Mammography?s Next Step: The Dawning of Breast Tomosynthesis

For decades, breast imaging has pivoted around one gold-standard technology—mammography. Although it is the standard in breast cancer detection, mammography is far from perfect. “Mammography is saving lives. The breast cancer mortality rate is down 25 percent since 1990, primarily because of early detection,” says Daniel Kopans, MD, director of breast imaging at Massachusetts General Hospital (MGH) in Boston. But the downsides to mammography are significant. The false-negative rate hovers in the 20 percent range; false-positives sit at 12 percent, and a radiologist cannot distinguish malignant and benign breast tissue by viewing a mammogram. Consequently, researchers are pursuing a number of avenues to improve the breast cancer detection process via mammography.

One technology close to the verge of reality is expected to address mammography’s shortcomings: breast tomosynthesis. GE Healthcare, Siemens Medical Solutions and Hologic, Inc. are developing breast tomosynthesis systems, and it’s likely that at least one vendor will receive Food and Drug Administration (FDA) approval in the next year, Kopans notes. As researchers demonstrate the promise of breast tomosynthesis, Health Imaging & IT offers a primer on the technology to help sites prep for the next wave of breast imaging technology.
 

A new view

Digital tomosynthesis acquires multiple projections of a compressed breast from different angles. Images are reconstructed and can be viewed individually or dynamically. The new and improved images address many of the problems associated with conventional mammography.

“Part of the problem with mammographic images is that the breast gets in the way,” says Kopans. That is, breast tissue can obscure malignancies. In some mammographic images, normal tissue can appear suspicious, setting off a chain of additional imaging and patient anxiety. False-positives cost money, impact workflow and frustrate patients and physicians. “Tomosynthesis eliminates problems caused by the superimposition of normal structures,” Kopans says. Tissue isn’t sandwiched; it can be viewed layer by layer, much like the slices of a multidetector CT or MRI study.

The ability to view tissue layer by layer translates into important benefits, says Liane Philpotts, MD, chief of breast imaging at Yale New Haven Hospital in New Haven, Conn. Most importantly, the ability to view individual layers is expected to reduce the callback rate because the radiologist’s view of any suspicious area is not obscured by breast tissue. In addition, radiologists should be able to detect more cancers, specifically those obscured (and missed) by overlapping tissue. Earlier detection is another strong possibility. “It is likely that radiologists will find more smaller cancers than they would with conventional mammograms,” says Philpotts.  

Over the last few years, Kopans and his colleagues at MGH have imaged 3,500 women with breast tomosynthesis and compared the findings to 2D mammography. The callback rate is reduced by 30 to 40 percent with tomosynthesis, mainly because tomosynthesis eliminates superimposition of the breast. Breast tomosynthesis also delivers image quality improvements. “Radiologists can see the margins of abnormalities with greater clarity,” says Kopans. Because cancers tend to be ill-defined masses and may have spiculations, clearer margins can help radiologists differentiate benign and malignant findings. Finally, tomosynthesis’ slices allows radiologists to see the exact location of an abnormality. “Breast tomosynthesis will be a great improvement over 2D mammography,” sums Kopans.


Breast tomosynthesis in practice

It’s fair to say that many radiologists’ heads are still swimming after adjusting (or considering the shift) from analog to digital mammography. Is tomosynthesis another major paradigm shift? Is it affordable? What does it mean for workflow?

It appears likely that tomosynthesis will be an easier—and could be a relatively affordable—transition. “Tomosynthesis images look like mammograms, but allow radiologists to scroll through the breast tissue. It’s similar to breast MRI,” says Philpotts. Because most radiologists have grown accustomed to 3D views in other modalities, the learning curve should be relatively short.

The workflow impact of breast tomosynthesis remains unclear. It takes Kopans about 40 seconds to review a breast tomosynthesis study, about the same amount of time as a mammogram. Philpotts provides a longer rough estimate, placing her digital mammography review time at one minute and tomosynthesis review time at two minutes. “It does depend on the patient. It takes longer to read a study if the patient has calcifications or masses,” adds Philpotts. In addition, the two radiologists prefer different protocols, which could partially explain the variability in review time.

Kopans believes it may be possible to provide an accurate interpretation of breast tomosynthesis studies based on medio lateral oblique (MLO) views of each breast. Philpotts disagrees. “It’s most likely that radiologists will need two views of each breast: the MLO and craniocaudal (CC),” she says.

Although the MLO view generally includes more tissue than the CC view, breast compression tends to be better during the MLO view. In fact, research conducted at MGH and presented at the Radiological Society of North America (RSNA) 2006 annual meeting indicated that subtle cancers may be missed in the MLO view. The study concluded that obtaining both views optimizes visualization of lesions. Ultimately, the workflow impact should be minimal with any increased review time cancelled out by a reduced callback rate.

The financial front remains a bit murky as well. Breast tomosynthesis systems are based on digital mammography detectors, which can cost two to four times as much as an analog system. But the additional cost of a breast tomosynthesis system should be marginal. That’s because most digital mammography vendors expect to offer digital mammography buyers an upgrade path to the tomosynthesis platform. “Administrators should insist on an upgrade path to tomosynthesis if they are in the digital acquisition process,” advises Kopans, who estimates that tomosynthesis will add 20 percent to the cost of a digital mammography system. But note that for sites going from analog to tomosynthesis, once available, will need to plan for a cost about four times that of an analog system.

Kopans adds another key cost point. “The initial equipment cost is a small component of the total cost of breast imaging.” Other factors such as radiologist and technologist time are larger elements of the cost equation. Plus, if preliminary data hold up, fewer women will require diagnostic mammograms, resulting in an overall cost-savings.

What’s more, tomosynthesis could provide a competitive advantage. “Tomosynthesis will be tremendous once it hits the market. Women will demand access,” predicts Philpotts. On the other hand, cost hurdles do exist. The FDA process is lengthy, and securing reimbursement for tomosynthesis, a prerequisite for return on investment, could take some time. Initially, tomosynthesis may be limited to academic or research sites, but as research demonstrates its advantages, penetration could occur fairly rapidly.

The other unknown in the tomosynthesis world is its role: screening or diagnostic breast imaging? “The exact role of breast tomosynthesis is yet to be determined. It will depend on how the FDA approves the first systems. Eventually, however, tomosynthesis will be used in place of mammography,” predicts Philpotts.


Time for tomosynthesis?

The breast tomosynthesis era is around the corner. The technology promises to improve on mammography by providing a clearer view of breast tissue and eliminating the challenges associated with overlapping tissue. The mammography recall rate should fall with tomosynthesis, and radiologists may be able to detect more cancers at earlier stages. The implementation process could be fairly smooth as more sites are equipped with digital platforms that form the backbone of breast tomosynthesis systems. Plus, tomosynthesis images look like other 3D datasets, which should translate into a manageable learning curve for radiologists. Smart providers are getting ready as breast tomosynthesis could leap from the research realm to clinical practice in the next year or two.