Computer-aided mammography doesn't improve breast cancer

The widely used mammography software known as computer-aided detection (CAD) doesn't improve detection of invasive breast cancer, new research suggests. But CAD does increase the chances that a woman will be called back for further testing, according to the study, published July 27 in the Journal of the National Cancer Institute. "All in all, we found very little impact of CAD on the outcomes of mammography," said study author Dr. Joshua J. Fenton, an assistant professor of family and community medicine at the University of California, Davis. With colleagues, Fenton analyzed 1.6 million screening mammograms from seven states and the records of the nearly 685,000 women who got them from 1998 to 2006. Three out of four mammograms done in the United States include CAD, the authors said. To determine if CAD aided breast cancer detection, they examined the detection rate when the software was used, and the stage and size of the cancer when diagnosed. They also looked at how many women were called back for further testing who didn't have breast cancer. "We found that CAD has a slightly increased chance [that] a woman would be recalled unnecessarily for further testing, but it did not increase the chance that the breast cancer would be detected at an earlier stage," Fenton said. The study indicated that for every 200 women who are screened with CAD who have a second mammogram, one additional woman is called back unnecessarily for further testing, he said. When Fenton's team looked only at early-stage cancer known as DCIS (ductal carcinoma in situ), they found a trend to increased detection with CAD, but it was not statistically significant. Overall, the detection rates with and without CAD were similar, the study said. Use of CAD, which identifies areas of potential abnormalities so the radiologist can take a closer look, has boomed in recent years, even though previous research has not demonstrated a clear benefit, Fenton said. Congress mandated that Medicare reimburse for CAD in 2001, and most private insurers followed suit, he said, explaining the growing popularity. Based on previous research, Fenton reported in the New England Journal of Medicine in 2007 that CAD was linked with reduced accuracy of mammogram interpretations but did not affect the detection rate of invasive breast cancer. In the current study, Fenton's team said the costs of CAD -- about an additional $12 per mammogram -- may outweigh the potential benefits. Annually, direct costs to Medicare top $30 million, a study published last year in the Journal of the American College of Radiology found. Dr. Carol Lee, a New York radiologist and spokesperson for the Breast Imaging Commission of the American College of Radiology, said the new study looks at CAD in real-life settings. "What this is saying is, in actual practice, when you look in the community, it doesn't seem to be living up to what earlier studies with different designs promised,"" she said. But, ""I don't think based on this study we should abandon the use of CAD," she added. For one thing, it's not clear whether the radiologists were trained to use CAD correctly, she said. The authors also point this out as a possible limitation. The trend to CAD picking up more early-stage cancers is worthwhile, she said. Lee also said a woman is unlikely to know whether her mammogram included CAD unless she asks the doctor. In its practice guidelines on mammography, the American College of Radiology says that CAD ''may slightly increase the sensitivity of mammographic interpretations. "It also notes that CAD may be linked with increased recall of patients, some unnecessary. Under the guidelines, CAD is not considered standard of care.