From DNA to Personal Health Record

Transcription

1 From DNA to Personal Health Record Report of the 2 nd ReDaPeD workshop 17 June 2013, Grand Hotel Karel V, Utrecht Commissioned by the Netherlands Genomics Initiative embedding project ReDaPeD (Towards Responsible Data Management for Personalised Diagnostics) Project Leader: Dr Maud Radstake, Centre for Society and the Life Sciences In cooperation with: Dr Terry Vrijenhoek, Centre for Genome Diagnostics Report by: Dr Daan Schuurbiers, De Proeffabriek

2 Introduction What are the conditions for responsible management of genome data in light of the rise of personal health records? That was the central question in the second workshop of the ReDaPeD project (Towards Responsible Data Management for Personalised Diagnostics) on 17 June 2013 in Utrecht. 1 Experts in genomics, bioinformatics, clinical genetics, patients perspectives, ethics, law and social science gathered to discuss responsible data management for personalised genome diagnostics. The workshop built on the main recommendation stemming from earlier findings of the ReDaPeD project: that the technical, legal and moral frameworks for responsible data management should be defined proactively in a transdisciplinary forum involving researchers, ICT professionals, medical experts, patients and citizens. This workshop aimed to establish such a hybrid forum. There is a growing body of literature on the ethical, legal and social aspects of -omics research and its translation to health care - e.g. legal aspects of informed consent, ethical aspects of next-generation sequencing, public perceptions of genome technologies etc. But overarching questions of data management remain relatively unexplored. The workshop therefore followed genome data through the cycle of diagnostics and research: how do nextgeneration sequencing data move from the patient in the doctor s office through the laboratory, the server room and the clinical geneticist s office back to the patient? And what does that imply for their collection, storage and use? From DNA to dossier: presentations and discussion Introductory presentations on the state of the art in next generation sequencing, patient participation and personal health records set the stage for the discussion. 2 Dr Maud Radstake, ReDaPeD project leader, introduced the central topic of the workshop: to reflect on the diagnostic application of next-generation sequencing (NGS) in the context of the emergence of personal health records.dr Radstake suggested that the two developments may converge around questions of data management. The central questions for the workshop were therefore: how can we envisage the integration of our DNA data in a digital personal health record? What does that imply from a technological and societal perspective? Which decisions are being made on storage and access to these data, and who is involved in these decisions? 1 The ReDaPeD project is funded by the Netherlands Genomics Initiative (NGI) and carried out by the CSG Centre for Society and the Life Sciences in collaboration with the Centre for Genome Diagnostics (CGD) and De Proeffabriek. The main aim of the project is to enable interdisciplinary collaboration on responsible data management for personalised diagnostics. Activities include a literature review, a series of expert interviews and three multi-stakeholder workshops. 2 Workshop presentations are available on request. 2

3 Results of the pilot project CARDIO Dr Terry Vrijenhoek subsequently presented recent experiences of the Centre for Genome Diagnostics with the diagnostic application of NGS in the pilot project CARDIO. He noted that much has happened since the first research papers appeared on the potential medical relevance of NGS. Due to the improved speed and accuracy of sequencing it is becoming easier to concretely apply NGS in a diagnostic setting - although it is not yet part of routine procedures and is the subject of much discussion. As the technology becomes more accurate and affordable, legal and ethical issues such as the question of reporting back incidental findings also become more common and pertinent. The discussion following a recent report of the American College of Medical Genetics and Genomics (ACMG) 3 is a case in point. The ACMG report suggested that laboratories performing clinical sequencing should actively seek and report relevant mutations within a specified list of genes. But the question if, how and when to report unsolicited findings is still a topic of heated debate. 4 The objective of the CARDIO project was to analyse genome diagnostic procedures in different academic centres in the Netherlands. Samples of ten patients with hypertrophic cardiomyopathy were sent out to the eight clinical genetics centres for diagnosis by way of NGS. All centres returned the right diagnosis even though they all used different platforms to get their results. In addition to studying the technological dimensions of NGS, CARDIO also explored organisational and policy issues. One important question is how to address the additional findings that are not directly related to the diagnosis but might be clinically relevant. How to communicate to patients the probabilities that these findings might be medically relevant? And can additional findings be used for further research? According to Dr Vrijenhoek, these questions point to the need for robust ethical and legal frameworks that address the grey area between research and care. As an example, legal analysis in the CARDIO project pointed out that data obtained for diagnostic purposes could in particular cases be used for research under the heading of experimental diagnostics. The right to know and the duty to inform Dr Cor Oosterwijk of the Dutch Genetic Alliance (VSOP) presented his views on patients right to know and the duty to inform. He brought into memory the heated debate on the management of genetic data a few years ago when the implementation of a National 3 American College of Medical Genetics and Genomics (2013) ACMG Recommendations for Reporting of Incidental Findings in Clinical Exome and Genome Sequencing. Available at: Anticipated_Recommendations_on_Incidental_Findings_in_Clinical_Exome_and_Genome_Sequencing.pdf 4 Workshop participants noted that the Dutch Society for Human Genetics (NVHG), the Public Population Project in Genomics and Society (P3G) and the Dutch Health Council (Gezondheidsraad) have already responded to the ACMG recommendations, or are preparing such a response. 3

4 Electronic Patient Record (EPD) was in the news in The Netherlands. Questions included: what happens to our genetic data? How to ensure responsible use of those data in the absence of a relationship of care? There was a certain ambivalence between enthusiasm for data sharing for research purposes and privacy concerns. Dr Oosterwijk noted that some are more worried about privacy and informed consent than others. He pointed out that patients generally want to be involved in decisions that concern them. 5 Patient organisations are therefore very willing to be engaged in research and policy debates. Dr Oosterwijk argued that the sooner patients are given the choice to have a diagnosis, the better. This prevents diagnostic delay, provides early access to research, treatment and reimbursement, avoids uncertainty and may guide reproductive choices. He also addressed the question of unexpected findings, noting that it can be very difficult for patients to be confronted with unexpected findings but allows them to prepare themselves and make an informed decision on the appropriate course of action. Dr Oosterwijk therefore welcomed the ACMG recommendations: if one can easily identify medically relevant genetic mutations, it could be considered a moral duty to seek and communicate those findings to patients. 6 Dr Oosterwijk s position sparked a debate during the workshop since the moral obligation to convey certain information may clash with patients right not to know. Rights and responsibilities vary across contexts: it may be morally reprehensible if a doctor does not notify the patient of medically relevant findings, but the relation between researchers and research participants is different. A discussion ensued about the responsibilities laid down by current laws and regulations and the validity of the norms underlying those laws given recent technological developments. Law is, in effect, solidified morality: it enshrines the societal values and norms that should guide our behaviour. As our ideas about these norms and values change in light of new developments (consider the ways that NGS blurs the boundaries between research and diagnostics for example), one could argue that the legal arrangements may also need to be revisited. Of course, changing the law is a complex process. Apart from the practical difficulty of rewriting Dutch or European laws (and the protracted political and legal processes that come with it), opening up existing laws and regulations will lead to much confusion about the rights and responsibilities of doctors, researchers and patients in current medical practice. According to Dr Oosterwijk (and several workshop participants) however, the traditional ethical and legal norms governing access to medical data and health information from research urgently need to be rethought. The current system is based too much on tutelage, 5 Dr Oosterwijk noted that the question who should be involved in policy decisions on access to genetic data is becoming more difficult to answer since the boundaries between patients and non-patients are blurring - we all have genetic variations that might at some point turn out to be medically relevant, so these decisions ultimately affect us all. 6 Dr Oosterwijk suggested that the term unexpected finding is somewhat misleading: if one decides to take a broader perspective then much less may be unexpected. The term incidental findings might therefore be more appropriate. 4

5 not on empowerment of the patient. The vision of patient organisations is that the patient is the primary owner of the data, so it is the patient who should be able to decide what happens to those data. From that perspective, a doctor or researcher should not know more about the data than the patient. This is why Dr Oosterwijk argued for the decompartmentalisation of research and diagnostics. He concluded that patients have a fundamental right to access the genetic data and information stored in medical and research records. Patients should be able to decide for themselves what they bring in and what they take from biobanks. Conversely, medical doctors and researchers have a moral duty to inform the patient of incidental findings with life relevance, irrespective of whether the findings are the result of research, screening or diagnostics. Personal health records As a third constitutive element for the overall discussion, Ms Bettine Pluut presented recent developments in personal health records in The Netherlands. Personal health records are lifelong tools for managing relevant health information and promoting health maintenance via an interactive set of electronic health information and e-health tools. Ms Pluut pointed to several trends in e-health, noting that personal health records are but one step in the digitization of health. She highlighted the current debate about the choice between electronic patient records and personal health records, explaining the differences between the two. Patient records are mainly directed by the care provider, centrally organised and designed to couple health care systems. By contrast, personal health records are focused on the individual patient and access to the dossier is organised by the patient. The main difference underlying discussions about implementation of electronic patient records versus personal health records is thus about the mandate: should the individual patient be primarily responsible for the dossier, or should it be the care provider? One important question is to what extent patients are capable of managing and understanding all the information contained in their personal health record - capabilities vary across different groups within society. Ms Pluut did not advocate a specific choice but suggested that doctors and patients should engage in dialogue about the ways in which information is to be shared. Ms Pluut highlighted several themes that characterise the discussion on personal health records. They offer opportunities for improving the quality and completeness of health records, building trust in the medical profession, improving patient-doctor communications, lowering the costs of health care, offering logistical support, enhancing compliance, encouraging behavioural change and prevention and improving the availability of medical data for research. She also identified a number of factors that influence the adoption of health 5

6 records including personal factors such as e-health literacy and communication preferences; technological factors such as the expected use, complexity and expected benefits; health factors such as attitudes towards self-care, complexity of the disease and the treatment; and environmental factors such as incentives and social influences. Ms Pluut concluded by identifying several important themes in the debate on personal health records, noting that public concerns over privacy are much more acute in the case of medical data as opposed to other forms of data (privacy concerns practically meant the end of the national electronic patient record in The Netherlands, while we happily put our personal information on Facebook). Another point of contention is whether the maintenance of health records in e-selfcare can be made obligatory. Are health records a personal service or do they also embody an obligation? This is related to the question how much can be asked of a patient. To what extent is the patient capable of interpreting, processing and managing the information in the health record? Debate over the choice between centrally organised patient records and individually organised personal health records will presumably continue for some time. Ms Pluut advocated a sensible combination of the two. Follow the data After lunch, Terry Vrijenhoek presented the case of patient 9, following one of the samples from the CARDIO project through the diagnostic cycle from the patient in the doctor s office through the laboratory, the server room and the clinical geneticist s office back to the patient. What data are being generated, and what information does that provide, as the sample moves through the diagnostic process? The cycle starts with a patient who enters the doctor s office with a medical condition, in this case arrythmia. The treating physician may suspect a genetic disorder, possibly hypertrophic cardiomyopathy or Fabry disease, and refers the patient to a clinical genetics centre for genome diagnostics. After patient counselling and signature of the informed consent form for the use of DNA for research purposes, a genetic test is requested. Before the introduction of NGS, the clinical geneticist requested a number of individual tests for suspected mutations on a request form. With NGS-based diagnostics, the requesting physician can simply request genome diagnostics for a particular disease instead of individual gene tests. The tests are subsequently run in the laboratory. How exactly the test is being performed varies between labs. Some labs sequence the entire genome and filter the relevant genes; others sequence a limited number of genes. Interestingly, even though different labs return the right diagnosis, they inevitably identify different variants because of the use of different equipment and technologies used. The sequencing machine creates images that plot nucleotides; the images are transferred into text files containing sequences of the nucleotide 6

7 letters A, C, T and G. The images are destroyed immediately after sequencing, but the text files along with the necessary metadata - are stored: these files are the raw data from which every diagnostic analysis begins. 7 The sequencing machine produces short reads : strands of nucleotides. Individual reads are positioned by comparing them against a reference. The reads are then analysed for mutations: the variations in nucleotide sequence are subsequently used for diagnosis. Of course there may be errors in each of these processing steps. There are quality criteria that determine when one can justifiably conclude that variations to the reference are indeed variants (and not a sequencing error). The art of genome diagnostics is to find the causal mutation (or deletion, or duplication) among the hundreds of thousands of sequenced nucleotides although the determination of variants is increasingly outsourced to automated software. Following this procedure, there was very high consistency in the diagnostic conclusions for patient 9. However, one centre missed a pathogenic mutation in a single patient due to low sequence coverage and this led to prioritisation of another variant as potentially pathogenic ( greedy interpretation ). Upon detection of the causal variant, the clinical geneticist receives a letter with the result, which is communicated to the patient along with a proposal for treatment. But the cycle doesn t stop there. There are several points of entry and exit. Feedback to the patient might include an advice for further testing of family members, who can then initiate a diagnostic procedure with their treating physician. Also, patients might request access to their data and submit it to commercial providers for further testing. The data could also be used for research purposes: the combined data from each of these individual diagnostic tests could lead to the identification of hitherto unknown variants. Dr Vrijenhoek s presentation pointed out that genome diagnostics presents significant opportunities for diagnosis and options for treatment, but there are still important questions of harmonisation, quality control, informed consent and patient involvement. Plenary discussion NGS data and personal health records Participants subsequently explored relevant considerations and decision-points about the storage, access and use of NGS data in light of their potential inclusion in a medical or personal health record. The following main questions and concerns were discussed (please see the appendix for a more elaborate description of the arguments, questions and concerns put forward during the discussion): 1. What data should be included in personal health records? What is to be considered relevant data: the sequencing readout, the analysed data, all variants or only those variants 7 Because of the earlier processing steps, one workshop participant appropriately renamed the raw data as al dente data. 7

8 currently known to be medically relevant? Participants noted that what is considered relevant is partly dependent on future technological developments (since future research may lead to the discovery of new variants). 2. How and where to store the data? What is manageable in terms of size? What should be the appropriate conditions for storage in light of safety, privacy and access? 3. How to regulate data sharing or transfer between research and diagnostic contexts? Opening up medical data to research offers significant promise for health benefits, but current law inhibits data exchange across contexts. In a legal sense it makes a crucial difference whether diagnostic data is used for health care or for research purposes in terms of informed consent, communication with the patient, access to data and so on. 4. How to organise patient access to diagnostic and research data? Data is not the same as information providing information is an act that includes interpretation and presentation of the data. What does a doctor s responsibility to ensure that patients are correctly informed imply with respect to providing access to genetic data? Should patients be given access to all the available data, or is it up to the medical professional to determine what and how the information is presented? The patient s right to information and the medical professional s duty to inform may clash with the patient s right not to know and the medical professional s duty of care. 5. Who is responsible for managing data in personal health records? Should the primary duty of care for the dossier lie with the patient, or with the care provider? How is privacy secured? And what should be the role and position of commercial companies offering individualised screening and diagnostic services? Conclusion The arguments, questions and concerns put forward in relation to the storage and use of genomic data in personal health records clearly show the need for further discussion. Towards the end of the meeting, participants were asked to suggest priorities for public debate on the future of diagnostics. Several participants called attention to the opportunities that genome diagnostics offer in terms of health benefits. Others stressed the need for clear rules for reporting incidental findings. Some asked what can be done to further personalise diagnostics, providing not only tailor-made diagnosis but also highly individualised counselling that takes the individual patient s personal situation into account. Advances in both genome technologies and personal health records raise complex technological and moral challenges. What became clear during the workshop is that these technological and normative considerations are deeply intertwined. Technological developments challenge normative assumptions: the opportunity to discover new variants through the comparison of large sets of diagnostic data for instance challenges the assumption 8

9 that research and diagnostics should be separated. Conversely, all technical decisions contain moral elements such as suppositions about which data are relevant, or presuppositions on what patients can and cannot understand. Decisions on who should manage health records are influenced by different normative views of the patient: as autonomous, rational agents who have a right to information (so the option would be to open up and personalise dossiers) - or as vulnerable beings who require protection (so control should remain with the care provider). Through design choices, moral positions thus become hardwired in infrastructures. The interplay of moral and technical norms in the design of digital infrastructures merits broader scrutiny in early stages of development. The workshop discussions pointed out that such broader conversations are possible: participants from various backgrounds contributed to the discussion from their own perspective. Those perspectives may converge through sustained dialogue. To be sure, the use of genome data in a clinical context is still in its infancy. Moral dilemmas occur on a limited scale; reporting of unsolicited findings is a matter of good counselling. But on the longer term systems should be in place that allows patients to make an informed decision on the management of their genetic data. Researchers, medical professionals, policy makers, patients and citizens have to engage in dialogue on the design principles of ICT-infrastructures as genomic technologies mature and become embedded in routine diagnostic procedures. ReDaPeD third workshop The results of this second workshop combined with the findings of the scoping interviews and literature survey will feed into the third workshop to be held in Brussels late This international workshop will establish a multi-stakeholder platform to define an agenda for the integration of societal values and challenges in the design of data infrastructures. The ReDaPeD project targets interdisciplinary collaboration and early engagement of stakeholders. Responsible data management' requires collaborative, transdisciplinary inquiry to guide our collective imagination on the possible futures for the use of genomic data in research and health care. 9

10 Appendix - arguments, questions and concerns on the storage and use of genomic data in digital dossiers as discussed during the workshop. 1. What data should be included in personal health records? What is to be considered relevant data? Under which conditions is it possible and desirable for a personal health record to include the sequencing readout, the analysed data, all variants or only those variants currently known to be medically relevant? What is considered relevant is partly dependent on future technological developments (future research leads to the discovery of new variants). 2. How and where to store the data? What is manageable in terms of size? In genome diagnostics, analysed data of patients is currently stored. Raw data is retained but not re-analysed. Data storage in hospitals consumes considerable resources (due to data stewardship prescriptions); storage at the university is less reliable but more manageable in temrs of size. Cloud computing could provide solutions for efficient storage and sharing but there are legal concerns, such as over privacy and third party access. At some point resequencing of original samples may be more convenient than storing the data. 3. How to regulate data sharing or transfer between research and diagnostic contexts? Opening up medical data to research offers significant promise for health benefits. Current laws and regulations however inhibit data exchange across contexts. From the perspective of future health benefits the firewalls between medical and research contexts may be unproductive. That said, the role responsibilities and liabilities of a doctor trying to cure a patient are very different from a researcher trying to find a new genetic variant. Technological developments challenge normative assumptions. As our moral convictions change, so may the laws that enshrine those convictions eventually need to be adapted. From diagnostics to research... It could be considered a moral duty to unlock the information in one s DNA if it might help others to find a cure for their disease. 10

11 Large data files from diagnostics are often not stored uniformly which makes it difficult to use them for research purposes. While sequencing is often applied to patients with a specific question, there seems to be a momentum towards sequencing more and more individuals (for general purposes). Access to health care should not be made dependent on obligatory knowledge about one s personal genetic traits. In a legal sense it makes a crucial difference whether diagnostic data is used for health care or for research purposes (in terms of informed consent, communication with the patient, access to data etc.) The use of NGS data in a diagnostic context demands appropriate guidance and care. The interpretation of data used to be dependent on expertise of lab specialist and clinical geneticist. But more and more pipelines - commercial software that assesses clinical relevance - are developed. Who then is to interpret the software results and to make final decisions on their relevance and treatment options? Who should be primarily responsible for interpretation of the data: the lab specialist or clinical geneticist? The clinical geneticist doesn t get the considerations, but simply receives the information letter from the lab. Who is ultimately responsible for the diagnosis? Not everything in genome diagnostics is communicated; it doesn t help if everything is shared with the clinical geneticist. To what extent should medical specialists be involved? Can lab technicians take over diagnostic interpretations from specialists? From research back to diagnostics... Research data are often good enough to make statements about large groups of people, but they often do not qualify for diagnostic purposes (because of a high rate of false positives and negatives). 4. How to organize patient access to diagnostic and research data? Data is not the same as information providing information is an act that includes interpretation and presentation of the data. In a medical context, doctors have a responsibility to ensure that patients are correctly informed. Does this mean that patients should be given access to all the available data, or is it up to the medical professional to determine what and how the information is presented? The patient s right to information and the medical professional s duty to inform may clash with the patient s right not to know and the medical professional s duty of care. 11

12 Holding back information can be considered paternalistic. Patients have a right to view all medically relevant outcomes (although in practice it is difficult to gain access to medical data). If patients have access to all the available medical data, does that include access to communication about the patient between doctors? How will this affect communications between professionals? What do patients want to know? What can patients understand? And what do they need to know or understand to be able to make informed decisions? Different groups of patients have different knowledge levels, values and capabilities to interpret the data offered to them. Patients generally cannot meaningfully interpret the raw outcomes of sequencing data. How to communicate the probabilities involved in potentially relevant variants? Without appropriate guidance, how many unnecessary concerns will be generated through false positives, and how many treatment opportunities missed because of false negatives? Rights and responsibilities vary across contexts. The legal and moral responsibilities of researchers to report findings that are relevant for a person s health differ from those of medical professionals; The moral duty to report back unsolicited findings from clinical NGS is contested (see ACMG Recommendations) Research participants participate in research for the common interest; they should be able to exercise their right not to know certain outcomes. Many research studies are designed precisely to report back findings to participants in cohorts. It may be technically impossible to design a research environment that eliminates the possibility of coupling data to their origin (recent publications show that anonymous genome data from studies can be traced back to individual participants). Many research findings are not reported back. Decisions on reporting back unsolicited findings could be organised with the help of dynamic consent (see Jane Kaye). 5. Who is responsible for managing data in electronic health records? Should the primary duty of care for the dossier lie with the patient, or with the care provider? How is privacy of digital health records secured? Concerns over privacy in relation to medical data receive considerably more attention in the public debate than in relation to other forms of personal data. 12

13 Can control over data storage and sharing be left to patients (legally and morally)? Are genome data personal? Every individual has a unique pattern. But the information contained in it also concerns one s family members, so individual choices necessarily affect others. What if a patient wants to share information with family members against their will? What should be the role and position of commercial companies offering individualised screening and diagnostic services? 13

14 List of respondents 1. Martin Boeckhout, Postdoc Biobank Governance, Radboud University Medical Centre 2. Dr. Annelien Bredenoord, Assistant Professor Medical Ethics, Julius Centre, University Medical Centre Utrecht 3. Dr. Christian Gilissen, Postdoc Bioinformatics, Radboud University Medical Centre 4. Karin Idema, Policy Advisor, De Hart & Vaatgroep 5. Dr. Marjolein Kriek, Clinical geneticist, Leiden University Medical Centre 6. Dr. Pieter Neerincx, Postdoc, Genomics Coordination Center Universitair Medisch Centrum Groningen 7. Dr. Cor Oosterwijk, Director, VSOP: Patient Alliance for Rare and Genetic Diseases 8. Mr. dr. Corette Ploem, Legal expert, Academic Medical Centre 9. Bettine Pluut, Independent consultant, Bettine Pluut Advies en Onderzoek 10. Dr. Maud Radstake, Project Manager, Centre for Society and the Life Sciences 11. Dr. Marc Rietveld, Project Leader Knowledge Platform Research ICT University Medical Centre Utrecht 12. Dr. Daan Schuurbiers, Independent consultant, De Proeffabriek 13. Dr. Erik Sistermans, Head Genome Diagnostics, Free University Medical Center 14. Dr. Eric Vermeulen, Researcher, Netherlands Cancer Institute 15. Dr. Terry Vrijenhoek, Programme Manager, Centre for Genome Diagnostics 14