Response to "Principles of the translation of ‘omics’-based tests from discovery to health care"
In August, the National Health and Medical Research Council in Australia called for comment on a draft of their "Draft Principles for the translation of ‘omics’-based tests from discovery to health care". The Human Variome Project International Scientific Committee submitted the following response.
Dear Professor Ward,
On behalf of the Human Variome Project International Scientific Advisory Committee, we congratulate you on the work of the National Health and Medical Research Council (NHMRC) Human Genetics Advisory Committee and the NHMRC Secretariat in producing the Principles for the translation of ‘omics’-based tests from discovery to health care and we welcome the opportunity to respond as part of the document’s public consultation. These are complex issues that many health systems and research funding bodies around the world are grappling with. We are pleased to have been invited by the NHMRC Secretariat to share the expertise of our members and provide an international perspective to these principles.
As you are well aware, the Human Variome Project is an international non-governmental organisation that is working to ensure that all information on genetic variation and its effect on human health can be collected, curated, interpreted and shared freely and openly.
The Human Variome Project provides a central coordinating function for national and international efforts to integrate the collection, curation, interpretation and sharing of information on variation in the human genome into routine clinical practice and research. We are an active and growing Consortium of over 1,100 individual researchers, healthcare professionals and policy makers and organisations from 81 countries that collaborate to develop and maintain the necessary standards, systems and infrastructure to support global scale genomic knowledge sharing. The Project itself is not directly involved in the development and operation of physical data storage and sharing infrastructure; that is the responsibility of international disease groups, national consortiums/health systems and individual members. Rather, the Project exists to assist these groups by:
- Collaboratively developing technical standards and harmonised, common approaches so that data from different sources can be easily shared in an interoperable manner that is sensitive to the ethical, legal and social requirements of both the data sources and consumers;
- Coordinating an international platform to facilitate discussion of Genomics in global health with the aim to foster necessary professional interaction and debate in the area of genomics, global health, and service delivery and safety;
- Linking world leading professionals and institutions with genomics professionals, researchers and academics in all parts of the world, facilitating knowledge exchange and interactive debate;
- Establishing a global evidence base for knowledge sharing in medical genetics and genomics and bringing relevant issues to the attention of Ministries of Health, Science and Technology and Education.
General comments on the document
The document is clearly well thought out and addresses the relevant areas of this complex issue. It should provide a useful basis for more detailed work on standards in this area. We are particularly pleased to see that this document presents a measured approach to the challenges presented by the rapidly developing technological landscape that we have found ourselves in and avoids “genohype” by explicitly recognising that the vast majority of the clinical applications of human genetics and genomics knowledge are, and will continue to be for some time, monogenic in nature. We are also pleased to see the explicit recognition that the translation of omics into healthcare requires a validated evidence base. In a rapidly advancing field like ours, there is always the danger of the technologies driving what we do rather than the health needs of our patients and societies.
However, in our opinion, the definition of omics testing utilised throughout the document is too restrictive: the definition specifies that and omics test is an assay that is “interpreted by a fully specified computational model.” This definition minimises the level of human knowledge and effort that goes into interpreting these tests, both in terms of interpreting the results of individual tests themselves and in constructing and curating the data resources that are required for the computational tools to function.
The document as read seems to be aimed solely on the development of omics tests for novel targets and is silent on how these principles should be applied to existing clinical tests carried out on older technology platforms. Will these tests be required to have their clinical validity and utility assessed again?
We would also recommend increasing the implied weight of each principle through the use of the word “must” in place of the word “should.”
Comments on individual sections
The introduction to the document states, “The full potential of omics will only be realised when big data can be synthesised into a usable format.” While this is true, format is only part of the problem, and a relatively downstream part of the problem at that. Assembling these “big data” datasets will require vast amounts of clinically validated data to be made available in a responsible manner through data sharing at the international level. Increasing rates of global mobility mean that many high-income countries with established health systems and functional genetic health services, such as Australia, will be treating patients from a diverse range of population groupings. Knowledge of particular genetic variants that are specific to certain population groups and that impart either protection from or increased risk of particular diseases will be vital to provide effective diagnosis and care. This kind of information can only be generated at the scale required in the countries from which these population groupings originate during the course of routine clinical testing.
In making data available for future use and reuse internationally, data providers will need to be aware of the ethical, legal and social issues that are inherent in data sharing relationships, as well as the associated equity and access issues that data sharing raises. Principle 4.2 addresses these concerns, but they are of significant import, and genuinely addressing them is so inextricably linked to the successful and responsible translation of omics testing into clinical settings that we feel they should be included in the introduction.
As an aside, the Human Variome Project exists to assist data providers, particularly those within national health systems and with a clinical focus, to develop their organisational and national capacity to enable exactly this scale of sharing.
2. Governing Principles
Beginning with a set of ethical principles sets the right tone for the overall document. We are particularly pleased to see the inclusion of principles covering collaboration and equity. These principles, particularly equity, are very important for the Human Variome Project as we work with health systems in low- and middle-income settings. We are pleased to see the document affirms that access to new tests should be based on health needs and not the ability to pay. This is of concern in many parts of the world where health systems are not as strong as in Australia. Direct to consumer genomic tests, especially those of a non-invasive nature, are particularly concerning, as while these sorts of testing services may be available to those people who are able to afford them, appropriate treatment and counselling services may not be locally available to provide adequate follow up after the results of the tests become known.
Education is also a very big issue for the Human Variome Project and we are pleased to see it included in the operating principles. We have found that almost all countries, even those with established health systems, are grappling with issues of education and the associated issues of establishing and maintaining a workforce that is fully across all aspects of these new technologies. It would be useful to articulate these aspects in the document, including, and we cannot stress enough the importance of this enough, the data management aspects.
We would also recommend articulating the specific aspects of public education that these new technologies will necessitate. These will include ensuring that patients and the broader community are aware of their rights, are able to distinguish good quality services from poor quality, ensure they are adequately protected and most importantly, be able to fully understand and assess their risk of disease and that of their family members.
We would also recommend including an additional operational principle: Interoperability. As the document makes clear, the scale of the challenge of omics based testing requires collaboration nd data sharing on an extremely large scale. This level of activity will only be possible if global standards for the description and processing of data are followed. We suggest the following language for this principle:
Interoperability – the complexity and multi-disciplinary nature of omics data requires compliance to data standards, including minimal information, use of structured vocabularies and ontologies to enable efficient data integration and data interoperability.
We applaud the document’s recognition of the tension inherent in the need for large amounts of clinically validated data to validate new omics tests and the requirement for new tests to be clinically validated before they can be used and generate data. To assist in addressing this tension, we would point out that Australia, like much of the world, could do more to facilitate the sharing of clinically validated data that has been generated on testing platforms that have already been clinically validated and are utilised every day in routine genetics and genomics healthcare. Likewise, the Human Variome Project believes that much more research needs to be done worldwide in relation to comparative effectiveness research in the practice of genomic medicine.
Validating genetic and genomic tests is difficult and time consuming. The Human Variome Project has many years of firsthand knowledge of this exact challenge and is well aware of the scale of data required to meet this challenge properly. Two international groups that the Human Variome Project promotes that have done extensive work on clinical validation of variant interpretation are the CFTR2 Project and the International Society for Gastrointestinal hereditary tumours. Both these groups have assembled impressive collections of molecular and clinical data, primarily resourced from clinical as opposed to research settings, and used them to determine the pathogenicity of a number of variants that are causative of cystic fibrosis and colorectal cancer respectively. These models can be readily adapted for many other diseases, but doing so will require more support made available for data sharing infrastructure, data curation, and the organisational structures to support these activities, at both the national and international level.
The Human Variome Project readily supports the document’s interpretation of the central issues in the translation of omics tests, namely, people who are offered a test should be able to:
- exercise choice in what they learn;
- decide who has access to the data and resultant information; and
- be protected from misuse of results.
4. Domain specific principles
The Human Variome Project supports each of the domain specific principles, and, is particularly pleased to support principles 1.3 (although we believe that, for the avoidance of doubt, specific reference should be made to the need for validation of the external data sources used as inputs to these models), 3.8, 4.2 and 4.3.
However, we believe the document could be improved by consideration of the following points:
- Domain 1: The document would benefit from an additional principle in this domain that deals with the use of common reporting standards so that data, once shared, can be universally understood and reused.
Reporting standards are crucial in omics to ensure data interoperability.
Clinicians and researchers (laboratory, bioinformaticians, statisticians).
- Domain 1, Principle 1.4: Why is it just research labs that “should” have data quality control standards in place? Why not clinical labs too? Specifying the laboratory class opens a loophole that could be exploited. We would also suggest that the document should promote the use of a core set of control samples that are readily available from cell line or DNA repositories to audit and assess uniform data quality control measures.
- Domain 2, Principle 2.2: A requirement for institutional support for clinical trials in regards to ethical, legal and social issues should be made explicit. Principle 2.8 deals with some ELS aspects of this type of testing, but not all. These principles should make it clear that ELS considerations are of critical importance at all levels of omics testing.
- Domain 3: The utility of data sharing to aid clinical practice should be mentioned, specifically to assist genetic and genomic pathologists interpret the results of tests and provide clinically relevant information to the requesting physician. This is especially relevant to Principle 3.4.
- Domain 3, Principle 3.4: While it would be useful to have uniform interpretive reporting standards, especially for VUS, a requirement to share data on variants that have been characterised in accredited laboratories would enable laboratories to more effectively collaborate on the interpretation of variants and distinguish true VUS from interpretable variants that are deemed to be of unknown significance due to a lack of data. We note the presence of the technical capacity to do this currently exists within Australia through the Human Variome Project Australian Node and we would urge the Council to support this valuable resource.
- Domain 3, Principle 3.7: The rationale and the Principle here do not seem to go together. The Principle seems valid, but the rationale does not seem to be related to the “healthcare” domain.
- Domain 4: We echo the points we raised under Domain 3, Principle 3.4: data sharing should be sharing of all data, including data generated during the course of routine clinical practice, not just data from research projects.
- Domain 4, Principle 4.1: Footnote 10 mentions only a single example, and not one from the Australian context. A more relevant example would probably be more illustrative for the intended audience of the document. The RCPA/HGSA/HVP database standards project (referred to elsewhere in the document), although still in development, would work well.
- Domain 4, Principle 4.3: The Human Variome Project maintains a curated list of data standards that might be useful to mention here.
- Domain 4, Principle 4.4: The rationale should specifically mention data backups and the need to ensure that the same policies apply to these as they do to the active/production data. This is an often overlooked point and it bears emphasising where possible.
5. Case Studies
- Case Study 1: The third commentary section states that “non-validated research findings should not be deposited into data repositories.” Some research repositories do accept the deposition of appropriately labelled data that clearly identifies its validation status. We would argue that this should be the case for all repositories and that researchers should be encouraged to submit such data.
Once again, we commend the NHMRC for tackling these important and complex issues and for engaging in genuine public consultation.
Timothy D. Smith
Coordinating Office Liaison - Human Variome Project International Scientific Advisory Committee
on behalf of
Human Variome Project International Scientific Advisory Committee*
* The Committee declares a potential conflict in that Ingrid Winship is a member of the Committee and a member of the NHMRC Human Genetics Advisory Committee. The Committee were aware of this during the preparation of their response.