Dear Mr. President,
I was very pleased to listen to your State of the Union address and learn of your interest in Precision Medicine. As I am sure you know, this has led to a number of commentaries about what this term actually means (here, here, here). I would like to provide yet another perspective, this time from someone who has practiced clinical medicine, led academic research teams and currently works in the pharmaceutical industry.
Let me start by acknowledging that I know very little about your plan, but that is because no plan has been announced. However, that inconvenient fact should not prevent me from forming a very strong opinion about what you should do. Similar behavior is observed in politics (which you know well) and sports radio (see for example “Deflate-gate”). So here it goes…
I want to clarify my definition of “precision medicine” (see here for my previous blog on how this is different from “personalized medicine”). In the simplest of terms, precision medicine refers to the ability to classify individuals into subpopulations based on a deep understanding of disease biology. Note that this is different than what clinicians normally practice, which is to classify patients based on signs and symptoms (which can be measured by clinicians as part of routine clinical appointments). Because clinicians cannot easily “see” this underlying biology during routine patient examinations, new technologies such as genome sequencing and molecular profiling are required to reveal these biological subpopulations.
But what are these technologies?
There are many technologies, all of which have strengths and limitations. I personally think that whole genome sequencing is one of the most promising, as it allows genetic differences among individuals to be linked to clinical outcomes such as disease risk and response to treatment. Once genetic data is linked with clinical data, it is possible to search for those special places in the genome that, in certain people, increase or decrease risk of disease. By studying those individuals further, it is possible to precisely define the genes and biological pathways that create subpopulations of disease.
Thus, my first piece of advice is that you provide funding for large-scale whole genome sequencing in individuals linked with clinical data who are consented to be re-called for further functional studies. As I have commented previously (see here), many thousands of genomes are needed to unlock the genetic basis of common and rare disease, so please be ambitious about the number of genomes that you decide to sequence.
How might society benefit from this information?
There are several applications. The most immediate benefit for patients today is to use newly defined subpopulations of patients to establish more accurate clinical diagnoses; to prescribe approved therapies that are more likely to safe and effective; and to predict clinical outcomes such as disease severity. As an example, autoantibodies can subset patients with different autoimmune diseases (e.g., rheumatoid arthritis, lupus, scleroderma, vasculitis) into molecular categories that differ in clinical outcomes. Other examples can be found in a 142-page document on precision medicine from the National Academy of Sciences (see here).
It is worth noting that many of these patient subpopulations were not defined with modern molecular tools such as genome sequencing and expression profiling. A skeptic may claim that there are few examples of modern precision medicine in action. In my view, this skeptical position has justification. However, a more positive perspective is that there are enough anecdotes – especially in oncology – to make the investment worthwhile.
What is the sequence of events that will unfold to realize precision medicine?
I see two paths: (1) define patient subpopulations with new molecular technologies and then use existing medical tools (e.g., approved medicines) to improve patient care; and (2) define patient subpopulations and then develop new tools (e.g., new medicines) based on these insights, which will over the course of many years improve patient care. While the two paths are not mutually exclusive, my prediction is that most benefit will come from the second (and much longer) path. Supporting this view is the current list of approved therapies with FDA pharmacogenetic labels, where many examples of targeted therapies were developed only after patient subpopulations were defined based on a deep understanding of disease biology (see my previous blog here).
With regards to the stages of drug discovery and development, here is how I see the future unfolding. The stages are: (1) discovering a drug target, (2) generating a therapeutic molecule against the target, (3) testing proof-of-concept in humans, and (4) testing efficacy and safety in large clinical trials.
In the first stage, drug discoverers will use natural experiments in humans to uncover “causal human biology”. The key to causal human biology is establishing a cause-effect relationship between perturbing a target and a desired effect on human physiology. This is the first place where precision medicine will have an impact, as patient subpopulations will be used as a starting for defining targets. In particular, I believe that large-scale sequencing in patient populations will help identify new drug targets (see here).
In the second stage, drug discoverers will generate a therapeutic molecule that reverses the molecular defect observed in human biology. The same molecular tools used to classify patient subpopulations will be used to develop molecular readouts of target modulation. That is, precision medicine tools will be used to develop robust molecular biomarkers that accurately measure causal human biology.
In the third stage, drug developers will administer therapeutic molecules in small patient populations to test therapeutic hypotheses. Here is where precision medicine becomes more obvious. The small patient populations should be defined based on molecular subtypes of disease. Further, the molecular biomarkers developed in the second stage can be used to establish proof-of-concept that a therapeutic molecule is doing what it should be doing in humans.
In the fourth stage, the concepts of precision medicine will be carried into larger patient populations to test efficacy and safety. Again, patient populations for these larger clinical trials should be defined based on molecular subtypes of disease.
If all four stages can be connected via precision medicine, then I predict that the success rate for new therapies will improve. Moreover, I predict that the drugs developed based on precision medicine principles will have a greater impact on improving human health. And if success rates and impact both improve, then medical costs go down.
Thus, my second piece of advice is that you preach patience, as the real benefits will take many years. Remember that that it takes more than 10 years to progress from discovering a new target to formal FDA approval. When your constituents demand more immediate impact, please emphasize the economic impacts that this infrastructure will bring. For example, if other countries invest heavily in a precision medicine infrastructure (e.g., Genomics England) but the U.S. does not, then it is likely that opportunities and jobs will shift to these other countries over time.
Thank you for taking the time to consider my point of view. I wish you luck in this ambitious endeavor.