UT group uses computers, economics to make transplant more effectiveWritten by John Krudy | | firstname.lastname@example.org
When Dr. Michael Rees sits down in a crowded coffee shop, he doesn’t see thirsty suburban professionals, or college students acting pretentious. He sees kidneys.
“Look at her,” he said pointing to a young woman in black suit, with neat, black hair. “She could need a kidney, but her family might not have a match.”
According to Rees, there are 70,000 people in the U.S. waiting for a kidney transplant. And many of them die before they ever get a chance at a new organ.
Rees, an associate professor of urology at the University of Toledo, decided to try something new. The challenge with kidney donation is that it requires near-perfect matches of many criteria, including blood type, age and comparative health, availability and proximity of donors, and hundreds of antibodies and proteins, which must match if the recipient’s body is going to let the new organ live.
“Those proteins are like a slot machine,” Rees said. “You pull the handle, and they all have to come up right if you’re going to have a match.” The number of potential donations is n(n-1)/2, where n equals the number of pairs.
That casino-like dilemma has to be resolved before pairs can be matched. It’s especially difficult with this new method, which tries for three- or four-way organ swaps. Rees admitted he spent hours at his kitchen table trying to match pairs, but that the math required was too tedious.
The problem is one of optimization, a concept more often studied in economics classrooms than tissue labs. Those deciding who gets a kidney must determine how the greatest number of lives can be improved the most, while keeping transplants equitable, workable and just. And that’s where paired donation outperforms the usual kind.
“Imagine a set of pairs needing kidney transplants: A, B and C,” said John Kopke, a research assistant at the Institute for the Study of Health at the University of Cincinatti. “If A gives to A, but the B and C pairs aren’t compatible, you only help one person. If A gives to B, and C can exchange with A, you’re helping two people instead. And we’d all agree that’s better.”
“Look at those guys,” said Rees, pointing across the coffee shop at two men in suits. “One might want to give a kidney to the other, his brother. If they’re incompatible, why not have them swap with someone else?”
When he first started work on the paired donation project, Rees asked the University of Cincinnati to help develop the computer software. Rees’ father, Alan Rees, did early work on the software, and Tuomas Sandholm, professor of computer science at Carnegie Mellon University in Pittsburgh, Pa., did much of the original work on the concept. Rees even worked with economist Alvin Roth on some of the game theory and probability solutions. But Kopke now enters the data, optimizes matches and works on new versions. He said the program runs on a standard desktop computer.
“The writing of the program is actually more tedious than complicated,” he said. “I think I was the first to actually make it work, but that’s not because of any particular genius.”
Multiplying in his head, Kopke said 125 people in the database will create 15,625 permutations of matches. But myriad exclusive factors create the need for a computer’s calculating assistance.
“One donor might be able to help ten people, but he only has one kidney to give,” Kopke said. “And three things can prevent incompatibility: a blood transfusion, which developed antibodies, a mother’s reaction to her husband, by pregnancy, or a previous transplant.” Kopke estimated that such factors cause some people to be sensitized to 95 percent of the nation, and that makes the proper organ especially hard to find.
“And I have to look at those six proteins,” Kopke said. “A “close” match is a problem, because that’s enough to trigger an immune reaction.” And the computer program can rank the utility of the matches, so there’s an objective ranking for each transplant.
Kopke said that “part of the program is wrong half the time” — but every time it runs, it records what matches failed, and so can optimize better the next time. The computer, Kopke said, never makes the final decision on a transplant, so none of those failures affect the patient.
It’s the human need, the sense of charity and the chance to help that drives Rees and surgeons like him. But he points out that the costs, too, are worth considering.
“One year of dialysis costs $80,000, and the government pays for that,” Rees said. “In its first year, a transplant costs $235,000 — but there’s no dialysis cost after that. We end up saving half a million by making a transplant. Shouldn’t the government be interested in that?” Rees estimated that with 3000 transplants, the U.S. could save $1.5 billion.
Kopke said the emotion and charity of the work has reached even him, “just a programmer.
“I sit and push keys in my air conditioned office, and I don’t see the human side,” he said. “But I got to that [transplant] conference, and I saw four people who’d had the transplants. I ran up and told them, ‘you’re so brave, to be the first to do this. I have pictures of you plastered all over my cubicle.’”