National Center for Advancing Translational Sciences, National Institutes of Health, United States of America.
A position paper is defined as “an essay that presents an arguable opinion about an issue” (Wikipedia). In that spirit, I will argue here that personalized medicine is (a) as intention, as old as medicine itself; (b) as generally applicable possibility, only a few decades old; and (c) as reality, already upon us, with all the promise and perils that were forecast.
Physicians since antiquity have realized that every patient is different and attempted to treat them accordingly – incorporated into the concept of diathesis and the curative potential of individualized surgery for tumors. The history of medicine is in many ways the history of gradual subdivision of symptoms (e.g., cough) and signs (e.g., fever) into categories that could be used to diagnose individual patients (e.g., cough + fever = pneumonia), and these diagnoses’ subdivision into categorical (e.g., bacterial vs viral pneumonia), the treatment of which was distinct (antibiotics for bacterial pneumonia, not for viral pneumonia) and eventually specific to etiology (e.g., penicillin for pneumococcal pneumonia vs. vancomycin for staphylococcal pneumonia) and increasingly personalized (e.g., penicillin for pneumococci sensitive to it, other antibiotics for penicillin-resistant pneumococci). This progression has been repeated in every area of disease, some more successfully (e.g. cancer) than others (e.g., mental illness).
What is new in the last several decades is therefore less a change in physicians’ intention to individualize (i.e., “personalize”) patients’ diagnosis and treatment and more a change in their ability to do so. This change has been enabled by the eradication of previously common, lethal (mostly infectious) diseases, transitions in views of the rights of individuals to health, and technological advances including sensitive analyses of analytes in body fluids, multiple imaging modalities, and genetic analyses from karyotypes to mutation detection to exome/genome sequencing. These revolutionary changes – which have all happened during the professional lifetimes of today’s practicing physicians – have allowed the much- heralded shift from medicine based on an individual practitioners’ experience and intuition (“experience-based medicine”) to medicine based on objective data (“evidence-based medicine”). Among the downsides of this otherwise positive evolution is the erosion of the doctor-patient relationship that can be at the heart of healing, since most practitioners are no longer facile with all possible diagnoses and treatments, requiring fragmentation of patients’ care among multiple specialists and frequently poor care coordination. The exploration of computer-driven artificial intelligence (AI) to more effectively evaluate diagnostic and therapeutic options for individual patients is the most recent adaptation of the health care ecosystem to the information- and personalization-explosion. It would be ironic indeed if the practical requirements of personalized medicine produce medical practice that lacks meaningful human interaction. No diagnostic or treatment algorithm, no matter how accurate or personalized, can hold the hand of an anxious patient who has just received life-altering, or life-ending, medical information. Personalized medicine is increasingly a reality; in our laudable headlong rush to adopt it, we must be sure to keep the “person” in personalized, or its potential to improve health may not be realized. We must hope that after this period of rapid technological adaptation, physicians will become sufficiently facile with the analyte detection, genetics, imaging, and AI that enable personalized medicine to enable these tools to augment their practice, rather than replace them.
But personalized medicine is not just a future state; it is the present state for the field of medicine that diagnoses and treats people living with rare genetic diseases. Among medical fields, none has been affected more dramatically and positively by the technologies that enable personalized medicine than the field of rare diseases. In fact, though it is remarkably underappreciated, it can be confidently asserted that the principles, practices, potential, and perils of personalized medicine are already instantiated in rare disease diagnosis and treatment.
The issues in rare diseases may be less familiar to the attendees of the Vatican conference, so brief background follows. The definition of a “rare” (sometimes also called “orphan”) disease varies somewhat among countries, but all approximate a population prevalence of less than 1/2,000. The United States is thought to have ~25 million people with rare diseases; Europe ~30 million, and 400 million people worldwide. The reason rare diseases can be cumulatively so common (about the same prevalence – 8% – as Type 2 diabetes) is the enormous multiplicity of rare diseases: currently ~7,000 rare diseases are known, and ~250 new rare diseases are being defined every year. Of these, only about 400 diseases have a treatment approved by a regulatory agency such FDA, EMA, or PMDA. The majority of rare diseases have their onset in childhood, and many (though certainly not all) are characterized by high degrees of morbidity and premature mortality. Most (~80%) are caused by a genetic lesion of large effect, and demonstrate mendelian inheritance (i.e., run in families according to rules first set out by Gregor Mendel, the 19th century Augustinian friar known as the “father of genetics”). Familiar examples are sickle cell disease, Tay-Sachs disease, cystic fibrosis, and Huntington’s disease.
Before exploring why rare disease medicine is personalized medicine being practiced today, it is important to reiterate what writers of other position papers have emphasized, which is that “personalized medicine” only rarely seeks to be truly “personalized” – i.e., applied to a single person. Rather, personalized medicine as generally envisioned is a technologically-enabled extension of the longstanding gradual segmentation of sickness from the general to the specific. For this reason, many (including the U.S. National Academy of Medicine and my own institution, the National Institutes of Health) have advocated for use of the term “precision medicine” rather than “personalized medicine”. The NIH defines precision medicine as “an emerging approach for disease treatment and prevention that takes into account individual variability in genes, environment, and lifestyle for each person” (1).
Though it was not intended to be so, this aspirational definition for common diseases unwittingly describes quite nicely the current reality for rare diseases. That is, rare diseases are exemplary of precision medicine because they are, by nature, precisely (or, if one prefers, personally) defined and treated.
Precision medicine endeavors to use in-depth genetic and environmental data to define a genotype- phenotype pairing that is based in definable physiological dysfunction and predicts treatment with a high likelihood of positive response. Rare diseases are defined and managed in just this way today. When successful, the “rare disease/precision medicine” paradigm is remarkably effective. Because rare diseases are frequently metabolic or mendelian genetic, advances in analytical chemistry and genome sequencing have revolutionized their diagnosis. And treatments can be truly transformational, as children treated for phenylketonuria, Gaucher disease, cystic fibrosis, metastatic melanoma, and spinal muscular atrophy have demonstrated, allowing diseases that were previously universally, and often rapidly, fatal to be stopped, reversed, and even perhaps cured. The same is true for some common diseases that have been segmented into biologically more uniform subsets with higher response rates to targeted therapies, HER2-positive breast cancer and ALK-positive lung cancer being two prominent examples. These experiences suggest that the hope invested in personalized medicine is warranted, even if its generalizability has yet to be determined.
Certainly not all scientific lessons from rare diseases will necessarily be applicable to segmented common diseases even if each results in patient populations of the same size. Though there remains active debate and investigation on this point – common diseases such as Type 2 diabetes appear to be caused by multiple genetic and environmental lesions each of which has a small effect, in contrast to rare diseases that tend to be caused by a small number of genetic and/or environmental lesions which each has a large effect. But the operational and ethical issues encountered in the application of precision medicine approaches to rare diseases likely translate to precision medicine applied to common diseases.
On an operational level, personalized medicine is frequently antithetical to current systems of care, which are tailored to detect and treat highly prevalent diseases in large populations. This system has served society well, as infectious and cardiovascular diseases effectively addressed with the “one size fits most” paradigm were conquered, resulting in a doubling of life expectancy during the 20th century. But this paradigm assumes that (for the purposes of the intervention) all people are essentially the same; personalized medicine assumes the opposite – that all people are essentially different. This plays out in rare diseases in what is known as the “diagnostic odyssey”: the migration of patients from one doctor to another over an average of 5-15 years without a diagnosis since each is trained to look for the common and does not recognize the different. In personalized medicine for common diseases, this same dynamic will be at play. Payment systems do not easily accommodate the additional costs of multiple tests required for rare diseases, and they are unlikely to do so for personalized medicine.
Research to understand and develop treatments for rare diseases is different and frequently more expensive due to the dispersed nature of the affected patients and the unusual clinical trial designs required to accommodate diseases that may only have 100-1000 affected individuals worldwide. Since many of the costs of developing a new drug are the same regardless of the prevalence of the disease it is intended to treat, but the costs of development must be recouped and profit generated by sales to a smaller number of patients in the case of rare diseases, prices of new rare disease drugs may be 10-100 times those of new drugs for common diseases in order to maintain ROI. It has been posited that personalized medicine means the “end of the blockbuster model” for drug companies. Evidence shows the opposite: rare disease medicines routinely garner over $1B in sales given the prices they can command. In fact, blockbusters have become so common that a new term has had to be coined for drugs selling multiple $BB/yr: the “megablockbuster”. There is no reason to believe that these pricing dynamics will be different for personalized medicine within the context of common diseases.
Figure 1. “Personalized Medicine” is obverse of “Rare Disease Medicine”
The figure illustrates the point, including issues of therapeutics pricing that have come to the fore so forcefully in the last decade that they have come to dominate some national agendas on drug pricing. Countries with fixed health budgets are today facing ethically difficult choices between paying for lifesaving treatment for a single child with a rare disease costing $500,000/year, and vaccinating large populations.
As such, the experience of rare diseases offers both exhilarating medical promise, and sobering operational and ethical realities, for personalized medicine. Rare disease medicine is personalized medicine – and it is just as miraculous, expensive, and ethically fraught as expected. Solutions to the myriad issues must incorporate new translational technologies that are personalized (e.g., tissue chips using iPSCs), address the relevant scientific and functional issues to decrease the cost of production of new therapeutics, and implement new operational models, including non-ROI models, to produce treatments for small populations. New personalized gene editing and gene therapy technologies, which are directed at specific genetic changes in individuals, raise additional challenges (and benefits), given that they may truly be useful for only a single person, and may only need to be administered once to provide lasting improvement, even cure. These technologies are being developed and applied in patients at a pace that is unprecedented in the history of medicine. While this is to be celebrated, society runs the risk of having the technologies applied in inappropriate, dangerous, and unethical ways, as was recently demonstrated by the reported creation of “CRISPR babies” (2, 3).
In our social media savvy era, these discussions are made even more complicated by the defining feature of personalized medicine, which is that it is personalized. Population health is anonymous, allowing dispassionate discussion based on hypothetical persons with broad public health benefits. Personalized medicine, instantiated today in rare disease medicine, is also media-friendly medicine, particularly since rare diseases disproportionately affect children. Rational discourse on ethical allocation of scarce resources or balancing the need of one ill person with the needs (or rights) of other ill persons, easily gives way to social media storms that prioritize damage control over thoughtful resolution. On one hand, this a manifestation of laudable and desirable care of individual human beings for other individual human beings. But it creates moral and ethical dilemmas that are likely to be increasingly common in the era of personalized medicine. This workshop of the Pontifical Academy of Sciences is an ideal forum for informed, mutually respectful dialogue on them.
REFERENCES
1. https://ghr.nlm.nih.gov/primer/precisionmedicine/definition
2. David Cyranoski D, Ledford, H. Genome-edited baby claim provokes international outcry. Nature
563, 607-608 (2018), doi: 10.1038/d41586-018-07545-0
3. Editorial. The blind babymaker. Nature Biotechnology 37, 1 (2019), doi: 10.1038/nbt.4341.