Temporal changes in mammographic breast density and breast cancer risk among women with benign breast disease

Breast Cancer Research

Table 2 Associations between standardized percent mammographic breast density change and breast cancer risk, overall and by BBD histological classification

Standardized MBD change (%, per 2 years)	Case n = 261	Control n = 249	p-value*	Multivariable Model	p-trend†
Standardized MBD change (%, per 2 years)	N (%)	N (%)	p-value*	OR (95 CI)†	p-trend†
BBD overall (N = 510)
Stable/Increase ≥ 0%	65 (24.9)	65 (26.1)	0.47	1.09 (0.68, 1.75)
Minimal decrease less than 5%	151 (57.9)	132 (53.0)		1.00 (Referent)
Decrease more than 5%	45 (17.1)	52 (20.9)		0.64 (0.38, 1.07)
Trend				0.77 (0.57, 1.04)	0.09
Continuous				0.99 (0.95, 1.02)	0.46
Non-proliferative diagnosis (N = 329)
Stable/Increase ≥ 0%	40 (26.5)	43 (24.2)	0.8	1.28 (0.71, 2.30)
Minimal decrease less than 5%	85 (56.3)	100 (56.2)		1.00 (Referent)
Decrease more than 5%	26 (17.2)	35 (19.7)		0.75 (0.39, 1.44)
Trend				0.77 (0.53, 1.11)	0.16
Continuous				0.96 (0.92, 1.01)	0.13
Proliferative diagnosis (N = 181)
Stable/Increase ≥ 0%	25 (22.7)	22 (31.0)	0.14	0.88 (0.37, 2.07)
Minimal decrease less than 5%	66 (60.0)	32 (45.1)		1.00 (Referent)
Decrease more than 5%	19 (17.3)	17 (23.9)		0.49 (0.20, 1.18)
Trend				0.75 (0.44, 1.29)	0.30
Continuous				1.03 (0.96, 1.09)	0.41

No interactions of follow-up time or BMI with percent density change (standard) after adjusting for matching factors, BMI, and baseline density (quartiles) were observed
Standardized change in percent mammographic density (i.e., percent density change per two years) was calculated using percent density change from baseline to follow-up mammogram and divided by time intervals between two mammogram exams (percent density at follow-up mammogram − percent density at baseline mammogram)/(calendar years between baseline mammogram and follow-up mammogram/2)
Missing values in breast cancer risk factors were imputed using multiple imputations. Result estimates were calculated in five imputed datasets, separately, and averaged using statistical procedures
BBD benign breast disease, BMI body mass index, CI confidence interval, OR odds ratio
*p values were calculated from Chi-Square tests
^†Point estimates and 95% confidence interval were determined from multivariable unconditional logistic regression models adjusting for continuous age at BBD and follow-up time between BBD and breast cancer diagnosis, BBD calendar year, baseline percent density (quartiles), and BMI

ISSN: 1465-542X