We need to make a hierarchy for personalized medicine

Personalizing medicine is among the salient themes of modern advance, and clearly among the more widely captivating. President Barack Obama cited this area as a priority for the nation in the context of the cancer moonshot he endorsed in his State of the Union Address in 2016. In the past year, two radical cancer treatment advances have been announced, one for leukemia, one for lymphoma, both involving the genetic reengineering of a patient’s own cells into customized chemotherapy.

The methods of customization and the means of delivering care have evolved considerably over recent years. Personalization at first relied much on what was called metabolomics, and at times proteomics (or proteinomics), collectively an effort to use assays of metabolites and proteins to create a personalized patient profile. The basic idea was that the identification of such “products” in blood could be traced back to their biochemical assembly lines, thus illuminating any aberrations there. Those, in turn, could (in theory) be targeted with drugs, or more often, nutrient supplements and lifestyle interventions, to restore native balance to the body’s protein factories.

Since the sequencing of the full human genome in 2003, excitement for the idea that “DNA is destiny” has abated with good reason, but general enthusiasm for genomic customization has not. There are at least two popular applications of genomics for therapeutic refinement: pharmacogenomics, and nutrigenomics. The latter involves identifying genetic variation, usually in the form of what are called SNPs (pronounced “snips”), which can be used to predict variable responses to diets, foods, and nutrients.

As an example, a study by colleagues at Stanford University showed that weight loss was roughly comparable for groups of people assigned to distinctly different diets. Whereas that would have been the answer in the past, it was just the question in the genomic age: why do some people do well, and others poorly, on each of these diet assignments? Can genomic profiling be used to determine who is most likely to succeed on a specific diet? That work is now on going.

Other means of medical customization are emerging. A fascinating study in Cell in 2015 showed varying glycemic responses to the same foods by individuals with differing microbiomes — and a company to customize dietary recommendations based on microbiomics has already been established. The influence of the epigenome, the bulk of our chromosomal real estate responsible for determining what our genes actually do, is known to reverberate across generations; and has been shown to be malleable in ways genes are not. Epigenomics thus figures among the next “big things.”

With science driven forward by the view through a microscope, I worry that we may personalize medicine while overlooking the person. Those famous blind men examining the elephant in its isolated parts are a precautionary tale about the liabilities of reductionism run amok.

I have a remedy in mind, and have coined the term “hierarchical personalization” to characterize it. In case it catches on, remember you heard it here first.

First, we cannot personalize care for genes and microbes, and leave out the person. So, the first tier of hierarchical personalization is holistic. I have the six key domains of importance to lifestyle medicine in mind: social connections and relationships; stress and mental health; sleep; toxic exposures (substances, chronic pain, harsh environments at work or home, etc.); physical activity; and dietary pattern. Fundamentally, we can’t hope to personalize care without knowing who a person is, in the context of their life.

Second, there is sequential personalization. This is how holistic care can go from platitude to method. Perhaps someone wants to improve their diet and lose weight — but they have severe chronic pain, or terrible insomnia, or a toxic marriage. Personalization means figuring out what actually needs attention first so the wherewithal to address the next goal, and the next is cultivated. The ascent to vitality does not involve a helicopter ride; it is made step by step, up a spiral stair.

Then, I think, and only then, can all of the actual and potential power of various “omics” be applied to personalize the best approach to relieving pain or stress, to improving sleep or diet. We can call this final tier specific personalization, as it contributes potential insights to domainspecific therapeutic approaches.

I fully appreciate the potential power, much of it still inchoate, of personalized medicine. But I know the perils of reductionism, too. Diverse approaches to personalizing medicine will only improve care if attention is not unduly diverted to the parts of parts, and away from the person in the room.