AJR: FFDM with breast tomosynthesis decreases recall rates 30%

The combined use of digital breast tomosynthesis with full-field digital mammography (FFDM) for breast cancer screening may result in a substantial decrease in recall rates by an average of 30 percent, according to a retrospective observer study in the August issue of the American Journal of Roentgenology.

David Gur, MD, from the department of radiology at the University of Pittsburgh, and colleagues sought to compare the diagnostic performance of full-field digital mammography (FFDM) with that of digital breast tomosynthesis.

Eight experienced radiologists interpreted images from 125 selected exams--35 with verified findings of cancer and 90 with no finding of cancer. The four display conditions included FFDM alone, 11 low-dose projections, reconstructed digital breast tomosynthesis images and a combined display mode of FFDM and digital breast tomosynthesis images.

Observers rated exams using the screening BI-RADS rating scale and the free-response receiver operating characteristic paradigm. Observer performance levels were measured as the proportion of exams prompting recall of patients for further diagnostic evaluation. The results were presented in terms of true-positive fraction and false-positive fraction.

According to Gur and colleagues, the use of the combination of digital breast tomosynthesis and FFDM was associated with a 30 percent reduction in recall rate for cancer-free exams that would have led to recall if FFDM had been used alone.

The researchers also found that use of digital breast tomosynthesis alone also tended to reduce recall rates, an average of 10 percent, although the observed decrease was not statistically significant. However, the authors noted that there was no convincing evidence that use of digital breast tomosynthesis alone or in combination with FFDM results in a substantial improvement in sensitivity.

In terms of study limitations, the authors noted that the reduction observed in this study "may be reduced in actual readings, in which women may be recalled for diagnostic evaluation of suspected abnormalities in either breast; in this study we assessed possible effects on single-breast-based diagnoses. We also recognize that this study was performed with one group of readers practicing at one institution. We presume that breast imaging practices are somewhat similar at other sites."

Based on an analysis of their findings, Gur and colleagues cautioned that conclusive data about possible improvement in sensitivity, if any, with digital breast tomosynthesis will have to be found and validated in studies in which selection and ascertainment bias is reduced, if not eliminated.