|
PMD reduction
|
Univariate
|
Adjusted
|
---|
|
≤10%
|
>10%
|
OR
|
L95
|
U95
|
Ptrend
|
OR
|
L95
|
U95
|
Ptrend
|
---|
+ Genotyped from whole blood
|
EM
|
101
|
21
|
1.00
|
Reference
|
0.009
|
1.00
|
Reference
|
0.020
|
hetEM/IM
|
127
|
9
|
0.34
|
0.15
|
0.78
| |
0.28
|
0.10
|
0.79
| |
PM
|
19
|
1
|
0.25
|
0.03
|
2.00
| |
0.29
|
0.03
|
2.62
| |
n = 278
| | | | | | | | | | |
GOF
| | | | | | | | | |
0.928
|
+ Genotyped from whole blood
|
+ Baseline PMD ≥10%
| | | | | | | | | | |
EM
|
63
|
21
|
1.00
|
Reference
|
0.010
|
1.00
|
Reference
|
0.021
|
hetEM/IM
|
79
|
9
|
0.34
|
0.15
|
0.80
| |
0.29
|
0.11
|
0.81
| |
PM
|
13
|
1
|
0.23
|
0.03
|
1.87
| |
0.29
|
0.03
|
2.57
| |
n = 186
| | | | | | | | | | |
GOF
| | | | | | | | | |
0.609
|
- Results of tests of association between CYP2D6 metabolizer variants and change in percent mammographic density (PMD) in subjects treated with tamoxifen for at least six months. Mammograms were taken ≤1 year prior to initiation of tamoxifen, up to three years after treatment. The mammogram taken closest to the second year after tamoxifen treatment was selected for follow-up. Odds ratios (ODs) and 95% confidence intervals (CIs) (lower bound of a 95% confidence interval (L95), upper bound of a 95% confidence interval (U95)) were estimated by applying a logistic regression model, using the group with 10% reduction as a reference. Analyses were adjusted for baseline PMD, age at diagnosis (years), and body mass index (kg/m2) at baseline. The Wald test was used to determine the statistical significance of an overall linear trend for the association between CYP2D6 metabolizer status, treated as a semi-continuous variable, and PMD change (Ptrend). All statistical tests were two-sided. Statistical analyses were performed using R (version 2.15.1) [18]. EM, extensive metabolizer; hetEM/IM, heterozygous extensive/intermediate; PM, poor metabolizer; GOF, P value for Hosmer-Lemeshow goodness-of-fit test for logistic regression.