Personalised health – current opinion
Many of us will recall the completion of the sequencing of the human genome at the beginning of the millennium. Enthused by the achievements of the scientists, Bill Clinton announced at that time that “genomic science will revolutionize the diagnosis, prevention, and treatment of most, if not all human diseases”. Genetic disorders driven by a single gene are, however, not common and it became clear that the information value of a string of nucleotides in clinical decision-making is limited to a few conditions. Today, successful examples of personalised therapies based on a genomic diagnosis exist primarily in oncology. The testing for specific genetic abnormalities has substantially changed cancer therapy over the last decade and led to new classifications and treatments of tumours: in lung cancer, for example, the anatomical and histological diagnosis has been refined by molecular testing of and other genetic markers, which allows more targeted and – at least transiently – more effective therapies.
One could argue, however, that the conduct of genome-wide association studies has told us (as yet) little about the genetic basis of common diseases such as heart disease, diabetes, etc. Outside oncology there are only a few such successful examples. One relates to the association between proprotein convertase subtilisin/kexin type 9 (PCSK9), low-density lipoprotein (LDL) receptor removal and familial hypercholesterolaemia: the identification of subjects with loss-of-function mutations of PCSK9 and reduced plasma LDL was at the origin of the validation of the biological target and ultimately led to development of novel drugs such as evolocumab. However, the hope that delineation of the mechanisms by which genetic factors cause monogenic disorders will provide widely important information about the basic pathophysiological processes that underlie common diseases has, so far, not really materialised. The translation of a particular pathway into a disease phenotype has proved to be challenging in practice. These are influenced by multiple factors and variations and, in addition, may affect different conditions. Mutations that, for example, encode nuclear lamins (LMNAs) can cause cardiomyopathy, muscular dystrophy, lipodystrophy and progeria. Furthermore, even when some mutations can be associated with disease, it is often difficult to evaluate the effective pathophysiological role, if there is one at all, for many of the genetic variants.
Despite the substantial investments in large-scale genomic sequencing and human tissue-specific proteomics, one has to conclude that health cannot be assessed in a deterministic way based on molecular diagnostics. It may, however, be only a question of time until the insights in -omics and, possibly, epigenetics, in conjunction with advances in bioinformatics, will more broadly result in a redefinition of the taxonomy of disease as postulated by Oxford Regius Professor Sir John Bell. One can speculate where the personalised health revolution that was proclaimed by Bill Clinton 15 years ago now stands in the “emerging technology hype cycle”. In that model, described by Gartner and used to describe the impact of breakthrough technologies, a “peak of inflated expectations” is typically followed by a “trough of disillusionment”: only thereafter full adaptation and productivity occurs (fig. 1). For personalised health specifically, the extent to which these advances will be increasingly useful and adapted, or will disrupt clinical practice, will depend on how the medical community is and will be prepared to cope with the exponentially increasing flow of data and information and – most relevantly – to deal with its interpretation.
Figure 1: Gartner hype cycle curve for emerging technologies.
It is fair to say that the phrase “personalised health” is currently “hype”  and is used by many stakeholders, sometimes, unfortunately, without a clear understanding of what it actually means. The terms “personalised medicine“, “precision medicine” and “personalised health” are often used interchangeably. “Precision medicine” is favoured in the US  and has a more technology-oriented connotation, with focus on large-scale biological databases, powerful methods for characterising patients (-omics, cellular assays, mobile health technologies) and computational sciences. It remains for the purists to debate how precise nonsurgical medical intervention can and will be, even with molecular “precision” diagnostics, or whether the term “precision medicine” is somewhat misleading. “Personalised medicine” as a term is used more often in the context of pharmacotherapy and has been defined as “getting the right drug to the right person at the right dose at the right time” . This approach and objective is intrinsically close to the interests of pharmaceutical companies and hence mobilises substantial industry funding. For best societal benefit, “personalised health” should be defined somewhat more broadly and should cover prevention, intervention and treatment with focus on benefits to patients and society. The concept of personalised health incorporates a growing set of health data that are generated by physicians and by patients themselves and hence extends beyond a focus on genomics. As a consequence, personalised health and associated decisions have to be based on probabilistic models including data from different sources. Hence, its implementation must involve different disciplines; information technology (IT) and “big data” scientists will have to play a central role in leveraging these technologies to patients. The interface between computational sciences, biology and medicine shapes new research fields, asks for new skill sets and will imminently call for adaptation of the educational curricula of physicians, biologists and computational scientists.
The argument that medicine has always been personalised is certainly valid. The concept of personalised health is, however, not reinventing the wheel but applying the physician’s focus on the individual patient on a larger, population scale. The ability to better measure, aggregate and make sense of previously hard-to-obtain epidemiological, clinical and molecular biometric data should enable better clinical decision-making for the individual. Concerns that healthcare will be reduced to a set of algorithmically derived probabilities have been raised. They are, however, unlikely to become true in reality; even when computer sciences move beyond routine processing and rapid repetitive arithmetical processes, and get better at complex pattern matching and learning integration. The digital revolution and Industry 4.0 is happening and has reached healthcare. It is not much use to doubt or contain this evolution. The question is rather: “How can we use and stay on top of these technologies as tools to serve citizens and patients rather than to be uncomfortable or even feel threatened by them?”. Patients, physicians, healthcare systems and payers, as well as the diagnostic, pharmaceutical, biomedical, IT and communication industries, must share a strong interest in the application and evolution of personalised health concepts.
These interests may not necessarily be fully aligned among these groups and are therefore liable to create tensions:
- Patients seek clearer understanding in prevention and management of their disease, prognosis and the most effective measures and treatments.
- Physicians and healthcare systems share these interests but must also balance individual patient needs with management of society’s healthcare utilisation.
- Payers are concerned that molecular diagnostic and individualised therapies and prevention measures will drive up healthcare expenditure. They remain sceptical that these costs will be offset by more selective, more effective and safer use.
- Industry aims to leverage the new technologies to introduce new products and to maintain profit margins.
Establishing clarity on legal, societal and ethical regulations is, therefore, of the highest priority in order to introduce personalised health successfully and to manage these intrinsic tensions. The topic is complex and its implications are challenging to assess, even for healthcare professionals. It is important, therefore, that objectives, benefits and boundaries are transparently explained and discussed with patients and other citizens. Scientists and clinicians need to engage in such dialogues to establish acceptance in a field that requires, next to clear regulations, trust in healthcare and research institutions. Such attention to communication is not necessarily intrinsic to a research and clinical community, but will be extremely important to create the necessary engagement of the target audience, ultimately patients and society.
Much of the research in personalised health will be led by or conducted in cooperation with public and academic organisations, thereby utilising healthcare data responsibly and in compliance with legal and ethics regulations. Personalised health should deliver benefits to patients and, therefore, intrinsically includes the need to translate research results into clinical applications. The interdisciplinary nature of the task challenges classic academic structures, which traditionally are split in faculties and disciplines and, to put it more critically, are separated into “knowledge silos”. Industry, for example, pharmaceutical, biotechnology and IT companies, is set up by necessity for interdisciplinary work in an organisational matrix and is by default better positioned to translate innovation into tangible outcomes and deliverables. Academic institutions that can adapt to the needs of interdisciplinary collaborations will probably be most successful in the emerging and rapidly accelerating domain of personalised health research. Such structures and organisations are currently emerging primarily in universities in the US (e.g., Stanford, Duke), the UK (e.g., Cambridge) and the Netherlands (e.g., Leiden), but also in Asia (e.g., Singapore A*STAR). Typically, these translational research projects in personalised health are capital-intensive and can be only partly financed by grants and public funding. In times of financial constraints in the public sector, entrepreneurship, third party funding and partnerships with industry are required.
Swiss academic institutions are in an excellent position with regard to access, networks and opportunities for collaboration with the local life-science industry. This advantage is often postulated, but, arguably even for our internationally top-league academic institutions, insufficiently leveraged relative to the potential that actually exists. Such collaborations must be professionally managed, also on the academic side, thereby ensuring transparency of the relationships and of the defined boundaries in order to retain public acceptance and trust. Anglo-Saxon countries, in particular the US and UK, but also the Netherlands and Scandinavia have been effective in managing these relationships without compromising their academic independence. Switzerland has the advantage of a high-quality healthcare system and internationally competitive academic research including basic and clinical science. Internationally, several large public initiatives had already been initiated during recent years. The UK has launched an effort to sequence completely the genomes of 100 000 British patients to characterise the genetic basis of their diseases . There are a number of other efforts in Europe, such as the Finnish Institute for Molecular Medicine, which is focused on leukaemia. In the US, a large number of initiatives have been created locally, at the state level, or nationally, coordinated partly by the National Institutes of Health. Centres and institutes have been established at Harvard, Stanford, Connecticut (the new Jackson Laboratory for Genomic Medicine) and many other places. President Obama has launched the Precision Medicine Initiative with substantial public funding for better and more individualised healthcare in the 21st century . Even though several of these initiatives are already more advanced, the Swiss Personalized Health Network (SPHN) initiative is still timely . A national effort that may include adaptation of current programmes and structures is, however, urgently needed to maintain Switzerland’s internationally competitive position in life-science research and healthcare. Today we are only beginning to realise the changes in medicine and in society that personalised health will drive:
- Progress in diagnostic technologies to determine and analyse the human genome, proteome and metabolome. Ability to assess a wide range of clinical and lifestyle biomedical data resources in large numbers of individuals.
- Capabilities to store, process and analyse these “big data” sets, which enable novel applications in clinical practice, healthcare management, research and economics.
- Development of preventive measures and therapeutic medicines that are tailored to the individual patient needs based on their specific molecular make-up.
The amount of healthcare data is expected to rise between 2015 and 2025 from 500 to 25 000 petabytes . High-confidence algorithms can predict workable interventions that will have direct implications for improving healthcare outcomes. Building such algorithms requires competence to integrate multiple factors, such as the ability to incorporate new data types and sources including contextual information; they require feedback loops that allow the models to rapidly learn and transparency around the variance of the predictions. It is important to realise, however, that research projects that explore and mine large data sets are primarily hypothesis-generating. (Pre-)mature conclusions on associations without causality should be drawn cautiously. The example of Google, who failed spectacularly to predict the peak of the influenza season in 2013, must not be interpreted as failure of the big data concept. It rather illustrates that it can be dangerous to rely for decision making on trends determined with such algorithms, even when they are constructed with methodological stringency. The assumption that big data are a substitute for traditional data collection and analysis can result in erroneous predictions and conclusions . Predictions driven by big data rather supplement than substitute controlled trials and well organised patient cohorts. Predictions based on big data algorithms will ultimately require validation in prospectively planned experiments before conclusions can reliably be applied in clinical medicine.
In dealing with new technologies, history tells that previously successful structures – societies and companies – failed to adapt and to leverage these opportunities because they were afraid to hurt their established way of operation. A classic such example is Eastman Kodak, which in 1975 developed the first digital camera, but decided to drop the product for fear it would threaten its own photographic film business. The negative impact of this decision became dramatically evident only a few years later: Kodak began to struggle financially in the late 1990s as a consequence of the market switching to digital photography. Finally, in January 2012, Kodak had to file for Chapter 11 bankruptcy protection in the district court for southern New York . The case illustrates that disruptive technologies can be ignored only for a short time, and can fundamentally undermine the sustainability and success of presumed well-established operations. The speed of technology-induced changes to business and society has today further exponentially accelerated. Silicon Valley, and the biotechnology companies and academic centres in the Bay area of San Francisco are examples of successfully dealing with breakthrough technologies . Such dynamics, including flexible, delivery-oriented structures with low interdisciplinary boundaries and the ability to mobilise venture capital at risk are needed. Although aspired by politicians and industry, this proves to be much harder in Europe, and challenges its competitiveness. Swiss innovation from academia and the life-science industry is today internationally competitive. For Switzerland, the decentralised system may be at the base of its power in innovation. Although this heterogeneity is a strength, it makes decision making for national alignment that is needed for a personalised health initiative complex. This complexity needs to be constrained and managed with a focus on advancement and true deliveries, while acknowledging the needs of a federalist system. Committees are important tools to advance initiatives but must not become self-fulfilling or platforms for individual ambitions. Tangible projects are needed and require a dedicated effort, effective tailored structures and openness to entrepreneurship.
Despite the quality and strengths of Swiss healthcare and academic institutions, there is absolutely no space for any complacency. As the physicist William Pollack sharply stated in the 1960s: “The arrogance of success is to think that what you did yesterday will be sufficient tomorrow.” The speed of changes related to personalised health technologies and applications is breathtaking. Agility and innovation in leveraging these technologies are needed to maintain a competitive, leading position and to provide optimal healthcare to citizens and patients.
Andreas Wallnöfer is General Partner at BioMed Partners Switzerland. As independent consultant he currently also advises boards of several biotech companies. Previously, he held several senior R&D positions at F. Hoffmann-La Roche Ltd., including Head of Clinical Research & Exploratory Development and Head of Cardiovascular and Metabolism Research. He served for more than a decade on the Roche R&D leadership team and on the portfolio committee and played a key role in the integration of Genentech and Roche’s R&D organisations. He currently chairs the Singapore A*STAR External Early Development Advisory board and is member of the Scientific Advisory Board of the Centre for Human Drug Research, University of Leiden, The Netherlands. During 2015 and 2016 he supported the University and University Hospital of Basel in launching the Personalized Health Program as part of the Swiss national initiative.
- Abifadel M, Varret M, Rabès JP, Allard D, Ouguerram K, Devillers M, et al. Mutations in PCSK9 cause autosomal dominant hypercholesterolemia. Nat Genet. 2003;34(2):154–6. doi:http://dx.doi.org/10.1038/ng1161.
- Capell BC, Collins FS. Human laminopathies: nuclei gone genetically awry. Nat Rev Genet. 2006;7(12):940–52. doi:http://dx.doi.org/10.1038/nrg1906.
- Lander ES. Cutting the Gordian helix--regulating genomic testing in the era of precision medicine. N Engl J Med. 2015;372(13):1185–6. doi:http://dx.doi.org/10.1056/NEJMp1501964.
- SciLifeLab. The Human Protein Atlas. Available from: http:/www.proteinatlas.org.
- Kola I, Bell J. A call to reform the taxonomy of human disease. Nat Rev Drug Discov. 2011;10(9):641–2. doi:http://dx.doi.org/10.1038/nrd3534.
- Gartner, Inc. [Internet] Available from: http:/www.gartner.com
- National Institutes of Health. Precision Medicine Initiative [Internet]. Available from: www.nih.gov/precision-medicine-initiative-cohort-program
- Hamburg MA, Collins FS. The path to personalized medicine. N Engl J Med. 2010;363(4):301–4. doi:http://dx.doi.org/10.1056/NEJMp1006304.
- Genomics England [Internet]. Available from: http://www.genomicsengland.co.uk
- Meier-Abt P, Egli F. Kräfte bündeln: “Swiss Personalized Health Network. SAMW Bulletin. 2016;1/16:1–5.
- The Future of Personalized Healthcare. Predictive Analytics. Available from: https://rockhealth.com/reports/predictive-analytic
- Lazer D, Kennedy R, King G, Vespignani A. Big data. The parable of Google Flu: traps in big data analysis. Science. 2014;343(6176):1203–5. doi:http://dx.doi.org/10.1126/science.1248506.
- The last “Kodak moment” Jan 2012. Available from: https://www.economist.com
- Money Tree Report PWC. 2014. Available from: https://www.pwcmoneytree.com