“Success in sight: The eyes have it!” Thus the scientific journal Gene Therapy greeted the news, in 2008, that an experimental treatment was restoring vision to 12 people born with a congenital disorder that slowly left them blind. Healthy genes were injected to replace the faulty mutations in the patients’ retinas, allowing an 8-year-old to ride a bike for the first time. A mother finally saw her child play softball. Every patient, the researchers reported, showed “sustained improvement.” Five years in, a book declared this “breakthrough” — a good-gene-for-bad-gene swap long pursued as a silver bullet for genetic conditions — as The Forever Fix.
Earlier this month, two of the three research teams running these trials quietly reported that the therapy’s benefit had peaked after three years and then begun to fade. The third trial says its patients continue to improve. But in the other two, all the patients tracked for five years or more were again losing their sight.
Not all gene therapy ends in Greek-caliber tragedy. But these trials serve as a sadly apt parable for the current state of human genetics. This goes especially for the big-data branch of human genetics called Big Genomics. In five years of talking to geneticists, biologists, and historians, I’ve found that the field is too often distinguished by the arc shown here: alluring hope, celebratory hype, dark disappointment.
We live in an age of hype. But the overselling of the Age of Genomics — the hype about the hope, the silence about the disappointments — gobbles up funding that we might spend better elsewhere, warps the expectations of patients and the incentives of scientists, and has implications even for people who pay genetics scant attention. Many hospitals, for instance, are now collecting genetic information from patients that they may market to “research partners” such as drug companies. Some take more care than others do to secure informed consent. (Had blood drawn lately? Read everything you signed that day?) It’s not just that they’re selling you this stuff. They may well be selling you. And the sale depends on an exaggerated picture of genetic power and destiny.
To be sure, medical genetics has chalked up some sweet victories. Our growing ability to spot rare mutations, for instance, is helping doctors diagnose and sometimes treat nasty rare diseases. Last fall, for instance, doctors in St. Louis sequenced an infant dying of liver failure, saw that he had inherited a rare mutation that both his parents happened to pass to him, devised a way to counter the mutation’s disruption of his immune system, and saved his life.
But when it comes to how genes shape the traits and diseases that matter most to us — from intelligence and temperament to cancer and depression — genetic research overpromises and underdelivers on actionable knowledge. After 110 years of genetics, and 15 years after the $3.8 billion Human Genome Project promised fast cures, after more billions spent and endless hype about results just around the corner, we have few cures. And we basically know diddly-squat.
I know — diddly-squat is rough talk. Yet this is hardly a radical claim. Geneticists and doctors outside of Big Genomics — people studying genetics in songbirds, sea urchins, monkeys, microbes, fruit flies, and roundworms, for instance — often voice it privately. Others are eager to tell us what genes can’t do or warn that “precision medicine” will let us down. One of the world’s most respected geneticists, Britain’s Steve Jones, gives quite an entertaining lecture on our humble state of knowledge.“The more we learn, the less we understand,” he says. “We know almost nothing of genetics.”
The press, of course, too often falls hard for ludicrous memes such as “the slut gene.” But much of the time, the media is simply amplifying the signal sent by Big Genomics. Big Genomics outfits like the National Institutes of Health and the Broad Institute regularly assure us that their careful reading of the genome’s text will find crucial misspellings that generate disease — and let us revise, delete, or write around those errors.
In doing so, they continue a tradition as old as genetics itself. Historian Nathaniel Comfort, in The Science of Human Perfection, calls the history of genetics “a history of promises.” Cambridge geneticist William Bateson coined the term genetics in 1905; by 1927, biologists made the first of many assertions that genetics would cure cancer. In 1940, Canadian physician and embryologist Madge Macklin promised “a world in which doctors come to their patients and tell them what diseases they are about to have, and then begin treatments before the patient feels even the first symptoms.” In 1967, Stanford geneticist Joshua Lederberg predicted gene replacement therapy — the kind that is now failing in the blindness trial — “within a few years.”
In 2000 the leaders of the Human Genome Project doubled down. Standing next to President Bill Clinton, they announced that the project had sequenced the human genome, exposing the full genetic code to view. “Personalized genetic medicine,” an accompanying White House Statement said, would soon “cure diseases like Alzheimer’s, Parkinson’s, diabetes and cancer by attacking their genetic roots.” Francis Collins, the project’s director (now head of the National Institutes of Health), said the genomic revolution could reduce cancer to zero and would make gene-tailored personalized medicine common by 2010.
A century of hype is a lot, but this is particularly inspirational ground. The gene, especially after Franklin, Watson, and Crick gave us a peek at DNA in 1953, looked promising as hell. For decades, the gene was seen as the key to all of biology — or as President Clinton would eventually put it, “the language in which God created life.” In its code we would read the story of life, evolution, disease, and death.
But when the Genome Project finally revealed the links in Franklin, Watson, and Crick’s deceptively simple structure, it found few of the strong gene-to-trait connections one might have hoped for. Instead, it found a mess. Our DNA held far fewer genes than expected, almost 20,000, which was confusing. Few held obvious function. Some seemed to do nothing. Some seemed to work fine one day but not the next, or to do one thing in one situation and another in another. And these genes were surrounded by vast stretches of DNA material that aren’t really genes, and which some geneticists called junk, starting a big fight.
To clarify this mess — to figure out what did what, and to identify medically relevant genes — researchers started using sequencing machines to scan the genomes of tens or even hundreds of thousands of people for gene variants that appear more often in people with some condition, disease, or trait. These overrepresented genes are then presumed to contribute to the condition or trait in question.
Unfortunately, GWAs seldom revealed the sort of the neat or consistent gene-to-trait relationships that allow decisive treatment. Instead, they usually found “many genes of small effect”: handfuls and sometimes hundreds of gene variants carried by most (but not all) people with the condition in question, whose effects were seldom clear, and whose presence in a given person did little to predict risk.
“Many genes of small effect” became a sort of tepid curse. I myself prefer the stronger, more memorable phrase “Many Assorted Genes of Tiny Significance,” or MAGOTS — a mass of barely significant genes explaining little.
MAGOTS infest most GWA studies for a simple, brutal reason: If a gene variant reliably plays a large role in causing disease, both the variant and the disease it causes tend to be rare, because its carriers tend to die without leaving offspring. This is why the genetic contributions for common diseases and conditions usually come from MAGOTS — the effects of which, it bears repeating, are usually maddeningly obscure and unpredictable. This applies even to diseases and traits that run in families. Take height: Hundreds of genes of small effect, few clues to how they contribute, and no real target to tweak if, say, you want to make someone tall. The best way to engineer a tall person? Tell two tall people to tango.
Similarly, deep digs at cancer, schizophrenia, heart disease, hypertension, diabetes, intelligence, bipolar disorder, and height have found mostly MAGOTS. The biggest schizophrenia study so far, for instance, published last July to great fanfare, found 128 gene variants that appeared to account for perhaps 7% of a given person’s actual risk.
The genomic age’s signature finding is not any great discovery. It is the yawning gap between the genetic contributions that geneticists assume exist and the genetic contributions they can spot. It is as if they cracked a safe they knew was packed with cash and found almost nothing. The money’s got to be somewhere. But where?
Researchers in the field are quick to point to one of the handful of effective drugs to come from genomic insight, such as Gleevec, a leukemia drug developed in 2001. But Gleevec, however potent, falls far short of the medical miracles forecast 15 years ago. As science writer Ed Yong pointed out in a recent Twitter conversation about this, “Treasure was promised. Gleevec’s a coin.”
At this point, the problem is not so much that genetics fell short of its early promises. The problem is that big genomics players keep making similar promises.
Take, for instance, that schizophrenia study rife with MAGOTS. When the study came out last July, John Williams, head of neuroscience and mental health at the Wellcome Trust, Britain’s biggest biomedical funder, saw it as cause for humility. “What this research screams to me,” he wrote, “is how little we know about schizophrenia, and how far we are from biological tests and treatments for mental health disorders compared to other major diseases.”
Yet last July, the Broad Institute, a genomics powerhouse that played a big role in that schizophrenia study, triumphantly unveiled it as part of an announcement that a donor had given Broad $650 million to expand research at its Stanley Center for Psychiatric Genomics. Broad’s director called the study part of “a revolution in psychiatric disease.” Francis Collins, apparently deaf to how closely his promises echoed those he’d made 15 years before, when the Human Genome Project was unveiled, said psychiatric genomics now stood “poised for rapid advances.” The promises were a decade old, the rhetoric a century. The only things new were the event’s over-the-top staging and production — it views like an awards ceremony — and how boldly, even after 15 years of the “genomic age” with little to show, the Broad conjured big money from thin results.
Big Genomics is converting hype to cash at unsettling speed. After the FDA told consumer genomics company 23andMe it could no longer sell people health data, the company began selling that data to drug and biotech companies. An entire industry, potentially fed by almost anyone who draws blood, spit, or biopsies from you, is emerging to do likewise. Its growth, along with the increasingly routine collection of genetic data by hospitals, will feed the genomics bubble while putting private genetic and health information at increased risk.
Meanwhile, it’s becoming routine for researchers and research centers to leverage genomic findings into industry jobs or startups. This happened, for instance, with the research team that ran one of the sight restoration gene therapy trials whose effects faded after three years.
At the peak of the trials’ promise and publicity, the Children’s Hospital of Philadelphia (CHOP), which ran one of the studies, provided seed money and office space to Spark Therapeutics, a new startup that aims to develop genetic therapies. Spark’s ticker symbol, $ONCE, reportedly referred to the company’s intention to make treatments that would correct genetic faults with a single application.
Bolstered by a partnership with pharma giant Pfizer, Spark went public this January. It promptly became the first quarter’s hottest IPO, returning 256% on first-day investments and reaching a value over $1.5 billion by March. The vision trial had gotten great press — and its president, co-founder, and chief scientific officer, Katherine High, was one of the scientists leading the CHOP trial. High’s integrity may be bombproof. But this sort of bench-to-boardroom practice hardly encourages the critical interrogation of one’s own work that good science depends on.
None of this is to say we should pull the plug on Big Genomics. Some suggest — and I agree — that we’d do well to take some of the billions spent chasing genes for conditions like Type II diabetes, heart disease, or stroke and spend it instead on finding ways to change risk-elevating behaviors like smoking, overeating, overdrinking, and avoiding exercise.
It would be responsible, however, for researchers to temper their hype — though this seems unlikely, because hype pays.
So let me offer a hype filter. This one comes courtesy of the oceanographer Henry Bryant Bigelow, who helped found Woods Hole Oceanographic Institute. A century ago, Bigelow opened a letter his brother had written him from Cuba. His brother reported that while weathering a hurricane there, he had seen, flying by, what he was almost sure was a donkey.
With three words, Bigelow gently told his brother he didn’t quite believe him — and stated a maxim for maintaining the ever-curious but ever-skeptical stance that marks the good scientist.
“Interesting if true,” he wrote.
David Dobbs, Buzzfeed