Patients selection
From November 2011 to December 2013,
a total number of 9121 women underwent digital mammography at our institution.
According to our protocol,
additional DBT and US were routinely performed for all patients showing density patterns 2,
3 and 4 according to the American College of Radiology (ACR),
as well as for all patients with lesions detected on digital mammography (using DBT and US as problem solving techniques).
An informed consent was given to all patients.
Study design
We conducted a retrospective study,
following the recommendations of our Institutional Review Board,
selecting an enriched sample of 1042 patients (1041 female and 1 male) who underwent the three imaging techniques: DM+DBT+US.
The selection criteria were: patients with biopsy proven malignancies (84 patients),
patients with biopsy proven benign lesions (258 patients) and patients with normal studies or benign conditions,
with no biopsy but at least one year follow-up without significant changes (700 patients). The informed consent for this retrospective study was waived.
Both DM and DBT studies were obtained with the Siemens Inspiration unit (Siemens Medical Solutions,
Erlangen,
Germany).
DM was performed using both 45 º mediolateral oblique (MLO) and craniocaudal (CC) views.
DBT was routinely performed using a single 45º MLO view,
but additional CC DBT views were used as problem solving technique when necessary.
The wide angle used by DBT (50º) induces a long acquisition time (20 seconds) limiting the routine use of DBT to a single view (usually MLO view).
The radiation dose of DM+DBT for each breast was 3.8 mGy for 45 mm PMMA (polimethilmethacrylate) (1.9 mGy for DM two views and 1.9 mGy for one DBT view),
well within the accepted limits (5mGy) (9).
The US study was performed using a MyLab 60 unit (Esaote,
Genoa,
Italy),
with a multifrequency (5-13MHz) linear array transducer.
One expert radiologist,
with more than 15 years dedicated to breast diagnosis,
evaluated retrospectively all the cases.
For each case,
the reader evaluated DM,
classifying it according to the BI-RADS categories.
Then,
with the information of DM,
the reader evaluated the additional DBT,
and reclassified the case (DM+DBT). This first lecture took about one month.
Three weeks later,
the same reader reviewed again DM,
maintaining the previous categorization,
as well as the additional US studies.
These US studies were evaluated in conjunction with the DM information (DM+US).
Finally,
there were three classifications for comparison: the BI-RADS classification of DM alone,
the BI-RADS classification of DM in conjunction with DBT and the BI-RADS classification of DM plus US.
The cases classified as BI-RADS categories 3,
4 or 5 were considered as positive,
whereas the categories 1 or 2 were considered as negative.
The reader was blinded to the final results.
Statistical Analysis
All the data were recorded using the SPSS software (20.0 version).
The sensitivity and specificity as well as the statistical significance of both were calculated using the PEPI software (4.04 version).
The Areas under the Curve (AUC) of the different combinations of techniques (DM; DM+DBT; DM+US; DM+DBT+US) were calculated and compared with the SPSS software by using a z test.
Statistical significance was established for p<0.05.