A trawl of existing data has identified drugs that seem to stall organ rejection in patients who have undergone transplants.
By crunching through large, publicly available data sets, researchers pinpointed a suite of genes involved in organ rejection. They were then able to identify drugs that affect the activity of these genes—with candidates including a widely prescribed statin, a class of drug used to lower blood-cholesterol levels. A subsequent analysis of thousands of medical records indicated that statins do in fact help transplant patients.
The paper, published October 14 in The Journal of Experimental Medicine, lays out a framework for developing hypotheses, says Pankaj Agarwal, a computational biologist at GlaxoSmithKline in King of Prussia, Pennsylvania, who was not involved in the study. “This is part of the informatic-centric world. You can use the data that is already out there to make a hypothesis about how a drug might affect a patient population.”
Studies showed that atorvastatin prolonged the survival of mice that received heart transplants. Untreated mice died within ten days of transplantation, whereas some treated with the statin lived for as longer than 30 days.After using data from several unrelated studies to identify genes likely to contribute to rejection across a range of organ transplants, researchers co-led by Atul Butte, a computational biologist at Stanford University in California, did a literature search for drugs known to damp down each gene’s activity. This pointed to one of the world's most highly prescribed drugs—atorvastatin, marketed by the pharmaceutical giant Pfizer as Lipitor.
New for old
To narrow down the clinical effects of atorvastatin and other, similar statins, the researchers worked with colleagues at the University Hospitals Leuven in Belgium. They had access to the electronic medical records, going back to 1989, of more than 2,500 patients who had undergone kidney transplantation. No patient had been prescribed a statin for reasons relating to their transplant, but the organs were significantly more likely to have survived in patients taking statins. Seven years after transplantation, more than 80 percent of patients taking statins were still alive, compared with fewer than 70 percent of those not taking statins.
Using bioinformatics to find potential new applications for existing drugs is becoming an established practice. The approach recently led to a clinical trial to test whether an antidepressant might work for a particularly hard-to-treat form of lung cancer. But this latest study used pre-existing data both to generate hypotheses about a drug and to analyze its effects. “It's another level to show not only that it could work, it's probably already been working,” says Butte.
Purvesh Khatri, one of the authors also at Stanford, suspected that a common mechanism lay behind the rejection of different organs, but realized that no individual collaborator would have the necessary tissue samples. So he turned to a public data repository, the Gene Expression Omnibus, to obtain results from separate studies of transplanted livers, lungs, hearts and kidneys. “It took me about thirty minutes,” says Khatri. “Honestly, it is scary how easy it seems now, in retrospect.”
Search for stability
Khatri and his colleagues searched across these studies for genes whose activities could best distinguish troubled transplants from stable ones. After identifying genes in some data sets and validating them in others, the researchers were left with a suite of 11 genes that were significantly overexpressed in biopsies from rejected transplants.
“This is a good story, and there is some promise for future directions,” Suthanthiran adds. “It will be nice to see these drugs evaluated in a prospective clinical trial.”“I like the fact that they leverage existing information to do an analysis to show that there is a shared gene-expression pattern across organs,” says Manikkam Suthanthiran, who studies transplantation immunology at Weill Cornell Medical College in New York and was not involved in the study. Such analyses can help to reveal common mechanisms behind conditions that would otherwise not be considered together. “To do all these experiments from start to finish would be a big undertaking,” he says.