Examination of fully automated mammographic density measures using LIBRA and breast cancer risk in a cohort of 21,000 non-Hispanic white women

Table 3 Hazard ratios per SD of breast density assessments and breast cancer risk, by time since mammogram

Density measure	≤ 2 years		> 2 to 5 years		> 5 to 10 years
Density measure	HR^a	95% CI	HR^a	95% CI	HR^a	95% CI
LIBRA (MLO^b only)
Dense area (DA)	1.34	1.12–1.61	1.42	1.25–1.61	1.40	1.19–1.65
Non-dense area (NDA)	0.92	0.78–1.08	0.87	0.73–1.03	0.79	0.66–0.93
Percent density (PD)	1.37	1.08–1.75	1.45	1.25–1.69	1.54	1.33–1.79
LIBRA (CC^b only)
Dense area (DA)	1.27	1.12–1.43	1.32	1.16–1.50	1.26	1.11–1.42
Non-dense area (NDA)	0.97	0.82–1.13	0.87	0.74–1.04	0.79	0.67–0.93
Percent density (PD)	1.30	1.12–1.52	1.35	1.16–1.57	1.38	1.19–1.60
Cumulus
Dense area (DA)	1.44	1.26–1.63	1.43	1.25–1.62	1.48	1.31–1.68
Non-dense area (NDA)	0.86	0.73–1.01	0.88	0.74–1.05	0.72	0.61–0.86
Percent density (PD)	1.55	1.25–1.91	1.46	1.25–1.71	1.67	1.43–1.96

Number of breast cancers for ≤ 2 years, > 2 to 5 years and > 5 to 10 years were 322, 290 and 315, respectively. LIBRA and Cumulus DA and PD were log-transformed, and LIBRA and Cumulus NDA were untransformed
^aHazard ratios adjusted for age at FFDM (spline), mammogram year (categorical), BMI (spline), parity, first-degree family history, and HRT use within 5 years prior to mammogram date. Cumulus analyses were also adjusted for image batch. HRs are per standard deviation of density based on distribution in full cohort. Meta-analysis was used to combine Hologic and GE results
^bAverage of measures on right and left breasts

ISSN: 1465-542X