Large Study Shows Mixed Results in Comparison of DBT and Digital Mammography

Digital breast tomosynthesis (DBT) provided significantly lower recall rates than digital mammography (DM) in three rounds of successive screening examinations, but no significant difference in the detection of interval or advanced cancers, according to newly published data drawn from 58 breast imaging facilities in the Breast Cancer Surveillance Consortium.

For the study, recently published in Radiology, researchers compared data from 523,485 DBT exams and 1,008,123 DM exams for 504,863 women over three rounds of successive breast cancer screening. The study cohort was comprised of women, ranging between 40 to 79 years of age, who had DBT or DM screening exams between 2011 and 2020, according to the study.

In the first round of screening, DBT had a recall rate of 7.5 in comparison to 10.9 for DM. The researchers also noted lower recall rates for DBT in the second round (6.9 vs. 8.6) and third round of screening (6.1 vs. 7.3).

“In our study, the difference in recall rate between the two imaging modalities was greatest in round 1. Although the difference in recall rate remained significantly different between modalities, it decreased over subsequent screening rounds,” pointed out Brian L. Sprague, Ph.D., who is affiliated with the Departments of Radiology and Surgery at the University of Vermont Cancer Center in Burlington, Vermont, and colleagues. “This attenuating pattern may be explained by the relatively high DM recall rate in round 1, which decreased on successive screening rounds – likely reflecting the value of comparison images from multiple prior screening rounds in reducing recall rate.”

In the last round of three successive digital breast tomosynthesis (DBT) exams (c), architectural distortion in a right craniocaudal DBT section led to a diagnosis of breast cancer. (Images courtesy of Radiology.)

(Editor’s note: For related content, see “Study: Emerging AI Platform for DBT Shows 23 Percent Increase in Breast Cancer Detection Rate,” “Large Mammography Study Reaffirms Benefits of Digital Breast Tomosynthesis” and “Study: AI Improves Cancer Detection Rate for Digital Mammography and Digital Breast Tomosynthesis.”)

The researchers noted that DBT had a higher cancer detection rate than DBT with three or more successive screenings (a difference of .6 per 1,000 examinations). However, they found no significant differences in the sensitivity rate with the modalities. The study authors also saw no differences between DBT and DM in the detection of interval or advanced cancers.

“It is possible that the elevated cancer detection observed with DBT may predominantly represent slow-growing cancers that would not have arisen as interval or advanced cancers before the next screening examination,” suggested Sprague and colleagues. “Alternatively, additional follow-up examinations and increased sample size may be required to detect small reductions in interval or advanced cancer rates between DBT and DM, particularly if the differences are limited to subgroups of women, such as those with extremely dense breasts and high breast cancer risk.”

In an accompanying editorial, Per Skaane, M.D., Ph.D., said DBT has “great potential” for reducing false positives and superfluous recalls, but the lack of difference between DBT and DM with respect to interval and advanced cancer detection remains a significant issue.

“DBT is replacing DM as the imaging modality of choice for routine screening in the United States, but this has not been the case in population-based European screening programs,” pointed out Dr. Skaane, a professor emeritus at Oslo University Hospital in Norway. “One reason for non-implementation of DBT as a routine screening tool for women at average risk in Europe has been the mixed results on statistically significant reduction of interval cancers and associated lack of a reduction in advanced cancer rates. Consequently, there might be a potential for over-detection (over-diagnosis) and over-treatment.”

In regard to study limitations, the researchers noted a lack of randomization and conceded that residual confounding may have affected the study findings. Acknowledging small sample sizes for interval and advanced cancer outcomes, the study authors said they could not exclude the possibility of small but clinically significant differences between interval and advanced cancer detection rates.