenotypes more widely and with greater robustness than previously possible. The relationships between imaging phenotypes across organ systems, for example, exploring the heart–brain axis or the relationships between measures of large and small vessel disease vascular disease, will become possible. Data relevant to cognition, mental health and systemic disease promise to enable a rapid expansion of research into biological bases for behaviour. With time, a growing number of cases of incident diseases of later life in the imaged population will afford the discovery of imaging and related measures that explain future disease risk and the emergence of hypotheses regarding potential new treatments. The precision and depth of the imaging phenotype combined with the mid- to late-life population and its size promise to shed light on what is happening in the brain years before presentation with dementia, stroke or other disorders. These data will contribute to raising the quality of neuroimaging science. The cost and complexity of neuroimaging studies is challenging for researchers with limited budgets. As a consequence, experimental imaging studies have most typically included small numbers of subjects, risking both type I and type II error (Button et al., 2013). Associations with genetic, lifestyle or environmental exposures, which typically demand larger datasets to identify the modest effects expected, have generally not been addressed with well-designed prospective studies, although some meta-analytic approaches have been possible (Thompson et al., 2014). This first data release is an important first step towards resolving these issues for many kinds of questions. At the same time, we also believe these new UK Biobank data will enable better design and interpretation of the much smaller, focused imaging studies of well-defined populations. Such studies will now benefit from the rich context of the UK Biobank dataset. This ‘big science’ project also contributes to individual investigator-led imaging-based discovery science by lowering the barriers to asking new questions. With the addition of this dataset to the other imaging datasets that also are openly available, researchers can test hypotheses rapidly, armed with no more than a good idea, appropriate software (much of which has been developed in academic centres and is downloadable at no cost) and access to the necessary computational resources. As we have seen previously with ADNI data (and are beginning to see with the HCP data), the UK Biobank data release should trigger a major step change in the number of informative new studies. In a related development, the UK MRC Dementias Platforms UK (DPUK) was funded over the last year to develop an informatics platform to promote access to data from >20 other prospective cohorts available for researchers. DPUK also intends to work with UK Biobank to allow at least 10 000 of the future imaged participants to be recalled for follow-up imaging, cognitive testing and other measures. This should allow measures of brain structural and functional change over time to be used as additional and, arguably, much more sensitive, prognostic indices. Our enthusiasm for the vision and ambition of the UK Biobank should be apparent. We have both had the privilege to be involved, although, individually, our contributions were modest. It took many, many people to reach where we are today. The first release of imaging data from UK Biobank represents a major success for the scientific community. We expect it to accelerate new discovery of risk factors and early disease mechanisms that will help neurology move towards a future focused on maintaining ‘brain health’ as much as on diagnosing and managing disease.