Drug-induced toxicity is the leading cause of acute liver failure in the United States. Traditional drug-screening tests sometimes fail to uncover potential toxicity problems before drugs reach, or even pass, clinical trials. This puts patients at risk and leads to recalls that are costly for pharmaceutical companies. Now two MIT groups that have been developing new systems for modeling the human liver in the lab are forming startups to bring their products to the market.

TE-bio, founded by Linda Griffith, Steven Tannenbaum, and Walker Inman will launch next year in collaboration with Dupont and is talking with Pfizer as a potential research partner. Their microscale liver tissues are three-dimensional. Hepregen, founded by Sangeeta Bhatia and Salman Khetani, has developed cell cultures that consist of plates with multiple wells, each of which contains two-dimensional, structured growths of liver cells surrounded by supportive cells. Hepregen is currently raising money and talking with Merck and Novartis. Both models function better than the traditional cell cultures used by drug companies because they attempt to mimic the structural complexity of the human liver.

"There is a growing recognition of the need for in vitro alternatives in toxicology," says Michael Shuler, a chemical-engineering professor at Cornell University. Only one of every ten compounds tested by pharmaceutical companies becomes a product, says Shuler, and half of the failures are due to toxicity.

Before a compound can be brought to clinical trials, it must be screened for toxicity on cells in culture and in animals, usually rodents. "There is currently no good way of predicting whether a compound is toxic in humans," says Tannenbaum, professor of toxicology and chemistry at MIT. "Testing in animals is never going to be able to predict all human toxicity." And the tests that are done in simple cell cultures also have major limitations.

"The liver is a complex organ that has many different cell types," says Tannenbaum. These cells exchange chemical signals and even exert mechanical forces on each other that help maintain their function; they form complex structures, including bile ducts. "In order to get any functionality [in a model], you have to have multiple cell types organized into a structure like a liver," he says. When cells are taken out of the liver and cultured using traditional means, their gene-expression profiles change very quickly, and they begin to deteriorate in a few days.

This week, Bhatia and Khetani published a paper in Nature Biotechnology that describes the liver-like functions of the cells in their cultures. They make the cultures by seeding liver cells on plastic plates that are micropatterned with circular spots of collagen. The cells congregate on the collagen and are then surrounded by support cells called fibroblasts. Liver cells arranged in this carefully controlled pattern are better mimics of the human liver than are liver cells growing on their own. For four to six weeks, these cells maintain gene-expression profiles comparable to those of liver cells in the human body; they continue to produce the enzymes that break down and modify drugs; and they even form functioning bile ducts, important transport systems in the liver. When the clusters of liver cells were exposed to known human-liver toxins, they exhibited the same relative toxic effects.