Skip Navigation


American Journal of Epidemiology Advance Access originally published online on April 2, 2008
American Journal of Epidemiology 2008 167(9):1027-1036; doi:10.1093/aje/kwn063
This Article
Right arrow Abstract Freely available
Right arrow FREE Full Text (PDF) Freely available
Right arrow All Versions of this Article:
167/9/1027    most recent
kwn063v1
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Email this article to a friend
Right arrow Related articles in Am. J. Epidemiol.
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Add to My Personal Archive
Right arrow Download to citation manager
Right arrow Search for citing articles in:
ISI Web of Science (1)
Right arrowRequest Permissions
Right arrow Disclaimer
Google Scholar
Right arrow Articles by Kelemen, L. E.
Right arrow Articles by Vachon, C. M.
PubMed
Right arrow PubMed Citation
Right arrow Articles by Kelemen, L. E.
Right arrow Articles by Vachon, C. M.
Social Bookmarking
Add to CiteULike   Add to Connotea   Add to Del.icio.us  
What's this?

American Journal of Epidemiology © The Author 2008. Published by the Johns Hopkins Bloomberg School of Public Health. All rights reserved. For permissions, please e-mail: journals.permissions@oxfordjournals.org.

ORIGINAL CONTRIBUTIONS

Age-specific Trends in Mammographic Density

The Minnesota Breast Cancer Family Study

Linda E. Kelemen1, V. Shane Pankratz1, Thomas A. Sellers2, Kathy R. Brandt3, Alice Wang1, Carol Janney1, Zachary S. Fredericksen1, James R. Cerhan1 and Celine M. Vachon1

1 Department of Health Sciences Research, Mayo Clinic College of Medicine, Rochester, MN
2 Division of Cancer Prevention and Control, H. Lee Moffitt Cancer Center and Research Institute, Tampa, FL
3 Department of Radiology, Mayo Clinic College of Medicine, Rochester, MN

Correspondence to Dr. Celine M. Vachon, Department of Health Sciences Research, Mayo Clinic, 200 First Street SW, Charlton 6-239, Rochester, MN 55905 (e-mail: vachon.celine{at}mayo.edu).

Received for publication December 19, 2006. Accepted for publication July 30, 2007.


   ABSTRACT
TOP
 ABSTRACT
INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 References
 
Mammographic density is a strong risk factor for breast cancer, yet few studies have evaluated density trends, and associated factors, over time. The authors retrieved and digitized mammograms (≥1 per woman) imaged in 1990–2003 to evaluate percent density (PD) in the Minnesota Breast Cancer Family cohort. Multivariable-adjusted, mixed-effects, repeated-measures models incorporating a natural cubic spline provided estimates of nonlinear trends in PD with age and were used to examine association with covariates. Overall, 5,698 mammograms from 1,689 women with covariate information were digitized. In descriptive analyses, the highest median PD was 33.1% (interquartile range, 21.8%; n = 230) among premenopausal women, 31.0% (interquartile range, 23.2%; n = 175) among women who transitioned from pre- to postmenopause, and 18.7% (interquartile range, 22.2%; n = 1,284) among postmenopausal women. On average, premenopausal compared with postmenopausal women had 1.9% (p = 0.001) higher PD. In repeated-measures analyses, greater declines in PD occurred with menopause and among women with higher baseline PD; current postmenopausal hormone use and higher body mass index modified these declines (p interaction < 0.001). No significant modification of the density change with age was seen with parity/age at first birth, age at menarche, oral contraceptive use, family history of breast or ovarian cancer in a first- or second-degree relative, educational level, smoking status, or alcohol intake were observed. These data suggest that menopause, baseline PD, postmenopausal hormone use, and body mass index predict changes in mammographic density trends during adult life.

breast; mammography; postmenopause; premenopause; radiographic image interpretation, computer-assisted; risk factors

Abbreviations: BMI, body mass index; FU1, first follow-up questionnaire; FU2, second follow-up questionnaire; PMH, postmenopausal hormone


   INTRODUCTION
TOP
 ABSTRACT
 INTRODUCTION
MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 References
 
Mammographic density reflects a higher proportion of stromal and epithelial tissue relative to fat and is a strong risk factor for breast cancer. In both Caucasian (1) and non-Caucasian (2, 3) populations, women with mammographic densities covering 75 percent or more of the breast compared with very low density have two- to sixfold increases in risk of breast cancer. This association has been seen consistently across 40 studies (1), including those using both qualitative and quantitative estimates (4–6), and is independent of other risk factors for breast cancer (1, 4, 7).

Most studies of mammographic density and breast cancer have involved a single measure of density taken 1–16 years before diagnosis. However, density is not a static trait. Studies on hormonal interventions and mammographic density have shown that combination (estrogen plus progestin) hormone use is associated with increased density (8, 9) and tamoxifen with decreased density (10, 11). Cross-sectional data have suggested that lower mammographic density is associated with increasing age, with some of the greatest decreases occurring with menopause (4).

Few studies have examined longitudinal trends in mammographic density and factors associated with these trends (12–15). Boyd et al. (12) reported an additional 3 percent decrease in density among women who reached a relatively early natural menopause (median age, 46 years) compared with age-matched controls who remained premenopausal during the same time period. Maskarinec et al. (15) used repeated measures of percent density taken over an average of 5 years but did not find a large effect of menopause on density. In that study (15), significantly slower declines in density over time were seen among women with a higher body mass index (BMI >25 kg/m2), experiencing late age at first birth, of Japanese ethnicity, and taking postmenopausal hormones (PMH).

In the current paper, we add to the limited literature by evaluating density trends among women of mammographic age and by assessing the influence on these trends of menopause, BMI, and other risk factors previously shown to be associated with cross-sectional assessments of mammographic density.


   MATERIALS AND METHODS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
RESULTS
 DISCUSSION
 References
 
Study population
Details of the baseline (16) and first follow-up questionnaire (FU1) (17) phases of the Minnesota Breast Cancer Family study have been described previously. Briefly, a family study of breast cancer was initiated in 1944 and was approved by the University of Minnesota's institutional review board. Breast cancer probands (n = 544) were women ascertained at the Tumor Clinic of the University of Minnesota Hospital between 1944 and 1952. From 1990 to 1996 (FU1), data on 426 (78 percent) families were updated; each proband's first- and second-degree female relatives and spouses of male relatives were contacted, and extensive risk factor data on 6,194 women (93 percent) were collected by telephone interview.

In 2001, a second follow-up questionnaire (FU2) was mailed to the 6,194 women who completed the FU1 survey. We excluded those who were deceased (n = 604, 10 percent), were lost to follow-up (n = 654, 11 percent), refused participation (n = 1,109, 18 percent), and were unable to complete the FU2 questionnaire (n = 84, 1 percent); age-ineligible women (aged <40 years, the age in the United States at which screening mammograms are not yet recommended) also were included in these categories and were excluded. Thus, remaining were 3,743 age-eligible women who completed the FU2 questionnaire, resulting in a response rate of 77 percent of those contacted and competent to complete the survey, and an overall participation rate of 60 percent of those who participated in FU1.

At FU2, we requested participant consent to obtain serial mammograms for all women who reported ever having received a mammogram. Of 3,743 age-eligible women, 2,939 (79 percent) reported having a mammogram and provided authorization; for 136 of them, breast cancer was diagnosed prior to mammogram and they were excluded, leaving 2,803 age-eligible women. We requested and received at least one mammogram for 1,914 women (65 percent of those providing authorization); the remaining 889 (32 percent) provided authorization but mammograms were either unavailable (n = 350) or not retrieved because of time constraints (n = 539). Of the 1,914 age-eligible women, 44 received a breast cancer diagnosis after their mammogram(s) and were included in analyses.

Risk factor information
We obtained covariate data from both the FU1 and FU2 questionnaires: covariates that would not be expected to change during the questionnaire interval such as height, educational level, age at menarche, parity, and family history of breast cancer were derived from the FU1 survey, while time-dependent variables such as age, BMI, menopausal status, reason for menopause, PMH use, and smoking were included from both surveys. Menopausal status was ascertained by the response to the question, Have you had a menstrual period within the last 12 months? If no periods were reported, women were asked the age at which menstrual periods stopped and the reason that periods stopped. Those who reported natural menopause, surgical menopause, or "no known reason why periods stopped" were considered postmenopausal. Lifetime frequency of alcohol intake was asked as a single question in FU1: For most of your life, how often did you usually drink alcohol? The possible responses were daily, weekly, less often than weekly, or never. We therefore treated this variable as a single measure of exposure in statistical models.

Mammographic density estimation
The number of mammograms digitized per woman was based on her age at the time of FU2. For example, all mammograms available prior to age 60 years were digitized; for mammograms performed after age 60 years, the one performed closest to this age and at 3-year intervals from this age was digitized. If only two mammograms were available after age 60 years, both were digitized. We applied this algorithm from observations that the greatest changes in mammographic density occur during the menopausal transition (12), with less change after menopause. When this algorithm was used, retrieved mammograms were imaged between 1990 and 2003. In the present population, natural menopause occurred at a median age of 50 years (interquartile range, 47–52) and is similar to the mean age at menopause observed in other US states (18).

We measured mammographic density by using the cranial-caudal or top-down view from the left breast digitized on a Lumiscan 75 scanner (Lumisys, Sunnyvale, California) with 12-bit grayscale depth. The right breast cranial-caudal image was used if the left breast image was unavailable. The pixel size was 0.130 x 0.130 mm2 for both the 18 x 24 cm2 and 24 x 30 cm2 films. Batch files were created that randomized mammograms within woman by date of image, resulting in the greatest precision of percent density estimates (19). Percent density was determined by using a validated computer-assisted thresholding program (Cumulus software; University of Toronto, Toronto, Canada) (20). All images were read by one trained technician who consistently maintained high reliability (intraclass r > 0.90) for percent density estimates from 5 percent duplicate images included in each batch file.

Statistical analysis
Descriptive data were summarized by using medians and interquartile distributions or counts and percentages for all women combined and were stratified according to menopausal status during the FU1 and FU2 questionnaire interval. To compare percent density distributions by risk factor information, we used the mammogram that corresponded in date closest to (either before or after) the date of questionnaire completion.

We used mixed-effects linear models to estimate the association of age with trends in percent density over time. These models included random intercepts and slopes to capture individual-specific, time-dependent trends in percent density and a spatial power correlation structure to account for varied time intervals between repeated measurements (21). To account for the nonlinear trend, a natural cubic spline was applied within the mixed-effects models to obtain nonparametric estimates of the trend in percent density over the range of observed ages (22, 23). These spline trends are formed by joining cubic polynomials such that their first derivatives are equal at the knots: interior knots at ages 50 and 65 years were chosen to bracket the menopause, while boundary knots at ages 43 and 77 years constrained the trends to be linear beyond ages at which there were relatively little data. The number of knots used to define the spline determined the smoothness of the curve. With more knots, the overall trend with age would have been retained but with more local perturbations about that trend. Selection of different interior knot locations would have had little effect on the resulting smoothed trend given the degree of smoothing induced by selecting only two interior knots. The final smoothed curve represents a cross-sectional, age-specific average based on all of the observed values of percent density, constrained to reflect individual-specific trends present in the data provided by each woman.

In this paper, we first describe age-specific trends within percentile distributions of percent density beginning at age 40 years, obtained by using quantile regression methods. These series of percentile curves can be interpreted as cross-sectional trends in density levels for the women in this study at a given age. To examine factors associated with longitudinal trends in density, we fit two models to the data. The first was a main-effects-only model that included the variables age (represented by three basis variables that generated the natural cubic spline), BMI, menopausal status, reason for menopause, age at menarche, parity/age at first birth, PMH use, oral contraceptive use, family history of breast or ovarian cancer in first- or second-degree relatives, educational level, smoking status, and alcohol intake. The second was a main-effects model augmented by two-way interactions between age and each covariate. The significance of interactions was evaluated by using a likelihood ratio test. We used backward elimination to remove interactions at p ≥ 0.01 and nonsignificant main effects at p ≥ 0.01 that did not comprise retained two-way interaction terms. The variables retained were age, BMI, menopausal status, reason for menopause, parity/age at first birth, PMH use, and the interactions of age and BMI, age and PMH use, and age and reason for menopause.

SAS version 8 (SAS Institute, Inc., Cary, North Carolina, 1999) and S-Plus version 7.05 (Insightful Corporation, Seattle, Washington, 2005) software was used for statistical analyses. Two-sided p values of <0.05 were considered statistically significant.


   RESULTS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
DISCUSSION
 References
 
The 1,914 age-eligible women were compared with the 4,280 remaining women from the original cohort of 6,194 for whom we did not have or obtain a mammogram. The groups were broadly similar. When we used data from FU1, we found that a greater proportion of the 1,914 women were college educated (17 percent vs. 13 percent), parous (91 percent vs. 87 percent), physically active (36 percent vs. 32 percent), alcohol consumers (87 percent vs. 83 percent), heavier (BMI 27.1 kg/m2 vs. 26.2 kg/m2), never smokers (55 percent vs. 53 percent), and a blood relative in one of the original proband families (63 percent vs. 52 percent).

The total number of mammograms digitized was 5,698, representing 1,689 women for whom covariate information was complete. Figure 1 illustrates the distribution of mammograms according to menopausal status across the age range studied (40–90 years). These and the following results are a statistical summary of change over the adult age span and should be interpreted with the caveat that no one woman contributed data over the entire age range. However, women who contributed at least two mammograms during this time constituted 88 percent of the sample (table 1).


Figure 1
View larger version (17K):
[in this window]
[in a new window]
[Download PowerPoint slide]
  		FIGURE 1. Distribution of mammograms by age and menopausal status, Minnesota Breast Cancer Family cohort, 1990–2003.

 

View this table:
[in this window]
[in a new window]

  		TABLE 1. Distribution of number of mammograms per woman, Minnesota Breast Cancer Family cohort, 1990–2003

 
Of the 1,689 women, 230 (14 percent) were premenopausal at both questionnaires, 175 (10 percent) transitioned from premenopause at FU1 to postmenopause at FU2, and 1,284 (76 percent) were postmenopausal at both questionnaires (table 2). Percent density was highest among premenopausal women and lowest among postmenopausal women. The greatest difference in percent density between the earliest and most recent mammogram occurred among women transitioning from pre- to postmenopause and represented a decrease of about 8 percent. Despite the longer time contributed to the study by these women, this was more than a twofold decrease when these women were compared with those who remained premenopausal or postmenopausal.


View this table:
[in this window]
[in a new window]

  		TABLE 2. Descriptive measures* of mammographic density by menopausal status, Minnesota Breast Cancer Family cohort, 1990–2003

 
Table 3 shows the distribution of covariates according to menopausal status. Compared with postmenopausal women, premenopausal women tended to be taller and gained more weight during the questionnaire interval. As expected, a large proportion (60 percent) of women transitioning to menopause initiated or continued PMH use. Within each group of women, the distribution of reproductive-related variables was similar, and differences between groups likely reflected changes in secular trends. For example, a greater proportion of younger (premenopausal and pre- to postmenopausal) compared with older (postmenopausal) women were former oral contraceptive hormone users at both follow-up periods, were college graduates, and had a natural menopause, and fewer had larger families.


View this table:
[in this window]
[in a new window]

  		TABLE 3. Distribution of covariates* of mammographic density between the FU1 and FU2 surveys stratified by menopausal status, Minnesota Breast Cancer Family cohort, 1990–2003

 
The curves in figure 2 describe the unadjusted, age-specific percentile distributions of percent density among women with one or more mammograms. Generally, density was inversely associated with age, with the largest declines observed between the ages of 45 and 60 years during menopause. Declines in percent density were smaller among women in the 5th (lowest) percentile distribution of density, whereas declines were larger among women in the 95th (highest) percentile.


Figure 2
View larger version (27K):
[in this window]
[in a new window]
[Download PowerPoint slide]
  		FIGURE 2. Percentile distributions of age-specific, cross-sectional trends in mammographic density, Minnesota Breast Cancer Family cohort, 1990–2003.

 
In analyses that adjusted for covariates, on average, premenopausal women had 1.9 percent (p = 0.001) higher density than postmenopausal women. Compared with women who gave birth to 1–2 children and whose age at first birth was ≥21 years (e.g., 1–2 children/≥21 years of age), nulliparous women had 2.7 percent (p = 0.01) higher percent density, women with 1–2 children/≤20 years of age had 2.9 percent (p = 0.02) lower percent density, and women with ≥3 children/≤ 20 years of age had 2.8 percent (p = 0.001) lower percent density. Interestingly, women with ≥3 children/≥21 years of age had only 0.9 percent (p = 0.24) lower percent density than the reference group, suggesting that timing of the first birth in addition to number of children might be an important factor when predicting density.

Figure 3 shows the average pattern of change in percent density with increasing age among postmenopausal women from the modifying effects of PMH use adjusted for other covariates (p interaction of age x PMH use < 0.0001). Before the age of approximately 50 years, percent density was higher among postmenopausal women who never used hormones compared with former and current users. After age 50 years, never hormone users had larger declines in percent density. In contrast, current hormone users showed smaller reductions in percent density with age and had higher levels than never and former hormone users for the remaining years.


Figure 3
View larger version (9K):
[in this window]
[in a new window]
[Download PowerPoint slide]
  		FIGURE 3. Longitudinal patterns of mammographic density by postmenopausal hormone (PMH) use and age among postmenopausal women, Minnesota Breast Cancer Family cohort, 1990–2003.

 
Figure 4 shows that higher BMI was associated with lower density among postmenopausal women at any age after adjusting for covariates, and this finding was observed regardless of PMH use. A more gradual decline in density at a younger age, however, was observed at a higher BMI (p interaction age x BMI < 0.0001).


Figure 4
View larger version (6K):
[in this window]
[in a new window]
[Download PowerPoint slide]
  		FIGURE 4. Longitudinal patterns of mammographic density for postmenopausal women by reported postmenopausal hormone use (A, never use; B, former use; C, current use) and body mass index (BMI; kg/m2), Minnesota Breast Cancer Family cohort, 1990–2003.

 
Figure 5 shows density changes with age among postmenopausal women according to the reason for menopause. Women who attained menopause by bilateral oophorectomy or other means (primarily hysterectomy without bilateral oophorectomy) had higher initial density than women attaining a natural menopause; they also showed steeper declines in density with age (p interaction age x reason for menopause = 0.001). Women who chose bilateral oophorectomy did not have an excess prevalence of first-degree family history of breast cancer (143/234 = 61 percent) compared with those who did not choose oophorectomy (932/1,455 = 64 percent). Although parity/age at first birth was associated with percent density at any given age, this variable did not modify the trends in percent density with age among postmenopausal or premenopausal women (data not shown).


Figure 5
View larger version (7K):
[in this window]
[in a new window]
[Download PowerPoint slide]
  		FIGURE 5. Longitudinal patterns of mammographic density for postmenopausal women by reported postmenopausal hormone use (A, never use; B, former use; C, current use) given constant body mass index (26 kg/m2) and parity (1–2 births), Minnesota Breast Cancer Family cohort, 1990–2003.

 
Because of the heritability of this trait (24–29), and because the participants in the present analyses were recruited from families with a breast cancer proband, we evaluated whether familial correlation confounded the results. Inclusion of a random effect for familial correlation did not change the effect estimate of any fixed effect (e.g., BMI) by more than 1.1 percent of the unadjusted coefficient's value (data not shown). Furthermore, family history of breast cancer was not associated with density in this (p = 0.96) or another (30) study.


   DISCUSSION
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
References
 
This longitudinal analysis observed that change in mammographic density appeared to be influenced by age, baseline density, BMI, and PMH use. We found that the transition from pre- to postmenopause was associated with the greatest reductions in density but that this decline was attenuated among current hormone users. Furthermore, women who had a surgical menopause were more likely than women who attained a natural menopause to have higher premenopausal density and a steeper decline in density with age. Although parity/age at first birth was associated with percent density, it did not modify density trends with age.

Our data support the findings of Maskarinec et al. (15) that PMH use and higher BMI at baseline predict slower declines in density with age. We also observed that postmenopausal women who never used PMH had higher percent density before age 50 years, showed greater declines in density after age 50 years, and remained at lower density at all subsequent ages when compared with former and current hormone users, who showed the opposite trend. Possibly, never users of PMH may have represented a unique subset of women medically indicated not to use PMH, perhaps because they had not experienced menopausal symptoms. We also observed that BMI modified density changes. Both BMI and PMH use have been positively associated with breast cancer risk (31–33). Interactions between BMI and PMH use with change in density may have implications for risk. This association may be further clarified when data become available on the nature of the association between density changes and breast cancer risk.

In this study, percent density was associated with greater declines during menopause (between ages 45 and 60 years), particularly among women in the highest percentile distribution of density and among those who attained menopause through nonnatural means. The decrease in density associated with menopause was observed in one longitudinal study (12), but not another (15).

Several reasons could explain our finding of larger declines in density associated with nonnatural menopause. First, one might anticipate that immediate cessation of ovarian production of estrogen and progesterone would impact density more strongly than a natural and graded cessation. Second, women who entered natural menopause may have started the perimenopausal period earlier and therefore had lower density at the outset, so that the declines in density were exaggerated. Although possible, our data may suggest otherwise. Approximately 820 mammograms (about 15 percent) in this study were from premenopausal women aged 40–50 years, which is up to 10 years prior to the median age at menopause in this sample and is an age range in which hormone fluctuations (and subsequent effects on density) might expect to be observed. Yet, a large difference in densities between the groups was still evident across this age range. The difference was not attributable to an excess of women with a family history of breast cancer in the surgical menopause group, who may have been motivated to have an oophorectomy. Third, women undergoing surgical menopause may have had higher premenopausal density levels than women experiencing natural menopause because these women represented an unusual group medically indicated for surgery. For these various reasons, our results should be interpreted cautiously and may not be generalizable to other populations.

The density level among premenopausal women in this study is lower than has been previously reported for the computer-assisted thresholding measure (34, 35). This difference could result in difficulty in detecting small changes in density over time; however, these small changes are less likely to be of clinical significance. Furthermore, the error in our assessment of percent density was uniform across all density levels. In the evaluation of repeated measures of 195 mammograms included in the batches of mammograms reviewed for this study (refer to the Materials and Methods section), the absolute difference between the two measures was up to 9 percent across all density levels (e.g., from low density to high density), with a relatively constant degree of error distributed across density levels.

Variables reported to be associated with percent density and circulating hormones in cross-sectional analyses, such as parity and age at first birth (36–40), alcohol intake (37, 41, 42), and age at menarche (39, 41–43), were not associated with changes in percent density over time in the present study.

The strengths of our study include the large sample of women and large number of digitized mammograms over a wide age range, the categorization of women by menopausal status, and the use of a quantitative measure of mammographic density.

There are also potential limitations. First, the final model obtained in our analysis was identified from backward selection methods by using variables previously shown or suspected to be associated with mammographic density in this sample (8, 17). Possibly, other variables including diet (15) may be important predictors of density. Second, we did not have detailed information, for example, from blood samples, to validate the occurrence of menopausal transition among women. Thus, it is possible that women undergoing a hysterectomy continued to produce ovarian steroids and were misclassified as "postmenopausal." In addition, we did not have information on PMH formulation. Because combined progesterone and estrogen has been shown to increase density, with relatively little change observed with estrogen-only formulations (44), it is possible that the higher density observed among former and current PMH users in this study is due mostly to combined progesterone and estrogen formulations. The potential misclassification of hysterectomized women as postmenopausal and the inability to differentiate PMH formulation each would have attenuated our between-group findings and may suggest that the association of menopausal status as well as PMH use with density is greater than observed.

Third, we were cautious in our interpretation of the findings according to type of menopause since we could not be certain that women undergoing natural menopause had a longer menopausal transition than those with surgical menopause. Among women with natural menopause, this difference may have resulted in lower density at the outset of, with attenuated declines during, the study period than that among women who had a surgical menopause. Fourth, the Lumiscan 75 scanner has a useful optical density range of 0.2–3.6, yet film background typically has an optical density of approximately 3.8. Thus, in the scanned image, the breast edge may be difficult to distinguish from the film background, possibly leading to a tissue threshold that excludes the far edge of the breast and could result in a slightly higher estimate of percent density. Furthermore, higher contrast images were introduced throughout the 1990s, which could result in better detection of the breast edge and consequently lower density compared with older, lower-contrast images. This possibility would attenuate between-group findings; however, there is no reason to think that these effects differ between different risk factor subgroups in our study.

In summary, we observed that menopause, baseline percent density, PMH use, and BMI are associated with changes in mammographic density during adult life.


   ACKNOWLEDGMENTS
 
Supported by National Institutes of Health/National Cancer Institute grant P01 CA82267.

Conflict of interest: none declared.


   NOTES
 
Editor's note: An invited commentary on this article is published on page 1037.


   References
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 References
 

  1. McCormack VA, Dos Santos Silva I. Breast density and parenchymal patterns as markers of breast cancer risk: a meta-analysis. Cancer Epidemiol Biomarkers Prev (2006) 15:1159–69.[Abstract/Free Full Text]
  2. Ursin G, Ma H, Wu AH, et al. Mammographic density and breast cancer in three ethnic groups. Cancer Epidemiol Biomarkers Prev (2003) 12:332–8.[Abstract/Free Full Text]
  3. Maskarinec G, Meng L. A case-control study of mammographic densities in Hawaii. Breast Cancer Res Treat (2000) 63:153–61.[CrossRef][ISI][Medline]
  4. Boyd NF, Lockwood GA, Byng JW, et al. Mammographic densities and breast cancer risk. Cancer Epidemiol Biomarkers Prev (1998) 7:1133–44.[Abstract]
  5. Oza AM, Boyd NF. Mammographic parenchymal patterns: a marker of breast cancer risk. Epidemiol Rev (1993) 15:196–208.[Free Full Text]
  6. Warner E, Lockwood G, Tritchler D, et al. The risk of breast cancer associated with mammographic parenchymal patterns: a meta-analysis of the published literature to examine the effect of method of classification. Cancer Detect Prev (1992) 16:67–72.[ISI][Medline]
  7. Saftlas AF, Wolfe JN, Hoover RN, et al. Mammographic parenchymal patterns as indicators of breast cancer risk. Am J Epidemiol (1989) 129:518–26.[Abstract/Free Full Text]
  8. Vachon CM, Sellers TA, Vierkant RA, et al. Case-control study of increased mammographic breast density response to hormone replacement therapy. Cancer Epidemiol Biomarkers Prev (2002) 11:1382–8.[Abstract/Free Full Text]
  9. Lundstrom E, Christow A, Kersemaekers W, et al. Effects of tibolone and continuous combined hormone replacement therapy on mammographic breast density. Am J Obstet Gynecol (2002) 186:717–22.[CrossRef][ISI][Medline]
  10. Cuzick J, Warwick J, Pinney E, et al. Tamoxifen and breast density in women at increased risk of breast cancer. J Natl Cancer Inst (2004) 96:621–8.[Abstract/Free Full Text]
  11. Brisson J, Brisson B, Cote G, et al. Tamoxifen and mammographic breast densities. Cancer Epidemiol Biomarkers Prev (2000) 9:911–15.[Abstract/Free Full Text]
  12. Boyd N, Martin L, Stone J, et al. A longitudinal study of the effects of menopause on mammographic features. Cancer Epidemiol Biomarkers Prev (2002) 11:1048–53.[Abstract/Free Full Text]
  13. Salminen TM, Saarenmaa IE, Heikkila MM, et al. Risk of breast cancer and changes in mammographic parenchymal patterns over time. Acta Oncol (1998) 37:547–51.[CrossRef][ISI][Medline]
  14. van Gils CH, Hendriks JH, Holland R, et al. Changes in mammographic breast density and concomitant changes in breast cancer risk. Eur J Cancer Prev (1999) 8:509–15.[ISI][Medline]
  15. Maskarinec G, Pagano I, Lurie G, et al. A longitudinal investigation of mammographic density: the multiethnic cohort. Cancer Epidemiol Biomarkers Prev (2006) 15:732–9.[Abstract/Free Full Text]
  16. Sellers TA, Anderson VE, Potter JD, et al. Epidemiologic and genetic follow-up study of 544 Minnesota breast cancer families: design and methods. Genet Epidemiol (1995) 12:417–29.[CrossRef][ISI][Medline]
  17. Vachon CM, Kuni CC, Anderson K, et al. Association of mammographically defined percent breast density with epidemiologic risk factors for breast cancer (United States). Cancer Causes Control (2000) 11:653–62.[CrossRef][ISI][Medline]
  18. Nichols HB, Trentham-Dietz A, Hampton JM, et al. From menarche to menopause: trends among US women born from 1912 to 1969. Am J Epidemiol (2006) 164:1003–11.[Abstract/Free Full Text]
  19. Stone J, Dite GS, Gunasekara A, et al. The heritability of mammographically dense and nondense breast tissue. Cancer Epidemiol Biomarkers Prev (2006) 15:612–17.[Abstract/Free Full Text]
  20. Byng JW, Yaffe MJ, Jong RA, et al. Analysis of mammographic density and breast cancer risk from digitized mammograms. Radiographics (1998) 18:1587–98.[Abstract]
  21. SAS Institute, Inc. SAS. OnlineDoc, version 8. (1999) Cary, NC: SAS Institute, Inc.
  22. de Boor C. A practical guide to splines. (1978) Berlin, Germany: Springer Verlag.
  23. Cheney W, Kincaid D. Numerical mathematics and computing. (1985) New York, NY: Brooks/Cole Publishing Co.
  24. Wolfe JN, Albert S, Belle S, et al. Familial influences on breast parenchymal patterns. Cancer (1980) 46:2433–7.[CrossRef][ISI][Medline]
  25. Kaprio J, Alanko A, Kivisaari L, et al. Mammographic patterns in twin pairs discordant for breast cancer. Br J Radiol (1987) 60:459–62.[Abstract]
  26. Boyd NF, Dite GS, Stone J, et al. Heritability of mammographic density, a risk factor for breast cancer. N Engl J Med (2002) 347:886–94.[Abstract/Free Full Text]
  27. Pankow JS, Vachon CM, Kuni CC, et al. Genetic analysis of mammographic breast density in adult women: evidence of a gene effect. J Natl Cancer Inst (1997) 89:549–56.[Abstract/Free Full Text]
  28. Byrne C. Studying mammographic density: implications for understanding breast cancer. J Natl Cancer Inst (1997) 89:531–3.[Free Full Text]
  29. Vachon CM, King RA, Atwood LD, et al. Preliminary sibpair linkage analysis of percent mammographic density. J Natl Cancer Inst (1999) 91:1778–9.[Free Full Text]
  30. Mitchell G, Antoniou AC, Warren R, et al. Mammographic density and breast cancer risk in BRCA1 and BRCA2 mutation carriers. Cancer Res (2006) 66:1866–72.[Abstract/Free Full Text]
  31. Chlebowski RT, Hendrix SL, Langer RD, et al. Influence of estrogen plus progestin on breast cancer and mammography in healthy postmenopausal women: the Women's Health Initiative Randomized Trial. JAMA (2003) 289:3243–53.[Abstract/Free Full Text]
  32. Anderson GL, Limacher M, Assaf AR, et al. Effects of conjugated equine estrogen in postmenopausal women with hysterectomy: the Women's Health Initiative randomized controlled trial. JAMA (2004) 291:1701–12.[Abstract/Free Full Text]
  33. Harvie M, Howell A, Vierkant RA, et al. Association of gain and loss of weight before and after menopause with risk of postmenopausal breast cancer in the Iowa women's health study. Cancer Epidemiol Biomarkers Prev (2005) 14:656–61.[Abstract/Free Full Text]
  34. Maskarinec G, Pagano I, Lurie G, et al. Mammographic density and breast cancer risk: the Multiethnic Cohort Study. Am J Epidemiol (2005) 162:743–52.[Abstract/Free Full Text]
  35. Torres-Mejia G, De Stavola B, Allen DS, et al. Mammographic features and subsequent risk of breast cancer: a comparison of qualitative and quantitative evaluations in the Guernsey prospective studies. Cancer Epidemiol Biomarkers Prev (2005) 14:1052–9.[Abstract/Free Full Text]
  36. Haiman CA, Bernstein L, Berg D, et al. Genetic determinants of mammographic density. Breast Cancer Res (2002) 4:R5.[CrossRef][Medline]
  37. Haiman CA, Hankinson SE, De Vivo I, et al. Polymorphisms in steroid hormone pathway genes and mammographic density. Breast Cancer Res Treat (2003) 77:27–36.[CrossRef][ISI][Medline]
  38. Lillie EO, Bernstein L, Ingles SA, et al. Polymorphism in the androgen receptor and mammographic density in women taking and not taking estrogen and progestin therapy. Cancer Res (2004) 64:1237–41.[Abstract/Free Full Text]
  39. Maskarinec G, Lurie G, Williams AE, et al. An investigation of mammographic density and gene variants in healthy women. Int J Cancer (2004) 112:683–8.[CrossRef][ISI][Medline]
  40. Russo J, Russo IH. Development of the human breast. Maturitas (2004) 49:2–15.[CrossRef][ISI][Medline]
  41. Lai JH, Vesprini D, Zhang W, et al. A polymorphic locus in the promoter region of the IGFBP3 gene is related to mammographic breast density. Cancer Epidemiol Biomarkers Prev (2004) 13:573–82.[Abstract/Free Full Text]
  42. van Duijnhoven FJ, Bezemer ID, Peeters PH, et al. Polymorphisms in the estrogen receptor alpha gene and mammographic density. Cancer Epidemiol Biomarkers Prev (2005) 14:2655–60.[Abstract/Free Full Text]
  43. Warren R, Skinner J, Sala E, et al. Associations among mammographic density, circulating sex hormones, and polymorphisms in sex hormone metabolism genes in postmenopausal women. Cancer Epidemiol Biomarkers Prev (2006) 15:1502–8.[Abstract/Free Full Text]
  44. Warren R. Hormones and mammographic breast density. Maturitas (2004) 49:67–78.[CrossRef][ISI][Medline]

Add to CiteULike CiteULike   Add to Connotea Connotea   Add to Del.icio.us Del.icio.us    What's this?

Related articles in Am. J. Epidemiol.:

Invited Commentary: Assessing Breast Density Change—Lessons for Future Studies
Celia Byrne
Am. J. Epidemiol. 2008 167: 1037-1040. [Abstract] [FREE Full Text]  



This article has been cited by other articles:


Home page
 Am J EpidemiolHome page
C. Byrne
Invited Commentary: Assessing Breast Density Change--Lessons for Future Studies
Am. J. Epidemiol., May 1, 2008; 167(9): 1037 - 1040.
[Abstract] [Full Text] [PDF]

This Article
Right arrow Abstract Freely available
Right arrow FREE Full Text (PDF) Freely available
Right arrow All Versions of this Article:
167/9/1027    most recent
kwn063v1
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Email this article to a friend
Right arrow Related articles in Am. J. Epidemiol.
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Add to My Personal Archive
Right arrow Download to citation manager
Right arrow Search for citing articles in:
ISI Web of Science (1)
Right arrowRequest Permissions
Right arrow Disclaimer
Google Scholar
Right arrow Articles by Kelemen, L. E.
Right arrow Articles by Vachon, C. M.
PubMed
Right arrow PubMed Citation
Right arrow Articles by Kelemen, L. E.
Right arrow Articles by Vachon, C. M.
Social Bookmarking
Add to CiteULike   Add to Connotea   Add to Del.icio.us  
What's this?