U.S. researchers have begun testing drugs using a microchip lined with living cells that replicates many of the features of a human lung, a technology that may one day help improve drug testing and reduce researchers' dependence on animal studies.

In 2010, researchers at Harvard's Wyss Institute for Biologically Inspired Engineering developed the so-called lung-on-a-chip technology that mimics the function of air sacs called alveoli, which transfer oxygen through a thin membrane from the lung to the blood.

For drug companies, the technology offers a way to better predict how drugs will work in people, ultimately reducing the cost of drug development by identifying problems before drugs are tested in clinical trials.

"Major pharmaceutical companies spend a lot of time and a huge amount of money on cell cultures and animal testing to develop new drugs, but these methods often fail to predict the effects of these agents when they reach humans," Dr. Donald Ingber, whose study was published on Wednesday in Science Translational Medicine, said in a statement.

Now the Wyss team is putting its artificial lung to the test, using the device to recreate pulmonary edema, a condition that causes fluid to leak into the air sacs of the lungs, and then treating it with an experimental drug from GlaxoSmithKline.

The device, which is about the size of a memory stick, is made of a flexible polymer that contains hollow channels.

These channels are divided by a thin, permeable membrane lined on one side with human lung cells and on the other with tiny blood vessel or capillary cells that are bathed in fluid to simulate blood flow. A vacuum is applied to recreate the way human tissue stretches during breathing.

For the study, the team treated the device with interleukin-2 or IL-2, a cancer drug that can cause pulmonary edema, a deadly condition in which the lungs fill with fluid and blood forms clots.

When injected into the blood channel of the device, the drug caused fluid to start leaking across the membrane, reducing the amount of volume of air in the other channel. Blood plasma crossed into the air channels and started to clot.

Dr. Geraldine Hamilton, co-author on the paper and the senior lead for the organs on chips program at Wyss, said the study is "providing us with a very exciting proof of concept for our ability to use organs on chips to create human disease models."

When the team turned on the vacuum to simulate breathing, fluid leakage increased, suggesting that breathing may make the condition worse.

"We learned more about the mechanisms by which this happens. Than really wouldn't have been possible through an animal model," Hamilton said.

The team next used their model to test a new class of drug being developed by GlaxoSmithKline called a TRPV4 channel blocker. They found that treating the tissues in the device with the Glaxo drug before exposing it to IL-2 prevented blood vessel leakage in the device.

To confirm this finding, Kevin Thorneloe, a scientist at GlaxoSmithKline, did a parallel study in which he tested the drug in the lungs of rodents and dogs with pulmonary edema caused by heart failure and found the drug improved lung function and reduced leakage, consistent with the chip finding.

"These findings suggest that TRPV4 blockers could be used to limit pulmonary edema in patients with heart or lung disease, and were an important step toward validating the lung on a chip model," said Thorneloe, whose companion study was published in the same journal.

"This technology is still in its early stages of development," he said, noting that several additional studies will be needed to further validate the chip device.

Although initially, such devices will be used to support early research seeking to get a better understanding of disease at the molecular level, Thorneloe said over time, they could be used to quickly study the impact of several drugs on lung function.

In July, Wyss entered a $37 million agreement with the U.S. defense department to help develop 10 engineered organs, all linked into one system.

The idea is to replicate a human body on a chip, which could be used to rapidly assess responses to new drugs and potential chemical threats.

Donna Dambach, a pre-clinical safety scientist at Roche's Genentech unit, said she thinks drug companies will be quick to adopt these technologies for their own internal decision-making.

But Dambach said it will likely take much longer for drug companies to replace animal models used for regulatory approval with these engineered organs.

"I think everyone would love that, but animals are complex, like human beings, and we really have to make certain that we're at least predicting some level of that complexity in these systems."