Research on a new blood test that apparently can distinguish people with cancer from those who are healthy or who have other diseases suggests that the test may someday provide a rapid, inexpensive way of detecting malignancy, a Harvard scientist reported here last week.
But many more studies will be needed to determine the large-scale screening value of the experimental test, which uses a sophisticated technique called nuclear magnetic resonance (NMR) spectroscopy, said Dr. Eric T. Fossel of Boston's Beth Israel Hospital.
Speaking at a symposium in Bethesda sponsored by the French Association for Cancer Research and the National Institutes of Health, Fossel said he and his colleagues have performed the blood test on more than 3,000 people, of whom about 1,000 were healthy and the others were suffering either from cancer or from other diseases. "We could distinguish with very good accuracy people with cancer and people without" cancer, he said.
Fossel's initial scientific report on the new blood test, published late last year, caused a flurry of excitement among cancer researchers, but some other scientists have had trouble duplicating his results, experts said last week. Fossel said such problems may have arisen because the temperature and other conditions under which the test is performed can affect its accuracy. He said that between 35 and 50 laboratories are working on the test, and that several are preparing to publish papers confirming his findings.
The test measures radio signals emitted by groups of carbon and hydrogen atoms contained in fat molecules in the blood, he said. The atoms emit characteristic signals when they are exposed to a powerful magnetic field. Fossel said the presence of cancer somehow changes the biophysical properties of the fat molecules, altering the pattern or spectrum that is created when the radio emissions are shown graphically by a computer.
Measuring the width of certain "peaks" that show up on the spectrum, Fossel and his colleagues found they were narrower in cancer patients than in healthy individuals or in those with other diseases, including benign tumors.
He said the test cannot distinguish between different kinds of cancers and does not appear to be affected by the stage of a malignant tumor's growth. Fossel said that when he tested women whose mammograms had raised the suspicion of cancer, the test was able to pick up cancer even in women who had very small breast tumors.
In two patients with leukemia, the initially abnormal test results became normal when the patients were treated with chemotherapy and went into remission, he said. When the patients suffered relapses, the test results became abnormal again before cancer symptoms recurred, he said.
Fossel said the "false positive" rate -- the frequency of abnormal blood tests in healthy people -- appears to be about 3 percent, based on studies of blood samples obtained from American Red Cross donors. He said about 6 percent of the time women with benign breast disease displayed the same test results as women with breast cancer, and there were similar "overlaps" when results in lung cancer patients were compared with those in patients with other forms of lung disease.
A few conditions frequently cause a "false positive" blood test, including pregnancy, very high levels of blood fats called triglycerides, and benign enlargement of the prostate gland, he said.
When cancers were induced experimentally in guinea pigs, the blood test became abnormal early in the course of the disease, suggesting it might provide an early warning of the presence of malignancy, Fossel said.
He said his research suggests the alteration in the NMR spectrum of blood fats arose from a chemical response of the body to cancer and might be caused by a substance called tumor necrosis factor, a natural product of the immune system that can affect fat metabolism.
Fossel said his research group next plans to study the use of the test in so-called "cancer-prone families," those with genetic conditions that cause very high cancer rates, and in citizens of a town where a cancer-causing dye was dumped into the water supply.
He said the two-minute test is performed by an NMR spectrometer, a machine that costs about $400,000. Because one spectrometer can process hundreds of blood samples a day, the blood test would probably cost a consumer less than $50.