Table 2.
Mammographic density and breast cancer risk
| Area-based methods | Volumetric-based methods | ||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Cumulus | ImageJ-based method | Volpara | Quantra | SXA | |||||||||||
| Percent density | Number | ORa | (95% CI) | Number | ORa | (95% CI) | Number | ORa | (95% CI) | Number | ORa | (95% CI) | Number | ORa | (95% CI) |
| Average of CC and MLO readingsb | 1,010 | 1.58 | (1.33, 1.88) | 891 | 1.45 | (1.21, 1.74) | 1,010 | 1.83 | (1.51, 2.21) | - | - | - | - | - | - |
| Only CC readingsb | 1,009 | 1.57 | (1.32, 1.87) | 888 | 1.33 | (1.11, 1.59 | 1,009 | 1.75 | (1.45, 2.10) | 975 | 1.40 | (1.19, 1.66) | 934 | 1.37 | (1.16, 1.63) |
| Only MLO readingsb | 1,008 | 1.56 | (1.31, 1.85) | 882 | 1.42 | (1.19, 1.69) | 1,008 | 1.75 | (1.44, 2.11) | - | - | - | |||
| Multiple imputationc | 1,095 | 1.63 | (1.37, 1.93) | 971 | 1.45 | (1.22, 1.73) | 1,095 | 1.82 | (1.51, 2.20) | 1,057 | 1.40 | (1.19, 1.64) | 1,014 | 1.35 | (1.15, 1.60) |
| Restricted to women aged <80 yrs | 948 | 1.58 | (1.32, 1.88) | 830 | 1.44 | (1.20, 1.72) | 948 | 1.78 | (1.47, 2.15) | 914 | 1.38 | (1.17, 1.63) | 874 | 1.36 | (1.15, 1.62) |
| Restricted to women with data for all quantitative methods | 797 | 1.55 | (1,29, 1.87) | 797 | 1.52 | (1.26, 1.84) | 797 | 1.64 | (1.33, 2.02) | 797 | 1.26 | (1.05, 1.55) | 797 | 1.48 | (1.22, 1.80) |
| Dense area/volume | Number | OR a | (95% CI) | Number | OR a | (95% CI) | Number | OR | (95% CI) | Number | OR | (95% CI) | Number | OR | (95% CI) |
| Average of CC and MLO readingsb | 1,010 | 1.57 | (1.33, 1.84) | 891 | 1.56 | (1.31, 1.84) | 1,010 | 1.46 | (1.25, 1.70) | - | - | - | - | - | - |
| Only CC readingsb | 1,009 | 1.58 | (1.35, 1.86) | 888 | 1.45 | (1.23, 1.71) | 1,009 | 1.47 | (1.26, 1.72) | 975 | 1.04 | (0.88, 1.22) | 934 | 1.26 | (1.08, 1.47) |
| Only MLO readingsb | 1,008 | 1.52 | (1.30, 1.79) | 882 | 1.49 | (1.26, 1.77) | 1,008 | 1.38 | (1.18, 1.61) | - | - | - | |||
| Multiple imputationc | 1,095 | 1.60 | (1.36, 1.87) | 971 | 1.54 | (1.31, 1.82) | 1,095 | 1.43 | (1.23, 1.66) | 1,057 | 1.02 | (0.87, 1.19) | 1,014 | 1.23 | (1.05, 1.43) |
| Restricted to women aged <80 yrs | 948 | 1.57 | (1.33, 1.84) | 830 | 1.54 | (1.31, 1.82) | 948 | 1.49 | (1.27, 1.74) | 914 | 1.06 | (0.89, 1.26) | 874 | 1.27 | (1.08, 1.49) |
| Restricted to women with data for all quantitative methods | 797 | 1.63 | (1.36, 1.95) | 797 | 1.62 | (1.36, 1.94) | 797 | 1.70 | (1.42, 2.04) | 797 | 1.23 | (1.02, 1.49) | 797 | 1.74 | (1.43, 2.11) |
| Non-dense area/volume | Number | OR | (95% CI) | Number | OR | (95% CI) | Number | OR | (95% CI) | Number | OR | (95% CI) | Number | OR | (95% CI) |
| Average of CC and MLO readingsb | 1,010 | 0.69 | (0.57, 0.84) | 891 | 1.13 | (0.93, 1.37) | 1,010 | 0.85 | (0.71, 1.03) | - | - | - | - | - | - |
| Only CC readingsb | 1,009 | 0.74 | (0.61, 0.90) | 888 | 1.15 | (0.95, 1.40) | 1,009 | 0.88 | (0.73, 1.06) | 975 | 0.80 | (0.66, 0.97) | 934 | 0.90 | (0.75, 1.08) |
| Only MLO readingsb | 1,008 | 0.66 | (0.54, 0.80) | 882 | 1.08 | (0.89, 1.30) | 1,008 | 0.84 | (0.69, 1.01) | - | - | - | |||
| Multiple imputationc | 1,095 | 0.67 | (0.56, 0.82) | 971 | 1.09 | (0.90, 1.31) | 1,095 | 0.83 | (0.69, 1.01) | 1,057 | 0.78 | (0.65, 0.94) | 1,014 | 0.87 | (0.73, 1.04) |
| Restricted to women aged <80 yrs | 948 | 0.69 | (0.57, 0.84) | 831 | 1.13 | (0.93, 1.37) | 948 | 0.88 | (0.72, 1.07) | 914 | 0.82 | (0.68, 1.00) | 874 | 0.91 | (0.76, 1.10) |
| Restricted to women with data for all quantitative methods | 797 | 0.93 | (0.74, 1.17) | 797 | 1.12 | (0.91, 1.37) | 797 | 1.10 | (0.88, 1.38) | 797 | 1.03 | (0.83, 1.28) | 797 | 1.01 | (0.81, 1.26) |
This table presents the change in breast cancer risk associated with one standard deviation (SD) increase in percent density, absolute density and absolute non-density associated with each one of the five quantitative methods. aOdds ratios (OR) and 95% CI as estimated by logistic regression models based on standardized values of transformed (log for volume-based methods and square root for area based methods adjusted for age, body mass index (BMI), menopausal status and parity (see Methods). There was no evidence of departure from linearity. bBased on density measurements taken from the unaffected breast for cases and a random breast side (left or right) for controls. Measurements for each of the two views were yielded by Cumulus, the ImageJ-based method and Volpara. Single energy x-ray absorptiometry (SXA) generated measurements only for the CC view. Quantra combined data from both views to produce a single overall measurement for each breast (Figure 1). cMultiple imputation methods used to impute values for women with missing data on age, BMI, menopausal status and/or parity (see Table 1 and Methods). CC, cranio-caudal view; MLO, medio-lateral oblique view; OR, odds ratios (with 95% confidence intervals).