Skip to main content

Table 4 Odds ratios per standard deviation increase in the log-transformed LIBRA and Quantra density measures (per-category in the ordinal BI-RADS density measure) adjusted for the Gail risk factors (i.e., age at menarche, parity, family history, age, race and number of prior biopsies), P values, and discriminatory capacity (AUC) of the continuous and categorical BI-RADS breast density estimates

From: Preliminary evaluation of the publicly available Laboratory for Breast Radiodensity Assessment (LIBRA) software tool: comparison of fully automated area and volumetric density measures in a case–control study with digital mammography

Density measure

Odds ratio

P value

AUC

Odds ratios at specific percentiles/Categories

10th

50th

75th

90th

LIBRA: absolute dense area, cm2

2.45

<0.001

0.75a

1

3.83

7.14

10.76

(Continuous)

(1.86–3.24)

(0.70–0.81)

(N/A - reference)

(2.54–5.82)

(3.9–13.17)

(5.18–22.58)

LIBRA: area percent density, PD %

1.64

<0.001

0.68

1

1.85

2.57

3.37

(Continuous)

(1.25–2.14)

(0.62–0.74)

(N/A - reference)

(1.32–2.58)

(1.53–4.28)

(1.73–6.49)

Quantra: absolute dense volume, cm3

2.40

<0.001

0.74a

1

2.95

5.57

8.79

(Continuous)

(1.80–3.21)

(0.68–0.79)

(N/A - reference)

(2.07–4.22)

(3.17–9.85)

(4.3–18.1)

Quantra: volume percent density, VD %

1.68

<0.001

0.68

1

1.96

2.69

3.39

(Continuous)

(1.09–1.72)

(0.62–0.74)

(N/A - reference)

(1.12–2.02)

(1.18–2.82)

(1.22–3.59)

Radiologist’s BI-RADS density (ordinal categories)

1.07 (0.74–1.56)

0.71

0.64 (0.58–0.71)

Predominantly fatty

Scattered fibroglandular densities

Heterogeneously dense

Extremely dense

1

1.07

1.15

1.23

(N/A - reference)

(0.74–1.56)

(0.55–2.43)

(0.41–3.80)

  1. Odds ratios computed at specific corresponding percentiles of the distribution of the LIBRA and Quantra density estimates relative to the 10th percentile and at the specific categories for BI-RADS density are also provided as a reference. Ninety-five percent confidence intervals of the odds ratios are also provided in parenthesis below the point estimate
  2. LIBRA Laboratory for Individualized Breast Radiodensity Assessment, BI-RADS Breast Imaging-Reporting and Data System, AUC area under the curve, N/A not applicable
  3. aindicates that the AUC of a model including the specific density measure has a statistically significant increase in discriminatory capacity over a baseline model including only standard risk factors (AUC = 0.64)