back to top

This Software Uses Your DNA To Prevent Medical Errors

Syapse is trying to figure out if you'll react badly to a drug — before you take it.

Posted on

When Hollis Klendenberg checked into Sanford Health with chest pains, just a few years after a stent procedure on his heart, his doctor began writing another electronic prescription for a blood-thinning drug, Plavix. But a software alert stopped him short.

It turns out that Klendenberg, 73, has a genetic mutation that makes his body more susceptible to clotting with Plavix, and he's better off with a different blood thinner. And his doctor knew to prescribe it for him, thanks to a new effort by Sanford Health, the largest health provider in the Dakotas, to collect its patients' DNA.

As patients produce an ever-growing amount of digital information about themselves, from their genetics to their medical history, health care providers like Sanford Health are turning to big data analytics to make sense of it and improve the care they provide. Their efforts are examples of personalized medicine, an emerging field supported by the Obama administration's new Precision Medicine Initiative.

In an ideal world, software could help prevent problems before they happen — not just after a patient gets sick or a drug's negative side effects take their toll. Sanford's software platform is Syapse, a Palo Alto bioinformatics startup, a partnership they described to BuzzFeed News.

While doctors are aware of the possible side effects of the medications they prescribe to patients, they often have little idea who will actually experience them. "Now we're finding out in gene testing that perhaps we can predict if a medication is going to be effective," said Dr. Anthony Tello, Klendenberg's internal medicine physician at Sanford Health's hospital in Bismarck, North Dakota, which is using Syapse. "We can also predict in some situations if the patient is going to have side effects from a medication."

Spotting problems early

Sanford is the latest health care provider to use software to examine patients' DNA before making prescription decisions, following Vanderbilt University Medical Center, St. Jude Children's Research Hospital, the University of Florida Health Shands Hospital, and NorthShore University Health System.

Syapse has its limits: It can only scan for the gene-drug relationships it's been programmed to detect. But it's a helpful arrow in the clinician's quiver, and one that more institutions will likely adopt, according to Philip Empey, assistant professor of pharmacy and therapeutics at the University of Pittsburgh School of Pharmacy. To spot potential problems early on, it's important "to look at patients coming in and say, 'This one meets the criteria, it looks like they might have a drug or might have a condition in the future,'" Empey said. "That's where the field is definitely going."

Actually, it's been headed there for a few years. From 2011 to mid-2013, St. Jude genotyped more than 1,000 young patients with diseases, searching for variations in four genes linked to negative responses to a dozen drugs. Researchers discovered that almost 80% had at least one high-risk genetic variant. They incorporated that information into the patients' medical records and adjusted treatment accordingly.

Sanford's executives say their system is unique because it treats a rural population and, ideally, will someday gather the DNA of all its primary care patients — the healthy ones and those who have diagnosed medical conditions or a risk of developing them.

Since July, more than 450 patients have consented to be genotyped at one of Sanford's federally certified labs and had their DNA scanned for eight genes linked by peer-reviewed research to known adverse drug reactions. Syapse uses that information to calculate the safety and side effects of various combinations of medications. Then, it adds that information to Sanford's electronic medical record system where it's used to flag possible drug-gene complications — like Klendenberg's reduced ability to process Plavix.

Syapse's software is also in place at the University of California at San Francisco and Sarah Cannon, the cancer research arm of the Hospital Corporation of America, among other institutions. Sanford's clinical information technology board, which reviews all such alerts, says the system has so far been error-free.

In Klendenberg's case, Syapse was spot-on and prompted his doctor to prescribe an alternative medication, Brilinta. "They found a different blood thinner and the hospital had me stable in two hours and I was feeling so good, I wanted to go home," Klendenberg told BuzzFeed News.

Sanford also tests for genetic mutations that can cause negative reactions to cholesterol-lowering statins, anticancer drugs, and painkillers, among other medications.

Challenges lie ahead

Klendenberg's case is relatively unique, though. According to a 2014 study in the Annual Review of Pharmacology and Toxicology, genetic testing can only definitively determine prescriptions for about 100 out of 1,000 approved drugs in the U.S.

Another hurdle for hospitals who want to follow Sanford's example is that reimbursement for genetic tests is inconsistent among states, private insurers, and Medicare and Medicaid. (Sanford, for its part, doubles as an insurer that covers members' tests, and a $125 million donation supports its genetics program.)

And tying genetic data to an individual's medical record can be problematic. Each health care provider has its own record system, which doesn't necessarily communicate with others and must be modified to include DNA information. "As a patient moves from one provider to the next, their genomic data, which are relevant over a lifetime, do not necessarily follow," the authors of the 2014 review wrote. Many clinicians, too, do not understand or trust DNA as a tool in health care: Less than 30% of 10,000 U.S. physicians reported any education in pharmacogenomics in a 2012 survey by Medco Health Solutions, a former pharmacy benefits manager.

More importantly, scientists are far from understanding the health implications of all 3 billion DNA base pairs in the human genome. Parsing genetic information along with other factors that can affect health conditions, like environment and lifestyle, requires a level of nuance that software often can't account for.

"The problem is you find out not only the things you're looking for, but all kinds of incidental and perhaps unwanted findings," Dr. Gene Hoyme, Sanford Health's chief of genetics and genomic medicine, said. "There's a lot of ethical and procedural issues with whole genome sequencing before we push it out for everyone to use in their health care."

That said, early efforts seem promising and beneficial for some patients, including Klendenberg.

"I'm about at the end of my trail, so when they changed the blood thinner, it kind of made me think, 'Well, maybe if they keep my arteries from plugging up, I might be able to run around for another 20 years,'" Klendenberg said. "You never know."

Stephanie M. Lee is a science reporter for BuzzFeed News and is based in San Francisco.

Contact Stephanie M. Lee at

Got a confidential tip? Submit it here.