We’ve all been hearing for years about how a number of technologies are poised to change the nature of medicine. One of those with the biggest promise is centered on the study of genetics, which is revealing the blueprints behind some of our most intractable diseases. Another, which proponents argue will be a disruptive force to advance preventative medicine, is Big Data—the analysis of massive amounts of information being collected on patients to find clues to detecting and treating disease.
This is not the first article to show that those years of dedication and work by scientists in a number of fields are starting to bear fruits. An article published late last month in the journal Science Translational Medicine reveals some of the latest findings. In it, a team used both genetics and Big Data to link patients’ seemingly unrelated traits with the onset of disease. Their work will one day point doctors to a developing disease before it could have been diagnosed in the past.
“Over the last 10 to 15 years, researchers have been doing lots of genetic studies,” says Dr. Atul Butte, a biomedical informatics researcher and Stanford professor. “These studies are indicating that certain diseases come from certain genes, but also that certain traits can predict a coming disease. Electronic medical records are feeding this, offering more data on patients along with their genome to see what spelling differences in DNA mean.”
Butte and scientists from Mount Sinai School of Medicine, Stanford and Columbia used previous gene research that linked a specific gene variant to a trait, like elevated levels of cholesterol or a certain enzyme, and a disease. They then analyzed electronic medical records to link the early appearance of that trait to a later disease diagnosis. Their effort uncovered five previously unknown associations after combing through as many as 610,000 anonymized patient records per linkage.
For patients who have a gene variant that has been associated with a specific trait and disease: One new link found that a measure of the average volume of red blood cells, called mean corpuscular volume, is elevated a year before a diagnosis of acute lymphoblastic leukemia, a cancer that most commonly afflicts children. Another revealed that the number of platelets—cells central to the control of bleeding— decreases well before a diagnosis of alcohol dependence is possible. An elevated level of alkaline phosphatase, an enzyme found throughout the body, is linked to the development of venous thromboembolism, a potentially fatal condition in which a blood clot in the body breaks loose and travels through the bloodstream. Magnesium levels of those gene variant carriers who will soon be diagnosed with gastric cancer are altered compared to the rest of the population. The same alteration in prostate-specific antigen precedes by a year a diagnosis of lung cancer.
“We crunched the data and, lo and behold, there is a direct correlation between some traits and diseases before we can even make a diagnosis—up to a year before we would have seen signs of the disease,” Butte tells Txchnologist. “These are earlier signs of the problem before it progresses and they’re associations we would have never predicted.”
These potential early indicators of disease have been hidden in plain sight because doctors might only deal with thousands of patients in their practice. The researchers, instead, crunched data from hundreds of thousands of patient records across three healthcare systems in California and New York. “Doctors are in the trenches and aren’t going to be able to see the link between a certain trait and a disease, so we need to do data science on it,” he says.
Butte says there is still work to do before these traits are used to give a disease diagnosis before a patient even realizes anything is wrong inside. But, with the cost of sequencing a person’s DNA continuing to fall and ubiquitous health data collection, he believes the predictive power of a person’s genome will be unlocked within a decade.
“We’re getting to a case where looking at genomes will give people a clue that they’ll get a disease, and they can take preventative steps sooner,” he says. “We’re in a world where doctors and hospitals are all incentivized to put data into medical records and there’s so much being collected—it’s at the exponential level. Now we’re entering an era when we’re going to make use of that data.”
