The Virtual Patient
If mice and rats were fortunetellers, we’d fire them. They only succeed in predicting whether a drug will be found toxic to humans 43 percent of the time—yet animal testing is the backbone of our safety system. Little wonder, then, that adverse drug reactions cause the death of one in a hundred hospitalized patients.
The trouble with animals is “they can hardly substitute for the variety of humans,” says Hartung, the Doerenkamp-Zbinden Endowed Chair in Evidence-Based Toxicology. Our sensitivities morph over time, and many people, especially the elderly, take several drugs at once. When vetting a single substance can require $10 million to $20 million with traditional toxicology tests, vetting combinations simply isn’t affordable.
What’s the alternative? Studying human toxicity in human cell systems, one pathway at a time. As a test case, Hartung is exposing induced pluripotent stem cells to chemicals suspected of leading to autism early in life. “We only get one small piece of info,” says Hartung, but his approach—identifying critical molecular interactions—yields insights that apply readily to new challenges.
Ultimately, Hartung’s vision is to compile a catalog of human toxicology to guide therapy decisions. Someday, he says, “[when] you come to the hospital, an avatar will be produced, with all of your genetic background, your disease, your physiology, your body weight.” Doctors will use this computer-based avatar to simulate treatments—heading off adverse drug reactions and fine-tuning dosages.
“The virtual patient project… will be a very big job,” he recently told the Dutch journal Medicines, “but I don’t think it belongs in the world of science fiction.” In fact, CAAT has been working with more than 140 other organizations to create a sophisticated virtual patient.