Computer tool may not boost mammograms' accuracy
Last Updated: 2013-04-17 15:45:04 -0400 (Reuters Health)
NEW YORK (Reuters Health) - Using a computer tool to help doctors analyze mammography images increases the number of early, non-invasive breast cancers that are caught, but also means more women without cancer have to undergo follow-up ultrasounds and biopsies, according to a new study.
Researchers found use of so-called computer-aided detection, or CAD, has spiked in recent years. CAD was used in more than 60 percent of screening mammograms performed among women on Medicare in 2006.
But it's unclear whether the tool - which alerts doctors to possible cancers on digitized images - ends up doing more good than harm, said the study's lead author, Dr. Joshua Fenton.
For one, he said, it's unclear whether identifying very early cancers, known as ductal carcinoma in situ (DCIS) and treating them aggressively is even beneficial - especially among women in their late 60s and older.
Women with DCIS are generally advised to have surgery, sometimes with radiation and anticancer drugs.
"At this time we just don't know whether detecting these cancers reduces a woman's chance of dying of breast cancer," Fenton, from the University of California, Davis, told Reuters Health.
And over-testing women who turn out not to have cancer provides no added benefit - only possible harms.
"CAD has been broadly adopted, largely due to Medicare coverage that began in 2001, and after 12 years and over a billion dollars spent on CAD it's still unclear if any benefits of CAD outweigh its risks," Fenton said.
CAD can be used both on digital mammograms and on digitized versions of conventional x-ray mammogram films, according to two doctors' groups, the American College of Radiology and the Radiological Society of North America.
The U.S. Preventive Services Task Force, a government-backed panel, recommends women age 50 to 74 be screened for breast cancer every other year.
Fenton and his colleagues analyzed data from a nationwide study of 163,000 women, age 67 to 89, who had an average of two or three mammograms each between 2001 and 2006.
The proportion of those mammograms that used a computer detection tool increased from 3.6 percent at the start of the study to 60.5 percent by the final year, Fenton's team reported in the Annals of Internal Medicine.
The researchers found using CAD increased the number of DCIS cancers identified by 17 percent.
There was no change in the number of invasive cancers detected using the computer tool. The study team saw a slight shift toward more early- and fewer late-stage cancers diagnosed in the year following a mammogram - though it's unclear what would explain that, Fenton said.
Women without breast cancer who were screened using CAD were 7 percent more likely to have a breast ultrasound and 10 percent more likely to have a biopsy after a so-called false positive.
CAD is still useful for picking up tiny calcifications in the breast, according to University of Pittsburgh radiology researcher Bin Zheng - but it over-diagnoses the larger masses.
"The consequence is, you need more recall, more additional imaging tests, more biopsies, and this is a problem that hurts individually the patient and society, as medical care costs will be high," Zheng, who wasn't involved in the new study, told Reuters Health.
"Even in studies that have included younger women, (CAD) has been associated with a higher rate of false positives and more biopsies," Fenton said.
Given that consistent risk of harm, and a "dubious benefit," he suggested women be given the choice to have their mammograms done using CAD, or analyzed by doctors alone.
Rather than phase out CAD altogether, Zheng said he is in favor of looking into strategies to reduce the rate of false positives with the technology - since doctors are still prone to over-diagnosing breast cancer on their own.
SOURCE: http://bit.ly/14tHOnB Annals of Internal Medicine, online April 15, 2013.
Copyright © 2013 Reuters Limited. All rights reserved.Republication or redistribution of Reuters content, including by framing orsimilar means, is expressly prohibited without the prior written consent ofReuters. Reuters shall not be liable for any errors or delays in thecontent, or for any actions taken in reliance thereon.Reuters and the Reuters sphere logo are registered trademarks andtrademarks of the Reuters group of companies around the world.