Cystic fibrosis, an inherited chronic disease linked to the CFTR gene, affects some 30,000 Americans. Today, their average life expectancy is 37 years.
In the United States, newborns are screened for the defective gene upon birth in order to begin early treatment. But there are some 1,700 rare mutations that can impact CF severity and its progression. If detailed genetic testing turns up these mutations, doctors are at a loss to tell patients what it means.
“It’s a serious problem,” says Rachel Karchin, the William R. Brody Faculty Scholar and associate professor of biomedical engineering at the Whiting School. “We have people making decisions about whether or not to end a pregnancy. But the disease implications of rare CFTR mutations are not understood.”
Now, Karchin and her colleagues have come up with a computer algorithm that could change that. She and assistant research professor David Masica, PhD ’09, both of the Institute for Computational Medicine, have teamed up with clinicians Patrick Sosnay and Garry Cutting, at the Johns Hopkins McKusick-Nathans Institute of Genetic Medicine, who have pioneered better understanding of the disease’s genetic underpinnings.
The researchers have used a large array of clinical and genetic data from CF patients to “teach” the Karchin lab computer algorithm to think like a doctor—making highly accurate diagnoses of specific symptoms and disease severity, based on the presence of these rare CFTR mutations.
“This is a completely new way of predicting disease status,” says Karchin. “In essence, we can begin to predict the severity of individuals’ likely symptoms from their genetic code.”