Researchers drew that conclusion after evaluating both mammograms and MRIs in 404 suspected cancer cases.

"This was an attempt to describe in a large trial how MRI is performing in a general sense, and then take that information and determine which patients in particular would be helped by MRI," said Dr. David A. Bluemke, an associate professor of radiology at Johns Hopkins School of Medicine and the study's lead author.

The study was the largest multi-center study to date to look at the performance of MRI in conjuction with mammography, with patients drawn from 14 centers.

The findings appear in the Dec. 8 issue of the Journal of the American Medical Association.

For years, doctors have known that mammography, the primary imaging method used to detect the disease, has limitations in finding cancer. They have been evaluating MRI to see what role it might play in early detection.

In the study, Bluemke and his colleagues from Hopkins and several other institutions evaluated 821 women referred for breast biopsy from June 1998 through October 2001. All had had suspicious findings on a mammogram, clinical exam, or ultrasound exam.

In the past, studies have suggested that lesions seen by MRI but not visible on mammogram have been reported in 27 percent to 37 percent of patients.

But "almost all the data was based on a single center," Bluemke said, and the findings were sometimes not applicable to other populations or other countries.

Of the 821 patients, 404 were found to have breast cancer. Of those, MRI correctly identified 356 as malignant, which translates to a sensitivity of more than 88 percent. Scientists consider a method sensitive if it detects most cases of a condition or disease.

"Eighty-eight percent is pretty good for any of the diagnostic tests we have," Bluemke said. In comparison, he added, mammography has a sensitivity of 90 percent, "but it picks up different lesions [than does MRI]."

Previous studies had found a sensitivy for MRI from 88 percent to 95 percent.

Bluemke's team also found that MRI correctly identified as non-cancerous 281 of the 417 women with suspicious mammograms. That translates to a specificity of 67.3 percent. A measure is specific if non-cases, in this instance cancer, are not misdiagnosed as cancerous most of the time.

The researchers also found that the accuracy of the MRI wasn't affected by the density of the breasts, as mammograms are. "If you have dense breasts, your mammograms won't be as accurate," he said.

The study findings suggest a role for MRI, Bleumke said, in specific populations of women, such as those with dense breasts and those with genetic mutations that put them at higher risk of developing breast cancer.

MRI could also be useful, he said, in patients whose mammograms show several areas that are hard for the radiologist to interpret.

As well as MRI performs, it won't eliminate the need for breast biopsy, he said.

In an accompanying editorial, Dr. Monica Morrow, chair of surgical oncology at the Fox Chase Cancer Center in Philadelphia, pointed out that the study results aren't good enough to either justify avoiding a breast biopsy or for the routine use of MRI. She also emphasized that MRI is better, as the study found, for detecting invasive cancers rather than smaller intraductal cancers.

"MRI isn't specific enough to eliminate biopsy," she said. "We've known that." And, she added, the new study confirms that.

The best advice for consumers? "If you have something abnormal on an MRI, don't assume it is cancer until you have had a biopsy," Morrow said.

Two of the study's authors have received research grants from MRI manufacturers.

More information:

To learn about the value of early detection of cancer, visit the American Cancer Society (www.cancer.org ).