Analytics & Data From the Magazine Science

Why We Still Can’t Rely on Genomic Medicine

Illustration by Jon Han for Techonomy

Genome-based medicine, a topic once confined to scientists in ivory towers, has reached the mainstream. In news reports and in conversations with their doctors, more and more people are hearing about the promise of using information gleaned from their own genes to deliver more targeted, customized healthcare. It’s gotten to the point that standard recommendations for some types of cancer now include getting the tumor sequenced. But how close is genomic medicine to becoming a routine part of modern medicine?

The answer is far more complicated than you might expect. Genetic testing is being used for more and more conditions. Scientists are making new discoveries all the time about natural genetic variations in people, and about the meaning of such variations when someone has a disease, especially cancer. So in that sense, the era of genomic medicine is all but upon us. However, we are still at the earliest stages of understanding what all this data actually means. As a result, we remain very far indeed from the day when incorporating genome information in healthcare is routine.

Success in genomic medicine will hinge on two things: accurately reading the genome, or at least the relevant portions of it; and interpreting how that information gets translated into well-being or disease. If you’ve heard about all the human genomes that have been sequenced so far, you probably think we’ve got the first problem licked.

In reality, 15 years after completion of the Human Genome Project, the world still has not seen a single fully-sequenced human genome. It turns out that the three billion letters that make up our genome are not uniformly easy to read—so while the vast majority of the human genome has been sequenced, sections remain indecipherable with the sequencing technologies of today. While we can make significant progress based on what we already know, we’ll obviously need to be able to decode the whole thing to gain a complete picture of the exact DNA we possess.

Even when we know what’s there, interpreting it remains the single greatest challenge for genomics in medicine today. Figuring out the biological function of each individual genetic variant is like solving a thousand Rubik’s Cubes at the same time. Today’s technologies of cloud services with giant data repositories and the growing capability of analytic tools make that possible, but still painstaking. But then scientists still have to figure out how all these variants work together, with any combination potentially affecting a person’s health.

We are at the earliest stages of this transition to genomic medicine. Many of the discoveries made today will be refined or found to be incorrect later, so a lot of today’s interpretations will ultimately be overturned. That may be hard for patients to accept. But regardless, it remains only a matter of time until a far more accurate and comprehensive version of genomic medicine becomes reality.

EARLY SUCCESSES

The low-hanging fruit in genome-based medicine was found in rare, often deadly, diseases. The genetic variants that cause such diseases are typically just as rare.

So if you compare the genome of an affected person to those of 100 unaffected people, the DNA culprit in these cases virtually jumps off the page. A number of rare diseases have been traced back to causative genetic variants, including Huntington’s disease and Alport syndrome. For people with such diseases, a genetic test can provide a conclusive diagnosis and sometimes assist in finding a path for treatment.

These successes fueled the fire for genomic insights, and institutions around the world invested more resources to find the genetic causes of other diseases. This led to genome-wide association studies, in which correlations are sought between genetic variants and specific diseases by scanning the DNA of thousands or even tens of thousands of people. Only a decade ago, studies of that size were inconceivable due to the then higher cost of DNA sequencing. Such research enabled a huge step forward in assigning biological functions to rare genetic variants—but scientists quickly found that they were generally not yielding unambiguous answers for common diseases, or for complex diseases that turn out to be caused by more than just one genetic variant.

ENCOUNTERING PROBLEMS

Though genomics has been a dedicated scientific field for decades now, researchers are still finding new genetic factors that influence disease. Every few years, there’s some unexpected discovery that not only explains a whole new set of genetic variants, but also reminds us how little we know about the inner workings of our code.

To make useful advances, scientists have called for bigger studies. In some cases that means plumbing the depths of a far larger number of genomes. That’s the rationale behind efforts like the Precision Medicine Initiative in the U.S., which has the goal of sequencing more than one million people. The concept is simply that if we compare enough genomes, otherwise hard-to-spot differences will eventually become obvious. A different approach is to study fewer people but get access to much more detailed data about both their genetic variants and their exact medical conditions. This idea gave rise to efforts like Health Nucleus, part of J. Craig Venter’s Human Longevity, Inc. (see following story for an interview with him). For each participant, HLI conducts whole genome sequencing in addition to many clinical tests, from brain scans to microbiome analysis and much more.

But most scientists agree that these approaches must be used together for maximum impact. At a conference, Venter told researchers, “We can’t tell much more today about my genome than we could 15 years ago. We literally need millions of genomes.”

Regardless of how much information such studies yield, though, correctly interpreting a set of variants in a given person’s DNA will still not be straightforward. “There’s a lot of subjectivity,” says Heidi Rehm, director of the Laboratory for Molecular Medicine at Partners Healthcare Personalized Medicine in Cambridge, Massachusetts. She has championed the development and implementation of better standards for clinical labs so analysts can generate more consistent interpretations. Nonetheless, she says, any time human judgment is a factor, it’s very difficult to control for variables.

WHERE WE ARE NOW

While the research community is quite familiar with the limitations on our ability to interpret genomes today, the same cannot be said for physicians or the public. And that is bound to cause problems.

Let’s say, for example, that someone goes in for a genetic test to learn whether she is at increased risk of breast cancer. While the genetic analysis turns up hundreds of variants, none of them is known to be associated with breast cancer. The patient breathes a sign of relief and moves on with her life. But two years later, one of the variants picked up in her scan is found by scientists to be implicated in hereditary breast cancer. Now an at-risk woman is walking around thinking she’s in the clear, while access to the revised scientific data might help save her life. How will the medical community handle this? With the crude medical record systems in use today, it is virtually impossible to track the status of knowledge about all those variants over time and update patients when information changes. But this will be urgently needed during this time of great medical transition.

Then you have cases when scientists believe variants should be causing a certain disease, but they don’t. No one knows this better than Rosalynn Gill, one of the first participants in the Personal Genome Project initiated by Harvard Medical School’s George Church. Gill’s genome scan in 2012 revealed a variant believed to be responsible for what’s called Long QT syndrome, a rare and dangerous heart condition. She had no family history for it and so was skeptical of the finding, but PGP scientists urged her to get checked. A thorough exam concluded that she does not have the syndrome. “We have a lot to learn,” says Gill, a veteran in the genetic testing field who believes her experience will be repeated as more people get their genome sequenced. For a range of diseases, she adds, even when “the early reports of mutations [had] really strong associations—as more data rolls in, the associations [have] become less strong.”

The technical explanation could be that the genetic variant is only somewhat associated with the disease, or that it is indeed causative but other protective variants can offset it. Or maybe the variant must work in conjunction with another variant to cause disease. There’s also the possibility that the variant must accompany external factors, such as those from environment or diet, before disease onset occurs. We simply don’t know. And this confusing set of possibilities could apply to the relationship between many genetic variants and medical conditions.

What we do know is this: When genomic medicine works, it works better than any other weapon in the arsenal. There are countless stories about last-resort patients with rare diseases or cancer whose lives have been saved or extended thanks to the use of genome-based information. But before we can get to the tantalizing future in which most patients see regular benefits, we need a lot more investment in research to gain sufficient scientific understanding of how the human genome works.

MEREDITH SALISBURY is a longtime genomics journalist and a communications consultant in life sciences.