Automated breast ultrasound comes up a winner in dense-breast screening

Twitter icon
Facebook icon
LinkedIn icon
e-mail icon
Google icon
 - Thinkingwoman

Women with dense breasts are significantly better served when imaged with a combination of screening mammography and supplemental 3D automated breast ultrasound (ABUS) than when they’re imaged with screening mammography alone.

That’s because the combo catches significantly more cancers than its solo counterpart, and it does so without driving up recalls and false positives, according to a study running in the June edition of the American Journal of Roentgenology.

Maryellen Giger, PhD, of the University of Chicago and colleagues looked at 17 radiologists interpreting a cancer-enriched set of 185 full-field digital mammography (FFDM) and ABUS exams of asymptomatic women with BI-RADS C or D breast density.

The radiologists first interpreted the FFDM studies alone, then FFDM combined with ABUS.

Of the 185 exams, 133 were negative and 52 had biopsy-proven cancers. Of the 52 cancer cases, the screening FFDM images were interpreted as showing BI-RADS 1 or 2 findings in 31 cases and BI-RADS 0 findings in 21 cases.

Comparing radiologist performance in terms of sensitivity, specificity and area under the curve (AUC) of the receiver-operating characteristic, Giger and team found:

  • When a cutpoint of BI-RADS 3 was used, the sensitivity across all readers was 57.5 percent for FFDM alone and 74.1 percent for FFDM with 3D ABUS—a relative increase in sensitivity of some 29 percent.
  • Overall specificity was 78.1 percent for FFDM alone and 76.1 percent for FFDM with ABUS.
  • The AUC was 0.72 for FFDM alone and 0.82 for FFDM combined with ABUS, producing a statistically significant 14 percent relative improvement in AUC.

Further, for the mammography-negative cancers, the average AUC was 0.60 for FFDM alone and 0.75 for FFDM with ABUS, yielding a statistically significant 25 percent relative improvement in AUC with the addition of ABUS.

Supplemental sonography continues to impress

In their discussion, the authors note that their study confirms the results of previous research showing that supplemental screening breast ultrasound turns up cancers that elude detection with mammography alone.

“These same studies have shown that most of these sonographically detected cancers are small, node-negative, invasive cancers that are clinically important to detect in their earliest stages for improved prognosis,” they write.

The addition of ABUS to screening mammography in the present study “showed a significant increase in cancer detection with a nominal insignificant decrease in specificity,” Giger et al. conclude. “Although these findings were in a research environment, one might expect a similar impact of screening ABUS in clinical practice.”

In fact, among the study limitations the authors acknowledge is its design as a reader study rather than a clinical trial, which may skew reader-performance results away from what they would be in a clinical setting.

“However, the performance of a reader study neutralizes the impact of a single radiologist's interpretation,” they counter, “because 17 radiologists interpreted the same examinations, albeit in different order.”

ABUS vs. handheld

Comparing ABUS technology with bilateral handheld ultrasound, the authors suggest that ABUS may go further to ease integration of ultrasound into the screening workflow environment.

Acquisition of each ABUS image requires approximately 1.5 minutes, they point out before adding that, with the usual three images per breast, the ABUS examination time “is far less than the 20 minutes reported for scanning with handheld ultrasound.”

“Also, if handheld scanning is performed by an ultrasound technologist, the significant findings are those detected and saved by the scanning technologist, not by the interpreting radiologist,” Giger and co-authors write. “In contrast, the automated image acquisition uncoupled from interpretation, as occurs with ABUS, allows the interpreting radiologist to analyze the entire dataset.”