Digital breast tomosynthesis may improve breast cancer detection rates

Authors

  • Mary Kay Barton MD


A recent prospective study has indicated that integrated 2-dimensional (2D) mammography and digital breast tomosynthesis (DBT) with 3-dimensional (3D) images may improve breast cancer detection and reduce false-positive readings (Lancet Oncol. 2013;14:583-589).

Standard 2D mammography is limited by the superimposition of breast tissue, which can interfere with reading by concealing cancers or making normal tissue appear abnormal. DBT technology uses several radiographs to reconstruct a pseudo-3D image. Investigators set out to determine whether this technology had advantages over standard mammography.

“I like to compare standard mammography–film/screen as well as digital–to a book with clear pages,” explains Daniel Kopans, MD, who is the senior radiologist in the breast imaging division at the Massachusetts General Hospital in Boston and the inventor of DBT, but was not involved in this study. “You can hold the book up to the light and see all the words, but they are superimposed on one another so that they are hard to read. DBT allows you to look at each page individually.”

The Screening with Tomosynthesis OR standard Mammography (STORM) trial, which was conducted from August 2011 through June 2012, was a prospective, population-based breast cancer screening study that compared mammography screen reading in 2 phases: standard digital 2D-only versus integrated digital 2D and digital breast tomosynthesis with 3D images. The patients had mammograms on a unit that takes 2D and 3D images at the same examination with one breast position and compression per view. Each 2D and 3D screening examination consisted of a bilateral 2-view mammogram. Therefore, each woman had paired results for each screening examination. The participants were asymptomatic women undergoing routine screening.

Screening tests were interpreted in 2 ways: a radiologist first looked at the 2D image and issued a report, and the same radiologist on the same day then issued another report based on integration of the 2D and 3D images. Radiologists had to decide whether to recommend recall of a patient after reading the 2D image alone, and then again after the integrated reading. Primary outcomes were the numbers of cancers detected and the numbers and percentage of false-positive recalls, as well as the incremental cancer detection rate attributable to integrated 2D and 3D screening.

Increased Detection Rates

Population-based breast cancer screening programs in Trento and Verona, Italy examined 7292 women (median age, 58 years) and detected 59 cancers (52 invasive and 7 ductal carcinoma in situ) in 57 participants. Thirty-nine cancers were detected on both the 2D reading and the integrated 2D and 3D reading. Another 20 cancers were found only by the integrated 2D and 3D reading. No cancers were detected by the 2D reading that were not also recognized on the integrated 2D and 3D reading.

Figure 1.

A comparison of (Left) standard 2-dimensional mammography and (Right) digital breast tomosynthesis shows that breast cancer (arrow) is more clearly seen in the digital breast tomosynthesis image.

Furthermore, significantly more cancers were detected with the integrated reading; 5.3 cancers per 1000 screens were found with 2D reading alone versus 8.1 cancers per 1000 screens for integrated 2D and 3D readings (P < .0001). This represented a 33.9% increase in the cancer detection rate. The increased detection rate was similar between women with low-density breast tissue and those with high-density breast tissue, but the small number of women with high-density breast tissue limited the density-stratified analysis.

A total of 395 screenings resulted in false-positive readings (5.5%). Of these, 181 occurred from both readings. However, 141 false-positive results were found with 2D screens, and 73 were found with integrated 2D and 3D screens (P < .0001).

“Overall, we need to keep in mind that although the data on DBT are very exciting, we have relatively little evidence from large screening trials, so one has to proceed with some caution,” says Nehmat Houssami, MPH, PhD, corresponding author and associate professor at the Sydney School of Public Health at the University of Sydney Medical School in Sydney, New South Wales, Australia.

Dr. Houssami and her colleagues noted that they studied comparative cancer detection between 2D-only and integrated 2D and 3D screens, not absolute screening sensitivity. They found significantly more breast cancers using the integrated technique as well as fewer false-positive readings. Based on these data, the authors noted that false-positive recalls could be reduced by approximately 20% using integrated 2D and 3D screening.

“Usually in order to detect more cancers (increased sensitivity) you have to lower your threshold of suspicion and that leads to more recalls (decreased specificity). DBT is unusual since it improves both sensitivity and specificity,” Dr. Kopans observed.

The STORM investigators also noted that no other screening trials to date have reported final results using DBT. There is a preliminary report of a trial performed in Oslo, Norway, that did demonstrate significantly increased detection rates using integrated 2D and 3D mammography as well as reduced false-positive findings in over 12,000 women screened during the first year of the trial (Radiology. 2013;267:47-56). The increased detection rate of 30% reported in the Oslo trial was similar to that in the STORM trial.

Furthermore, in the STORM trial, the 3D images were not interpreted independently of the 2D images, and therefore conclusions regarding DBT with 3D images alone cannot be made. “We have no evidence that the 3D mammogram can substitute for the 2D mammogram, and radiologists rely on the 2D images to form an opinion,” says Dr. Houssami. “Small studies, clinical series-type studies, have looked at this issue and have not come up with consistent findings.”

Limitations of the STORM study include the fact that repeat screenings were not evaluated and therefore the increased cancer detection rate may not continue (or may be lower) with repeated screenings. Likewise, the study did not measure whether increased detection decreased the breast cancer mortality rate, nor did it examine whether interval cancer rates decreased at subsequent screenings.

Another consideration is radiation dose. The average dose from 3D imaging is approximately the same as 2D imaging, and therefore integrated 2D and 3D imaging roughly double radiation exposure. However, there is technology recently available that allows 2D images to be reconstructed from the 3D images, thus eliminating the need for separate 2D images to be taken.

Should DBT Be a Standard Test?

Given the STORM trial findings, Dr. Houssami recommends that more studies are needed before DBT with 3D imaging is made a standard test for breast cancer screening.

“Recommendations for population screening should be based on solid, consistent evidence coming from several large screening trials,” Dr. Houssami says. “On this basis, we recommended awaiting replication of our findings from other studies as the next step. So far, we have results from only 2 large screening studies: our study based on double-reading as practiced in European screening programs, and interim results from the Oslo study. Additional screening trials using DBT are in progress, and are expected to provide relevant results in the near future,” says Dr. Houssami.

However, DBT is being offered in the community. Dr. Houssami says she would support the decision to have the integrated 2D/3D screening over standard mammography if a woman is fully informed of the available data.

By contrast, Dr. Kopans does not think it is premature to recommend digital breast tomography for general breast cancer screening. “Mammography has been shown in randomized, controlled trials to reduce the death rate from breast cancer,” Dr. Kopans adds. “DBT is a better mammogram. It simply makes sense to find more cancers early while decreasing the recall rate.”

Ancillary