January Monthly Topic:

Mammography CAD

Since 1998, computer-aided detection (“CAD”) technology has been FDA-approved for use in mammography screening.  This relatively recent technological advance can assist radiologists to identify suspicious findings on mammograms by effectively functioning as a “second pair of eyes.”  After the radiologist has made an initial interpretation of a patient’s mammogram, the CAD software is used to detect and mark any breast abnormalities or “regions of interest” (“ROI”) on the film.  The radiologist can then re-review the mammogram to determine whether the areas marked by the CAD software are suspicious and require additional patient follow-up.  When using CAD, the radiologist still makes the final interpretation of the mammogram. (1) 

Although routinely used in many practices, CAD has recently come under scrutiny, which is likely a consequence of several recent publications that raised questions regarding CAD’s clinical utility.  As a result, the Food and Drug Administration (“FDA”) has demonstrated increased scrutiny of marketing applications for CAD technology.

The FDA’s reaction to CAD appears to be partially grounded in the inconsistencies regarding the technology’s clinical utility, as described in the scholarly literature.  While some research has suggested that CAD is more helpful to inexperienced radiologists and may be effectively utilized as an educational tool (2), other studies suggest that the benefits of mammography CAD may be outweighed by its disadvantages.  Specifically, a recent study published in the New England Journal of Medicine, by Fenton, et. al., found that women who received mammograms at facilities using CAD were more likely to be given false positives, which increased the likelihood of call-back visits and additional biopsies.(3)  The study found that due to the use of CAD, although 32 percent more women were recalled for more tests and 20 percent more women received breast biopsies, the technology had no clear impact on the early detection of breast cancer.  The study suggests that for every additional case of breast cancer detected, the use of CAD software will lead to 157 additional patient recalls and 15 additional biopsies.  These additional procedures could increase the national costs of mammogram screening by approximately $550 million each year. (4)  Critics of CAD technology argue that the benefits of the technology are not outweighed by the increase in health care costs.

On the other hand, many CAD advocates, including much of the radiology community, believe that the Fenton study does not fairly assess CAD technology.  Many argue that the CAD software used by Fenton is obsolete and that since the time of the study (2002), there have been numerous advances in CAD technology, which have improved its accuracy. (5)  It is also stressed that the primary purpose of mammography CAD is to detect breast cancer at an earlier, lower stage, a parameter that was not measured by Fenton.  Another parameter left unexplored in the Fenton study is that of the time it takes radiologists to adjust to the use of CAD technology.  According to manufacturers, this process can take from weeks to years.  Other studies have shown that proper training on the use of CAD technology has a statically significant effect on user performance in CAD-aided mammogram reading. (6) 

Critics and supporters alike seem to agree that CAD is particularly adept at detecting ductal carcinomas in situ, an early stage, noninvasive cancer, which currently accounts for approximately 25 percent of the breast cancer diagnosed in the United States. (7)  However, critics also highlight that the use of CAD technology may not reduce the mortality rate for breast cancer if it primarily aids in the identification of noninvasive—rather than invasive—cancer. (8)

In light of the competing concerns and the debate being played out in the scholarly literature, FDA is likely to be cautious in its assessment of mammography CAD and related CAD technologies, particularly with respect to new medical devices in this space.  The considerable public health implications of CAD are also likely to drive FDA to seek formal guidance from clinical experts in the form of an Advisory Panel.  Such a panel is likely to meet sooner rather than later, with some suggesting a meeting could be convened as early as this spring.