Tomosynthesis with synthesised two-dimensional mammography yields higher cancer detection compared to digital mammography alone, also in dense breasts and in younger women: A systematic review and meta-analysis

Abstract

ObjectivesThis systematic review and meta-analysis focuses on breast cancer screening performance outcomes stratified into breast density, age, and reading procedure using ‘digital breast tomosynthesis (DBT) with synthesised two-dimensional mammography (s2D)’ compared to ‘digital mammography (DM) alone’. MethodsStudies comparing ‘DBT with s2D’ and ‘DM’ were searched in PubMed and Cochrane library. Pooled risk ratios (RR) using fixed or random effects models (F-/REM) for cancer detection rates (CDR), recall rates, interval cancer rates (ICR), biopsy rates, and positive predictive values (PPV) 1–3 were calculated. Outcomes were stratified into breast density (non-dense and dense), age (<60, ≥60), and reading procedure (double-/non-double reading). Risk of bias was assessed using the Quality Assessment of Diagnostic Accuracy Studies (QUADAS)-2 tool. ResultsWe identified 13 studies. Using DBT plus s2D compared to DM alone resulted in a higher increase in CDR for dense ([number of studies included, FEM RR, 95% confidence interval (CI)]; n = 3, 1.60, 1.16–2.22) versus non-dense breasts (n = 3, 1.32, 1.08–1.61). Recall rates were lower in dense (n = 2, 0.84, 0.75–0.94), but much lower for non-dense breasts (n = 2, 0.65, 0.59–0.72). Age stratification resulted in small differences in CDR (<60: n = 2, 1.64, 1.18–2.29 / ≥60: n = 2, 1.56, 1.19–2.05). After screening with DBT plus s2D compared to DM alone the risk of being recalled was less in non-double (n = 3, 0.57, 0.54–0.60) than in double reading (n = 5, 0.95, 0.81–1.11) and the risk of cancer detection was higher in double reading (n = 6, 1.53, 1.40–1.67) than in non-double reading (n = 4, 1.17, 1.02–1.33). ConclusionSince only few studies are available for meta-analyses statistical significance strongly depends on single study results. Taking this into account, the most important results concern the increase of CDR in women with dense breasts, the increased CDR in double reading, and the lower recall rates particularly with non-double reading.