Tomosynthesis Versus Digital Mammography in a Population-based Screening Program

Overview

Preliminary results from prospective screening trials show that Digital Breast Tomosynthesis (DBT) is more sensitive and more specific than conventional digital mammography (DM). However, large randomised controlled trials (RCTs) are needed to confirm these promising results. Furthermore, potential side effects of DBT in the screening setting have to be properly evaluated. In particular, the major concern is about overdiagnosis and its consequences; overdiagnosis refers to the detection of cancers at screening, which would not have become clinically apparent in the woman's lifetime. This RCT was designed to compare benefits and harms of DBT in a population-based screening program with conventional DM. The primary outcome parameter will be the difference between the two tests in rates of advanced cancers detected at the subsequent round and interval cancers (within the first two years after the study screening round). Secondary outcomes measures will be diagnostic performance indicators for organised breast screening (i.e, cancer detection rate, recall rate, false positive rate, positive predictive value), parameters used in cost-effectiveness analysis, pathologic and biological characteristics of screen-detected cancers.

Design of the study This trial will involve several hospital centres within the organised breast screening in the Italian region of Piedmont. An invitation letter to participate in a randomised screening trial comparing DBT versus DM will be mailed to all women, 46-68 years old, living in the Piedmont Region, and eligible for invitation to the regional mammography screening program. Women who attend the centres for screening and consent to the study will be randomly allocated either to DBT or DM arm. Women in the DM arm will be screened with bilateral, two-views digital mammography examination. Women in the DBT arm will be screened with the bilateral two-views combo mode (DM and DBT images acquired in a single compression). In both screening arms, examinations will be independently read by two expert screening radiologists. In the DBT arm, each reading will be randomly allocated (on a per-case basis) to one of these three modes: DM plus DBT (DBT is read only after initial DM interpretation), DBT plus DM (DM is read only after DBT interpretation) and "concurrent mode" (both DBT and DM are read simultaneously). In both screening arms, participants will be recalled for further examination if indicated by at least one radiologist (without consensus or arbitration). At the subsequent round all participants will be invited to DM only. Interval cancers and advanced cancers screen-detected at the next round will be recorded. Pathological and biological parameters of screen-detected cancers (e.g, stage, grading, hormone receptors, HER2) in both screening arms will be measured to investigate overdiagnosis. With a sample size of 23,000 and 69,000 women in the DBT and DM arm respectively, the investigators will be able to observe a significant decrease in interval cancers and advanced cancers rate of at least 50% (power 80%, two tails). The collected data will be also used to perform analyses on radiological work-flow and workload, and cost-outcome and cost-effectiveness.