The researchers took blood samples from the three volunteers and ran hundreds of tests on each, measuring the reactions of their blood to various combinations and concentrations of chemicals that are involved in the clotting process. They then used that data to create a neural network model — a sophisticated computer simulation — of each person’s blood.
The resulting research paper was published online last month in a journal named, fittingly, Blood.
Using the computer models for each blood donor, the team simulated blood clots of the sort that would occur when plaque inside an artery ruptures, which can lead to a heart attack. The model was then used to simulate treatment with aspirin and three other anticlotting agents. The scientists then tested their predictions on actual samples of the volunteers’ blood, and determined that they were accurate, said senior author Scott L. Diamond, a professor of chemical and biological engineering at Penn.
"The way each patient responds to something like aspirin can be highly variable," he said.
The donor whose blood did not respond to aspirin had a mutation that, in effect, made his clotting system act as if he were already taking aspirin all the time, Diamond said. It is the equivalent of a mild bleeding disorder but did not place the donor in any immediate harm, he said.
The research may have other uses beside tailoring a drug regimen to a particular patient. The researchers hope the work might also be useful in designing implantable devices such as valves and pumps that do not induce clotting.
— Tom Avril
