Researchers are closing in on highly sensitive devices that trap bits of cancer in the blood, offering a "liquid biopsy" that would give doctors and drug companies a better way to see if cancer drugs are working.

Scientists have long known that cancer cells can break off from the main tumor and circulate in the blood, but these cells are so rare and so fragile they have been hard to capture and study in a meaningful way.

Johnson & Johnson has the only approved test on the U.S. market, but researchers say its usefulness is limited because it only traps a few of the cells.

So, several teams have gone back to the drawing board and are now close to developing more sensitive tests. If they work, they could usher in a new age of personalized cancer treatment in which doctors use cancer cells from a patient's blood to guide treatment.

"Once you have these cells, you now have the very cancer you are trying to attack in the palm of your hand," said Dr. Bob McCormack of J&J's Veridex unit. "You can begin to direct therapy from a test tube."

So far, there are about 25 different technology companies vying to develop tests, ranging from small labs to very large companies.

Researchers say a better test could be worth billions.

"The huge interest in this is that this is blood," said Dr. Minetta Liu of Georgetown University Hospital in Washington. "It is not sticking a needle in an organ," she said.

Most tests for cancer cells circulating in the blood look for epithelial cells, which form the outer layer of organs. These cells do not belong in blood, but are found there in 85 to 90 percent of cancers.

"The supposition is these circulating cells are the very cells that kill," said Dr. Mehmet Toner, a biomedical engineer at Massachusetts General Hospital and Harvard Medical School.

FROM KILLER DISEASE TO CHRONIC

If they can be detected in the blood before they attack a new organ in the body, doctors may be able to keep cancer from spreading, turning a killer like lung cancer into a chronic disease. "That could really turn the whole cancer field upside down," Toner said in a telephone interview.

His team last month published a study showing that their new silicon chip spotted more than 90 percent of cancer cells the team added to blood samples.

The device, an improved version of a chip developed in 2007, has microscopic posts coated with antibodies that act like glue to trap epithelial cells but not blood cells.

"With the current technology, you may find one cell in an 8 milliliter tube of blood. We find hundreds or thousands of cells in the same amount of blood," Toner said.

Once trapped, scientists can run any number of tests to see what type of cancer it is and what drugs might best attack it.

Instead of capturing cells, scientists at the Scripps Research Institute in California are using digital imaging and high-tech computing to count and characterize the cells.

"We go at this from the other end of the spectrum," Dr. Peter Kuhn, a physicist, said in a telephone interview.

With his approach, researchers arrange a single layer of blood cells on a glass slide and label them with a fluorescent marker that lights up epithelial cells. Using a fluorescent microscope, they take thousands of digital images, eating up 100 gigabytes of data with each test.

STILL SKEPTICAL

Kuhn's group is working with Microsoft's high performance computing team to manage the data and has licensed the technology to Epic Sciences of California.

So far, drug companies are interested in the potential of the new tests, but they remain skeptical, says David Shalinsky, who helps assess new cancer technologies for drug giant Pfizer Inc.

Some of the skepticism comes from the lack of sensitivity of the Veridex test, which has failed to live up to the promise of being a useful tool for drug research.

But Shalinsky, whose company is advising Epic Sciences on how to validate its test, is encouraged, too.

"The concept of a liquid biopsy is attractive," he said in a telephone interview. "That one could have a surrogate for what is happening in the tumor ... would give us confirmation that we are hitting our target."

At the Harvard lab, most of the testing has been in lung and prostate cancer, but Toner said they have also worked with breast, pancreatic and brain cancers.

He is part of a so-called "dream team" funded by the entertainment industry group Stand Up to Cancer aimed at speeding new cancer discoveries.

Toner expects to have a system ready for testing at partner labs next year. He wants to get the science right before licensing the technology, but thinks the chip could be on the market in three to five years.

It may be longer for the Scripps team. Kuhn credits Veridex for bringing the field along, but he wants to take the technology to the next level, making it possible to give doctors a fresh cancer biopsy each time a patient walks into the office.

"This will change the way we manage every individual cancer patient," Kuhn said.