A new study published by the Journal of the National Cancer Institute revealed that computer-aided detection (CAD) for analyzing and interpreting mammograms does not improve accuracy. The United States currently uses CAD technology for analyzing three out of four mammograms. The technology recognizes patterns linked to breast cancers and tags potential abnormalities for radiologists to consult before making a final diagnosis.

Between 1998 and 2006, Joshua J. Fenton, M.D. and his colleagues from the University of California, Davis, analyzed data from over 1.6 million film screening mammograms at facilities in seven states, to determine whether CAD offers more accurate readings of mammograms. The data quality has been strictly monitored by the Breast Cancer Surveillance Consortium, in which all facilities participate.

The study period consisted of 25 out of a total of 90 facilities using CAD for an average of 27.5 months. Researchers compiled information on mammograms obtained from women with and without CAD technology, irrespective of those diagnosed with breast cancer or not within a year of screening.

The study revealed, that CAD readings resulted in more false positive identifications, labeling tumors incorrectly. No improvement was shown in detecting invasive cancers. In addition, the cancers detected through CAD were no more likely to be smaller, less advanced or have less lymph node involvement than those detected without using the technology. The results did not vary after adjusting patient parameters, such as age, breast density, use of hormone replacement therapy and other factors that might have an influence on mammography findings.

The cost of using CAD costs Medicare more than $30 million per year. The author points out:

“As currently implemented in U.S. practice”, CAD screenings produce questionable health benefits, as they appear to increase the risk for women having to be recalled for further testing after their CAD mammography screenings.”

Donald Berry, Ph.D. of the M.D. Anderson Cancer Center in Houston states in an accompanying editorial that any benefit to CAD is likely to be so inconsiderable, that it would be difficult to detect even in a very large randomized study. In addition, improving the sensitivity of CAD may find less aggressive tumors or others that would show up between mammograms. He furthermore writes, that an early detection of such tumors is unlikely to have much of an impact on breast cancer mortality.

The editor concludes, that CAD needs more research to be of real value, stating:

“. . . . this should happen in an experimental setting and not while exposing millions of women to a technology that may be more harmful than it is beneficial.”

“Effectiveness of Computer-Aided Detection in Community Mammography Practice”
Joshua J. Fenton, Linn Abraham, Stephen H. Taplin, Berta M. Geller, Patricia A. Carney, Carl D’Orsi, Joann G. Elmore, William E. Barlow
J Natl Cancer Inst (2011) doi: 10.1093/jnci/djr206

Written by Petra Marcus