When conventional digital mammography is combined with digital breast tomosynthesis (DBT) for interpretation by two radiologists, cancer detection is significantly improved, especially for invasive cancers. But interpretation times nearly doubled for the combined exam.
These findings, as well as the fact DBT reduces false-positive rates by 18%, were published online on 4 April in European Radiology. This is the second peer-reviewed journal article from an interim analysis of results of the Oslo Tomosynthesis Screening Trial (OTST).
The first article, which reported that single reading of full-field digital mammography (FFDM) combined with DBT identified 40% more invasive cancers than single reading of FFDM alone, was published online on 7 January in Radiology.
OTST is an ongoing, population-based prospective study undertaken by Oslo University Hospital to determine if the use of tomosynthesis in screening can increase cancer detection rates. During its first year of implementation, from November 2010 through December 2011, 12,631 women had their biennial mammography screening exams as part of the Norwegian Breast Cancer Screening Program (NBCSP).
The women either had a conventional two-view screening mammogram (Selenia Dimensions, Hologic) or an exam that added the tomosynthesis procedure with a single breast compression per view. The tomosynthesis images took approximately 10 seconds to acquire, and the radiation dose of the tomosynthesis study was set to produce approximately the same dose as a single mammographic view.
Eight radiologists whose experience in reading mammograms ranged from two to 31 years interpreted the exams. Four radiologists independently interpreted each mammogram to meet the requirements of the clinical trial: single reading with and without DBT, and double reading with and without DBT. Interpretation times were automatically recorded into the NBCSP database.
Exams were rated on a five-point scale. If an exam had a score greater than 1 (benign), at least two radiologists subsequently discussed the findings to reach a consensus as to whether the patient required additional imaging, according to principal investigator and lead author Dr. Per Skaane, PhD, and co-authors.
The research team compared cancer detection rates, positive predictive values, and recall rates for double readings of mammography alone with mammography and DBT. A total of 1,382 FFDM-only exams received a score of 2 or more; of these, 366 women were recalled for additional imaging and 90 cancers were confirmed. A total of 1,175 FFDM plus DBT exams received a score higher than 1, and of the 463 recalled, 119 cancers were detected.
|FFDM plus DBT for cancer detection
Thirty-one cancers were only detected by the combined exam. Two cancers were not detected when tomosynthesis was added. However, 24 of the additional 29 cancers detected with the combined exam were node-negative invasive cancers. Twenty-one of these were speculated masses and/or distortions.
Positive predictive values were 24.7% for FFDM and 25.5% for FFDM + DBT for every 100 patients recalled. Recall rates were 2.9% and 3.7% respectively.
"The optimal method of using tomosynthesis in mammography screening needs to be addressed," the authors wrote. Average interpretation times nearly doubled with the combined exam (89 seconds compared with 48 seconds for FFDM only).
The authors also recommended meetings by interpreting radiologists be conducted when double reading protocols were utilized and scores differ. Joint discussion of independent findings substantially reduced the number of patients who needed to be recalled.