In 1892, Sir William Osler published the first of many editions of his famous textbook ‘The Principles and Practice of Medicine’ Citation[1]. This text, more than any other, celebrated medicine as a science and in doing so, highlighted the importance of bringing together the different disciplines of clinical observation and examination, pathology and physiology. At that time the available treatments were few and comprised surgery, fraught with risk or the use of traditional medicines mostly derived from plants. This empirical approach to the treatment of illness, much of which was never systematically tested and subject to the whims and fancies of the practicing physician, was of dubious efficacy and with a high chance of serious side effects.
The interaction of chemistry with biology: the birth of pharmacology
The advent of pharmacology, at the turn of the 20th century, and its link with chemistry began the long journey of understanding disease pathophysiology at a molecular and cellular level. Exciting discoveries followed such as adrenaline, acetylcholine, cortisone, digoxin, aspirin and mercurial diuretics. While drugs such as adrenergic agonists, anticholinergics and antihistamines were developed progress was slow. The concept of structure–function relationships as the basis of modern pharmacology gradually involved using well-defined isolated tissue (e.g., smooth muscle contraction) or whole-animal responses to refine structure-activity responses. Even with the discovery of penicillin by Pasteur, it took over 75 years before sufficient synthetic penicillin became available from Biochemie (now Sandoz/Novartis) for its clinical uptake as a revolutionary treatment for infections. One should remember that even 60 years ago the physician still only had a handful of chemically synthesized drugs, most diseases being managed with herbal remedies containing undefined pharmacologically active agents. However, pharmacology as we now know it, was born and led to a rapid expansion in industry willing to take on the challenges of treating disease with the idea of maximizing efficacy and reducing side effects.
The next 50 years witnessed an explosion in modern drug development with spectacular success in the form of blockbuster drugs. Examples include corticosteroids, adrenergic agonists and Β-blockers, a range of antihypertensives, cyclo-oxygenase inhibitors, statins, oral contraceptives, oral hypoglycemic and a myriad of antimicrobials to name but a few. The pharmaceutical industry thrived on this and companies such as Merck, Burroughs Wellcome, Allen and Handburys, Eli Lilly, Roche, Chemical Industries Basel (CIBA) and Imperial Chemical Industries (ICI) went on to become major economic drivers in many western countries. The randomized clinical trial became the standard for assessing efficacy while the new discipline of clinical pharmacology emerged to gain insight into drug pharmacokinetics and pharmacodynamics.
Difficulties in maintaining the delivery of new drugs into the clinic
As companies became more and more successful, mergers and acquisitions began to occur. Thus, during the latter part of the 20th century large new megacompanies emerged, examples including GlaxoSmithKline (GSK), Novartis, AstraZeneca, Pfizer and Merck. It was becoming apparent that this merger and acquisition strategy was in large part driven by shrinking drug pipelines and the dramatic increase in costs for drug development, not least being driven by the high failure rate of drugs once they were selected for clinical trial. Looking back over the last 50 years or so, it is salutary to note that many new drugs were first tried in humans as the test ‘guinea pig’.
“The advent of pharmacology, at the turn of the 20th century, and its link with chemistry began the long journey of understanding disease pathophysiology at a molecular and cellular level.”
However, rising concerns over safety and toxicity placed increasing emphasis on animal models of disease as suitable surrogates. One can appreciate the rational for this especially since cancer, degenerative and inflammatory diseases could be reproduced in animals using various breeding selection procedures, exposures and manipulations. Recapitulation of a disease was most frequently defined in terms of pathology and a limited range of phenotypic outputs that resembled the human disease in question. Although having the appearance of the human disease, when assessed by these criteria alone, interventions that were highly effective in these ‘animal models’ of human disease (including nonhuman primates), when trialed in humans, the failure rate was enormous and in some diseases absolute. It is not that animal models were not helpful in drug or biologic selection, they were, but it is of great importance to recognize that they are most frequently a long way from the human disease, especially in reflecting disease chronicity, relapse and the many environmental and lifestyle factors contributing to these different disease manifestations. To put it simply, ‘complex disease is, by definition, complex involving interactions of many genes and environmental factors across the life course’ Citation[2].
Another factor contributing to the decline of drugs being developed into clinical use includes the increasing concern that society has experienced over side effects that the public is willing to accept, leading to increasingly stringent regulation for drug or biologics approval. Over reliance on high-throughput drug screening, combinatorial chemistry and computerized drug/target prediction has also contributed to the failure to create innovative treatments, although they have certainly found their place for improving the actions of already identified pharmacological activities and possibly reducing side effects. However, in the final analysis, the realization that although we have learnt much about the origins and progression of complex disease, this is insufficient to provide the robust leads needed for sustaining the blockbuster drug strategy. This strategy was soundly based upon large proportions of the population responding to an individual intervention, but now we are beginning to appreciate that blockade of a single pathway is insufficient to make a large difference in most complex diseases. However, the addition of each new intervention, by affecting a proportion of subjects to a major extent or most subjects to a minor extent, when added together leads to a substantial health gain at a population level. This amounts to stratified medicine.
Emergence of pathway-specific diagnosis & treatment: the beginning of personalized medicine
At its more refined level, stratified or personalized medicine within the context of heterogeneous complex disease is aimed at identifying those pathways that contribute to the individual disease phenotype, developing diagnostic tests to uncover these and then targeting appropriate interventions to those pathways and monitoring outcomes with the appropriate pathway-specific biomarker. As referred to repeatedly in Personalized Medicine, cancer is leading the way in large part owing to molecular phenotyping of cancer. Spectacular successes have been obtained in leukemia and lymphoma, some stromal tumors, breast cancer and melanoma, with promising therapies emerging in lung and gastrointestinal cancers all based upon this approach. It is in the inflammatory and degenerative diseases that a similar approach needs to be adopted.
“Looking back over the last 50 years or so, it is salutary to note that many new drugs were first tried in humans as the test ‘guinea pig’.”
The need to change behavior to deliver personalized medicine
To achieve this, a different type of relationship is needed between patients, physicians and healthcare providers, academic researchers and the drug discovery and diagnostics industry as well as the research funders. Research needs to return to being patient oriented with clinical, physiological, imaging and laboratory phenotyping being linked to discovery platforms capable of uncovering causative pathways most likely provided by the ’omics technologies. By converging these activities and utilizing systems approaches, it is possible to identify biochemical and immunological ‘nodes’ where causal pathways linked to specific subphenotypes converge and are likely to prove to be critical points at which interventions need to be aimed at. New in vitro pathway-based human and animal models need to be developed in order to validate the effect(s) of manipulating a specified target. All of this ultimately will require well run biobanks, novel statistical approaches to phenotyping (e.g., machine learning methods) and construction of disease consequences in virtual patients to enable greater accuracy in predicting outcome with and without interventions across the life course.
“…a different type of relationship is needed between patients, physicians and healthcare providers, academic researchers and the drug discovery and diagnostics industry as well as the research funders.”
The development of computational tools to reliably integrate patient-specific clinical and lifestyle data with ’omics data needs very large, well-characterized reference datasets that have follow-up over extended time intervals. Europe has for a long time invested in large epidemiological cohorts and more recently in biobanks, which provide a platform for constructing such reference datasets. Projects that provide full genome variation and other ’omics data in very large study samples are in progress. Scandinavian countries, which have long traditions in well-characterized epidemiological study collection and national registers with harmonized, nationwide health data, provide a unique resource for such studies. An example of such an initiative is the Sequencing Initiative Suomi (SISu) project that aims to generate near full-genome variation data of all 200,000 samples in the Finnish National Biobank representing approximately 4% of the entire Finnish population. These samples are linked to the national registers providing decades of follow-up data to analyze disease frequencies and prognoses in a nationwide manner. This, and similar information from other Nordic countries, will then be integrated to expand the knowledge base for a roadmap by algorithms and computational tools towards needs of stratified and personalized medicine.
The European Science Foundation Forward Look on personalized medicine
In recognizing the importance of Personalized Medicine the European Medical Research Councils (EMRC) commissioned a Forward Look to gain insight into the needs for research programs and infrastructures, policy and education. Forward Look enables Europe’s scientific community, in interaction with policy-makers, to develop medium- to long-term views and analyses of future research developments with the aim of defining research agendas at a national and European level. As described above, the challenges ahead are significant, and it is becoming clear that biological systems operate in a far more complex way than we might have previously thought. Nevertheless, this approach is starting to yield results and it is predicted that these technologies will generate innovative therapies, limit adverse effects of treatments, increase the quality of clinical care, create an optimal fit between a patient and a treatment, and decrease the costs of healthcare. But personalized medicine will have a broader impact on society as increased transparency of individual biological inequalities will affect basic issues such as risk, responsibility and solidarity, raising complex social and ethical questions. For these reasons the Forward Look is a collaborative effort between all standing committees under the auspices of the European Science Foundation (ESF): life, earth and environmental sciences; humanities; social sciences; and physics and engineering. The overall aim of this Forward Look is to analyze, in a systematic way, the complex and constantly moving field of personalized medicine to provide timely policy advice that will help prepare Europe for the likely changes in how society deals with health and disease.
Technology platforms are crucial to dissect out complex biochemical and other pathways and interactions that contribute to complex disease pathobiology. The first meeting dedicated solely to these technologies and methodologies was therefore chosen to initiate the foresight process. It took place in London on 19–20 September 2011. ESF then commissioned a series of articles to further explore the current state-of-the-art: how far the expectations of personalized medicine have been realized, where the bottlenecks might lie and how as a society we are going to move towards such an approach (if at all). We are extremely grateful to those meeting participants who kindly accepted to contribute these articles.
Ivano Bertini and colleagues from the Magnetic Resonance Center at the University of Florence emphasizes the importance of the integration of the different ’omics data to obtain combined information for health and disease status for an individual at the whole body level and to build a ‘virtual patient’ model Citation[3]. They then use metabolomics as an example of how such molecular phenotyping technology might be applied to distinguish between healthy and diseased phenotypes and identify novel disease biomarkers. Building upon this, Jukka Corander (University of Helsinki, Finland) and three colleagues from Finland and the Wellcome Trust Sanger Institute in the UK having been carrying out work concerning the massive computational and statistical challenges that personalized medicine holds Citation[4]. They emphasize that this grand challenge facing medicine calls for both integrated experimental–computational approaches utilizing all relevant data sources and a paradigm shift in computational and statistical thinking of how to model and integrate the ’omics data to enable useful predictions to be made requiring data harmonization, computational efficiency and rigorous statistical inference.
Although we almost always consider complex human disease as the substrate for more personalized medical approaches, Francesc Palau from the Instituto de Biomedicina de Valencia, CIBERER in Valencia, Spain focuses upon rare diseases Citation[5]. He draws our attention to the different philosophies of human disease as proposed by Osler (broken machine and patients represent classical cases) and Garrod (patient individuality). In a highly informative table he draws out the differences that exist between common and rare diseases. Using this contrast, he then makes the important point that few biomarkers are available for the diagnosis of rare diseases and seeks dynamic biomarkers to define the prognosis and therapeutic response for both common and rare diseases and that in both cases these need to be related to the biological pathways involved in the disease process.
Our final commentary by Carsten Carlberg from the Department of Biosciences, Kuopio, Finland draws our attention to the need of education in personalized medicine Citation[6]. In particular, he focuses upon the need of the healthcare community to be sufficiently informed about this new approach of healthcare professionals to enable them to interpret results and generate meaningful interactions with the public and patients. He makes the point that to achieve this, personalized approaches to healthcare will require a broadly based education and training for multidisciplinary purpose.
The need to plan for this new way of practicing medicine is essential if such a revolution is to move ahead in a productive way. We hope that these commentaries add further light on this rapidly emerging field. We are indebted to the individual authors of each commentary for the time and effort they have spent in preparing their contributions and to the ESF for financial support.
Financial & competing interests disclosure
The authors have no relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript. This includes employment, consultancies, honoraria, stock ownership or options, expert testimony, grants or patents received or pending, or royalties.
No writing assistance was utilized in the production of this manuscript.
References
- Bliss M. William Osler: A Life in Medicine. Oxford University Press, Inc, NY, USA (1999).
- Yang Q , KhouryMJ, FriedmanJ, LittleJ, FlandersWD. How many genes underlie the occurrence of common complex diseases in the population? Int. J. Epidemiol. 34(5), 1129–1137 (2005).
- Bertini I , LuchinatC, TenoriL. Metabolomics for the future of personalized medicine through information and communication technologies. Per. Med.9(2), 133–136 (2012).
- Corander J , AittokallioT, RipattiS, KaskiS. The rocky road to personalized medicine: computational and statistical challenges. Per. Med.9(2), 109–114 (2012).
- Palau F. Personalized medicine in rare diseases. Per. Med.9(2), 137–141 (2012).
- Carlberg C. The need of education in personalized medicine. Per. Med.9(2), 147–150 (2012).