Meta-analysis of prospective studies evaluating breast cancer detection and interval cancer rates for digital breast tomosynthesis versus mammography population screening

Nehmat Houssami; Sophia Zackrisson; Katrina Blazek; Kylie Hunter; Daniela Bernardi; Kristina Lång; Solveig Hofvind

doi:10.1016/j.ejca.2021.01.035

Meta-analysis of prospective studies evaluating breast cancer detection and interval cancer rates for digital breast tomosynthesis versus mammography population screening

Eur J Cancer. 2021 May:148:14-23. doi: 10.1016/j.ejca.2021.01.035. Epub 2021 Mar 9.

Authors

Nehmat Houssami¹, Sophia Zackrisson², Katrina Blazek³, Kylie Hunter⁴, Daniela Bernardi⁵, Kristina Lång², Solveig Hofvind⁶

Affiliations

¹ Sydney School of Public Health, Faculty of Medicine and Health, University of Sydney, Sydney, Australia. Electronic address: nehmat.houssami@sydney.edu.au.
² Department of Translational Medicine, Diagnostic Radiology, Lund University, Skane University Hospital, Malmö, Sweden.
³ Sydney School of Public Health, Faculty of Medicine and Health, University of Sydney, Sydney, Australia.
⁴ NHMRC Clinical Trials Centre, University of Sydney, Sydney, Australia.
⁵ Breast Imaging and Screening Unit, Department of Radiology, Humanitas Research Hospital, Milan, Italy.
⁶ Section of Breast Screening, Cancer Registry of Norway, Oslo, Norway.

PMID: 33706172
DOI: 10.1016/j.ejca.2021.01.035

Abstract

Introduction: Breast cancer (BC) screening using digital breast tomosynthesis (DBT) has been shown to increase cancer detection compared with mammography; however, it is unknown whether DBT impacts interval cancer rate (ICR).

Methods: We systematically identified prospective DBT studies reporting data on screen-detected and interval BCs to perform a study-level meta-analysis of the comparative effect of DBT on ICR in population screening. Meta-analysis of cancer detection rate (CDR), ICR, and the differences between DBT and mammography in CDR and ICR pooled estimates, included random-effects. Sensitivity analysis examined whether study methods (imaging used, comparison group design, interval BC ascertainment) affected pooled estimates.

Results: Five eligible prospective (non-randomised) studies of DBT population screening reported on 129,969 DBT-screened participants and 227,882 mammography-only screens, including follow-up publications reporting interval BC data. Pooled CDR was 9.03/1000 (95% confidence interval [CI] 8.53-9.56) for DBT, and 5.95/1000 (95% CI 5.65-6.28) for mammography: the pooled difference in CDR was 3.15/1000 (95% CI 2.53-3.77), and was evident for the detection of invasive and in-situ malignancy. Pooled ICR was 1.56/1000 DBT screens (95% CI 1.22-2.00), and 1.75/1000 mammography screens (95% CI 1.46-2.11): the estimated pooled difference in ICR was -0.15/1000 (95% CI -0.59 to 0.29) and was not substantially altered in several sensitivity analyses.

Conclusions: Meta-analysis shows consistent evidence that DBT significantly increased CDR compared with mammography screening; however, there was little difference between DBT and mammography in pooled ICR. This could suggest, but does not demonstrate, some over-detection. Meta-analysis using individual participant data, randomised trials and comparative studies quantifying cumulative detection and ICR over repeat DBT screen-rounds would provide valuable evidence to inform screening programs.

Keywords: Breast cancer; Interval cancer; Mammography; Population screening; Tomosynthesis.

Publication types

Comparative Study
Meta-Analysis
Review

MeSH terms

Breast Neoplasms / diagnosis*
Diagnostic Tests, Routine / methods*
Early Detection of Cancer / methods*
Female
Humans
Mammography / methods*
Prognosis
Prospective Studies