Personalized medicine holds great promise as a basis for patient-centered care. But the exponential growth in the past decade of scientific publications focused on its expansion is closely intertwined with a similar growth of a more critical literature that pays close attention to the multiple issues it raises.
Actually, there are several reasons to challenge the ability of personalized medicine to provide patient-centered, individualized medicine, correlative of two different moments in its development. As already pointed out, in its infancy stage, personalized medicine intended to take advantage of the results disclosed by genetic research, especially those of the Human Genome Project, in order to improve healthcare. At that time, the underlying view of genetics was by and large monocausal and deterministic in its understanding of the relationship between genes and disease. This understanding is reflected in the early focus of personalized medicine, that is, anticipate drug responses based on newly discovered genetic markers, which will then enable more targeted therapies and help us to choose more beneficial treatments for specific patients with fewer adverse drug reactions. This new goal of the right drug for the right patient at the right time stood in dramatic contrast to the focus of conventional medicine and pharmacy on the development of new blockbuster drugs.
A good example of this early focus of personalized medicine is the drug Abacavir used by HIV patients. It was relatively safe for nearly all patients but induced a life-threatening allergic reaction in 6% of patients. In 2002, two independent research groups correlated this toxic reaction with a single genetic variant of the major histocompatibility complex class I (HLA-B*5701) [4, 5]. That, in turn, led to the screening of HIV patients for the genetic variant before Abacavir is prescribed. This impressive direct association constitutes a persuasive example of what personalized medicine has wanted to achieve. Furthermore, it illustrates that initially, personalized medicine intended to produce a more individualized understanding of health and diseases on the basis of genetic patterns, shifting from symptom-based to genome-based approaches.
However, in this initial form, personalized medicine failed to achieve its promise of truly personalizing health care (e.g., [6,7,8,9,10]). Rather than treating the whole person, personalized medicine was able to achieve its goal of the right drug for the right person at the right time only by reducing patients to their genetic profiles, that is, to their objective bodies. Moreover, in order for the patient’s genetic profile to yield actionable information, it had to be matched to a particular category of patients all sharing the same marker, a process known as stratification. It is not clear how being treated in the same way as everybody else in a pre-defined group of patients, even a relatively small group of patients who all share one important genetic trait, amounts to truly personalized treatment. In fact, this process of stratification is not novel. Rather, it is a refinement of the existing tools for diagnosis, management and prevention of disease. And those, in turn, presume a somewhat naive and underdeveloped understanding of what really is a person, which, at best, equates personhood with the objective body, or, at worst, with a single genetic trait shared with a cluster of other patients.
More recently, personalized medicine has taken an important turn by expanding the number and type of health-related data used to determine what is the right treatment for the right person at the right time. Capitalizing on the vastly increased IT storage and data processing capacities now available in all developed countries at dramatically reduced costs, personalized medicine has moved beyond merely genetic data to include the full range of available patient information, from a molecular scale (proteome, transcriptome, metabolome, etc.) to an epidemiological one (foodome, sociome, environtome, etc.). Thus, big data science is changing the epistemic base of personalized medicine.
First, the old “one gene – one disease” model is being exchanged for a much more complex, multi-causal model of disease, requiring the integration of vast amounts of different types of data, including but not limited to genetic data. That also means that it is no longer possible to accurately predict what will happen next based on a patient’s genotype, as the monocausal and deterministic framework of genetic personalized medicine attempted to do. It is the statistically determined correlations that become decisive. Thus, big data personalized medicine does not seek a true diagnosis as traditionally defined, that is, the real and certain presence of a disease state that is hence amenable to therapeutic interventions. Instead, big data personalized medicine generates probabilities about a particular person exhibiting certain characteristics. For example, it may tell us that Mr. P has a 21% developing asthma due to his living in a neighborhood with postal code 68564, augmented with a 23% chance risk of developing asthma or COPD based on his genealogical ancestry, as well as a 20% relative increase in suffering from asthma symptoms due to high levels of air pollution in the city, with a 65% chance of responding favorably to drug X due to his genotype. None of this is epistemologically analogous to diagnosing Mr. P as having asthma. This, in turn, raises the question how the ultimate goal of personalized medicine to provide the right treatment for the right person at the right time, can be met when big data can only supply us with population-based information about the effects of treatments; when all patient characteristics (so-called markers) have meaning only within strata of patients sharing that characteristic; and when all other determinants of health and illness have only correlative significance. As long as all information used to develop a care plan gains relevance only when and to the extent that it is population-based, the patient as a person is missed.
The failure to develop a robust understanding of personhood not only has serious epistemic consequences; it also skews the ethical discussions. Consider the often discussed topic of predictive information. Advocates of personalized medicine insist that providing people with more predictive information will empower them to better manage their own health and render them more autonomous in making health care related decisions. Critics counter that patients’ freedom may actually be reduced by the predictive information concerning incurable or particularly dreadful conditions for which there is no effective treatment available; and even if preventative life-style changes may delay their onset or lessen the symptoms, people’s capacity to initiate and sustain life-style changes is often limited. This dueling pair of perspectives suggests that the ethical quandary at hand is how to weigh the value of predictive information. However, this understanding of the ethical quandary is fundamentally flawed, and we are now in a better position to explain why.
Notwithstanding the apparent irreconcilable disagreement between advocates and critics, they both assume that the information given by personalized medicine to the individual person is directly relevant to and has meaning for that unique individual. It could neither benefit nor harm if it did not. But we have just shown that there are good reasons to doubt that assumption. So even if we were to find a happy medium and develop a prudent policy or adopt a well-crafted new law that somehow reconciles the two opposing views on the value of information summarized above, the fundamental underlying ethical problem will be completely missed. Instead of fostering genuine personalization, such policies and laws reinforce the mistaken idea that respect for the uniqueness of each individual patient can be reduced to a scientific understanding of the body supplemented with a variety of correlations derived from population-based strata.
Some advocates of personalized medicine have acknowledged this limitation. But rather than seeking to overcome it, they have proposed to circumvent the problem by exchanging the goal of personalizing care for the more limited goal of developing stratified medicine or precision medicine . Albeit honest, this position does not suddenly undo the reality that the public’s hope, the support of many national governments, and the vast investments by industry in this new field of medicine are fueled by the promise of truly personalized health care. Do we have to conclude that this hope, support and investment are bound to be in vain?
We do not think so. But in this paper we will argue that the relationship between personalized medicine and the patient’s personhood has to be reversed: Rather than adapting medicine to the unicity of individual persons, personalized medicine may contribute to the process by which a person is becoming him- or herself. To clarify and justify this unexpected reversal, we need to conduct a philosophical analysis of the concept of personhood and of the complex relationship between the person and the objective face of the world, to which belong the countless data generated by personalized medicine. We propose to undertake this analysis by drawing on the twentieth century philosophical tradition known as phenomenology, and more specifically phenomenological explorations of the nature of human subjectivity.