Human Limits

The announcement in early June that the National Institutes of Health (NIH) was launching a 12 year 4.7 Billion dollar Initiative to study the brain caught my attention and spurred a bit of reflection about the NIH’s last big initiative – the Human Genome Project (HGP). We seem to be in an era of “Moonshot Science” and maybe it is time to ask how the clinical reality of the HGP has done compared to the original moonshot vision promulgated by its many advocates? Have we gotten to the clinical equivalent of the moon?

One of the main ideas behind the HGP was simple: if we know how the DNA based genetic code varies between people with and without certain diseases we will gain insight into the causes of disease. More importantly we can then screen for the genetic variants associated with diseases and offer pre-emptive and preventive interventions to those at greater risk. In some cases we might also be able to tailor drug therapy based on the individual’s genotype. For things like hypertension, heart disease, diabetes, obesity and some cancers this led to what has been termed the common-disease common-variant hypothesis.

At one level the HGP has been a massive test of the common-disease common-variant hypothesis. What are the results so far?

First, for most of the big killers in the developed world mentioned above, clear cut patterns of genetic risk have not emerged. Instead hundreds of genetic risk variants with very small effect sizes have been discovered and the distribution of these risky gene variants is the same in people with and without the disease of interest.   Further, when information about these risky gene variants is plugged into risk prediction scoring systems commonly used by Drs, the predictive ability of the scoring systems do not get better. In fact for things like heart disease, diabetes, and hypertension simple knowledge of a patient’s height and waist size tells you far more about disease risk than genetics does.

Second, for common diseases the idea that we could get a gene test and then decide which drug is best for each patient is not working out as neatly as anticipated either. In fact for things like high cholesterol and high blood pressure the guidelines seem to be moving away from the idea that a specific drug might be picked for a specific patient based on genetics.   For the commonly used blood thinner Warfarin, which can be tricky to dose, there was hope that information about the genetics of its metabolism would lead to better dosing algorithms, but unfortunately the clinical trials testing this idea failed.

Third, there is increasing evidence that it is difficult for both patients and providers to put information about genetic risk in context. Tell people they are at lower risk and some pay less attention to behavioral factors like diet and exercise that might help them prevent a variety of conditions. Tell them they are at increased risk and at least some get fatalistic and might also pay less attention to behavioral factors like diet and exercise that might help them prevent a variety of conditions. There is also evidence that the perception of increased risk can lead to more tests, biopsies and interventions which all cost money and at least some of them like preemptive surgery have their own set of risks.

Fourth, the HGP seems to have had the “side effect” of encouraging the development of animal models (mostly mice) where an engineered genetic variant leads to a predictable pattern of disease in the animals.   This can then lead to the development of drugs that cure disease in animal models but fail clinical trials. One example is Alzheimer’s disease and the idea that it is all about amyloid. A number of anti-amyloid drugs work in the animal models designed to generate a buildup of amyloid in the brain, but twenty plus have failed clinical trials in humans.   Additionally, there seems to be a disconnect between the reductionist ideas about amyloid as the cause of Alzheimer’s and the epidemiology data showing the major risk factors for Alzheimer’s include things like diabetes, hypertension, and physical inactivity. In some animal models, the disease of interest does not even show up if the animals are given access to minimal amounts of exercise.

So, based on the issues outlined above I would say some skepticism about moonshot biomedical science including the brain initiative is warranted. This skepticism also seems warranted because so many biomedical breakthroughs seem to be reverse examples of Yogi-Berra’s line about losing by making too many wrong mistakes. In medicine sometimes we win by making the right mistakes. A good example is the drugs that were designed to block the growth of blood vessels in tumors and “cure” cancer. Their effects on cancer have been modest but they have been vision saving in macular degeneration. Likewise drugs like Viagra started out as treatments for heart disease and their effects on erectile dysfunction were an unexpected and profitable surprise for the drug companies. The back story on Viagra is even more interesting because the fundamental observations ultimately responsible for the drug (which led to a Nobel Prize) were the result of a “mistake” made by a lab technician.

In the early 1970s the physiologist Julius Comroe and his anesthesiologist colleague Robert Dripps catalogued the 100 or so key discoveries needed to do then cutting-edge open heart surgery and concluded that about half of them happened by serendipity. Comroe and Dripps questioned the wisdom of too much goal directed big science just at the time the “war on cancer”, an even more dramatic metaphor than moonshot, was starting. As Gina Kolata reported in the Times in 2009, victory in that war is nowhere in sight.

However, there is hope. What the HGP has not revealed along with new ideas from the field evolutionary biology are leading to much more nuanced views of the role that DNA has in influencing the fate of animals including humans.   These ideas are showing that DNA is not a simple read only code or program but that it operates in a way that can actually adapt to the environment. My colleague Denis Noble from Oxford has argued that the genetic code is in fact not a code at all but more like a musical keyboard that can be played by other parts of the body and even the environment, behavior and culture.   So perhaps we need to stop thinking about biomedical problems as moonshots or wars and more like music. With that mindset maybe we can get to the “right mistakes” generated by all of this big science and find the clinical insights a little bit faster.