This retrospective,
single-centre,
lesion-based study was approved by the IRB.
277 cancers,
histologically confirmed,
were reviewed from October 2011 to September 2016.
Out of them,
30 pure intraductal cancers were excluded.
Our final sample consisted on 247 invasive cancers.
All of them were previously studied with mammography,
tomosynthesis ans US.
Imaging evaluation
Both DM and DBT studies were performed using a Siemens Mammomat Inspiration unit (Siemens Medical Solutions,
Erlangen,
Germany).
Standard 45 mediolateral oblique (MLO) and craniocaudal (CC) DM views were acquired.
Given that our DBT system uses a wide angle (50) and requires around 25 s for the acquisition,
we only performed one single view to save radiation dose and time.
Additional CC view was performed if the index lesion was better depicted on CC views,
which occurred in approximately 5% of patients.
US examinations were performed using a MyLab60 unit (Esaote,
Genoa,
Italy) with a multi-frequency (5–13 MHz) linear array transducer.
All US exams were performed after reading only the DM,
without the information of DBT.
Second look US were performed in cases with initially normal US but positive DBT.
Study design
One expert breast radiologist,
blindly reviewed all the mammographic (MX),
ultrasound (US) and tomosynthesis (DBT) examinations.
Initially the radiologist reviewed mammography and US,
and 4 weeks later reviewed again mammography and DBT.
One expert breast radiologist,
with more than 17 years of experience (AE),
retrospectively evaluated all 247 breast cancers blinded to the final diagnosis.
Initially,
MX were evaluated and classified according to the BIRADS categories (1–5).
Four weeks later,
the same reader then evaluated and classified the combination of DM and DBT into BI-RADS categories (1–5),
as well as the combination DM and US.
For the analysis,
those cases classified as BI-RADS categories 3,
4,
and 5 were considered positive,
whereas categories 1 and 2 were considered negative.
Additional cancers were those detected by DBT and/or US but not visible on DM.
A McNemar test (IBM SPSS 21.0 ) was used to compare the sensitivity of the combination of different techniques in the three immunohistochemical patterns.