Digital Breast Tomosynthesis May Improve Detection, Outcomes
Tuesday, Dec. 01, 2015
Digital breast tomosynthesis (DBT) used with digital mammography may over time detect more cancers, and more clinically significant cancers, than digital mammography (DM) alone, the current standard of care, according to a study presented Monday by researchers from the University of Pennsylvania
The study sought to build on previous studies showing that DBT/DM screening leads to decreased recalls and increased cancer detection. Because few studies have evaluated the sustainability of these outcomes, the researchers compared DBT/DM with DM alone at one, two and three years. The study found that DBT screening resulted in increased cancer detection and positive predictive value over time. It also showed a decrease in interval cancer rate within one year of screening, suggesting that DBT detects more clinically significant cancers earlier.
"These improvements in outcomes directly address the major criticisms of mammographic screening: too many false positives and too few cancer detections," said presenter Emily Conant, M.D., chief of the division of breast imaging at the Hospital of the University of Pennsylvania (HUP). "By improving these outcomes, DBT tips the benefit-risk ratio even further in supporting routine DBT screening of women for breast cancer."
Since 2011, HUP has used a DM/DBT combination for all screening mammograms. The study analyzed 33,000 screenings in the three years—about 11,000 per year—after the conversion, and compared results with prior DM rates on several measures: recall, cancer detection, positive predictive value, biopsy rates, and interval cancer rates within one year. A positive screen was defined as a recall prompting a biopsy recommendation. Network cancer registry data was used to determine interval cancer rate.
The study population had a mean age of between 56 and 57, and was about half black and 40 percent white. About 56 percent had scattered fibroglandular breast density and another 30 percent had heterogeneous density.
Recall rates decreased over all three years. The baseline DM rate was 10.4 percent, and the DBT rate was 8.8 percent for the first year, 9 percent for the second year, and 9.2 percent for the third. Cancer detection rates per 1,000 screened continued to increase from a baseline DM rate of 4.6 to 5.5 for DBT year 1, 5.8 for year 2, and 6.1 for year 3.
The biopsy rate remained relatively stable, and positive predictive value continued to increase over the three years. The interval cancer rate decreased from 0.9/1000 screened for DM to 0.5 for DBT year 1, and 0.1 for DBT year 2. There is not yet adequate follow-up to calculate interval cancer rate for DBT year 3.
Dr. Conant said she would like to see the study repeated in multiple centers to greatly expand the volume of data.
An audience member queried Dr. Conant about the danger of reader "burnout" from constant exposure to the vast amounts of information generated by DBT studies.
The HUP group, which reads 70 to 80 screening studies a day interspersed with about 60 diagnostic studies, has found that the learning curve for interpreting DBT information may be as long as two years, Dr. Conant said. However, the volume of information has not, in itself, presented a problem.
"Over time, DBT becomes a very task-driven tool," she said. "You look at the 2-D images for global things and they look very differently through the DBT stack," Dr. Conant said. "So we have not experienced burnout. We are still thrilled with DBT and we think It's the new standard of care."