I have previously expressed my concern about the surge of interest in precision medicine, and how it may conflict with our core purpose of promoting the health of the public. In May 2016, Dr. Muin Khoury, director of the Office of Public Health Genomics at the Centers for Disease Control and Prevention (CDC), and I had an online point-counterpoint “debate” in the form of a webinar sponsored by the Precision Medicine and Population Health Interest Group at the National Cancer Institute. The audio presentations and slides of this dialogue are available here. After the debate, we teamed up to lay out the arguments on both sides of the issue. We addressed the current tension between the enthusiasts and the naysayers, aiming to forge an enhanced medicine–public health collaboration in the precision medicine era and a consensus on an approach to precision medicine that can contribute to improving the health of the public. We subsequently summarized the points we made in our debate in a paper that was just published in the Journal of the American Medical Association. Below is a somewhat expanded presentation of the arguments made in that paper.
Point: Precision Medicine Is Unlikely to Improve Population Health
We start with my concerns about the rise of precision medicine and why it may not bode well for population health. There are, in my view, three fundamental reasons that precision medicine may not improve the health of populations.
First, disease pathogenesis—particularly pathogenesis for non-communicable diseases that will be the dominant causes of ill health in the coming decades—is extraordinarily complex. There is no question that genetics contribute to the production of health, but that contribution is almost certainly small. There is abundant evidence for the association between specific genes and health indicators, including, for example, hypertension or different cancers. However, in almost all cases, the associations between genetic variants and health have small effect sizes, suggesting that the contribution of the specific genetic variant to health is much less robust than the contribution of well documented behavioral and social factors. Small genetic effects are further complicated by the multimechanistic link between genotype and phenotype, with incomplete penetrance and differential expression of particular genes. This substantially diminishes enthusiasm for an “all-in” approach that aims centrally at identifying particular genotypes, as a way of improving health.
Second, one of the central promises of the precision medicine agenda is the identification of predictors, genetic or molecular, that can be used to identify the right “personalized” treatment for disease. This may indeed prove to be the case for some diseases, especially cancer, where an understanding of the molecular pathogenesis for particular mutations can allow individual “fingerprinting” for a specific cancer, but is unlikely to be the case for most other complex diseases. Centrally, the challenge arises from the mathematical foundations of genetic epidemiology. Large population samples are used to identify associations between genotype and phenotype. These associations (nearly always small, as noted, above) have, however, limited and, in most cases, minimal capacity to predict phenotype in individuals; the ultimate goal of precision medicine. It would take substantially stronger associations—several orders of magnitude greater than what we typically identify in genetic epidemiologic studies—to pinpoint genetic markers that can fruitfully predict the health of individuals.
Third, one of the potential benefits of a precision medicine approach, even if we could not identify genetic or molecular markers that predict individual phenotype, would be if individuals, alerted to their belonging to a higher risk group, would change their behavior, hence lowering their risk. While this may seem intuitively plausible, current data suggest that this is not the case, and the best evidence suggests that individuals do not much change their behavior even if alerted to their being a part of a higher risk group.
These three challenges do not, in any way, diminish the potential of a precision medicine approach to help elucidate particular mechanistic pathways that can, over time, shed light on disease pathogenesis. They do, however, diminish one’s enthusiasm for the potential of precision medicine as an approach to improving the health of populations. Does this, however, suggest that an unstinting focus on precision medicine, given the potential for mechanistic insight, poses a challenge to the health of populations? Three reasons suggest why this might be the case.
First, the United States faces extraordinary challenges to the health of its population. We have over the past 30 years fallen behind all our high-income peer nations in health attainment, and have a patchwork system of health care delivery which leaves us doing far more poorly than these same peer nations on a number of metrics that assess access to health care. The country is characterized by persistent, and growing, gaps between health haves and health have-nots, stratified by axes of income and race/ethnicity. The solution to these challenges lies not in an ever-greater focus on the individual, but rather in a focus on the social, economic, and structural drivers of population health that are ubiquitous and ineluctably linked to our national health achievement. There is no prima facie reason why redoubled investment in these forces cannot coexist with a precision medicine effort that aims to identify mechanistic links that can inform our understanding of pathogenesis. However, there is little doubt that the centrality of the precision medicine effort to our national health research agenda stands to distract and detract from efforts that aim to label the foundational causes of health—without action on which we shall have little, if any, success in reversing the trends in the country’s poor achievement in population health.
Second, and relatedly, the precision medicine agenda has, inevitably, commanded resources, both towards its specific goals, and, more importantly, towards shifting what is funded by health research agencies. Funding for grants with a population health or public health goal has dropped over the past ten years at the NIH, while funding for genomic research has grown proportionally. These are simply sentinel markers of the broader shift in the field, being steered in no small part by the precision medicine agenda. Unfortunately, this may lead to an emerging generation of health scientists who see the world through an individualist lens, who are not engaged with the factors that we must engage with in order to improve the health of populations, and who could be ill-prepared to help tackle the very real challenges that have caused our health indicators to consistently lag behind our peer nations for the past several decades.
Third, the precision medicine agenda rests on the promise of better health for all. This promise is enormously seductive. Who, after all, would not like her or his health predicted, and treatments tailored towards improving individual health? Unfortunately, this promise breeds other promises, including most recently the cancer moonshot, which may echo previous promises that may not have lived up to the expectations they created. These promises could breed disillusionment in the goals of health science, with potential long-tail consequences for public confidence and public investment in our science and our goals.
Counterpoint: Precision Medicine Can Improve Population Health
By contrast, Dr. Muin Khoury argues that there are three fundamental reasons that advances in precision medicine could improve population health.
First and foremost, population health is improved by complementary individual and public health level approaches to health care and disease prevention. Improving health requires a multifaceted approach which includes access to effective treatments and disease prevention efforts, as well as policy and environmental interventions. There is no argument that a focus on the wider environmental, structural, and social determinants of health is of great importance for improving health and addressing health inequities. However, by pitting the health of individuals against the health of populations, we risk widening an unnecessary schism between medicine and public health. Population health planning and prevention require that efficient use of resources be directed at those most in need. In addition to other characteristics such as age, gender, and location, genetic susceptibility factors could allow for the stratification of populations into risk groups for multiple chronic diseases and could thus provide for much more efficient and effective prevention and treatment strategies. For example, a recent large population study suggests that the use of age and polygenic risk score (even though individual genetic variants have small effect sizes) could be used to stratify women into risk categories for breast cancer screening much better than age alone. Moreover, there is a profound ongoing change in patient engagement and community empowerment in health matters. Placing individual citizens at the center of health care and disease prevention will require that they be given much more information about themselves in shared health decision making. Ignoring the importance of advances in genomics and digital technology is not only counterproductive but potentially harmful to population health.
Second, even without the novel discoveries that will result from the US Precision Medicine Initiative, we have plenty of evidence that a genetically targeted personalized approach to health has already demonstrated a population health impact. After all, newborn screening is the largest established “precision medicine” public health program in this country and around the world. Currently, more than 4 million babies are screened every year in the US for more than 30 genetic, metabolic, and other conditions to find 10,000 babies who require immediate medical preventive interventions that can save lives and prevent morbidity and disability. While the US Precision Medicine Initiative will lead to numerous discoveries of new genome-based associations and possible interventions in adults, it will take time to yield these dividends. In the meantime, we have a real opportunity for near-term population health impact by implementing in practice evidence-based genomic interventions. The CDC Office of Public Health Genomics has created a three-tiered evidence-based classification schema of genomic applications based on evidence for their use. At the top of the evidence pyramid are “tier 1” genomic applications that have sufficient synthesized evidence for clinical validity and clinical utility to provide meaningful and actionable information to providers, patients, and the public. In addition to an increasing number of tier 1 cancer genomic applications that influence treatment choices, there are several examples of tier 1 conditions for which genetic testing is recommended in healthy people at high genetic risk—conditions such as hereditary breast and ovarian cancer syndrome, Lynch syndrome and familial hypercholesterolemia, common autosomal dominant conditions associated with preventable premature death from cancer, and heart disease. An estimated 2 million people in the US have one of these conditions, and most are not aware of their risk. A shared clinical-public health challenge is how to implement genomic knowledge in practice to save lives now. In the US precision medicine cohort, there will be thousands of undiagnosed, unrecognized patients with these disorders alone, who can be identified through genomic sequencing and clinical data. These individuals and their relatives can take advantage of interventions to reduce their risk—an immediate population health impact from this study. Also, numerous pharmacogenomic traits with varying levels of evidence on drug effectiveness and side effects are available for dozens of currently used drugs. Finally, there is the possibility of expanding carrier and prenatal testing for various genetic disorders, a significant reproductive health issue in the US and around the world.
Third, and perhaps most importantly, precision medicine is not just about “genes, drugs, and disease.” The same technologies and “big data” that are propelling precision medicine forward are leading to a new era of “precision public health” that goes beyond personalized treatment of sick individuals. The word “precision” in the context of public health has been described as improving our ability to prevent disease, promote health, and reduce health disparities in populations by: 1) applying emerging methods and technologies for measuring disease, pathogens, exposures, behaviors, and susceptibility in populations; and 2) developing policies and targeted implementation programs to improve health. One already impactful example of precision public health is the use of genomics in the investigation and control of infectious diseases. Pathogen whole genome sequencing is rapidly changing both clinical and public health microbiology. These technologies deliver more precise information to enhance public health’s ability to detect, track, and respond to infectious disease threats. Since 2014, CDC has integrated applied genome sequencing and bioinformatics with traditional epidemiology in outbreak detection and response for numerous pathogens, leading to earlier and more effective interventions that save lives and reduce cost.
Forging a Consensus
It goes without saying that there are clear tensions at the intersection of precision medicine and public health. We suggest in the paper, however, that there are ways forward in which precision medicine could enhance collaborations between medicine and public health to address population health problems and disparities.
Much of the current focus of precision medicine involves developing new drugs for personalized treatment of cancer and other diseases. Moving forward, we must emphasize joining biological with social/environmental determinants of health to develop “precision” approaches beyond pharmacologic interventions, which can be implemented at both the individual and population levels. For example, biological knowledge of genetic susceptibility to environmental and occupational exposures could lead to population-wide policy protection based on thresholds determined by the most susceptible individuals in the population, rather than individual genetic testing with exposure avoidance only in susceptible individuals. It is also the case that a fundamental concern for population health must be with measuring and addressing health inequities. As with all new technologies, genomic technologies have the potential for widening the divide between the haves and the have-nots.
A major challenge going forward is how to use emerging information from multiple levels—from reductionist molecular markers (genomics, omics, etc.) to holistic macro-level risk factors (behavior, environment, policies)—in order to develop a better understanding of the determinants of health. “Precision” interventions should be based on evidence that links population data to measurable outcomes in subpopulations, stratified by persons, place, and time, capitalizing on emerging data and new technologies. Critically, even with millions of points of biological data collected on individuals, it may well be that population-level interventions on housing, nutrition, poverty, access to resources, and education may better benefit health than individualized interventions. It is, in fact, more likely that a combination of approaches, ranging from population-wide interventions to specific interventions tailored to higher risk groups, will be required to efficiently improve population health and narrow health disparities.
Since the paper’s publication, some commentary has been published challenging some of the notions introduced in the piece. I am less sanguine than Professor Zimmern is in his blog that precision medicine presents a way of bringing together individual and population approaches that have been apart for some time, but I hope that this work with Dr. Khoury may help to find consensus and a way forward for population health thinking in the precision medicine era.
I hope everyone has a terrific week. Until next week.
Sandro Galea, MD, DrPH
Dean and Robert A. Knox Professor
Boston University School of Public Health
Acknowledgements. This Dean’s Note was co-authored by Dr. Muin Khoury. A parallel version of this note appears on Dr. Khoury’s blog.
Previous Dean’s Notes are archived at: https://www.bu.edu/sph/tag/deans-note/