On personalized medication selection...

I attended a talk last night (Tuesday, June 2nd; 2015-06-02) on the future of medicine.  Doctors Hood and Srivastava gave an interesting presentation about such things as the history of gene sequencers (which Dr. Hood had much to do with), a model for thinking about medical care called "p4 medicine", wherein care and treatments are "Predictive, Preventive, Personalized and Participatory" (sorry, I'd add an Oxford comma, but I'm quoting that website directly), and a number of other technical and research advancements that are influencing what Dr. Srivastava described as a coming revolution in how medical care will be done.

And he (Srivastava) spent a bit of time talking about how this has been getting promised for some time now, but that he really felt that it was getting to an inflection point, where the difference between now and 20 years from now would be dramatically greater than the difference between 20 years ago and today.

I really hope he's right.  Because here's my take:

What we're doing today is primitive enough and problematic enough, in places (and great, in other places, don't get me wrong), that it will be looked back on at some point or other (I'll make no predictions of my own about time frame) as being utterly barbaric and cruel.

Example from my own life:

I had a horribly negative reaction to several different anti-depressant medications. In particular, Zoloft, Effexor, and Prozac.  They gave me akathisia.  They increased my suicidality (something I've ever struggled with) to the point (this being relatively rare in my life) of actually attempting something, at least once while on each of the above medications (so at least 3 attempts), and a couple more times (5 or 6 attempts, total) after this round of "therapies", which I still attribute in large part to the fact that I had been on them.  The effects of these medications (I believe) even had me bordering on homicidal, directed specifically towards my psychiatrist at the time, who, besides prescribing me not one but two other meds somewhat closely related to the first one I had the bad reaction to, with relatively little care taken to making sure I was well supported in avoiding or quickly stopping subsequent reactions, also would literally fall asleep during the course of our 15 minute sessions.  Regularly.  (And I heard from someone I met during a hospital stay around that time (October of 2000), who was also a patient of his, that he'd done the same to her.)  How can someone possibly be giving good care in such a manner?

Anyway, I could probably go on on the horrors of that time, and in particular my reactions to those specific medications.  The point, though, is that my reactions were horrible.  And yet, many people seem to report good results with them.  There was a big part of me that wanted SSRIs to be completely banned, on the premise that I wasn't the only one to be experiencing this (we may have lost comedian Phil Hartman to akathisia).

And while that (banning a med) hasn't happened with Prozac, Doctor Srivastava relayed (after me asking a question about this) the fact that a number of drugs have been completely pulled off the market, because of things like causing heart arrhythmias or liver failures or some such.  Which means that that people who were actually getting solid benefit from the drug (which is a little more questionable in Prozac's case than it seemed to be in some of the other drugs Srivastava was talking about) could no longer get it!  This has the potential, with some meds, to have just about as bad an effect from not having meds as those few who have a negative reaction to taking them (deadly disease, deadly med; which death will you get?).  But who can weigh the two?  Is it worth sparing a few folks from the negative experience, if that costs thousands or even millions of people the chance at a benefit?  What if (as I believe is the case with some medications) the negative experience is almost certain to be a fatal one, and yet, the beneficial experience is one that can save lives?  Would you kill one person to save five?  Not an easy question to find a satisfactory answer to.

But, there's a way to address it, I think, that is more prone to a satisfactory answer: Personalized prescriptions. In their talk, the speakers last night mentioned treatments that involved taking cells from someone's skin or blood, turning them into a stem cell line (because that's doable, these days), possibly modifying them along the way (for example, "teaching" T-cells how to detect a particular form of cancer cells, which they can then destroy with methods already available to them), then re-specializing the line again into the cells for a particular organ type, and re-inserting these new cells (or even whole organs made from them) back into the patient that the cell line originally came out of.  Pretty personalized!

And yet there's an even simpler version of this, that I've been thinking of (and deeply wanting) for a long time now, and which I asked about during the Q&A: It strikes me as obvious that, in gathering the kinds of data they're talking about gathering (whole gene sequences for an individual, as well as numerous other factors), and taking a systems approach to this (an idea that Dr. Hood in particular spoke to), we may at some point find (when correlating this data with other individualized data on patient responses) that there are markers that can help us tell in advance whether a particular drug is likely to cause a particular reaction, positive or negative, in a particular patient.  In his response to my question, Dr. Srivastava even took this idea a step further than I'd imagined, in mentioning that the embryonic stem cell lines that can now be cultured from skin or blood cells, could then be used to grow, say, heart cells, and, for the case of medications that create heart problems, you could actually test reaction of the individual patient's cells from the particular organ that matters, with the actual drug... all without actually putting the drug into the patient's body itself.  If a negative reaction happens, it happens to cells in a culture dish.  (Which hopefully most of us don't consider to be sentient life, and therefore have fewer qualms about damaging – though I must admit, especially once you start talking about growing whole organs in a lab, one starts to wonder where the line is drawn there... I want to mention that quandary, though, without going into it; that's a discussion for another day, I just don't want to pretend it's not one worth having – though I don't believe it was actually mentioned in the talk, sadly.)

Anyway, one way or another, it strikes me that we have the opportunity to know, with at least a high degree of confidence, whether a particular patient will have a positive, neutral, or negative experience with a particular medication.  And if we can know that – either as a stronger probability (changing "1 in 10 have a negative reaction" to "you have markers that indicate a 99.9932% probability of such-and-such reaction"), or as a relative certainty ("in a lab, your cells responded in such-and-such a manner", and note that the decision about which drugs to even test in this way could be shaped by the stronger probabilities, to start with) – then we can choose which patients to give Prozac to, which to give Wellbutrin to, which to give something else to, and who knows?  Maybe even which ones to send to a sleep study because they're likely to have sleep apnea, which, if properly treated, might rid them of the depression.  (More generally: we might have a better chance of treating or even curing underlying causative conditions, rather than merely treating symptoms.)

Which gets me, roughly, to the actual question that I asked.  While I don't remember my exact wording (perhaps a recording will be released at some point, and you can listen for yourself), the thrust of what I was attempting to get at was this:

It seems to me that culturally, current health care providers simply have no interest or willingness to even begin to try to do tests like this.  What roadblocks are there to gaining acceptance of personalized medicines amongst health providers, and what things can we do to help facilitate the cultural change that may be necessary to go along with all these technology changes, so that we can actually have the revolution in medicine that Dr. Srivastava was predicting for a mere 20 years hence?

There was discussion, too, about n=1 experiments.  Dr. Hood in particular mentioned how many dismiss these as not useful... And of course, as predictors of how an overall population will respond to something, indeed, they often are not!  (Though see link for some ways they might have broad applicability, even still.)  But in terms of predicting how that one individual might respond to something, it seems to me they're probably the very best way to go!  There could also even be considered to be n>1 experiments in the sense of having multiple cultures having the same experiment done to them, and in doing so, figure out some sort of general probability based on not just a single result, even if all the results are on cells from a single patient.  And that's to say nothing of the fact that many different experiments could also be done from a single culture line, thus giving a list of likely outcomes for a list of available treatments.

And I believe that's now set the stage for the thing that I'm really wanting to express:

What could be better, here, is if medical providers had the tools, time, and inclination to bring more science directly into their practice of medicine.  To put quantitative tools to work, whether that's a personal activity tracking device (and/or heart monitor, etc.; these got some discussion in the talk), or blood tests (which, who knows, may someday be done by personal tracking devices! There's certainly been attention on measuring blood glucose levels in such ways, though apparently that's not quite there yet), or DNA sequencing.  To start gathering that data, even if we don't yet know what to do with it, because we haven't yet figured out which markers correlate with, say, which responses to a medication.  But we'll never know, if we don't gather the data.  So, one thing I'd like there to be more of is data gathering around this stuff.

Now, it's worth mentioning that there are concerns around this.  Privacy concerns stand out to me in particular, especially if we're trying to gather large amounts of data, and then share it in ways to allow research to be possible on correlations (and, eventually, causation research) between various markers and various responses.  This is something that I hope will be given a lot of careful thought along the way for this.  And yet, I think that there are ways to handle such things that will at least be helpful, and... anyway, I hope it's not cheating too much to defer that discussion to another day, as well.  Because here's the thing:

With hard data, and analysis of such data, we can begin to make much better predictions about outcomes for various treatments.  We could prescribe Prozac to some depressed patients, CPAP to others, and a small specific dietary change (say, avoiding sucrose) to still others, and have them all respond better than they might to other interventions, because the intervention chosen was chosen based on data specifically about them.  We could know who not to give a particular medication for, say, diabetes, because we know that that particular person would have an adverse liver reaction (say).  And we could thus instead give them some different medication, which their liver would handle just fine.  Or a third medication, which their liver would also handle just fine, but would also be more efficacious for them in particular!  Or figure out what specific foods they need to eat more or less of, that would make them no longer need any other treatment at all!

Stuff like that.  So... let's get our data on, shall we?

No comments:

Post a Comment