We reviewed Changes 1-3 as outlined in the previous section,
and how these changes might impact breast density assessment in our clinic using automated software (VolparaDensity,
Matakina Technology Ltd,
Wellington,
New Zealand).
Change 1 – Using letters rather than numbers for the density categories
The manufacturer of the volumetric software already allow configuration of their software to conform to either the 4th or 5th Edition categories.
An example clinical scorecard,
using letter density categories is shown in Figure 2.
Change 2 – Removal of quantitative quartile ranges of area-based percent density
As described earlier,
the ACR does not expect U.S.
radiologists’ interpretation of BI-RADS to change with removal of the quartile ranges of percent density.
Figure 3 shows the distribution of BI-RADS categories amongst 3,865,070 U.S.
women undergoing screening mammography from 1996 - 2008,
who participate in the Breast Cancer Surveillance Consortium.
The ACR used this figure to demonstrate that despite the addition of the quantitative quartile ranges in 2003 (i.e.
in the BI-RADS 4th Edition),
the distribution across density categories has largely remained unchanged.
Hence,
removal of the quartile ranges is not expected to alter U.S.
radiologists’ assessment of breast density in clinical practice.
The automated software assigns a BI-RADS grade based on the correlation of volumetric breast density with U.S.
radiologist-assigned BI-RADS scores.
Having never used area-based thresholds,
and given that U.S.
radiologists are not expected to change their interpretation of density,
removal of the quartile ranges is not expected to impact density assessment using VolparaDensity software.
Change 3 – Increasing the density category for “breasts containing coalescent areas of fibroglandular tissue that are sufficiently dense to obscure small masses”
The ACR updates emphasized that predominantly fatty breasts may still contain regions in the breast that are sufficiently dense to obscure small masses (i.e.
“focal densities”) and in such cases,
that the density category should be increased to reflect the potential masking risk.
We have noted in our clinical practice that VolparaDensity tends to read high for “focal densities”.
Six mammographic screening cases (2-view or 4-view) were selected,
which had been processed using the automated software.
All six cases comprised largely fatty breasts with “focal densities”,
and the volumetric breast density findings were reviewed for their compliance with the BI-RADS 5th Edition (Figures 4 – 9).
Assessed visually,
all cases would have been categorized as BI-RADS 2 density grade according to a strict interpretation of BI-RADS 4th Edition.
Due to the potential masking risk of the “focal densities”,
these cases should all be categorized as BI-RADS c,
according to the 5th Edition.
Owing to the sheer volume of dense tissue of these “focal densities”,
these cases were categorized as BI-RADS c using either the average breast density or the denser breast,
as determined by the automated volumetric software.
Additional Compliance with BI-RADS 5th Edition – Using the denser breast to make the final assessment of breast density
We noted that the manufacturer of the automated software uses the average volumetric breast density from the left and right breasts to assign a final BI-RADS category.
If breasts differ in density,
the ACR stipulates that the denser breast be used to categorize breast density.
To align better with the BI-RADS 5th Edition,
the manufacturer should be using the density of the maximum breast,
rather than the average.
We received confirmation from the manufacturer that the software could be easily configured to use the denser breast and it was important to assess how this might impact breast density assessment in our clinic.
From our own data,
we compared the same population of women using the average volumetric breast density (i.e.
the current method of VolparaDensity) versus using the volumetric breast density of the denser breast to determine the BI-RADS score.
Tables 1 and 2 show the number of women that would change BI-RADS category or classification from “Non-dense” (i.e.
BI-RADS 1 or 2) to “Dense” (i.e.
BI-RADS 3 or 4).
Using a four-category scale,
using the denser breast would result in less than 10% of women in our population increasing in BI-RADS category.
3.5%,
3.9% and 2.2% of women would be re-classified from 1 to 2,
2 to 3,
or 3 to 4,
respectively,
if the denser breast was used (rather than the average density).
Using a two-category scale,
using the denser breast would result in 3.9% of women being re-classified from “Non-dense” to “Dense”.