Skip to main content

Table 6 Logistic regression models and discriminatory capacity before and after inclusion of the breast density estimates to the Gail risk factors

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

Risk factor

Standard risk factors only

Gail risk factors plus area percent density

Gail risk factors plus area and volumetric density with feature selection

 

Odds ratio

P value

AUC

Odds ratio

P value

AUC

Odds ratio

P value

AUC

Demographics

  

0.64 (0.58–0.71)

  

0.68a (0.62–0.74)

  

0.77a (0.71–0.82)

 Age

0.96

0. 59

1.02

0.73

1.06

0.41

(0.84–1.10)

(0.90–1.17)

(0.92–1.23)

 Age at menarche

0.56

0.01

0.54

0.008

0.64

0.06

(0.36–0.89)

(0.34–0.85)

(0.40–1.02)

 Number of benign biopsies

1.08

0.67

1.02

0.94

0.97

0.90

(0.75–1.55)

(0.70–1.47)

(0.65–1.45)

 First-degree family history

0.91

0.72

0.83

0.49

0.72

0.26

(0.54–1.52)

(0.49–1.40)

(0.42–1.26)

Parity

   

 

 Nulliparous

1

1

1

(N/A)

(N/A)

(N/A)

 Prior to age 30 years

2.22

0.01

2.49

0.005

3.33

0.001

(1.18–4.16)

(1.31–4.73)

(1.67–6.64)

 Age 30 years or older

2.01

0.05

2.12

0.04

2.46

0.02

(0.99–4.08)

(1.03–4.35)

(0.–4.08)

Race

      

 Caucasian

1

1

1

(N/A)

(N/A)

(N/A)

 African-American

0.54

0.04

0.58

0.08

0.39

0.005

(0.30–0.97)

(0.32–1.06)

(0.21–0.75)

 Asian

0.45

0.33

0.50

0.40

0.55

0.49

(0.09–2.19)

(0.10–2.52)

(0.10–2.98)

 Other

0.80

0.76

0.90

0.88

1.04

0.96

(0.20–3.27)

(0.21–3.83)

(0.23–4.74)

Continuous density estimates

      

LIBRA: absolute dense area

1.84

<0.001

(1.32–2.54)

LIBRA: area percent density

1.64

<0.001

(1.25–2.14)

Quantra: absolute dense volume

1.67

0.003

(1.19–2.34)

Quantra: volume percent density

  1. Odds ratios, area under the curve (AUC) of the receiver operating characteristic and 95 % confidence intervals for models based on standard risk factors alone, standard risk factors plus area percent density, and a combined model with standard risk factors, and both area and volumetric density measures after feature selection are provided. Dashes indicate features not included in each model
  2. N/A not applicable, LIBRA Laboratory for Individualized Breast Radiodensity Assessment
  3. aindicates that the predictive capacity of a model including density is statistically significant than the baseline model including only Gail risk factors