Innovative regulation: keeping pace with genomics and software devices
16 December 2021
The Medicines and Healthcare products Regulatory Agency (MHRA) invited views on possible changes to the regulatory framework for medical devices in the United Kingdom (UK). The PHG Foundation welcomes the review and has provided a comprehensive response which can be viewed here.
In mid-September, the Medicines and Healthcare products Regulatory Agency (MHRA) published proposals for amendments to the existing medical device regulations. The welcome proposals were needed following the UK’s Regulatory Horizons Council and Baroness Cumberlege’s reports, where current safety and transparency standards were deemed inadequate. The MHRA proposes a wide range of updates to the entire framework for medical and in vitro diagnostic devices.
Do genetics tests require a tailored approach?
There is no doubt that medical device technology has moved at a rapid pace in recent years. Some devices allow consumers to unlock health and genetic information without professional oversight, whether through personalised healthcare apps or home testing kits, and are challenging to regulate. Consequently, the new proposals on software and genetic testing devices are some of the most innovative and exciting developments set out for consultation.
It has been twenty years since the Medical Device Regulations 2002 (MDR 2002) were transposed into UK law. ‘Knowledge’ of personal genomics is no longer restricted to those with rare disorders, it is relatively easy to send off your samples to companies offering DNA analysis for a variety of reasons, including health assessments. Therefore, it is necessary to ask: are genetic or genomic tests exceptional (compared with other forms of medical testing or devices) and should they require professional oversight?
The application and significance of genetic tests vary widely; some may be more akin to non-genetic tests and others have profound implications for patients and their families. The current regulations treat such devices as non-exceptional. They are categorised as low-risk and hold no device-specific requirements, leading to concerns that current regulations are inadequate. However, the new proposals include an amendment that would require users to be informed as to the ‘nature, significance and implications’ of such tests.
Due to the potentially wide range of implications arising from genetic tests, the question of what information should be provided may need case-by-case consideration. For example, companion diagnostic tests are specifically mentioned in the proposals and are notably tricky to regulate. They test for biomarkers that help predict a patient’s likely response to therapeutic drugs and have typically been classified as low risk. However, their classification and the level of clinical evidence required to place these products on the market has been criticised, making an argument for exceptional treatment. The requirement for further information should also help mitigate any physical or psychological harm that might result from reliance on them. However, the proposed provision concerns information for patients only and could have been bolstered by requiring that tailored information also be provided for health professionals, as some tests are highly specialised and front-line staff may require this further support to properly interpret their implications for patients seeking advice, particularly for self-administered tests. This should be introduced as an MDR legal requirement as it would support healthcare professionals to meet their own legal obligation to tailor information to individual patients, both before and after testing.
Therefore, a nuanced approach is needed: some genetic tests may need device-specific requirements due to the risks that they present, but not all. We welcome the proposals’ approach, whereby device classifications are tailored proportionately to their risk. However, classifying them on the basis of presence of ‘risk’ may oversimplify the issue; perhaps ‘risk’ as ‘severity of potential harm’ might be a better factor for classification than ‘presence of risk’ alone. The PHG Foundation’s interpretability by design framework for black box algorithms is largely transferable to the classification of genetic tests, where factors such as incompleteness, ground truth and risks of use could further aid proportionate classifications.
Getting the balance right for software and AI
Another dramatic change since the existing regulations came into effect has been in the development of software and AI medical devices. Changes since the MDR 2002 include personalised health tracking via easily accessible apps that claim to provide information on, or monitor consumers’ health– some of which use DNA analysis for a more targeted approach. Moreover, software devices (SaMDs) are wide-ranging; some process data via the traditional input-output method and others, such as machine learning models, present a moving target for regulators as the software adapts in response to data. Consequently, targeted regulation that can keep up with novel software developments is needed.
The new proposals make a good attempt at navigating the delicate balance between making devices accessible and protecting consumers from harm. The MHRA’s innovative methods give it the flexibility to strike an appropriate balance as it grapples with the almost borderless and closed-loop market of health apps and consumer-driven genetic testing. To increase patient safety, the proposals have revised the definition of ‘placing on the market’ to include app store and website SaMDs. Likewise, tougher restrictions on distance sales should bring cross-border genetic testing within its scope. To mitigate the impact of more stringent regulation, the MHRA has proposed some innovative routes to market: the airlock classification rule would temporarily permit ‘game-changing’ software to circulate on the market under stringent requirements, and the Pathway for Innovative MedTech’s focus on small patient populations promises to ease accessibility of novel software devices for patients such as those with rare genetic diseases. However, there may be resourcing challenges to overcome. App store and website restrictions are likely to require significant resources to police effectively, and given that there are only three UK approved bodies to assess the safety of such devices, there may also be an impending bottleneck that could considerably undermine the MHRA’s efforts to maintain device accessibility.
Overall, the proposals suggest vital updates to the framework for genomics and software devices as their application expands beyond traditional care settings, to home use with less professional oversight. The updates and tailored rules for software and genetic testing devices are necessary because of their potentially wide-ranging harms for individuals. It will be very interesting to see which of these innovative proposals are adopted in the coming months.