Tension in “Translational Medicine”
A couple weeks ago, I attended a “Genomic and Personalized Medicine Forum" at Duke. The presentation by Sandeep Dave focused on the heterogeneity of cancers and challenges that heterogeneity of cancer poses in research and, of course, in treatment. As is often the case with well attended seminars, the Question-and-Answer discussion was very stimulating. One theme that popped up in Q&A has stuck with me.
Fitzgerald’s test: “… hold two opposing ideas in mind at the same time and still retain the ability to function.” The issue of power and scaling of studies came up. What would a researcher (like Sandeep) do with larger scale studies — instead of a few thousand “subjects” what about 100,000 or more? Would that give power enough to tease out heterogeneity? Victoria Christian asked the question; and, as she posed it, it struck me that it fit quite well. I know an oncologist — a prominent researcher of ovarian cancer — who had run into the Impenetrable Wall of Power in his studies. He and his international collaborators could no longer confidently refine their list of genetic variants relating to the onset of ovarian cancer, because their (prodigious) data no longer could support further claims that he felt were “just on the edge” of validity.
Of course, number and statistical power are very, very relevant.
But, nonetheless, there was something jarring about the 100,000-subject question in my mind. The talk, after all, explored the heterogenous quality of cancers — a heterogeneity that in effect makes every cancer unique. I think, in fact, Sandeep underplayed the heterogeneity of cancer genomes perhaps to simplify matters, saying that those who suffer from cancer “have two genomes” — the one they were born with and the one in their tumors. (In the Q&A, Jeff Marks remarked that there is evidence that tumors in fact are things with multiple genomes, so it’s not just two but God-knows-how-many genomes that cancer sufferers carry.) Cancer often drives cells into bizzaro-world imbalance, and genome stability staggers as stabilizers (like drivers of cell death) falter. Sandeep, of course, elaborated on Jeff’s observation to note that genomes themselves have other modes of modification, and not just DNA sequence variations.
So, I was sitting there in the room, wondering how I could hold the thought of the huge combinatorics of cancer heterogeneity in mind — combinatorics that molecularly make every cancer patient unique. And, at the same time, I was wondering how it was possible to find 100,000 patients for a cancer study. How do you group 100,000 people, if they bear the weight of a disease that, molecularly speaking, distinguishes each one of them from the other 99,999? We talk about “molecular phenotypes,” but where are lines drawn in a disease like cancer?
Maybe cancer is actually a plural noun — a grouping, true, but a grouping of individuals.
Certainly, we can group by whatever part of the body is afflicted or use the categories of disease that pathologists and oncologists use every day. They’re handy, but — relative to the range of molecular phenotypes — these also seem a rather gross method of categorization. Breast cancer, for example, affects the breast, though breast cancer for today’s cancer researchers and physicians falls into molecularly defined “subtypes.” One can say that breast cancer is really not one disease, but many, and thank God there are treatments available for different molecularly defined breast cancers. Still, though they are informed by molecular qualities, breast cancer subtypes are themselves groupings of genomic dysfunction and alteration.
There’s a heck of a lot of variety in there.
You often see what you’re looking at. The Q&A session highlighted for me some of the cultural stances that animate interactions among so-called “basic scientists” or (sometimes uttered disparagingly) “discovery” scientists and clinical scientists. The distinction is — and has probably always been — a matter of interest and focus: the basic scientists seek to understand processes and the clinical scientists seek to shape and alter processes of disease. It is apparent what underlies these two perspectives and how each is ennobled in its own right.
It’s also pretty easy to see how each would conflict with the other.
Vicky’s question about the 100,000-patient study and my qualms about even that sized study’s adequacy call upon two different frames of reference, each rooted in personal experience. The one (Vicky’s) nobly presses toward relief of disease and toward treatments that actually matter; the other (probably, mostly, tentatively my own) is an equally noble and valid press toward the understanding of processes of disease and its contrast with the processes of health.
But it’s not just that overly wrought distinction either, since each frame of reference relies upon and encourages the other — each just places an end point differently: the one with successful treatment and the other with clear understanding. This contrast (conflict?) is a gap that traditional organization of research has avoided, and yet in the context of translation medicine, the contrast simply can’t be ignored. Translational medicine posits a reality that assumes the two frames of reference to be compatible, somehow compatible.
And they are, I believe and hope, in the forums like those that took place a couple of weeks ago.
Where we’re at and where we might be. Perhaps, in the end, the scientific situation now is that we have too little information — despite the prodigious amounts of data that cancer researchers are producing in sequencing projects. And as a result, for clinicians, the picture is both positive and negative. “The positive is that the current low response rate to treatment of some cancers may well be a function of not choosing the right drug for the right molecular type of cancer,” blogged Ewan Birney. “By ‘typing’ the cancer better, there can be better tailored treatment. The negative is that the high heterogeneity between patients means that doing well structured trials is hard…. At some level this heterogeneity is daunting — it is going to take a lot of samples with careful analysis to sort out what is going on biologically in cancers and then how to leverage that knowledge into improving treatments.”
Where we end up is in a room together, looking at the same issues, the same questions. And we acknowledge our frames of reference — embrace them, actually — and teach one another how we see.
In the end, I think, the seeing is what will matter most.