An Oregon Health & Science University Cancer Institute researcher is among a group of scientists who have found that the use of computer applications initially designed to improve the interpretation of mammography is less accurate than the standard way of interpreting mammograms and can increase unnecessary biopsies, according to a study published Thursday, April 5 in the New England Journal of Medicine. 

The use of computer-aided detection in clinical practice has increased after the U.S. Food and Drug Administration  approved the technology in 1998 and Medicare began reimbursing its use. Computer-aided detection (CAD) was intended to help radiologists find cancer with greater accuracy. It is like having a second interpreter read the image to be sure nothing is missed.

"We are all paying for the use of this technology and it appears it doesn't work as it was intended.  It results in more biopsies and other unnecessary workups," said Patricia Carney, Ph.D., associate director for Population Studies at the OHSU Cancer Institute and professor of family medicine, OHSU School of Medicine.

Carney estimates that the use of computer-aided detection in the United States could increase the annual costs of screening mammograms by approximately $550 million.

The researchers studied 222,135 women, a total of 429,345 mammograms, from 1998 through 2002 at 43 facilities in three states, seven of which implemented CAD during the study time period. The researchers compared measures among facilities that did and did not use CAD. They evaluated the sensitivity, specificity, positive predictive value, cancer detection rate, biopsy rate, and overall accuracy of screening mammography with and without the use of computer-aided detection. The number of cancer detected for every 1,000 screening mammograms was 4.15 before CAD was implemented and 4.20 after CAD was implemented, indicating cancer detection did not improve significantly. The number of breast biopsies performed per 1,000 screening mammograms was 14.7 before CAD was implemented and 17.6 after CAD was implemented.  Because more cancers were not detected using CAD, many of these extra biopsies were likely not needed.

Carney collaborated with Joshua Fenton, M.D., M.P.H., University of California, Davis, Sacramento; Joann Elmore, M.D., M.P.H., University of Washington; and Stephen Taplin, M.D., M.P.H., National Cancer Institute, and others.

A related editorial in the journal called the results of this research  "a substantial hit to this technology."  Though new tools for breast imaging are under development, and technologies such as magnetic resonance imaging (MRI) have shown promise in detecting cancers that cannot be seen on mammography or found during a clinical breast examination, mammography remains the current mode for screening.

However, according to Carney: "If a woman is advised to have a biopsy based on a CAD finding, she may want to ask that a second radiologist review the image before a decision to biopsy is made."
      
The OHSU Cancer Institute is the only cancer center designated by the National Cancer Institute between Sacramento and Seattle. It comprises some 120 clinical researchers, basic scientists and population scientists who work together to translate scientific discoveries into longer and better lives for Oregon's cancer patients. In the lab, basic scientists examine cancer cells and normal cells to uncover molecular abnormalities that cause the disease. This basic science informs more than 200 clinical trials conducted at the OHSU Cancer Institute.