Whilst a similar trial is required to confirm these findings in a screening population, this trial suggests that synthetic 2D mammography could be cost-effective compared to combination 2D/3D mammography.
—Simon D. Holt, MD
In screening for breast cancer, results from a study in the United Kingdom suggest that the use of “synthetic” two-dimensional (2D) mammography, rather than 2D/three-dimensional (3D) combinations, could save radiologists’ time and patients’ exposure to radiation as well as result in many fewer recalls. With this approach, a 3D image is obtained, but only the enhanced, reconstructed, synthetic 2D image is interpreted unless the results are concerning.
This screening strategy was described by Simon D. Holt, MD, of Prince Phillip Hospital in Carmarthenshire, United Kingdom, at the 2015 San Antonio Breast Cancer Symposium.1
If the synthetic 2D is reported as M3, 4 or 5 on the BI-RADS (breast image-reporting data system) classification, only then is the 3D, or digital breast tomosynthesis, reviewed by the radiologist, he explained. Although the 2D/3D combination remains the most sensitive x-ray screening method, it has the disadvantage of doubling both the radiation dose and the reading times over the standard 2D alone as used by most screening programs at present.
“Screening for breast cancer with 3D, and reading the derived synthetic 2D films only, with selected review of 3Ds before recall, could be a cost-effective way of significantly increasing the sensitivity and specificity of screening,” Dr. Holt suggested.
Explaining the terminology, he said that 3D mammography is also called digital breast tomosynthesis, 2D mammography is full-field digital mammography, and synthetic 2D mammography produces 2D pictures derived entirely from 3D data, combining the individual optimally enhanced 1-mm slices of a digital breast tomosynthesis image.
“Synthetic 2D can present in one image some details normally best seen on the 2D films such as calcifications and at the same time select out other features of cancer normally best seen on the 3D, where overlapping normal breast parenchyma is less of a problem,” he said.
Rationale for Combination Screening
Combined 2D/3D (digital breast tomosynthesis) mammography is superior to 2D mammography alone in the diagnosis of breast cancer, as shown by a number of large studies. One large U.S. study of 454,850 patients found that combined 2D/3D screening detected 5.5 cancers per 1,000 women, compared with 4.3 cancers for 2D screening alone—a 28% higher detection rate and a 16% lower recall rate.2
However, combined 2D/3D screening roughly doubles both the radiation dose and the radiologist’s reading time. Synthetic 2D mammography, derived from 3D data without the need for additional radiation, retains the advantages of combined screening while reducing the disadvantages, revealed Dr. Holt.
“The hypothesis we wished to test was, If the synthetic 2D is normal/benign, is there any advantage in also viewing the digital breast tomosynthesis? Could reading the 2D synthetics with selective reading of the 3Ds be an answer to the problem of significantly increased reading times?” he said. The study used Hologic Selenia Dimensions system digital mammographic images and C-view synthetic 2D software to answer this question.
The investigators prospectively collected data on 2,500 unselected cases presenting symptomatically or at follow-up, all of whom underwent digital breast tomosynthesis. From the 3D data sets, synthetic 2D mammograms were constructed; they were read by one breast radiologist with 15 years of experience in interpreting mammograms and 5 years of experience with digital breast tomosynthesis. The radiologist reviewed the synthetic 2D films and reported them separately before then examining the digital breast tomosynthesis and issuing a final report.
The mammograms were reported M1 to M5 using the standard BI-RADs criteria. The BI-RADs scores for each breast were recorded prospectively. Similarly, the breast density as assessed by eye was recorded (fatty/average density/dense).
A total of 4,589 individual mammogram sets were reported. Researchers correlated the findings from the 2D synthetic and the 3D images.
If synthetic 2D had been used and the 3D films viewed only if 2D was BI-RADS M3/4/5, then only two cancers would have been missed, Dr. Holt reported.
“The correlation is very close, but there were 11 patients in whom the synthetic 2D image was reported normal or benign (M1 or M2), but the digital breast tomosynthesis image was reported as M3. Of them, 10 were benign on assessment and one, malignant. There was one patient reported as M2 on synthetic 2D image but M4 on digital breast tomosynthesis image. Assessment confirmed the malignancy,” he said.
Sixteen cases reported as suspicious of malignancy (M3/4) by synthetic 2D image were subsequently downgraded to benign after review of the digital breast tomosynthesis image.
This approach detected 144 cancers, of which 41 may have been missed on 2D imaging alone. Sixteen fewer patients would have required recall, which at 15 minutes per patient would have saved a total of 4 hours. The process would require reading of only 280 of the 3D mammograms. At 45 seconds per digital breast tomosynthesis image, this amounts to only 3.5 hours of the radiologist’s time, Dr. Holt calculated.
Of the 4,589 examinations, 1,131 (25%) were assessed as fatty, 1,851 (40%) were assessed as average density, and 1,607 (35%) were assessed as dense. One cancer was missed in an averagely dense breast and one, in a dense breast.
“Whilst a similar trial is required to confirm these findings in a screening population, this trial suggests that synthetic 2D mammography could be cost-effective compared to combination 2D/3D mammography,” Dr. Holt concluded. ■
Disclosure: Dr. Holt reported no potential conflicts of interest.
Emily Sedgwick, MD, Associate Professor of Radiology at Baylor College of Medicine, Houston, indicated that the study by Holt et al is the first description of the synthetic two-dimensional (2D) screening approach and has the strength of being prospective in design. Future evaluations would benefit ...