Surgeons and radiologists work side by side in the evaluation and management of breast diseases, so it is essential that surgeons remain well informed regarding novel breast-imaging modalities. Ongoing dialogue between specialists in these two fields will ensure that women receive clinically relevant breast cancer surveillance with appropriate interpretation of results and follow-up.
Digital breast tomosynthesis (DBT) is a promising new technology that has received a great deal of attention thanks to Food and Drug Administration approval of a manufacturer’s unit, and thus it is an example of a screening strategy that surgeons will need to incorporate into their armamentarium of breast-imaging options.
By Dr. Mark Helvie
The primary weakness of conventional mammography is detection of noncalcified cancers in patients with radiographically dense breasts. DBT attempts to address this weakness by providing thin-slice reconstructions of the breast. These reduce the camouflage of significant densities by overlapping structures in a way that is analogous to CAT scan imaging. In particular, dominant masses should be more apparent and pseudo lesions should be rendered inapparent. Radiologists then interpret these images by scrolling through a series of images for each view. For example, if a 60-mm-thick breast is reconstructed at 1-mm slice thickness, the radiologist would review 60 images for each projection, such as the cranial caudal view.
There are many questions to be answered after FDA approval, including DBT accuracy in clinical practice, physician training, interpretation time, ideal acquisition method, biopsy capability, radiation dose, and reimbursement.
Different manufacturers have approached clinical DBT in different fashions. The Hologic FDA application consisted of four views of each breast, compared with the conventional two views. The DBT images were incremental to the conventional two views. The radiologist interpreted both conventional mediolateral oblique (MLO) and cranial caudal (CC) views as well as the DBT MLO and DBT CC views. Other manufacturers may choose to use fewer views. Using this study design, Hologic submitted to the FDA two different reader studies showing a 7% improvement in the receiver operator curve (ROC)/area under the curve (AUC). In both studies, ROC AUC improvement occurred because of improved specificity, but not improved sensitivity – recall rates were lower, but cancer detection was not higher.
With training, radiologists should be able to improve sensitivity as well as specificity, based on the ROC curve, although this was not proven. However, we know that the attainment of additional images – even conventional images – results in improved diagnostic performance. This is why recalls are obtained. The unique benefit of DBT (compared with conventional mammography) is unknown. Conventional views may remain the optimal strategy in the detection of calcifications, while the strength of DBT may be for masses.
The diminution in recall rate obtained in research situations may or may not translate into clinical practice. It is unlikely to benefit calcification recalls. DBT should be able to prospectively resolve certain findings, such as asymmetry into true mass versus overlapping tissue, and allow detection of previously unseen masses. Incremental physician training and experience will be necessary prior to clinical reading – a fact that the FDA emphasized. While physicians may feel comfortable with DBT because it is similar to conventional mammography, lesion characterization may mandate different biopsy thresholds. Because more than double the information will be presented to the radiologist, reading times will take two to three times longer than conventional read times.
Lacking non-manufacturer larger trials showing sensitivity improvements, insurers might be reluctant to provide incremental reimbursement. Clinical adoption may mirror the experience with digital mammography, which was slow to be accepted until 2005, when DMIST (Digital Mammographic Imaging Screening Trial) showed a significant sensitivity improvement for young women with dense breasts. Compounding these issues is the recent U.S. Preventive Services Task Force controversy regarding screening mammography in general. Given that the USPSTF did not even recommend digital mammography, its acceptance of DBT may be expected to be similarly lacking.
Radiation dose, while overall considered very low for digital mammography, will be greater with DBT than with conventional mammography using the FDA-approved system. Doses will be within current FDA limits (3 mGy) per view. It is likely however, that dose reduction can be achieved over time. Different manufacturers are addressing these issues in different ways. Lesions detected solely by DBT will require biopsy technology unique to DBT.
Competing against DBT are other imaging technologies as well as nonimaging technologies. Whole-breast ultrasound screening, although effective in improving sensitivity, has achieved limited success with the application process because of the time needed for scanning, low specificity, and reimbursement issues. Magnetic resonance (MR) screening is sensitive although not specific, is costly, and requires intravenous contrast administration.