Personalized medicine starts to hit its stride

The flexible genome [pic from nature.com]

I recently attended a conference at Harvard Medical School on big data and translational medicine.  Translational medicine is the discipline that links scientific discovery (bench insights) to patient care (at the bedside, hence the term that is sometimes used..."bench to bedside").  The general idea is that we never have a truly clean slate of health.  We are conceived, born, live our lives and eventually die.  During this time, we are in a constant state of change. While we have  a solid set of genes in our personal genetic code, they are under considerable pressure from other elements such as regulatory genes and epigenetic signals that are influenced by an individual's internal and external environment.  Some of the changes wrought by these elements are permanent and others are transient but either way, they affect the expression of the our genes in real time throughout our lives and constantly nudge us towards disease.  Add to the mix DNA repair mechanisms that also become less effective as we age, and the scene is set for our gradual demise from before we are even born. At any given time, we have a number of mutations and damaged physiological systems that do not constitute enough for overt disease.  Over time, these effects multiple and at some point we will experience a symptom or two and eventually, a diagnosis.  By the time the symptoms appear, the disease has become quite complex and pervasive, and because of this it is much more difficult to treat. If it could be caught in the earlier stages where there are fewer factors involved, and fewer compensatory systems triggered, it could potentially be nipped in the bud. This is one major goal of translational medicine- to identify the unique signals that show disease or disease risk at a stage where treatment is likely to be more targeted and more successful.

Everything above is old news, but the conference revealed exciting new directions for translational medicine.  For the first time, I have hope that personalized medicine is really starting to become a reality.  Large data sets are being collected, not by physicians or pharmaceutical companies, but by patients.  Over 95% of these patients are allowing their data to be used for massive projects that will attempt to connect early signs and symptoms with the risk of various chronic diseases.  Linking seemingly insignificant phenotypic changes to chronic disease development will eventually allow serious diseases to be detected before they become fully fledged and more entrenched.  For instance, already we know that slow blink rate is related to Parkinson's Disease and this can be used as a flag to look for additional symptoms in patients who are at risk of  Parkinson's.  Whether medicines can be developed and given to patients at these very early stages remains to be seen, but a critical step is incorporating some of these phenotypic or 'patient-reported-outcomes' (PROs) into clinical trials so that the more subtle signs associated with disease can be used to monitor effectiveness of treatments in early stages.  Big data is crucial here, and that patients are willing to share their data at such an unprecedented rate is remarkable.  I have had ideas about epigenetic disease triggers, PROs as trial endpoints, and very early disease intervention for many years, and to see it start to come together as translational medicine is absolutely thrilling to me.
I believe we are on the edge of a precipice and that this science will now begin to accelerate on a logarithmic scale.  Astra Zeneca just signed a nice deal with PatientsLikeMe, which is a strong indication that personalized medicine is about to go mainstream.  I can't image a more exciting time to be in healthcare.  Now, if we can also figure out the economics of the system and make that work in favor of the patient versus the insurers, we would be firmly on the path to better health for all.

The weight of health information

Lately I've been speaking with several folks who are in possession of large amounts of health information for one reason or another.  They all believe their data is valuable and are looking for ways to leverage it to make a business or to advance medical understanding, or both.  Noble ambitions indeed but a question struck me as I listened to the the most recent of them this last week.  Could it be that the weight of all that information is actually crushing innovation rather than stimulating it?  It seems to me that we have a plethora of data looking for a problem to solve.  If we recall the old proverb, 'Neccessity is the Mother of Invention', and then consider that some of our most recent enduring inventions were developed locally and ground-up with little intellectualization at the start (some of the more successful social media forums for eg) we might conclude tha Nike may have it right in their newer proverb, 'Just Do It'.  Serve a local, immediate and acknowledged need; if it works scale it.  This avoids the political nightmares of having to first make a business case to those who don't believe there is one.  The latter is a very difficult way to start a business but we do it all the time.

If we start from the ground up, we all want better health but what does that really mean?  Here are some possibilities:  Better ways to know if we are sick, or going to get sick; better treatments to prevent or cure at costs we can afford; more control over our general health; less hassle in managing health for ourselves and those we care for; a more peaceful existence.  To me, the latter is perhaps the most important of all and to some degree relies on success in the former four.

There are numerous elements to each of the outcomes but it seems like the last place to start in attempting solutions might be in collecting the data.  There's no harm in it per se, provide one doesn't expect the answers to then simply reveal themselves.  We must exert some energy up front if we are to make sense of the problem and move toward solutions in a short time frame and at reasonable cost. However, given that my colleagues have started with the data,  let's look at the process from the data-first perspective.  We are forced to ask questions in the following way: 1) what data do you have and where does the data comes from 2) what problems can you apply it to 3) who might have a vested interest in either the data or a solution.  Very time consuming and like put the cake ingredients into a bowl and then asking what we might make with it.  If we think nationally or globally then the problem is magnified and can be prone to costly red herrings.

If we look at the problem first, then understanding what to do with the data becomes that much easier because the need is already understood. To further simplify, solving a problem locally first allows one to talk directly with those that have the problem rather than having to resort to assertions or assumptions about the population as a whole.  Locally, trial and error can be conducted at low cost and low risk.  Even for those that already have large data sets, perhaps as a side effect of another part of their business (pharmacies or drug companies for instance), this approach is a sensible way to determine how the data might be leveraged more broadly.

Starting with a large amount of data and a set of assumptions means starting with a muddy board.  The danger of solving the wrong problem is great, provided you can get out from under the weight of all that data.   Starting with a white board and a diverse group of passionate individuals with first-hand knowledge of the problem is an innovation waiting to happen.

So.....

1. Identify the problems before you collect/look at the data
2. Become a local hero first.

Anyone have anything to add?