3D Mammography May Improve Invasive Breast Cancer Detection

Laurie Barclay, MD

January 15, 2013

Screening with 2-dimensional (2D) mammography plus 3D reconstruction using tomosynthesis significantly increased detection of invasive breast cancers by 40% and reduced false-positive rates by 15%, according to findings of a large, prospective screening trial published online January 7 in Radiology.

"Digital mammography–based tomosynthesis produces cross-sectional sections by using multiple, low-dose acquisitions with total radiation exposure and breast compression similar to that used in conventional mammography," write Per Skaane, MD, PhD, from the Department of Radiology, Breast Imaging Center, Oslo University Hospital Ullevaal, Norway, and colleagues. "A number of small retrospective studies in which investigators evaluated tomosynthesis primarily in laboratory settings by using cancer-enriched populations demonstrated the potential for decreasing recall rates and possibly increasing cancer detection rates; however, none of these studies was performed in a manner that could affect treatment decisions."

This ongoing screening study compares digital mammography alone and combined with tomosynthesis in terms of detection rates for invasive and noninvasive breast cancers, false-positive rates before arbitration, and positive predictive values for women recalled after arbitration.

Four radiologists independently read mammograms alone and combined mammograms plus tomosynthesis for all participants. The present report is of a preplanned interim analysis of 12,631 women who underwent exams between November 22, 2010, and December 31, 2011. Statistical analyses included marginal log-linear models for binary data, which allowed accounting for correlated interpretations and adjustment for performance levels of each radiologist, using a 2-sided significance level of .0294.

"Very Significant Increase"

For mammography alone, the rate of detection of invasive and in situ cancers was 6.1 per 1000 examinations. Adding tomosynthesis to mammography resulted in a significant increase to a rate of 8.0 per 1000 examinations (27% increase; P = .001 after adjustment for reader). The increase in cancer detection occurred in women of all breast tissue densities, from dense to fatty.

"The combination of 2D full-field digital mammography and tomosynthesis detected more cancers than 2D alone," Laurie Margolies, MD, associate professor of radiology at the Icahn School of Medicine at Mount Sinai and chief of breast imaging at Dubin Breast Center, New York City, told Medscape Medical News when asked for comment. "This is a very significant increase in the detection of breast cancer. The rate of cancer is higher than at US centers since screening is done only every other year." Dr. Margolies was not involved in the current study.

Mammography plus tomosynthesis detected 25 additional invasive cancers compared with mammography alone, which represented a 40% increase (P < .001 after adjustment for reader). The combination study did not increase the detection of noninvasive ductal carcinoma in situ, which some critics of mammography screening claim may potentially be overdiagnosed.

"The overwhelming majority of the cancers found were invasive cancers — those that have the potential to spread outside the breast," Dr. Margolies added. "Forty percent of those found only on tomosynthesis were grade 2."

Furthermore, there was a significant 15% decrease in false-positive rates before arbitration (61.1 per 1000 examinations with mammography alone and 53.1 per 1000 examinations with mammography plus tomosynthesis; P < .001 after adjustment for reader).

"The combination exam also decreased the callback rate," Dr. Margolies noted. "The recall rate in the US is significantly higher; if a similar percentage decrease was observed, many fewer women would have the expense and anxiety associated with a recall."

For recalled patients with cancers subsequently verified, positive predictive values after arbitration were not significantly different with mammography alone (29.1%) or with mammography plus tomosynthesis (28.5%; P = .72).

Mammography plus tomosynthesis had a longer mean interpretation time (91 seconds) than mammography alone (45 seconds; P < .001).

More Studies Needed

Strengths of this study, according to Dr. Margolies, include the prospective design, large sample size, and use of a screening population. In contrast, many prior tomosynthesis studies have been reader studies of enriched populations.

However, she also notes study limitations, including differences in the performance of screening mammography in Oslo and in the United States.

"Women in this study were screened every 2 years, rather than yearly. This likely increased the number of invasive cancers in this screening population," Dr. Margolies said. "This study was performed at just 1 institution with 4 radiologists. It would be important to know that a large number of radiologists would have similar results."

In terms of additional research, Dr. Margolies recommends a similar study involving multiple institutions and many radiologists across a broad spectrum of facility types. She also suggests comparisons of mammography alone, mammography plus tomosynthesis, mammography plus ultrasound, mammography plus molecular breast imaging, and mammography plus magnetic resonance imaging. This would help determine whether ultrasound, molecular breast imaging, and/or magnetic resonance imaging would detect additional cancers, and if so, how many, compared with mammography plus tomosynthesis alone.

"At this time, the only drawback [of 3D mammography for breast cancer screening] for the patient is the additional radiation; this issue may be solved in the near future if the [US Food and Drug Administration] approves a synthesized 2D image," Dr. Margolies concluded. "The exam does not take significantly longer for the patient but does generally take longer for the radiologist to read and requires more expensive equipment and larger storage archives."

The study authors note an additional concern, namely that it is "virtually impossible" to determine overdiagnosis in studies of less than 10 to 20 years' duration.

"We believe that many of the additional abnormal findings detected with mammography plus tomosynthesis in our study are the very types of cancers one would hope to detect early and treat," the study authors conclude. "Perhaps our most important observation is that with the mammography-plus-tomosynthesis arm, the actual benefit, in terms of possibly improving outcome owing to earlier detection, may be larger than merely the difference in the total count or number of detected cancers."

Hologic supported this study by providing tomosynthesis equipment and financial support for additional interpretations. Hologic also employs 2 of the study authors. The other authors and Dr. Margolies have disclosed no relevant financial relationships.

Radiology. Published online January 7, 2013. Abstract