Abstract
Background
The V measure captures grayscale intensity variation on a mammogram and is positively associated with breast cancer risk, independent of percent mammographic density (PMD), an established marker of breast cancer risk. We examined whether anthropometrics are associated with V, independent of PMD.
Methods
The analysis included 1,700 premenopausal and 1,947 postmenopausal women without breast cancer within the Nurses’ Health Study (NHS) and NHSII. Participants recalled their body fatness at ages 5, 10, and 20 years using a 9-level pictogram (level 1: most lean) and reported weight at age 18, current adult weight, and height. V was estimated by calculating standard deviation of pixels on screening mammograms. Linear mixed models were used to estimate beta coefficients (ß) and 95% confidence intervals (CIs) for the relationships between anthropometric measures and V, adjusting for confounders and PMD.
Results
V and PMD were positively correlated (Spearman r=0.60). Higher average body fatness at ages 5–10 years (Level≥4.5 vs. 1) was significantly associated with lower V in premenopausal (ß=−0.32, 95% CI=−0.48 to −0.16) and postmenopausal (ß=−0.24, 95% CI: −0.37 to −0.10) women, independent of current body mass index (BMI) and PMD. Similar inverse associations were observed with average body fatness at ages 10–20 years and BMI at age 18. Current BMI was inversely associated with V but the associations were largely attenuated after adjustment for PMD. Height was not associated with V.
Conclusions
Our data suggest that early-life body fatness may reflect lifelong impact on breast tissue architecture beyond breast density. However, further studies are needed to confirm the results.
Impact
This study highlights strong inverse associations of early-life adiposity with mammographic image intensity variation.
Keywords: somatotype, body size, body fatness, adiposity, obesity, overweight, BMI, body mass index, breast density, mammogram, mammographic texture, V, parenchymal, fibroglandular tissue
INTRODUCTION
Anthropometrics such as height (1–3) and body fatness (3–11) are associated with breast cancer risk. Height is thought to indicate early-life nutritional status (12) and correlates with timing of puberty and early-life exposure to endogenous growth hormones. A meta-analysis of 159 prospective cohorts estimated a 17% elevated breast cancer risk associated with every 10-cm increase in adult height (2). Higher childhood and adolescent body fatness and young adult body mass index (BMI) are consistently inversely associated with both pre- and postmenopausal breast cancer risk (4–8,13), whereas later adult body fatness is positively associated with postmenopausal breast cancer risk (3,9–11). Although the mechanisms underlying the relationships between anthropometrics and breast cancer risk are unknown, one of the hypothesized mechanisms is that exposure, particularly during childhood and adolescence, may influence development of breast tissue structures and determine breast density. Percent mammographic density (PMD) is a strong breast cancer risk factor, which has been associated with four- to six-fold higher risk of breast cancer (14–16).
Multiple studies of PMD, which refers to the relative amount of dense (fibroglandular) vs. non-dense (adipose) tissue in the breast, have reported associations with early-life and adult anthropometrics (17–22). However, other features within a mammogram may provide additional information beyond PMD. Recently, novel algorithms have been developed to measure the heterogeneity in patterns of mammographic density, the features referred to as ‘texture’. Independent studies from the U.S. (23,24) and the UK (25) have shown that various measures of texture features, ranging from simple distribution of grayscale intensity values to the spatial relationships between intensities on a mammogram, are associated with breast cancer risk, independent of PMD. In the Nurses’ Health Study (NHS) and NHSII, we quantified the V metric, a measure of texture that captures variation in grayscale intensity values on a mammogram, using automated techniques. Using these data, we previously found that higher V, indicating greater intensity variation, was associated with a 41% increased breast cancer risk in premenopausal women and a 21% increased risk in postmenopausal women, after adjustment for PMD (24). These studies have provided evidence that, even among women with similar PMD, texture features such as V can further differentiate and predict women who are at increased risk of developing breast cancer later in life.
However, little is known about the factors associated with the V. Investigating the associations of established breast cancer risk factors with V may provide new insights into the mechanisms through which these risk factors influence breast cancer risk. Using automated techniques, we examined the associations of early-life and adult anthropometrics (childhood and adolescent body fatness, BMI at age 18, current adult BMI, change in BMI since age 18, predicted body fat mass, and adult height) with V among pre- and postmenopausal women in the NHS and the NHSII.
MATERIALS AND METHODS
Study population
The NHS began in 1976 among 121,700 female registered nurses at ages 30–55 years. The NHSII began in 1989 among 116,429 female registered nurses at ages 25–42 years. In both cohorts, baseline and subsequent biennial follow-up questionnaires were used to collect information on participants’ health behaviors, anthropometric and lifestyle factors, reproductive factors, medical histories, and disease diagnosis (26).
This analysis includes participants who served as controls in a nested case-control study of breast cancer within the Nurses’ Health Study (NHS) and NHSII blood and cheek cell collection subcohorts. Details of this nested case-control study have been previously described (27–29). In brief, blood samples were collected from 32,826 NHS participants aged 43–70 years in 1989–1990 and 29,611 NHSII participants aged 32–45 years in 1996–1999. Up to two controls were matched to incident breast cancer cases on age, menopausal status at blood draw and diagnosis, current postmenopausal hormone therapy use, month, time of day, fasting status at time of blood collection, and luteal day (NHSII timed samples only). From the cases and controls, mammograms imaged as close as possible to the date of blood draw (or 1997 for cheek cell collection subcohort) were collected (NHS: 1,379 cases and 2,514 controls; NHSII: 758 cases and 1,832 controls).
For this analysis, we restricted to controls and excluded women with missing information on V or PMD (n=338), anthropometrics (childhood and adolescent body fatness, BMI at age 18 years, current adult BMI, adult height) (n=291), and menopausal status (n=70). A total of 3,647 women (1,700 premenopausal, 1,947 postmenopausal) were included in this analysis. The study protocol was approved by the institutional review boards of the Brigham and Women’s Hospital and Harvard T.H. Chan School of Public Health, and those of participating registries as required.
Exposure assessment
Body fatness at ages 5, 10, and 20 years were recalled in 1988 (NHS) and 1989 (NHSII) using Stunkard’s nine-level pictogram (level 1 to 9: most lean to most overweight) (30). Recalled body fatness using this pictogram by women at ages 71–76 years has shown a good correlation with measured BMI at ages 5–20 (Pearson r = 0.60–0.66) (31). Average body fatness at ages 5 and 10 years and ages 10 and 20 years were used to represent childhood and adolescent body fatness, respectively. Extreme levels (≥4.5) were collapsed into a single category because there were fewer women in those levels. Height and weight at age 18 were reported via the baseline questionnaire (in 1976 for NHS, 1989 for NHSII). Current weight was reported via questionnaires administered prior to but around the time of mammogram. BMI (kg/m2) was calculated by dividing weight (kg) by baseline height (m) squared. Change in BMI since age 18 (kg/m2) was estimated by subtracting BMI at age 18 from current adult BMI. To account for limitations of BMI as a measure of adiposity, we also calculated predicted body fat mass (kg) and percent fat mass on the basis of age, race, height, weight, and waist circumference using the equations developed and validated in the National Health and Nutrition Examination Survey (32). The analyses of predicted body fat mass and percent fat mass were restricted to 1,078 premenopausal and 1,464 postmenopausal women with information on waist circumference. All continuous exposure variables (BMI at age 18, current adult BMI, change in BMI since age 18, predicted body fat mass, and adult height) were categorized based on quintiles.
Mammographic density measurement
Details of mammographic density measurement were described elsewhere (33). Briefly, a Lumysis 85 laser film scanner was used to digitize the craniocaudal views of both breasts for all mammograms in the NHS and for the first two batches of mammograms in the NHSII. The third batch of mammograms in the NHSII was scanned using a VIDAR CAD PRO Advantage scanner (VIDAR Systems Corporation; Herndon, VA) using comparable resolution of 150 dots per inch and 12 bit depth. We measured absolute dense area and total area using the Cumulus software for computer-assisted thresholding (34). PMD was estimated by dividing the dense area by the total area and then averaged that of both breasts. All images were read by a single observer (within person intra-class correlation >0.90). We adjusted for batch variability in the NHSII as previously described (33).
Mammographic texture variation measurement (“the V metric”)
The V metric is an automated measure that captures mammographic image intensity, as described previously (35,36). Briefly, the breast area is detected and eroded (by 25% [‘V75’] and 35% [‘V65’] along a radial direction) to eliminate the proportion of the breast that was not in contact with the compression paddle during the image acquisition, which is an approximation. The standard deviation calculated from pixels within the eroded breast region produces V. Images were digitized with different equipment and were from various time frames. To account for resolution and intensity scale differences, mammograms were normalized before V was calculated (37). Normalization processing involved spatial normalization, feature distribution normalization, and resolution estimation (Supplementary Methods and Materials). Example images of breasts with high vs. low V values given similar PMD are shown in Figure 1.
Statistical analysis
Because mammographic density and texture measures varied by menopausal status, all analyses were stratified by menopausal status at time of mammogram. To account for correlations among the controls within the matched sets, we performed linear mixed models with a compound symmetry correlation structure to estimate beta coefficients (ß) and 95% confidence intervals (CI) for the relationships between anthropometrics and V. Multivariable (MV) models included age, race, personal history of benign breast disease, family history of breast cancer, parity/age at first birth, alcohol use, smoking, and breastfeeding. In postmenopausal women, age at menopause and postmenopausal hormone use were additionally adjusted. For change in BMI since age 18, models additionally adjusted for BMI at age 18. For all analyses, we used covariate information reported via questionnaires assessed prior to but around the time of mammogram. We performed tests for trend using the median of the exposure categories as a continuous variable in the regression models. To evaluate whether the associations between anthropometrics and V are driven by their correlations with other mammographic measures, we compared the models after additional adjustment for PMD, dense area, and non-dense area, separately. For early-life anthropometrics, we also evaluated the role of potential mediators by comparing the models with and without adjustment for BMI at age 18 and current adult BMI. Additional adjustment for age at menarche did not change the results and thus are not shown in this paper. To assess whether the associations vary by the levels of PMD, we stratified analyses by low vs. high PMD (below vs. above the menopausal-specific median PMD) and performed tests for interaction using a Wald test for interaction terms. We also examined the relationships with PMD, absolute dense area, and absolute non-dense area by performing models on square-root transformed density measures, adjusting for V. In primary analyses, we used the data from V75 breast erosion strategy and low resolution images. In sensitivity analyses, we repeated the analyses using an alternate breast erosion strategy (V65) and after restricting to data from high resolution images only (n=1,242 premenopausal, 850 postmenopausal). We also stratified by digitization methods to account for measurement differences in V by these methods.
RESULTS
Participant characteristics
The mean age at mammogram was 45.9 years in 1,700 premenopausal women and 58.0 years in 1,947 postmenopausal women. Most women were White (97%) and parous (89%). On average, women with greater average body fatness at ages 5–10 years (level ≥4.5 vs. 1) were less likely to be non-White (1.4% vs. 4.9%) and have personal history of benign breast disease (14% vs. 22%) and more likely to have higher BMI at age 18 (23.8 vs. 19.6 kg/m2), higher current adult BMI (28.7 vs. 24.2 kg/m2), and younger age at menarche (11.9 vs. 12.8 years) (Table 1). Women with greater average body fatness at ages 5–10 years were also more likely to have lower PMD and absolute dense area and higher absolute non-dense area.
Table 1.
Childhood body fatness (average at ages 5–10 years) | |||||
---|---|---|---|---|---|
Level 1 (N=835) | Level 1.5–2 (N=1063) | Level 2.5–3 (N=851) | Level 3.5–4 (N=565) | Level ≥4.5 (N=333) | |
Mean (SD) or Percentage | |||||
Age at mammogram*, years | 55.0 (9.1) | 51.8 (8.9) | 51.1 (8.6) | 51.7 (8.4) | 51.9 (8.4) |
Non-White, % | 4.9 | 3.2 | 4.4 | 2.4 | 1.4 |
Height, inches | 64.7 (2.4) | 64.8 (2.4) | 64.8 (2.5) | 64.7 (2.4) | 64.7 (2.5) |
BMI at age 18 years, kg/m2 | 19.6 (2.0) | 20.4 (2.2) | 21.5 (2.6) | 22.8 (3.0) | 23.8 (3.2) |
BMI at mammogram#, kg/m2 | 24.2 (4.1) | 25.0 (4.7) | 26.3 (5.3) | 27.9 (6.2) | 28.7 (6.3) |
Predicted body fat mass at mammogram#, kg | 24.3 (7.0) | 25.7 (7.9) | 27.4 (9.0) | 30.4 (10.4) | 31.1 (9.7) |
Predicted percent body fat mass at mammogram# | 36.8 (3.9) | 37.5 (4.3) | 38.3 (4.7) | 40.0 (5.6) | 40.4 (5.2) |
Age at menarche, years | 12.8 (1.4) | 12.5 (1.4) | 12.4 (1.4) | 12.2 (1.4) | 11.9 (1.3) |
Parous#, % | 88.5 | 88.8 | 90.5 | 88.1 | 85.9 |
Among parous women# | |||||
- Paritya# | 2.8 (1.3) | 2.8 (1.3) | 2.7 (1.3) | 2.8 (1.3) | 2.8 (1.3) |
- Age at first birtha#, years | 25.5 (3.6) | 25.5 (3.9) | 25.7 (4.0) | 25.4 (3.8) | 25.5 (4.0) |
- Total breastfeedinga#, % | |||||
<1 month | 35.7 | 32.3 | 31.5 | 32.8 | 28.3 |
1–6 months | 22.0 | 18.7 | 20.0 | 17.8 | 22.0 |
7–12 months | 14.3 | 13.4 | 12.6 | 13.9 | 18.6 |
≥13 months | 28.0 | 35.6 | 35.8 | 35.5 | 31.0 |
Postmenopausal#, % | 53.2 | 54.7 | 52.5 | 53.6 | 53.8 |
Among postmenopausal women# | |||||
- Postmenopausal hormone useb# | |||||
Never users | 33.7 | 31.4 | 38.8 | 34.9 | 41.3 |
Former users | 19.8 | 18.8 | 18.1 | 19.9 | 19.8 |
Current users | 46.5 | 49.8 | 43.0 | 45.2 | 38.9 |
- Age at menopauseb#, years | 45.2 (6.6) | 44.9 (7.5) | 46.1 (6.8) | 45.2 (7.2) | 45.4 (7.4) |
Alcohol intake#, g/d | |||||
0 | 32.2 | 33.0 | 35.8 | 35.4 | 35.4 |
0.1–4.9 | 33.0 | 34.7 | 32.4 | 30.2 | 34.6 |
5.0–14.9 | 19.2 | 19.7 | 19.8 | 20.8 | 16.3 |
≥15.0 | 8.4 | 7.0 | 6.6 | 7.7 | 8.3 |
Missing | 7.3 | 5.6 | 5.5 | 5.9 | 5.3 |
Smoking#, % | |||||
Never | 56.9 | 60.8 | 57.5 | 56.1 | 51.1 |
Former | 33.5 | 31.6 | 34.1 | 35.5 | 37.3 |
Current | 9.6 | 7.5 | 8.4 | 8.4 | 11.6 |
Personal history of benign breast disease#, % | 21.9 | 21.2 | 19.1 | 19.4 | 14.0 |
Family history of breast cancer#, % | 11.0 | 11.1 | 9.4 | 11.3 | 7.4 |
Percent mammographic density# | 35.9 (19.4) | 35.8 (19.2) | 31.0 (19.6) | 27.8 (19.1) | 23.6 (18.7) |
Absolute dense area#, cm2 | 43.3 (26.8) | 43.7 (27.9) | 40.1 (27.6) | 38.8 (28.5) | 32.9 (26.6) |
Absolute nondense area#, cm2 | 91.5 (61.8) | 93.8 (64.9) | 111.8 (78.8) | 124.8 (77.8) | 135.7 (78.4) |
V measures | 0.12 (0.91) | 0.09 (0.95) | −0.12 (0.98) | −0.27 (1.00) | −0.58 (0.90) |
Values are means(SD) or percentages, standardized to the age distribution of the study population, except for age. Values of polytomous variables may not sum to 100% due to rounding.
Value is not age adjusted
At the time of mammogram
Among parous women only
Among postmenopausal women only
Note: V measures = V75 erosion, low resolution; V measures can take negative values because of the mapping.
Compared with postmenopausal women, premenopausal women had higher V (mean=0.15 vs. −0.26), PMD (mean=39.5% vs. 26.0%), and absolute dense area (mean=45.7 vs. 37.4 cm2) and lower absolute non-dense area (mean=80.6 vs. 128.3 cm2). Among both pre- and postmenopausal women, V was positively correlated with PMD (Spearman r=0.49 in premenopausal, 0.63 in postmenopausal) and absolute dense area (Spearman r=0.39 in premenopausal, 0.48 in postmenopausal) but negatively correlated with absolute non-dense area (Spearman r= −0.38 in premenopausal, −0.53 in postmenopausal) within a mammogram.
Early-life body fatness
Greater childhood body fatness (level ≥4.5 vs. 1) was statistically significantly associated with lower V in multivariable models among both premenopausal (β= −0.74, 95% CI= −0.92 to −0.57, p-trend<0.01) and postmenopausal women (β= −0.59, 95% CI= −0.75 to −0.43, p-trend<0.01) (Table 2). After additional adjustment for current BMI and PMD, the associations were substantially attenuated but remained statistically significant (premenopausal: β= −0.32, 95% CI= −0.48 to −0.16; postmenopausal: β= −0.24, 95% CI= −0.37 to −0.10; p-trends≤0.01). Similar inverse associations were observed for adolescent body fatness (level ≥4.5 vs. 1: β= −0.41, 95% CI= −0.64 to −0.18 in premenopausal; β= −0.34, 95% CI= −0.50 to −0.18 in postmenopausal) and BMI at age 18 years (≥23 vs. <19 kg/m2: β= −0.30, 95% CI= −0.45 to −0.16 in premenopausal; β= −0.25, 95% CI= −0.37 to −0.14 in postmenopausal) in multivariable models including current BMI and PMD (all p-trend<0.01). The associations between early-life body fatness and V were only slightly attenuated when absolute dense or non-dense area was included in the models instead of PMD (all p-trend<0.01).
Table 2.
Beta coefficients (95% confidence intervals) | ||||||||
---|---|---|---|---|---|---|---|---|
Premenopausal women (N=1,700) |
Postmenopausal women (N=1,947) |
|||||||
N | MV-adjusted a | MV + Current BMI b | MV + Current BMI + PMD c | N | MV-adjusted a | MV + Current BMI b | MV + Current BMI + PMD c | |
Childhood body fatness (average at ages 5–10 years) | ||||||||
Level 1 | 303 | 0 (Ref) | 0 (Ref) | 0 (Ref) | 532 | 0 (Ref) | 0 (Ref) | 0 (Ref) |
Level 1.5–2 | 507 | −0.09 (−0.22, 0.04) | −0.03 (−0.16, 0.10) | −0.04 (−0.16, 0.07) | 556 | 0.03 (−0.08, 0.14) | 0.05 (−0.06, 0.15) | 0.004 (−0.09, 0.10) |
Level 2.5–3 | 454 | −0.20 (−0.33, −0.06) | −0.07 (−0.20, 0.06) | −0.04 (−0.16, 0.08) | 397 | −0.25 (−0.37, −0.13) | −0.17 (−0.28, −0.05) | −0.11 (−0.21, −0.01) |
Level 3.5–4 | 275 | −0.45 (−0.60, −0.30) | −0.23 (−0.38, −0.08) | −0.12 (−0.26, 0.01) | 290 | −0.31 (−0.45, −0.18) | −0.17 (−0.30, −0.04) | −0.11 (−0.22, 0.001) |
Level ≥4.5 | 161 | −0.74 (−0.92, −0.57) | −0.46 (−0.63, −0.28) | −0.32 (−0.48, −0.16) | 172 | −0.59 (−0.75, −0.43) | −0.43 (−0.58, −0.27) | −0.24 (−0.37, −0.10) |
P-trend | <0.01 | <0.01 | <0.01 | <0.01 | <0.01 | <0.01 | ||
Adolescent body fatness (average at ages 10–20 years) | ||||||||
Level 1 | 72 | 0 (Ref) | 0 (Ref) | 0 (Ref) | 173 | 0 (Ref) | 0 (Ref) | 0 (Ref) |
Level 1.5–2 | 427 | −0.06 (−0.29, 0.17) | −0.03 (−0.25, 0.19) | −0.05 (−0.25, 0.15) | 585 | −0.09 (−0.25, 0.07) | −0.06 (−0.21, 0.09) | −0.10 (−0.23, 0.03) |
Level 2.5–3 | 608 | −0.20 (−0.42, 0.02) | −0.09 (−0.30, 0.13) | −0.07 (−0.27, 0.13) | 614 | −0.20 (−0.35, −0.04) | −0.11 (−0.26, 0.04) | −0.12 (−0.25, 0.01) |
Level 3.5–4 | 405 | −0.49 (−0.72, −0.27) | −0.28 (−0.51, −0.06) | −0.19 (−0.39, 0.02) | 358 | −0.43 (−0.59, −0.26) | −0.29 (−0.46, −0.13) | −0.25 (−0.39, −0.11) |
Level ≥4.5 | 188 | −0.95 (−1.20, −0.70) | −0.58 (−0.83, −0.33) | −0.41 (−0.64, −0.18) | 217 | −0.79 (−0.97, −0.60) | −0.53 (−0.71, −0.35) | −0.34 (−0.50, −0.18) |
P-trend | <0.01 | <0.01 | <0.01 | <0.01 | <0.01 | <0.01 | ||
BMI at age 18 years, kg/m2 | ||||||||
<19.0 | 342 | 0 (Ref) | 0 (Ref) | 0 (Ref) | 382 | 0 (Ref) | 0 (Ref) | 0 (Ref) |
19.0–19.9 | 308 | −0.15 (−0.29, −0.01) | −0.10 (−0.23, 0.04) | −0.09 (−0.21, 0.04) | 313 | −0.12 (−0.25, 0.02) | −0.08 (−0.21, 0.06) | −0.13 (−0.25, −0.02) |
20.0–20.9 | 315 | −0.14 (−0.28, −0.002) | −0.03 (−0.16, 0.11) | 0.004 (−0.12, 0.13) | 368 | −0.17 (−0.30, −0.03) | −0.08 (−0.21, 0.04) | −0.10 (−0.21, 0.01) |
21.0–22.9 | 411 | −0.36 (−0.49, −0.24) | −0.19 (−0.32, −0.06) | −0.12 (−0.24, 0.003) | 478 | −0.31 (−0.43, −0.19) | −0.15 (−0.27, −0.03) | −0.14 (−0.25, −0.03) |
≥23.0 | 324 | −0.84 (−0.98, −0.70) | −0.47 (−0.63, −0.32) | −0.30 (−0.45, −0.16) | 406 | −0.70 (−0.82, −0.57) | −0.42 (−0.56, −0.29) | −0.25 (−0.37, −0.14) |
P-trend | <0.01 | <0.01 | <0.01 | <0.01 | <0.01 | <0.01 | ||
Per 1−kg/m2 increase | −0.13 (−0.14, −0.11) | −0.07 (−0.09, −0.05) | −0.04 (−0.06, −0.02) | −0.10 (−0.12, −0.09) | −0.06 (−0.08, −0.04) | −0.03 (−0.05, −0.02) |
MV-adjusted model includes age (years, continuous), race (white/nonwhite), personal history of benign breast disease (yes/no), family history of breast cancer (yes/no), parity/age at first birth (nulliparous, 1–2 births/<25 years, 1–2 births/25–29 years, 1–2 births/≥30 years, >=3 births/<25 years, >=3 births/≥25 years, parous/missing age at first birth), alcohol use (0, 0.1–4.9, 5–14.9, ≥15 g/d, missing), smoking (never, former, current), breastfeeding (<1, 1–6, 7–12, ≥13 months), and, in postmenopausal women, age at menopause (<=44, 45–49, 50–54, ≥55 years, missing) and postmenopausal hormone use (never, former, current).
Additionally includes current BMI (kg/m2, continuous) in the MV-adjusted model.
Additionally includes current BMI (kg/m2, continuous) and percent mammographic density (square root transformed, continuous) in the MV-adjusted model.
Note: V metric = V75 erosion, low resolution
Abbreviations: BMI=body mass index, CI=confidence interval
Current adult body fatness
In pre- and postmenopausal women, current adult BMI (≥30 vs. <21 kg/m2) was significantly inversely associated with V, adjusting for childhood body fatness and potential confounders (premenopausal: β= −0.74, 95% CI= −0.89 to −0.59; postmenopausal: β= −0.74, 95% CI= −0.88 to −0.59; all p-trend<0.01) (Table 3). The associations were substantially attenuated after additional adjustment for PMD (premenopausal: β= −0.12, 95% CI= −0.27 to 0.03; postmenopausal: β= −0.08, 95% CI= −0.22 to 0.05); however, the inverse trend remained statistically significant. Similar patterns were observed when adjusted for absolute non-dense area instead of PMD. Adjustment for absolute dense area only slightly attenuated the associations. Similar results were observed with change in BMI since age 18 (≥7 vs. ≤0 kg/m2: β= −0.03, 95% CI= −0.18 to 0.12 in premenopausal; β= −0.04, 95% CI= −0.17 to 0.09 in postmenopausal), predicted body fat mass (≥35 vs. <20 kg: β= −0.05, 95% CI= −0.23 to 0.13 in premenopausal; β= −0.12, 95% CI= −0.26 to 0.03 in postmenopausal), and percent body fat mass (≥43% vs. <35%: β= −0.19, 95% CI= −0.38 to 0.001 in premenopausal; β= −0.10, 95% CI= −0.24 to 0.05 in postmenopausal) in multivariable models including childhood body fatness and PMD.
Table 3.
Beta coefficients (95% confidence intervals) | ||||||||
---|---|---|---|---|---|---|---|---|
Premenopausal women (N=1,700) |
Postmenopausal women (N=1,947) |
|||||||
N | MV-adjusted a | MV + Childhood body fatness b | MV + Childhood body fatness + PMD c | N | MV-adjusted b | MV + Childhood body fatness a | MV + Childhood body fatness + PMD c | |
Current adult BMI, kg/m2 | ||||||||
<21.0 | 285 | 0 (Ref) | 0 (Ref) | 0 (Ref) | 243 | 0 (Ref) | 0 (Ref) | 0 (Ref) |
21.0–22.9 | 351 | 0.08 (−0.06, 0.22) | 0.08 (−0.05, 0.22) | 0.21 (0.09, 0.34) | 335 | −0.05 (−0.20, 0.09) | −0.03 (−0.18, 0.12) | 0.11 (−0.02, 0.24) |
23.0–24.9 | 329 | 0.03 (−0.11, 0.17) | 0.07 (−0.07, 0.21) | 0.30 (0.17, 0.43) | 390 | −0.20 (−0.34, −0.06) | −0.17 (−0.31, −0.02) | 0.08 (−0.05, 0.20) |
25.0–29.9 | 429 | −0.29 (−0.43, −0.16) | −0.25 (−0.38, −0.11) | 0.14 (0.01, 0.27) | 601 | −0.48 (−0.61, −0.34) | −0.43 (−0.56, −0.29) | −0.01 (−0.13, 0.11) |
≥30.0 | 306 | −0.85 (−1.00, −0.71) | −0.74 (−0.89, −0.59) | −0.12 (−0.27, 0.03) | 378 | −0.83 (−0.97, −0.68) | −0.74 (−0.88, −0.59) | −0.08 (−0.22, 0.05) |
P-trend | <0.01 | <0.01 | <0.01 | <0.01 | <0.01 | 0.01 | ||
Per 1-kg/m2 increase | −0.07 (−0.08, −0.06) | −0.06 (−0.07, −0.05) | −0.02 (−0.03, −0.01) | −0.07 (−0.08, −0.06) | −0.06 (−0.07, −0.05) | −0.01 (−0.02, −0.003) | ||
Change in BMI since age 18, kg/m2 | ||||||||
≤0 | 0 (Ref) | 0 (Ref) | 0 (Ref) | 206 | 0 (Ref) | 0 (Ref) | 0 (Ref) | |
0.1–2.0 | 193 | −0.10 (−0.26, 0.06) | −0.09 (−0.25, 0.08) | −0.01 (−0.15, 0.14) | 291 | −0.05 (−0.21, 0.11) | −0.05 (−0.21, 0.11) | 0.02 (−0.11, 0.16) |
2.1–5.0 | 326 | −0.07 (−0.22, 0.08) | −0.07 (−0.22, 0.09) | 0.14 (−0.002, 0.28) | 618 | −0.10 (−0.24, 0.05) | −0.11 (−0.25, 0.04) | 0.10 (−0.02, 0.23) |
5.1–7.0 | 560 | −0.28 (−0.45, −0.11) | −0.27 (−0.44, −0.10) | 0.11 (−0.05, 0.27) | 300 | −0.19 (−0.36, −0.03) | −0.20 (−0.36, −0.04) | 0.10 (−0.04, 0.25) |
>7.0 | 239 | −0.46 (−0.62, −0.31) | −0.45 (−0.60, −0.29) | −0.03 (−0.18, 0.12) | 532 | −0.52 (−0.67, −0.37) | −0.52 (−0.67, −0.37) | −0.04 (−0.17, 0.09) |
P-trend | 382 | <0.01 | <0.01 | 0.47 | <0.01 | <0.01 | 0.09 | |
Per 1-kg/m2 increase | −0.05 (−0.06, −0.03) | −0.04 (−0.06, −0.03) | −0.004 (−0.02, 0.01) | −0.05 (−0.06, −0.03) | −0.05 (−0.06, −0.04) | −0.01 (−0.02, 0.001) | ||
Premenopausal women (N=1078)
f |
Postmenopausal women (N=1464)
f |
|||||||
Predicted body fat mass, kg | ||||||||
<20.0 | 272 | 0 (Ref) | 0 (Ref) | 0 (Ref) | 250 | 0 (Ref) | 0 (Ref) | 0 (Ref) |
20.0–24.9 | 331 | 0.15 (0.0001, 0.29) | 0.17 (0.02, 0.31) | 0.32 (0.19, 0.45) | 427 | −0.11 (−0.25, 0.03) | −0.10 (−0.24, 0.05) | 0.05 (−0.07, 0.18) |
25.0–29.9 | 201 | −0.21 (−0.37, −0.04) | −0.15 (−0.31, 0.02) | 0.16 (0.01, 0.32) | 342 | −0.32 (−0.47, −0.17) | −0.29 (−0.44, −0.14) | 0.003 (−0.13, 0.13) |
30.0–34.9 | 115 | −0.39 (−0.59, −0.19) | −0.32 (−0.51, −0.12) | 0.08 (−0.11, 0.26) | 197 | −0.57 (−0.74, −0.40) | −0.52 (−0.69, −0.35) | −0.12 (−0.27, 0.03) |
≥35.0 | 159 | −0.75 (−0.93, −0.57) | −0.64 (−0.82, −0.45) | −0.05 (−0.23, 0.13) | 248 | −0.83 (−0.99, −0.67) | −0.75 (−0.91, −0.58) | −0.12 (−0.26, 0.03) |
P-trend | <0.01 | <0.01 | 0.06 | <0.01 | <0.01 | 0.01 | ||
Per 1-kg increase | −0.03 (−0.04, −0.03) | −0.03 (−0.04, −0.02) | −0.01 (−0.01, 0.0004) | −0.04 (−0.04, −0.03) | −0.03 (−0.04, −0.03) | −0.01 (−0.01, −0.001) | ||
Predicted percent body fat mass | ||||||||
<35.0 | 397 | 0 (Ref) | 0 (Ref) | 0 (Ref) | 299 | 0 (Ref) | 0 (Ref) | 0 (Ref) |
35.0–36.9 | 234 | 0.13 (−0.02, 0.27) | 0.15 (0.01, 0.29) | 0.29 (0.15, 0.42) | 317 | −0.04 (−0.19, 0.10) | −0.02 (−0.17, 0.12) | 0.14 (0.02, 0.26) |
37.0–39.9 | 207 | −0.38 (−0.53, −0.23) | −0.34 (−0.49, −0.19) | 0.03 (−0.12, 0.18) | 363 | −0.25 (−0.39, −0.11) | −0.22 (−0.36, −0.08) | 0.07 (−0.05, 0.20) |
40.0–42.9 | 117 | −0.45 (−0.64, −0.27) | −0.38 (−0.57, −0.20) | −0.01 (−0.18, 0.17) | 233 | −0.57 (−0.73, −0.41) | −0.52 (−0.68, −0.36) | −0.06 (−0.20, 0.08) |
≥43.0 | 123 | −0.92 (−1.10, −0.73) | −0.80 (−0.99, −0.61) | −0.19 (−0.38, 0.001) | 252 | −0.83 (−0.99, −0.67) | −0.75 (−0.91, −0.59) | −0.10 (−0.24, 0.05) |
P-trend | <0.01 | <0.01 | 0.02 | <0.01 | <0.01 | 0.01 | ||
Per 1% increase | −0.07 (−0.08, −0.06) | −0.06 (−0.07, −0.05) | −0.02 (−0.03, −0.002) | −0.07 (−0.08, −0.06) | −0.06 (−0.07, −0.05) | −0.01 (−0.02, −0.003) |
MV-adjusted model includes age (years, continuous), race (white/nonwhite), personal history of benign breast disease (yes/no), family history of breast cancer (yes/no), parity/age at first birth (nulliparous, 1–2 births/<25 years, 1–2 births/25–29 years, 1–2 births/≥30 years, >=3 births/<25 years, >=3 births/≥25 years, parous/missing age at first birth), alcohol use (0, 0.1–4.9, 5–14.9, ≥15 g/d, missing), smoking (never, former, current), breastfeeding (<1, 1–6, 7–12, ≥13 months), and, in postmenopausal women, age at menopause (<=44, 45–49, 50–54, ≥55 years, missing) and postmenopausal hormone use (never, former, current). Models for change in BMI since age 18 additionally include BMI at age 18 (kg/m2).
Additionally includes childhood body fatness (level 1–4.5+, continuous) in the MV-adjusted model.
Additionally includes childhood body fatness (level 1–4.5+, continuous) and percent mammographic density (square root transformed, continuous) in the MV-adjusted model.
Predicted body fat mass and percent fat mass at mammogram were calculated using the formula developed in the NHANES on the basis of age, race, height, weight, and waist circumference. Because 622 premenopausal and 483 postmenopausal women were missing information on waist circumference, 1,078 premenopausal and 1,464 postmenopausal women were included in these analyses.
Note: V measures = V75 erosion, low resolution
Adult height
In both pre- and postmenopausal women, adult height (≥67.0 vs. <63.0 inches) was not associated with V (MV model: β= −0.03, 95% CI= −0.16 to 0.11, p-trend=0.67 in premenopausal; β=0.08, 95% CI= −0.05 to 0.21, p-trend=0.61 in postmenopausal) (Table 4).
Table 4.
Beta coefficients (95% confidence intervals) | ||||
---|---|---|---|---|
Premenopausal women (N=1,700) |
Postmenopausal women (N=1,947) |
|||
N | MV-adjusted a | N | MV-adjusted a | |
Height, inches | ||||
<63.0 | 294 | 0 (Ref) | 377 | 0 (Ref) |
63.0–64.9 | 518 | 0.001 (−0.13, 0.13) | 601 | 0.20 (0.08, 0.32) |
65.0–65.9 | 232 | −0.06 (−0.21, 0.10) | 278 | 0.10 (−0.04, 0.24) |
66.0–66.9 | 220 | 0.001 (−0.16, 0.16) | 258 | 0.13 (−0.01, 0.28) |
≥67.0 | 436 | −0.03 (−0.16, 0.11) | 433 | 0.08 (−0.05, 0.21) |
P-trend | 0.67 | 0.61 | ||
Per 1-inch increase | −0.005 (−0.03, 0.02) | 0.005 (−0.01, 0.03) |
MV-adjusted model includes age (years, continuous), race (white/nonwhite), personal history of benign breast disease (yes/no), family history of breast cancer (yes/no), parity/age at first birth (nulliparous, 1–2 births/<25 years, 1–2 births/25–29 years, 1–2 births/≥30 years, >=3 births/<25 years, >=3 births/≥25 years, parous/missing age at first birth), alcohol use (0, 0.1–4.9, 5–14.9, ≥15 g/d, missing), smoking (never, former, current), breastfeeding (<1, 1–6, 7–12, ≥13 months), childhood body fatness (1–4.5+ level, continuous) and, in postmenopausal women, age at menopause (<=44, 45–49, 50–54, ≥55 years, missing) and postmenopausal hormone use (never, former, current).
Note: V measures = V75 erosion, low resolution
For body fatness measures, the associations with V tend to vary by levels of PMD (low vs. high; all p-interaction<0.05) (Supplementary Table 1). For all anthropometric exposures evaluated, similar results were found with high-resolution V (Supplementary Table 2) and when we considered alternate breast erosion percentages (V65) for mammographic texture measurement and stratified by digitization method of mammograms.
Associations of early-life and adult anthropometrics with mammographic density measures (PMD, dense area, non-dense area) with and without adjustment for V are presented in Supplementary Table 3–4. After adjustment for V, the associations of body fatness measures with PMD and absolute non-dense area persisted (Supplementary Table 3 for premenopausal, Supplementary Table 4 for postmenopausal). Height was positively associated with PMD adjusting for V in premenopausal women but not in postmenopausal women.
DISCUSSION
To the best of our knowledge, this study is the first to examine the relationships between anthropometrics and intensity variation on a mammogram. Among both pre- and postmenopausal women, we observed a strong inverse association of early-life body fatness with adult mammographic intensity variation measure V after adjustment for current BMI and PMD. Our data suggest that the correlations of V with other mammographic features such as PMD cannot fully explain the association and provide evidence for an independent association. Current adult BMI was also inversely associated with V but the association was largely attenuated after adjustment for PMD. Adult height was not associated with V despite its positive association with premenopausal PMD. We also confirmed inverse associations of childhood and adult body fatness with PMD and positive associations with absolute non-dense area, independent of V.
Our findings of inverse associations between early-life body fatness and V are consistent with those from a British cohort of premenopausal women (18) that examined the associations with the Wolfe grade (38), a qualitative classification of parenchymal patterns that considers both breast density and some texture features (23). The strong inverse associations of early-life body fatness with both PMD and V shown in the present study also mirror the similar inverse association of early-life body fatness with breast cancer risk that has been consistently reported in previous studies (4–8). Our data support the notion that early-life body fatness may influence the lifelong risk of breast cancer through reducing intensity variation of dense tissue as well as the overall breast density. In a previous report in the NHS and NHSII, we observed that the association between early-life body fatness and breast cancer risk may be partially mediated by PMD (approximately 71%−82% in premenopausal and 26–98% in postmenopausal women) (39). It is possible that the breast density-mediated pathways may explain a greater proportion of the early-life body fatness and breast cancer relationship than what was previously estimated if texture variation (e.g., grayscale intensity variation, spatial arrangement of pixels) in mammographic density were also taken into account. Early-life body fatness is likely to have lifelong effects on breast tissue architecture and composition not mediated by adult exposures, as the associations with V and PMD were independent of adult BMI and other adult risk factors. These data are consistent with the hypothesis that the early-life period is a window of susceptibility to breast cancer during which breast tissues are particularly susceptible to stimuli. As shown in our previous report (40), early-life body fatness is also associated with lower adult levels of circulating insulin-like growth factor (IGF)-I, a known breast cancer risk factor. The possible lifelong IGF-1 reduction associated with early-life body fatness may represent a potential link between early-life body fatness and adult mammographic features (e.g., PMD, V). Although largely unknown, other links (e.g., tissue-specific hormone receptors, inflammation) may also exist to alter breast development and tissue composition, leading to lifelong changes in breast density and breast cancer risk. Further investigations are needed to elucidate the underlying pathomorphological characteristics and biological mechanisms related to mammographic intensity variation measures.
In contrast to our findings with early-life body fatness, we observed the associations of current adult body fatness with V were largely attenuated after adjustment for both childhood body fatness and PMD, suggesting that the associations were largely driven by correlations with childhood body fatness and PMD. However, the associations of PMD with adult body fatness persisted after adjustment for V and childhood body fatness. Our data suggest that body fatness during adulthood modifies the overall breast density but may not independently modify the intensity variation of dense tissues within the breast. Further investigations are needed to clarify the associations of current body fatness with mammographic intensity variation.
Adult height is believed to indicate early-life exposure to growth hormones that may contribute to increased breast density at young ages (21,22). However, in the present analysis, height was not associated with V, independent of PMD, suggesting that V may not provide additional information on biological pathways of height.
We acknowledge this study has limitations. All anthropometric measures were self-reported and thus measurement error is likely to occur. However, early-life body fatness data have shown good correlations with measured data in a validation study (31). All anthropometric data were also collected prior to the mammogram and thus any resulting measurement error is likely to be non-differential with respect to mammographic features. For mammographic density assessment, we used semi-automated Cumulus, an operator-assisted thresholding method, and thus the measurements are prone to intra- and inter-reader variability. However, many studies have consistently demonstrated that Cumulus produces a measure that is robust and a strong predictor of breast cancer risk (41). Digitized mammograms came from different platforms and time periods. These differences may also induce measurement variations in the V metric. However, given that these variables are unlikely to be correlated with the assessment of exposures (self-report of early-life and adult anthropometrics), it is unlikely that these variables confound the main associations but could still result in non-differential measurement error that bias the results towards the null. Further, V indicates image intensity variation on mammograms but not necessarily the spatial arrangement of density and thus future studies are needed to further investigate the associations with texture variations using various other measures including those that capture spatial arrangement of dense tissue. Lastly, our study population was mostly white and thus our results may not be generalizable to other populations, particularly Asian women with dense breasts.
Despite these limitations, this study has important strengths. This is the first study to examine the associations of early-life and adult anthropometrics with texture variation on a mammogram. Using automated techniques and quantitative assessment of V, we reduced measurement error and increased reliability of this measure. Using data from different imaging resolutions and different cutpoints for image processing in sensitivity analyses, we observed similar results, which further supports the robustness of our results. We also adjusted for related mammographic density measures including PMD in the regression models to evaluate the associations with V independent of these correlated features. Additionally, we had a large sample size and were able to carefully adjust for a range of potential confounders.
In summary, we observed a strong inverse association of early-life body fatness with adult mammographic texture measure V, independent of current BMI and PMD. Current body fatness was suggestively associated with lower V, independent of childhood body fatness and PMD. Adult height was not associated with V. These findings suggest that early-life body fatness may have lifelong impact on breast tissue architecture beyond breast density (dense vs. non-dense tissue).
Supplementary Material
ACKNOWLEDGEMENTS
We would like to thank the participants and staff of the Nurses’ Health Study and Nurses’ Health Study II for their valuable contributions as well as the following state cancer registries for their help: AL, AZ, AR, CA, CO, CT, DE, FL, GA, ID, IL, IN, IA, KY, LA, ME, MD, MA, MI, NE, NH, NJ, NY, NC, ND, OH, OK, OR, PA, RI, SC, TN, TX, VA, WA, WY. The authors assume full responsibility for analyses and interpretation of these data.
Financial support: This work was supported by the National Cancer Institute at the National Institutes of Health (R.M.T., grant number CA131332, CA175080; M.S., grant number UM1 CA186107 and P01 CA087969; W.W., grant number UM1 CA176726, J.J.H., grant number U01 CA200464; E.T.W., grant number K01CA188075), Avon Foundation for Women, Susan G. Komen for the Cure®, and Breast Cancer Research Foundation. H.O. was supported by the National Research Foundation of Korea (NRF) grant (2019R1G1A1004227), Korea University Grant (K1808781), and Korea University Research Institute of Health Sciences Grant. K.A.B. was supported by the Dahod Breast Cancer Research Program at Boston University School of Medicine.
Footnotes
Authors declare no conflict of interest.
REFERENCES
- 1.Kabat GC, Heo M, Kamensky V, Miller AB, Rohan TE. Adult height in relation to risk of cancer in a cohort of Canadian women. Int J Cancer 2013;132:1125–32 [DOI] [PubMed] [Google Scholar]
- 2.Zhang B, Shu XO, Delahanty RJ, Zeng C, Michailidou K, Bolla MK, et al. Height and Breast Cancer Risk: Evidence From Prospective Studies and Mendelian Randomization. J Natl Cancer Inst 2015;107 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.van den Brandt PA, Spiegelman D, Yaun SS, Adami HO, Beeson L, Folsom AR, et al. Pooled analysis of prospective cohort studies on height, weight, and breast cancer risk. Am J Epidemiol 2000;152:514–27 [DOI] [PubMed] [Google Scholar]
- 4.Baer HJ, Tworoger SS, Hankinson SE, Willett WC. Body fatness at young ages and risk of breast cancer throughout life. Am J Epidemiol 2010;171:1183–94 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Premenopausal Breast Cancer Collaborative G, Schoemaker MJ, Nichols HB, Wright LB, Brook MN, Jones ME, et al. Association of Body Mass Index and Age With Subsequent Breast Cancer Risk in Premenopausal Women. JAMA Oncol 2018;4:e181771. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Ahlgren M, Melbye M, Wohlfahrt J, Sorensen TI. Growth patterns and the risk of breast cancer in women. N Engl J Med 2004;351:1619–26 [DOI] [PubMed] [Google Scholar]
- 7.Fagherazzi G, Guillas G, Boutron-Ruault MC, Clavel-Chapelon F, Mesrine S. Body shape throughout life and the risk for breast cancer at adulthood in the French E3N cohort. Eur J Cancer Prev 2013;22:29–37 [DOI] [PubMed] [Google Scholar]
- 8.Weiderpass E, Braaten T, Magnusson C, Kumle M, Vainio H, Lund E, et al. A prospective study of body size in different periods of life and risk of premenopausal breast cancer. Cancer Epidemiol Biomarkers Prev 2004;13:1121–7 [PubMed] [Google Scholar]
- 9.Gaudet MM, Carter BD, Patel AV, Teras LR, Jacobs EJ, Gapstur SM. Waist circumference, body mass index, and postmenopausal breast cancer incidence in the Cancer Prevention Study-II Nutrition Cohort. Cancer Causes Control 2014;25:737–45 [DOI] [PubMed] [Google Scholar]
- 10.Rohan TE, Heo M, Choi L, Datta M, Freudenheim JL, Kamensky V, et al. Body fat and breast cancer risk in postmenopausal women: a longitudinal study. J Cancer Epidemiol 2013;2013:754815. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Neuhouser ML, Aragaki AK, Prentice RL, Manson JE, Chlebowski R, Carty CL, et al. Overweight, Obesity, and Postmenopausal Invasive Breast Cancer Risk: A Secondary Analysis of the Women’s Health Initiative Randomized Clinical Trials. JAMA Oncol 2015;1:611–21 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Schroeder DG, Martorell R, Rivera JA, Ruel MT, Habicht JP. Age differences in the impact of nutritional supplementation on growth. J Nutr 1995;125:1051S–9S [DOI] [PubMed] [Google Scholar]
- 13.Warner ET, Hu R, Collins LC, Beck AH, Schnitt S, Rosner B, et al. Height and Body Size in Childhood, Adolescence, and Young Adulthood and Breast Cancer Risk According to Molecular Subtype in the Nurses’ Health Studies. Cancer Prev Res (Phila) 2016;9:732–8 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Saftlas AF, Hoover RN, Brinton LA, Szklo M, Olson DR, Salane M, et al. Mammographic densities and risk of breast cancer. Cancer 1991;67:2833–8 [DOI] [PubMed] [Google Scholar]
- 15.Boyd NF, Byng JW, Jong RA, Fishell EK, Little LE, Miller AB, et al. Quantitative classification of mammographic densities and breast cancer risk: results from the Canadian National Breast Screening Study. J Natl Cancer Inst 1995;87:670–5 [DOI] [PubMed] [Google Scholar]
- 16.Byrne C, Schairer C, Wolfe J, Parekh N, Salane M, Brinton LA, et al. Mammographic features and breast cancer risk: effects with time, age, and menopause status. J Natl Cancer Inst 1995;87:1622–9 [DOI] [PubMed] [Google Scholar]
- 17.Jeffreys M, Warren R, Gunnell D, McCarron P, Smith GD. Life course breast cancer risk factors and adult breast density (United Kingdom). Cancer Causes Control 2004;15:947–55 [DOI] [PubMed] [Google Scholar]
- 18.McCormack VA, dos Santos Silva I, De Stavola BL, Perry N, Vinnicombe S, Swerdlow AJ, et al. Life-course body size and perimenopausal mammographic parenchymal patterns in the MRC 1946 British birth cohort. Br J Cancer 2003;89:852–9 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Sellers TA, Vachon CM, Pankratz VS, Janney CA, Fredericksen Z, Brandt KR, et al. Association of childhood and adolescent anthropometric factors, physical activity, and diet with adult mammographic breast density. Am J Epidemiol 2007;166:456–64 [DOI] [PubMed] [Google Scholar]
- 20.Andersen ZJ, Baker JL, Bihrmann K, Vejborg I, Sorensen TI, Lynge E. Birth weight, childhood body mass index, and height in relation to mammographic density and breast cancer: a register-based cohort study. Breast Cancer Res 2014;16:R4. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Sala E, Warren R, McCann J, Duffy S, Luben R, Day N. High-risk mammographic parenchymal patterns and anthropometric measures: a case-control study. Br J Cancer 1999;81:1257–61 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Dorgan JF, Klifa C, Shepherd JA, Egleston BL, Kwiterovich PO, Himes JH, et al. Height, adiposity and body fat distribution and breast density in young women. Breast Cancer Res 2012;14:R107. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Manduca A, Carston MJ, Heine JJ, Scott CG, Pankratz VS, Brandt KR, et al. Texture features from mammographic images and risk of breast cancer. Cancer Epidemiol Biomarkers Prev 2009;18:837–45 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Rice M, Bertrand KA, Heine JJ, Rosner B, Tamimi RM. Abstract 2595: Texture variation on a mammogram and risk of breast cancer. 2016; New Orleans, LA: Cancer Research; p 2595. [Google Scholar]
- 25.Wang C, Brentnall AR, Cuzick J, Harkness EF, Evans DG, Astley S. A novel and fully automated mammographic texture analysis for risk prediction: results from two case-control studies. Breast Cancer Res 2017;19:114. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Bao Y, Bertoia ML, Lenart EB, Stampfer MJ, Willett WC, Speizer FE, et al. Origin, Methods, and Evolution of the Three Nurses’ Health Studies. Am J Public Health 2016;106:1573–81 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Schernhammer ES, Holly JM, Pollak MN, Hankinson SE. Circulating levels of insulin-like growth factors, their binding proteins, and breast cancer risk. Cancer Epidemiol Biomarkers Prev 2005;14:699–704 [DOI] [PubMed] [Google Scholar]
- 28.Tworoger SS, Sluss P, Hankinson SE. Association between plasma prolactin concentrations and risk of breast cancer among predominately premenopausal women. Cancer Res 2006;66:2476–82 [DOI] [PubMed] [Google Scholar]
- 29.Hankinson SE, Willett WC, Colditz GA, Hunter DJ, Michaud DS, Deroo B, et al. Circulating concentrations of insulin-like growth factor-I and risk of breast cancer. Lancet 1998;351:1393–6 [DOI] [PubMed] [Google Scholar]
- 30.Stunkard AJ, Sorensen T, Schulsinger F. Use of the Danish Adoption Register for the study of obesity and thinness. Res Publ Assoc Res Nerv Ment Dis 1983;60:115–20 [PubMed] [Google Scholar]
- 31.Must A, Willett WC, Dietz WH. Remote recall of childhood height, weight, and body build by elderly subjects. Am J Epidemiol 1993;138:56–64 [DOI] [PubMed] [Google Scholar]
- 32.Lee DH, Keum N, Hu FB, Orav EJ, Rimm EB, Sun Q, et al. Development and validation of anthropometric prediction equations for lean body mass, fat mass and percent fat in adults using the National Health and Nutrition Examination Survey (NHANES) 1999–2006. Br J Nutr 2017;118:858–66 [DOI] [PubMed] [Google Scholar]
- 33.Rice MS, Tworoger SS, Bertrand KA, Hankinson SE, Rosner BA, Feeney YB, et al. Immunoassay and Nb2 lymphoma bioassay prolactin levels and mammographic density in premenopausal and postmenopausal women the Nurses’ Health Studies. Breast Cancer Res Treat 2015;149:245–53 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Boyd NF, Stone J, Martin LJ, Jong R, Fishell E, Yaffe M, et al. The association of breast mitogens with mammographic densities. Br J Cancer 2002;87:876–82 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Heine JJ, Fowler EE, Flowers CI. Full field digital mammography and breast density: comparison of calibrated and noncalibrated measurements. Acad Radiol 2011;18:1430–6 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.Heine JJ, Scott CG, Sellers TA, Brandt KR, Serie DJ, Wu FF, et al. A novel automated mammographic density measure and breast cancer risk. J Natl Cancer Inst 2012;104:1028–37 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37.Warner ET, Rice M, Zeleznik O, Fowler EE, Murthy D, Vachon C, et al. Automated percent mammographic density, texture variation on a mammogram, and risk of breast cancer (Submitted). 2019. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38.Wolfe JN. Breast patterns as an index of risk for developing breast cancer. AJR Am J Roentgenol 1976;126:1130–7 [DOI] [PubMed] [Google Scholar]
- 39.Rice MS, Bertrand KA, VanderWeele TJ, Rosner BA, Liao X, Adami HO, et al. Mammographic density and breast cancer risk: a mediation analysis. Breast Cancer Res 2016;18:94. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40.Poole EM, Tworoger SS, Hankinson SE, Schernhammer ES, Pollak MN, Baer HJ. Body size in early life and adult levels of insulin-like growth factor 1 and insulin-like growth factor binding protein 3. Am J Epidemiol 2011;174:642–51 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 41.Astley SM, Harkness EF, Sergeant JC, Warwick J, Stavrinos P, Warren R, et al. A comparison of five methods of measuring mammographic density: a case-control study. Breast Cancer Res 2018;20:10. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.