Prospective Trial of Digital Breast Tomosynthesis (DBT) in Breast Cancer Screening.

The accuracy of two view digital x-ray mammography (2DDM), the standard test for breast cancer screening, is limited because of the effect of superimposition of normal breast structures onto a two dimensional image. The mammography signs of breast cancer may be obscured, particularly in women with dense fibro glandular breast tissue, resulting in delay in diagnosis of cancer. National interval cancer data shows that up to 4000 women per annum (2.88 per 1000 screened) are diagnosed with breast cancer in the interval between screens. Conversely, superimposition of normal tissues may produce features on mammography which are suspicious for cancer and lead to unnecessary recall for further diagnostic tests. National screening data for 2012/2013 show that of 2.3 million women screened, 79000 (3.4%) without breast cancer were recalled to specialist diagnostic assessment clinics.

DBT is an x-ray mammography technique which involves acquiring multiple low dose projections over a limited angular range (less than 50 degrees). These projection images are reconstructed into a set of images consisting of parallel planes, typically 1mm apart throughout the breast. Three dimensional information is provided for the reader, overcoming many of the interpretation problems associated with 2DDM. Several studies have demonstrated improvement in the accuracy of DBT in the diagnostic setting. Published data from studies of DBT combined with 2DDM in screening have demonstrated increased rates of invasive cancer detection, cancers detected at a smaller size and decreased false positive recall rates. However, there may be increased costs associated with the technology, image data storage and longer reading time.

As a result of recent technical developments, a two dimensional mammography image can be produced by processing the tomosynthesis image data. Synthetic 2D (S2D) trial data show that the accuracy of DBT + S2D may be the same as that achieved with DBT + 2DDM.

The primary aim of this trial is to assess the impact of the addition of DBT to 2DDM in routine breast cancer screening compared to 2D mammography. 100 000 women will be recruited over twelve months using NHS screening sites with DBT ready mammography equipment. At each site, through a clinic randomisation process, half the trial participants will undergo standard 2DDM (the control group) and half will undergo 2DDM and DBT (the study group). Synthetic 2D images will be produced for all women in the intervention group. A secondary objective of the trial is to compare the accuracy of DBT + 2DDM with DBT + S2D. For each woman in the intervention group, one reader will view DBT + 2DDM, and the other reader will view DBT + S2D. This will enable a non-inferiority test to be applied to DBT + S2D vs DBT + 2DDM.

The investigators plan to use DBT in addition to 2DDM in the intervention group because:

• Most previous trials demonstrated increased sensitivity and specificity by using DBT in combination with 2D imaging rather than as a replacement.
• By using both 2D and DBT in the study group, the investigators will be able to review the display of the same cancer in the two techniques.

One concern about using DBT in conjunction with 2D imaging is the increase in radiation dose. The investigators conservatively estimate the mean glandular radiation dose of 4mGy in the control group and 9mGy in the study group. In other words, DBT imaging requires slightly more than double the dose of 2D imaging. A recent study of Yaffe \& Mainprize 2011 has shown that benefits of screening with mammography greatly exceed the detriment due to the radiation. Using similar calculations the investigators estimate that the ratio of lives saved and lost due to radiation in the 2DDM screening arm will be in the range 150-300 to 1. The higher dose in the study group is expected to lead to greater benefit and greater detriment. In order for the extra benefit to exceed the extra detriment of using DBT in the study group there will need to be a very small (0.45% to 0.9%) increase in lives saved. In practice the investigators expect a larger benefit that this.

• Health need: the effectiveness of 2DDM in detection of invasive breast cancer has improved over the last twenty years as demonstrated by a decrease in interval cancer rates (cancers occurring between routine three yearly screens). This improvement is a result of the introduction of two view mammography, digital mammography and improvements in film reading performance. However, the most recent published interval cancer rates of 2.85 per 1000 women screened demonstrates the limited sensitivity of 2DDM in cancer detection. Published data from studies of DBT and 2DDM in screening in Europe and North America demonstrate an increased invasive cancer detection rate of 40-48% compared to 2DDM alone. Data from this study will demonstrate whether such an improvement in sensitivity is achievable in the NHSBSP, whether cancers can be detected at a smaller size, and prognostic features of screen detected and interval cancers will be used to model whether there would be a greater reduction in breast cancer mortality from screening with 2DDM plus DBT compared to 2D alone. Data will also demonstrate the effect on false positive recalls in the NHSBSP. Published results show that this may be decreased by 15-20% with the use of DBT.
• Economic analysis: The study will demonstrate whether replacing standard 2DDM with DBT (and S2D) represents an efficient use of NHS resources, either because DBT reduces overall cost or because the additional health benefits of greater sensitivity and specificity justify the additional cost.
• Sustained interest: DBT may have a major impact on the effectiveness and the cost of screening. Data from this study will help decide the optimum strategy for using DBT in routine screening in the NHS.
• Development of research area: results from this study will help identify which groups of women will benefit from screening with DBT.
• Scientific knowledge: this study will advance our understanding of the effect of DBT on the characteristics of screen detected cancers, whether more biologically significant cancers are found when smaller and at an earlier stage, and whether more low grade cancers are diagnosed, increasing the problem of over diagnosis.