Data sharing for genomic medicine is gaining ground as a hot topic – running through both the Chief Medical Officer’s (CMO) Annual Report – Generation Genome, and the report published this week by the National Data Guardian (NDG) for health and social care on Developing consensus on data sharing to support NHS clinical genetics and genomics services.
The CMO in her report makes clear that the effective use of genomic data - its collection, storage, and appropriate sharing at scale - is fundamental to achieving the CMO’s ambitions to improve patient outcomes and maintain the UK’s leadership role in genomics. But this no trivial matter, as the CMO acknowledges, given the sheer volume and complexity of genomic information, as well as the ethical and legal context surrounding the use of patient data.
An entire chapter is devoted to the considerations around ‘solving data challenges’ in genomics -from collating different sources of data, to developing the resources to enable the management and analysis of that data. Specific reference is made to the centralisation of genomic variant-level intelligence. This is because pooling and sharing information on variants is critical to understanding their clinical significance - i.e. their association with disease and relevance to a patient’s condition.
As the chapter notes, there are many cases that could be readily solved if better use was made of existing variant level intelligence by centralising variant data from across all UK molecular diagnostic laboratories. Moreover, for ‘unsolved’ patients with extremely rare diseases it will only be possible to make genomic diagnoses through the systematic centralisation of genomic and phenotypic data. Our own work on data sharing to support clinical genomic services, emphasises this point. Data sharing is critical not only to reaching accurate diagnoses, but also to reducing delays in the time taken to reach a diagnosis, and to reducing misdiagnosis, which can result in inappropriate care.
Data sharing – what’s holding us back?
Given the implications for patient safety and care it is reasonable to ask why variant data hasn’t been systematically centralised to date. In 2015 The PHG Foundation and Association of Clinical Genomic Science (ACGS), held a joint workshop to examine the challenges to data sharing. Among the most prominent issues were
- lack of a clearly designated infrastructure for sharing data
- lack of practical support for data curation
- legal uncertainty on the legitimacy of data sharing, which contributes to a lack of confidence by healthcare professionals to share data
It was specifically around the aspect of legal uncertainly that our organisations engaged with the NDG.
In her 2016 Review on data security, consent and opt-outs the NDG acknowledged genomic data sharing and the work of the PHG Foundation and ACGS, but without going into detail. Subsequently in October 2016 an evidence session was held by the NDG’s Office with collaboration from the ACGS and PHG Foundation. A range of stakeholders including clinical geneticists and clinical scientists, legal and health policy experts explored the reasons for the legal ambiguity surrounding genomic data sharing within the NHS for routine service delivery, and how the NDG might help address concerns around legitimacy.
A resulting paper from the NDG suggests the need for (i) further work to explore appropriate consent for routine NHS clinical genetics and genomics services and (ii) a Privacy Impact Assessment (PIA) process to examine the suitable arrangements for genomic data sharing within the NHS.
While the paper is not, at this point, a policy position or official advice from the NDG, it is an attempt to help the genomics community gain clarity.
Data sharing for patient benefit – a brighter future?
There is no doubt that the importance of genomic data sharing is high on the agenda. The reports by the CMO and NDG are testament to the recognition that responsible and safeguarded genomic data sharing in the interest of patients is imperative and so the challenges to data sharing must be addressed. Good progress is being made by the professional genomics community, with ongoing engagement on this issue. Many laboratories have been sharing variants within the NHS consortium within the DECIPHER database. The professional groups have also agreed to, and rapidly supported the adoption of (the American College of Medical Genetics and Genomics) guidelines for classifying variants. The use of common guidelines helps to improve the consistency of variant interpretations and to compare interpretations and resolve differences when these arise.
Plans afoot to redesign genomic services in England present a great opportunity to ensure the most effective systems to support data sharing are embedded in future services to allow patient data to be interpreted with the best and most up-to-date evidence available. However, many questions remain. For example,
- What support and resources will be put in place to facilitate data sharing across future genomic laboratory services?
- What will be done to ensure existing variant-intelligence is integrated into any future systems for sharing knowledge?
- How will the proposals set out in the NDG’s 2016 Review (recently accepted by the Government) to introduce an opt-out for data sharing apply to genomic data sharing across laboratories?
Whilst these points are yet to be resolved, if there was ever an opportune time to address data sharing, it is now – generation genome depends on it.