The association of mammographic density with ductal carcinoma in situof the breast: the Multiethnic Cohort
© Gill et al.; licensee BioMed Central Ltd. 2006
Received: 3 March 2006
Accepted: 30 May 2006
Published: 23 June 2006
It is well established that women with high mammographic density are at greater risk for breast cancer than are women with low breast density. However, little research has been done on mammographic density and ductal carcinoma in situ (DCIS) of the breast, which is thought to be a precursor lesion to some invasive breast cancers.
We conducted a nested case-control study within the Multiethnic Cohort, and compared the mammographic densities of 482 patients with invasive breast cancer and 119 with breast DCIS cases versus those of 667 cancer-free control subjects. A reader blinded to disease status performed computer-assisted density assessment. For women with more than one mammogram, mean density values were computed. Polytomous logistic regression models were used to compute adjusted odds ratios (ORs) and 95% confidence intervals (CIs) for two measurements of mammographic density: percentage density and dense area.
Mammographic density was associated with invasive breast cancer and breast DCIS. For the highest category of percentage breast density (≥50%) as compared with the lowest (<10%), the OR was 3.58 (95% CI 2.26–5.66) for invasive breast cancer and 2.86 (1.38–5.94) for breast DCIS. Similarly, for the highest category of dense area (≥45 cm2) as compared with the lowest (<15 cm2), the OR was 2.92 (95% CI 2.01–4.25) for invasive breast cancer and 2.59 (1.39–4.82) for breast DCIS. Trend tests were significant for invasive breast cancer (P for trend < 0.0001) and breast DCIS (P for trend < 0.001) for both percentage density and dense area.
The similar strength of association for mammographic density with breast DCIS and invasive breast cancer supports the hypothesis that both diseases may have a common etiology.
As a result of increasing early detection efforts, breast carcinoma in situ (CIS) constitutes more than 20% of newly diagnosed breast cancer cases in the USA . Although breast CIS shares a number of risk factors with invasive breast cancer, and ductal carcinoma in situ (DCIS) of the breast is considered to be a precursor to some invasive breast cancers , it is not clear to what extent breast CIS and invasive breast cancer have the same etiology. Several case-control studies have confirmed that mammographic density is associated with risk for breast cancer ; women with high breast density have a three-fold to six-fold greater risk for developing breast cancer than women with low breast density [3, 4]. Most studies of mammographic density included only invasive breast cancer cases  or combined invasive and CIS cases [6–10]. One study investigating breast CIS observed that DCIS was more likely to occur in mammographically dense areas , and another study reported an increase in breast hyperplasia or atypia/CIS in women with greater than 50% breast density . This analysis examines the association between mammographic density and risk for breast DCIS in comparison with invasive breast cancer and breast cancer-free control subjects.
Materials and methods
The data for this analysis were collected using a nested case-control study within the Hawaii component of the Multiethnic Cohort, which was established between 1993 and 1996 . As described in detail elsewhere , all female members diagnosed with primary breast cancer between cohort entry and December 2000 were identified as potential cases (n = 1,587). A similar number of randomly selected control subjects (n = 1,584) who were not known to have breast cancer were frequency matched to the distribution of ethnicity and 5-year age groups of the cases. Cases and controls with a previous diagnosis of breast cancer, a history of breast augmentation or reduction, and no mammogram were excluded. Approximately 13% of breast cancer cases and 4% of controls were ineligible primarily because of death or pre-existent breast cancer. Of the 1,396 cases eligible to participate, 52.6% responded to the mailings and gave full consent. Of the 1,500 eligible controls, 48.7% responded to the mailings and gave full consent. After removing women who did not have suitable mammograms, the final sample consisted of 607 breast cancer cases and 667 control subjects. The original cohort and the nested case-control study were approved by the Committee on Human Studies at the University of Hawaii. All participants provided informed consent to participate in both studies.
Details of the study procedures were reported previously . In brief, information on demographics, medical history, reproductive behavior, hormone replacement therapy (HRT) use, and body mass index (BMI) were collected with an extensive questionnaire at entry into the cohort during the period from 1993 to 1996 . As part of the nested case-control study, a one-page Breast Health Questionnaire (BHQ) was completed to elicit information on menopausal status, previous breast surgery, mammography, and HRT medications .
The mammographic films were retrieved from clinics located throughout the State of Hawaii using the authorization forms signed by the study participants. The original cohort study had no records on mammography use except for one item in the baseline questionnaire. At that time, 90% of Caucasian and Japanese women and 75% of Native Hawaiian women reported previous mammography . Only craniocaudal views were digitized using a Kodak LS 85 Film Digitizer with a pixel size of 260 μm. If available, mammograms for every second or third year were scanned with the goal being to cover as wide a time period as possible for each woman. For cases, only mammograms taken before treatment for breast cancer were selected. However, the image of the contralateral breast taken at the time of diagnosis was used for five cases. The scanned images for both breasts were assessed for densities using Cumulus108 software  by one reader (GM), who was blinded to case status and time sequence of the mammograms. After the reader determined a threshold for the edge of the breast and for the edge of the dense tissue , the computer calculated the total number of pixels in the digitized image that constituted the total area and the dense area and computed the ratio between the two values as percentage density. Because readings for the right and left breast were very similar (correlation coefficient >0.90), we averaged the values for both to obtain one measure of total breast area, dense area and percentage density.
A random sample of 410 mammograms was read in duplicate to assess the reliability of the mammographic readings. The intraclass correlation coefficients  were 0.96 (95% confidence interval (CI) 0.95–0.97) for the size of dense area and 0.996 (95% CI 0.995–0.997) for the total breast area, resulting in an intraclass correlation coefficient for percentage density of 0.974 (95% CI 0.968–0.978).
To test for differences in covariate values across breast cancer cases and control subjects, we performed t tests for continuous variables and χ2 tests for categorical variables. We used unconditional polytomous logistic regression modeling with the SAS software  to compute odds ratios (ORs) and 95% CIs for the risks for DCIS and invasive cancer associated with breast density . All P values reported are two sided. Breast cancer cases were divided into CIS and invasive based on information provided by the state-wide Hawaii Tumor Registry, a member of the National Cancer Institute's Surveillance, Epidemiology and End Results program. Of the 125 breast CIS cases, 119 were classified as having DCIS.
We chose two measures of mammographic breast density as our exposure variables: size of the dense area and percentage density. Percentage density was grouped into four commonly used categories: <10%, 10–24.9%, 25–49.9%, and ≥50%. The size of the dense areas was classified as follows: <15 cm2, 15–29.9 cm2, 30.0–44.9 cm2, and ≥45 cm2. Study participants were grouped into four categories: Japanese, Caucasians, native Hawaiians and others (mostly Filipinos). To maximize the number of participants per group, women of mixed ancestry – regardless of admixture – were assigned to one ethnic category according to the following priority ranking: native Hawaiian, Japanese, Caucasian, and, finally, other . We created an HRT variable using the responses from the questionnaire at cohort entry and from the BHQ at enrollment into the breast density study. A comparison of the HRT information from the two questionnaires exhibited good agreement for overlapping years when both questionnaires reported HRT use. If a woman indicated that she had used HRT at any time but the write-in field in the BHQ was empty, we assigned the type of HRT from the cohort questionnaire completed at baseline. For the women with missing HRT type information (5.4%), we imputed the type based on hysterectomy status: estrogen only for women with a hysterectomy and combined therapy otherwise.
All models were adjusted for the following covariates that are known to be associated with breast cancer and mammographic density: mean age of all mammograms (continuous), ethnicity, BMI (<22.5, 22.5 to <25, 25 to <30, or ≥30 kg/m2), parity (0–1, 2–3, or ≥4), age at menarche (<13, 13–14, or ≥15 years), age at first live birth (<21, 21–30, >30 years, or no children), menopausal status (pre- or postmenopausal), family history of breast cancer (breast cancer in a first-degree relative or no history), and HRT use (never, estrogen only, or estrogen + progestin). Tests for trend were performed by fitting a variable representing ordinal categories (described above) of percentage density or dense area.
We were also interested in comparing how well percentage breast density and size of dense area predicted invasive breast cancer and breast DCIS. We performed unconditional binary logistic regression and examined the area under the receiver operating characteristic (ROC) curve, which is a method used in sensitivity-specificity analyses. The ROC curve assesses the ability of the model to distinguish between two groups (for example, diseased and disease free). If the model is able to separate the two groups perfectly, then the area under the ROC curve is equal to 1; if the model performs no better than chance, then the area will be 0.5.
Characteristics of the study population by disease status
Invasive breast cancer
Age at diagnosis (years)
Age at recruitment (years)
Mean age at all mammograms (years)
Body mass index (kg/m2)
Time from first mammogram to diagnosis (years)
Family history of breast cancer (%)
Age at first birth (years)
Age at menarche (years)
Number of children
Ever HRT use (%)
Number of mammograms
Breast percent densityd
Breast dense aread (cm2)
Total breast aread (cm2)
Mammographic breast density and risk estimates for breast DCIS and invasive breast cancer
Invasive breast cancer (n)
Breast DCIS (n)
Invasive versus controls
DCIS versus controls
DCIS versus invasive
Mean percentage density (%)
P value for trend
Mean breast dense area (cm2)
P value for trend
Because of sample size limitations, we were only able to analyze mammographic density and risk for breast DCIS by ethnicity for Japanese and Caucasian women (results not shown). Both ethnic groups exhibited similar trends; risk for breast DCIS increased with increasing percentage density and dense area. However, the CIs were wide and, except for the highest category of percentage density and dense area in Caucasian women, all intervals included 1.
The area under the ROC curve was similar for percentage density and dense area. For the adjusted invasive breast cancer model, both values were 0.74, whereas for the adjusted breast DCIS model the areas under the ROC curve were 0.67. When modeled without adjustment for covariates, the area under the ROC curve for invasive breast cancer was 0.59 for both percentage density and dense area. A similar unadjusted model for DCIS yielded values of 0.59 for dense area and 0.61 for percentage density.
Our analyses revealed that mammographic breast density, as measured by percentage density and the size of dense area, is associated with breast DCIS as well as invasive breast cancer. The association was slightly weaker for breast DCIS than for invasive breast cancer. Comparisons of areas under the ROC curve indicated that breast density and dense area performed equally well in distinguishing breast DCIS and invasive breast cancer cases from controls. The ORs for breast DCIS in the present study were not as large as those reported in a study of a cohort of Canadian women , which estimated relative risks between 7 and 12 for detecting atypia/CIS in biopsy specimens from women with more than 50% density. However, the cell sizes within the strata of density were small in that study and cases were selected from a group of women with biopsies. A British case-control study  found much lower risks for breast CIS, and the estimates were similar in strength to those for invasive breast cancer. Women with Wolfe parenchymal patterns P2 or DY had 70% greater risk (95% CI 1.1–2.6) of screening detected in situ breast cancer compared with controls and a 30% greater risk (95% CI 1.2–1.8) of screening detected invasive breast cancer compared with controls.
Other studies of breast DCIS support the idea that there is a relation between in situ lesions and mammographic density. An investigation of 28 mammograms confirmed that breast DCIS occurs to a greater extent in areas of the breast that exhibit high mammographic density . The relative risk for a second breast cancer was found to be three times higher among women with primary DCIS who had mammographic densities of 75% and greater compared with women with less than 25% density .
Although it is known that some breast DCIS lesions will develop into invasive breast cancers , and that DCIS is often found near invasive breast cancer lesions , it remains problematic to distinguish DCIS lesions that may progress to invasive breast cancer from those that may not. Given the growing number of breast CIS cases with increasing mammography screening , this question has important implications when decisions must be made regarding the aggressiveness of treatment . A comparison of risk factors for CIS (ductal and/or lobular) and invasive breast cancer found mixed results. Established invasive breast cancer risk factors shown to be consistently associated with breast CIS include family history of breast cancer [24–30], low BMI among premenopausal women [24, 29, 30], and nulliparity [24, 28, 30, 31]. However, other invasive breast cancer risk factors such as early age at menarche, late age at menopause, increased endogenous estrogen levels, and alcohol intake were associated with breast CIS in some studies [26, 27, 29, 32], but not in others [24–27, 30, 33].
Our study and other reports mentioned above indicate that mammographic density is associated with breast CIS and with invasive breast cancer. However, because some studies describe differential effects for some risk factors, more research needs to be done. Studies with larger samples of breast CIS must be performed to assess whether factors such as postmenopausal obesity, growth factors (insulin-like growth factor-I), exogenous hormones, and cell proliferation biomarkers can help in elucidating the association between breast density, breast CIS, and progression to invasive breast cancer. New molecular techniques may have the ability to identify factors that are responsible for the progression from DCIS to invasive cancer .
A limitation of the present study is the lack of data on the frequency of mammography use. Although this population has high screening rates , we cannot rule out bias toward the null in estimates of DCIS risk because of the possibility of undetected breast DCIS among controls. We also had relatively low participation rates, although a comparison of participating and eligible women revealed that they had very similar characteristics . Our assessments of HRT and BMI are limited because we had to rely on self-report, and assumed that their values remained constant between the time they were reported and when the study was completed, but an examination of BMI from a follow-up questionnaire five years after cohort entry showed that the mean BMI changes by only 0.50 kg/m2 during that time. Therefore, differences in BMI are unlikely to change the results materially. The use of BMI in categories is unlikely to have confounded our results because analyses with continuous values of BMI gave nearly identical results. We had limited power to estimate the risk for DCIS; there were nearly four times as many invasive breast cancer cases as breast DCIS cases. Nevertheless, the cohort design, the multiple mammograms, and the frequent mammography use in this population must be considered strengths of this project. The high rate of mammography participation decreases the probability that a large number of cancers were missed in this population.
In the present study similar patterns of association for mammographic density with DCIS and invasive breast cancers add to the growing body of evidence that certain breast CIS and invasive breast cancers may share etiologic factors. At the same time, it appears likely that factors that have not yet been identified may influence the effects of breast density and other known breast cancer risk factors in the progression of breast CIS to invasive breast cancer.
Breast Health Questionnaire
body mass index
carcinoma in situ
ductal carcinoma in situ
hormone replacement therapy
receiver operating characteristic.
The case-control study was funded by a grant from the National Cancer Institute (R01 CA 85265). The Multiethnic Cohort Study has been supported by USPHS (National Cancer Insitute) grant R37 CA 54281. Dr Gill was supported by R25 CA 90956. We are very grateful to Jihae Noh for her outstanding work in mammogram retrieval and scanning, to Andrew Williams for the excellent database, and to Maj Earle for providing information from the Multiethnic Cohort Study.
- American Cancer Society SR2: Cancer Facts and Figures 2005. 2005, Atlanta, GA: American Cancer Society, Inc.Google Scholar
- Bernstein L: The epidemiology of breast carcinoma in situ. Ductal Carcinoma In Situ of the Breast. Edited by: Silverstein MJ, Recht A, Lagios MD. 2002, Philadelphia: Lippincott, Williams and Wilkins, 2Google Scholar
- Boyd NF, Rommens JM, Vogt K, Lee V, Hopper JL, Yaffe MJ, Paterson AD: Mammographic breast density as an intermediate phenotype for breast cancer. Lancet Oncol. 2005, 6: 798-808. 10.1016/S1470-2045(05)70390-9.PubMedView ArticleGoogle Scholar
- Maskarinec G, Pagano I, Lurie G, Wilkens LR, Kolonel LN: Mammographic density and breast cancer risk: the Multiethnic Cohort. Am J Epidemiol. 2005, 162: 743-752. 10.1093/aje/kwi270.PubMedView ArticleGoogle Scholar
- Ursin G, Ma H, Wu AH, Bernstein L, Salane M, Parisky YR, Astrahan M, Siozon CC, Pike MC: Mammographic density and breast cancer in three ethnic groups. Cancer Epidemiol Biomarkers Prev. 2003, 12: 332-338.PubMedGoogle Scholar
- Torres-Mejia G, De Stavola B, Allen DS, Perez-Gavilan JJ, Ferreira JM, Fentiman IS, Dos Santos Silva I: 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: 1052-1059. 10.1158/1055-9965.EPI-04-0717.PubMedView ArticleGoogle Scholar
- Maskarinec G, Meng L: A case-control study of mammographic densities in Hawaii. Breast Cancer Res Treat. 2000, 63: 153-161. 10.1023/A:1006486319848.PubMedView ArticleGoogle Scholar
- Byrne C, Schairer C, Wolfe J, Parekh N, Salane M, Brinton LA, Hoover R, Haile R: Mammographic features and breast cancer risk: effects with time, age, and menopause status. J Natl Cancer Inst. 1995, 87: 1622-1629.PubMedView ArticleGoogle Scholar
- Boyd NF, Byng JW, Jong RA, Fishell EK, Little LE, Miller AB, Lockwood GA, Tritchler DL, Yaffe MJ: Quantitative classification of mammographic densities and breast cancer risk: results from the Canadian National Breast Screening Study. J Natl Cancer Inst. 1995, 87: 670-675.PubMedView ArticleGoogle Scholar
- Saftlas AF, Hoover RN, Brinton LA, Szklo M, Olson DR, Salane M, Wolfe JN: Mammographic densities and risk of breast cancer. Cancer. 1991, 67: 2833-2838.PubMedView ArticleGoogle Scholar
- Ursin G, Hovanessian-Larsen L, Parisky YR, Pike MC, Wu AH: Greatly increased occurrence of breast cancers in areas of mammographically dense tissue. Breast Cancer Res. 2005, 7: R605-R608. 10.1186/bcr1260.PubMedPubMed CentralView ArticleGoogle Scholar
- Boyd NF, Jensen HM, Cooke G, Lee Han H: Relationship between mammographic and histological risk factors for breast cancer. J Natl Cancer Inst. 1992, 15: 1170-1179.View ArticleGoogle Scholar
- Kolonel LN, Henderson BE, Hankin JH, Nomura AMY, Wilkens LR, Pike MC, Stram DO, Monroe KR, Earle ME, Nagamine FS: A Multiethnic Cohort in Hawaii and Los Angeles: baseline characteristics. Am J Epidemiol. 2000, 151: 346-357.PubMedPubMed CentralView ArticleGoogle Scholar
- Hawaii State Department of Health: Hawaii Behavioral Risk Factor Surveillance Report 1994–2000. 2002, Honolulu: State of HawaiiGoogle Scholar
- Byng JW, Boyd NF, Fishell E, Jong RA, Yaffe MJ: The quantitative analysis of mammographic densities. Phys Med Biol. 1994, 39: 1629-1638. 10.1088/0031-9155/39/10/008.PubMedView ArticleGoogle Scholar
- Shrout P, Fleiss J: Intraclass correlations: uses in assessing rater reliability. Psychol Bull. 1979, 86: 420-428. 10.1037/0033-2909.86.2.420.PubMedView ArticleGoogle Scholar
- SAS Institute Inc: SAS OnlineDoc 9.1.2. 2004, Cary, NC: SAS Institute IncGoogle Scholar
- Hosmer DW, Lemeshow S: Applied Logistic Regression. 2000, New York: John Wiley & SOns Inc, 2View ArticleGoogle Scholar
- Sala E, Solomon L, Warren R, McCann J, Duffy S, Luben R, Day N: Size, node status and grade of breast tumours: association with mammographic parenchymal patterns. Eur Radiol. 2000, 10: 157-161. 10.1007/s003300050025.PubMedView ArticleGoogle Scholar
- Habel LA, Dignam JJ, Land SR, Salane M, Capra AM, Julian TB: Mammographic density and breast cancer after ductal carcinoma in situ. J Natl Cancer Inst. 2004, 96: 1467-1472.PubMedPubMed CentralView ArticleGoogle Scholar
- Page DL, Dupont WD, Rogers LW, Jensen RA, Schuyler PA: Continued local recurrence of carcinoma 15–25 years after a diagnosis of low grade ductal carcinoma in situ of the breast treated only by biopsy. Cancer. 1995, 76: 1197-1200.PubMedView ArticleGoogle Scholar
- Ernster VL, Barclay J, Kerlikowske K, Grady D, Henderson IC: Incidence of and treatment for ductal carcinoma in situ of the breast. JAMA. 1996, 275: 913-918. 10.1001/jama.275.12.913.PubMedView ArticleGoogle Scholar
- Leonard GD, Swain SM: Ductal carcinoma in situ, complexities and challenges. J Natl Cancer Inst. 2004, 96: 906-920.PubMedView ArticleGoogle Scholar
- Kerlikowske K, Barclay J, Grady D, Sickles EA, Ernster V: Comparison of risk factors for ductal carcinoma in situ and invasive breast cancer. J Natl Cancer Inst. 1997, 89: 76-82.PubMedView ArticleGoogle Scholar
- Gapstur SM, Morrow M, Sellers TA: Hormone replacement therapy and risk of breast cancer with a favorable histology: results of the Iowa Women's Health Study. JAMA. 1999, 281: 2091-2097. 10.1001/jama.281.22.2091.PubMedView ArticleGoogle Scholar
- Trentham-Dietz A, Newcomb PA, Storer BE, Remington PL: Risk factors for carcinoma in situ of the breast. Cancer Epidemiol Biomarkers Prev. 2000, 9: 697-703.PubMedGoogle Scholar
- Claus EB, Stowe M, Carter D: Breast carcinoma in situ: risk factors and screening patterns. J Natl Cancer Inst. 2001, 93: 1811-1817. 10.1093/jnci/93.23.1811.PubMedView ArticleGoogle Scholar
- Brinton LA, Hoover R, Fraumeni JF: Epidemiology of minimal breast cancer. JAMA. 1983, 249: 483-487. 10.1001/jama.249.4.483.PubMedView ArticleGoogle Scholar
- Longnecker MP, Bernstein L, Paganini-Hill A, Enger SM, Ross RK: Risk factors for in situ breast cancer. Cancer Epidemiol Biomarkers Prev. 1996, 5: 961-965.PubMedGoogle Scholar
- Weiss HA, Brinton LA, Brogan D, Coates RJ, Gammon MD, Malone KE, Schoenberg JB, Swanson CA: Epidemiology of in situ and invasive breast cancer in women aged under 45. Br J Cancer. 1996, 73: 1298-1305.PubMedPubMed CentralView ArticleGoogle Scholar
- Lambe M, Hsieh CC, Tsaih SW, Ekbom A, Trichopoulos D, Adami HO: Parity, age at first birth and the risk of carcinoma in situ of the breast. Int J Cancer. 1998, 77: 330-332. 10.1002/(SICI)1097-0215(19980729)77:3<330::AID-IJC3>3.0.CO;2-P.PubMedView ArticleGoogle Scholar
- Missmer SA, Eliassen AH, Barbieri RL, Hankinson SE: Endogenous estrogen, androgen, and progesterone concentrations and breast cancer risk among postmenopausal women. J Natl Cancer Inst. 2004, 96: 1856-1865.PubMedView ArticleGoogle Scholar
- Zeleniuch-Jacquotte A, Gu Y, Shore RE, Koenig KL, Arslan AA, Kato I, Rinaldi S, Kaaks R, Toniolo P: Postmenopausal levels of sex hormones and risk of breast carcinoma in situ: results of a prospective study. Int J Cancer. 2005, 114: 323-327. 10.1002/ijc.20694.PubMedView ArticleGoogle Scholar
This article is published under license to BioMed Central Ltd. This is an open access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.