Over the last few years I’ve written a lot about genomic medicine and its various components, including sequencing, data analysis, ethics, regulatory upheaval, etc. Each advance and new initiative carries with it implications for researchers, doctors and sometimes patients. Also, the interactions between 23AndMe and the FDA have raised many questions about personal and/or direct-to-consumer genomics, genome data ownership and stewardship, and, again, what it might mean for patients.
Now we have the (sort of) $1k genome, non-invasive prenatal sequencing, CRISPR/cas and easier genome editing, “niche” markets for certain kinds of sequencers and sequencing, etc. For researchers, the ability to investigate previously unfeasible topics is growing each week. For the medical system, the issues around genomic medicine are becoming far more immediate and important. For someone with a genetic disease or cancer, the importance already is, or at least will soon be, obvious.
This week’s announcement by genome pioneer J. Craig Venter of a massive new project on human aging—built upon sequencing human genomes by the hundreds of thousands—ups the ante. It follows a recent announcement by fellow forward thinker Lee Hood and the 100k Wellness Project. The pharmaceutical company Regeneron also weighed in this January with a large-scale whole exome (all protein-coding genes) sequencing and patient data management project within a single regional healthcare system.
But what does it all mean? Are we truly at the threshold of figuring out enough about the genome to robustly improve human health? To answer that, please let me back up a bit and take a very general look at what we’re dealing with here.
Back to basics
I’ll start with phenotype. Take my phenotype for example. For those not familiar with biology jargon, it may sound like having a phenotype is a bit exotic. It’s not. We all have phenotypes. A phenotype is all of our traits, some obvious right away (height, for example), some not obvious but readily measured (blood pressure), some hard to quantify (behaviors) and so on. When we have a disease, it’s part of our phenotype as well.
I also have a genome. It’s composed of all my genetic material—the totality of my DNA. It has other associated components, such as histones, the proteins around which DNA wraps in highly convoluted structures, and modifying chemicals such as methyl groups, which can affect how it’s regulated. But DNA, all 3.2 billion base pairs, is what encodes my genome’s information. And the genome’s structure, function and regulation are mind-bogglingly complex, to put it mildly.*
My genome determines or influences my phenotype in countless ways. Some influences are straightforward—for example I have the gene variant that makes cilantro taste like an inferior brand of soap, which I find a nuisance as the use of cilantro skyrockets. But most genomic impacts are still poorly understood, including how they contribute to my susceptibilities for various common diseases. And then you have the traits that are strongly shaped by environment and my behavior within that environment, adding layers of near infinite variability.
Probabilities versus determinism
So while my genome contributes to all of my phenotype’s traits, it determines only a percentage of them. This is an important but often misunderstood concept, and it’s a vital distinction for genomic medicine. Most common diseases are complex, meaning more than one and perhaps dozens or even hundreds of genes play a role in the disease phenotype in addition to behavior and environment. Our genomes provide us with probabilities and susceptibilities, which may be greater or lesser than those of our peers. They aren’t deterministic for complex disease, however, meaning they don’t directly dictate whether we will get the disease.
Take, for example, my respiratory health. How would it be now if I’d smoked cigarettes instead of running cross-country as a teen, in part because running helped relieve my childhood asthma? It’s impossible to know, because my asthma phenotype helped shape both my behavior and the environments in which I placed myself. This sort of conundrum is pervasive, and it can be extraordinarily difficult if not impossible to tease out all the factors interacting with our genomes to ultimately yield our phenotypes as we go through life.
While the predictive power of genome sequencing and analysis for later disease will likely grow over time, it’s hard to say whether it will ever provide reliable predictions for complex disease. And many of the most important data about your health and wellness are obvious (sometimes painfully so) in regular day-to-day life. To use a personal example once more, I don’t need to have my DNA sequence to tell me I’m a short, middle-aged Caucasian male with high cholesterol (in spite of OK diet and rather good exercise habits, darn it) but still quite healthy overall despite a family history of Parkinson’s disease, cardiovascular disease and cancer. Perhaps my genome will one day reveal my true susceptibilities as I age with more accuracy and/or with more nuance than a family history and a physical can today, but if so it is many years away.
What is the value?
So going back to the original question, what does all the recent progress mean and how will these large-scale clinical sequencing efforts affect medicine? After all, 23AndMe cannot provide medical interpretation for its inexpensive direct-to-consumer genetic testing kits now, because the FDA determined that the company had over-marketed the kits’ benefits. Will the huge new sequencing projects yield significant insights and benefits?
It’s a valid question. Frankly, sequencing a single healthy person is still mostly an exercise in curiosity. Some people will find something surprising—unexpected ancestry, one or more of the few strong predictors for later disease, etc.—but most will receive a lot of data that adds up to very little in the real world. And while I personally find all this fascinating and enjoy knowing about it, for some it might be troubling, particularly if they have even slightly increased susceptibilities for incurable diseases, such as Alzheimer’s.
Will greatly expanding human sequencing numbers reveal patterns from the complexity? Can genomes ultimately provide far more value and less uncertainty, to the point that they can be preventive and predictive in healthcare and even slow the aging process so that we can have longer, less-diseased health and life spans?
The answer? I don’t have a definitive one, I fear. I hope it’s a “yes,” but I’m not convinced, and I'm not the only one with doubts. Clinical genomics is already valuable and will grow to be more so in oncology, rare diseases and many other applications. But the bar is higher for predictive medicine and longevity. We are the world’s worst experimental system, creatures of bewildering biology with few limitations on where we go and what we do. Our expanding capabilities are formidable, but so are the obstacles to tangible benefits in these areas.
*The Wellcome Collection in London has a large bookshelf holding more than 100 bound volumes, each averaging about a thousand pages long and labeled on the spine with a number (1-23) or an X or a Y. That’s right, the books contain a single human genome sequence, more than 3 billion DNA bases printed in type barely large enough to be legible. But even that full printout is of very little use on its own. It needs to be assembled and analyzed in a computer to tell us what the sequence means.