Study questions accuracy of breast CAD

A study appearing in today's issue of The New England Journal of Medicine reports that computer-aided interpretation of mammograms is less accurate than interpretation without a computer's help. Researchers concluded that computer-aided detection software designed to improve mammogram interpretation may actually make it worse.

The results of the study showed that women who got screening mammograms at centers using CAD were more likely to be told their mammogram was abnormal and they should thus undergo a biopsy to rule out breast cancer.

Dr. Joshua J. Fenton and colleagues at the University of California, Davis, Health System and the Breast Cancer Surveillance Consortium found that the use of CAD resulted in 32% more women being recalled for additional tests and 20% more women undergoing a breast biopsy, even though most ultimately did not have breast cancer.

The researchers also found that the use of CAD had no clear impact on the early detection of dangerous invasive breast cancers. In fact, they said the software may promote the detection of ductal carcinoma in situ, the least dangerous of breast cancers. The effect of in situ cancers on breast cancer mortality remains unknown, and some evidence suggests that not all such lesions develop into serious cancers, according to the study.

The investigators looked at the use of screening mammography in 222,135 women who had 429,345 mammograms. The observation, from 1998 through 2002, took place at 43 facilities in Colorado, New Hampshire, and Washington.

Seven facilities, representing 16% of the study sites, implemented CAD during the study period. The current study included 2351 women who received a diagnosis of breast cancer within one year after screening and also received a mammogram that did or did not use CAD.

The authors estimate that for every additional woman diagnosed with breast cancer on the basis of CAD, 156 women are falsely recalled for more tests and 14 undergo unnecessary biopsies to exclude cancer. Every time the CAD software marks a real cancer, a radiologist has to consider about 2000 additional false-positive marks, making it very difficult to distinguish between real cancers and marks that are not cancer.

"The study is partially right and partially wrong," said Dr. William Poller, associate director of the Breast Care Center at Allegheny General Hospital in Pittsburgh. "It's already been shown that CAD is more helpful to less experienced readers and less helpful to more experienced readers. In this study, did researchers break down the data according to radiologists and their experience? That's the key."

Data in the study were not broken down by radiologists' level of experience.

Poller expressed concern that insurers could cut off reimbursement for breast CAD based on these findings. CAD, as well as MRI, helps bring profitability to breast centers, he said.

Has the use of CAD changed his center's false-positive rate? It was 68% from biopsy before CAD, and it remains about the same, he said.

"The bottom line is that you have to be willing to say a mark by CAD isn't important. CAD is only telling the radiologist to look at this place again," Poller said.

CAD was approved by the FDA in 1998 and has been incorporated into many mammography imaging practices, but its effect on the accuracy of interpretation has been unclear. This study does not address whether the risks of using CAD outweigh the benefits of its use. That kind of study could provide better guidance about what women should do.

"Within three years of FDA approval, 10% of the mammography facilities in the country were using CAD," Fenton said. "There had been no large-scale community-based review of CAD efficacy despite the rapid adoption of this technology, so we did this study to see if CAD was proving to be beneficial."

The authors estimated that if all mammography facilities adopt CAD, the annual cost of mammograms in the U.S. could increase 18%, or an additional $550 million nationwide.

There are several manufacturers of FDA-approved CAD devices, including R2 Technology (ImageChecker) and iCAD (SecondLook). The R2 Technology product was the first to receive FDA approval, and that company manufactured the devices used by facilities in this study. In one comparative study, the R2 Technology and iCAD products had similar effects on radiologists' interpretation of mammograms.

For more information from the Diagnostic Imaging archives:

Imagers prove too hasty in dismissing breast CAD

CAD technology cools its heels amid skepticism

Report from ARRS: CAD gives radiologists a slight boost in mammography

Breast CAD succeeds where human imagers may fail