No, Personalized Medicine Isn’t Going To Save $600 Billion Over 50 Years By Preventing Heart Disease
The hype over personalized medicine has now reached astonishing new heights. In an article published in the Lancet, Victor Dzau, the new president of the Institute of Medicine, and coauthors write that personalized and precision medicine (PPM) could deliver hundreds of billions of dollars worth of improved health in the US over the next 50 years. (Click here for the Lancet press release.)
They used a health simulation model to estimate the effect of improved screening and risk prediction to treat people at high risk for 6 diseases: cancer, diabetes, heart disease, high blood pressure, lung disease, and stroke. They then calculated the resulting gains in life expectancy and quality-adjusted life expectancy.
They calculated that reducing heart disease by 50% “would generate a staggering $607 billion in improved health over 50 years.” More modest reductions of less prevalent diseases would produce smaller but still quite impressive benefits: even a 10% reduction in diabetes and cancer would generate $96 billion and $70 billion, respectively, they write.
I asked Sanjay Kaul to comment on the paper:
The projections from the modeling exercise are predicated on two key assumptions for which there is little or no evidentiary support. First, the claim that genetic testing can lead to identification of individuals at extraordinary risk of developing CV disease in whom “aggressive preventive and interventional strategies have a much greater likelihood of success” rests on a shaky foundation. Take the example of 9p21, the best studied genetic variant for CHD. It is associated with approximately 40% increased risk, a relatively small effect size which is clearly not commensurate with “an extraordinary risk of developing CV disease.” Given the feeble prognostic utility, it should not come as a surprise that there is precious little evidence to show that genetic variants, either singly or in combination, offer incremental improvement in risk discrimination or net reclassification compared to conventional risk models. Second, empirical support for the notion that genetic knowledge of increased susceptibility to CV disease will motivate individuals to modify their lifestyle and improve health is currently lacking. Thus, the predictive utility of genetic tests in therapeutically modifying CV risk remains an unanswered question. The recent announcement of the Precision Medicine Initiative by President Obama has created quite a buzz within the medical and research community. However, the transition of Precision Medicine from a buzzword to clinical reality should be driven by tangible evidence of benefit, rather than some rosy predictions of value derived from a modeling exercise.
Kaul’s points are entirely valid, in my opinion, but I think this paper is even more deeply flawed than his measured response suggests. In addition to the absence of evidence for any of the paper’s main points, as Kaul points out, the paper is completely unbalanced in its extreme optimism. The authors fail to consider or even mention any potential negative effects or unintended consequences of PPM.
For instance, the authors write:
Most personalised therapies currently on the market are indicated for slowing tumour growth or orphan diseases. However, the promise has spawned a rapidly growing industry in which genetic markers of disease and treatment responses are searched on a larger scale.
It is by no means certain whether the successes of personalized medicine in some cancers and a few orphan diseases can or will be replicated in the far more resistant lifestyle diseases of heart disease, diabetes, and obesity. It is possible that we will see some major breakthroughs in this area but Dzau and colleagues don’t even acknowledge that there may be fundamental limitations to PPM in these important areas. Instead, the authors offer a pie-in-the-sky view of the future.
Public health measures versus personalized medicine– The paper by Dzau et al includes an important, nearly hidden assumption that I think requires a much fuller and more open discussion. The authors simply assume that the biggest health advances in the future will come from PPM and that public health measures (which are the opposite of PPM since they target populations and not individuals) will inevitably be less successful. Forget trying to get everyone to exercise more or eat better, they say. Instead, they write,
PPM innovation allows identification of the subset of patients for whom intervention is most valuable. So, for example, although diet and exercise interventions might lower the risk of heart disease among the population in general, adherence to such programmes is notoriously poor in the general population. However, aggressive preventive and interventional strategies targeted to patients whose genetic tests identify them as having extraordinary risks of developing cardiovascular disease have a much greater likelihood of success.
As Kaul suggests, there is absolutely no evidence that the aggressive measures that don’t work in the general population will work any better in a highly targeted population. It’s possible this is true but it is also easy to imagine that this strategy could backfire in myriad ways. Some high risk individuals may well respond like good patients who are eager to learn their more precise risk status and change their behavior appropriately. But others may react quite differently. There may well be a significant backlash from people who can’t or won’t handle the information in the way the PPM experts anticipate. These people can’t be ignored or wished away, and they may well constitute a large percentage of the population.
More generally, I am always amazed by the willingness of the medical elite to completely write off lifestyle interventions. Indeed, success with lifestyle programs has been elusive. But let’s be clear: a gargantuan proportion of medical resources has been devoted to developing pills, devices, and procedures to treat or vitiate lifestyle diseases. By comparison, we’ve spent peanuts on efforts to improve the American diet or help make exercise an attractive activity for everybody.
As Atul Gawande recently noted, just the waste in our health-care spending, which now equals $750 billion a year, is greater than our entire national budget for K-12 education. Imagine if we spent even a small portion of those wasted health care dollars on studying and funding innovative lifestyle programs to encourage exercise and nutrition programs in our schools and elsewhere.
The Missing Downside of PPM– Dzau and colleagues also fail to take into account some of the pitfalls of PPM. Theoretically PPM will deliver highly precise estimates of risk and bring targeted interventions only to the right people, but the reality will inevitably be much murkier. There has never been a perfect test and there never will be. With more testing will come more false positives and false negatives. There will be many people who will not have longer and healthier lives, but those lives will be far more full of worry and concern over their genetic disposition to catastrophe.
The Money Pit– I wonder if any of the Lancet editors had a second thought about this statement:
Although we do a good job reimbursing for therapeutics, we do a much worse job reimbursing for diagnostic tests.
Perhaps, since they are in the UK, this passed the editors by. But no one in the US who has not been asleep for the last generation or so would seriously state that we’ve done a good job reimbursing for therapeutics (aside, of course, from those who make or deliver the therapeutics, since they’ve done quite well for themselves over the years).
But since Dzau and colleagues are so determined to show the multibillion dollar benefits of PPM, they might at least take a moment to consider the possibility that somewhere along the line there might be some companies and physicians and institutions that might take advantage of PPM to line their pockets. Undoubtedly PPM will be good for them. Whether it will also be good for the rest of us and our descendants is another matter entirely.