The purpose is to determine breast cancer risk factors and correlates of mammographic parenchymal patterns among Alaska Native women. A retrospective review was performed of mammograms and mammogram records among 528 sequential screening mammogram examinations performed in Anchorage, Alaska. Mammogram density was classified by American College of Radiology (Breast Imaging Reporting and Data System) density patterns 1–4 (fat→dense) and by percent density. Clinical data, including risk factors, ethnic group (Indian, Aleut, or Eskimo), and smoking status were obtained. Results were analyzed by univariate and multivariate analyses. Of 528 women, 164 were Indian, 155 were Aleut, and 209 were Eskimo. Mean age at first birth was lower and parity higher compared with published data in white women. Breast cancer risk factors were similar across ethnic groups. In multivariate analysis, patient age, parity, hormone replacement therapy, hysterectomy, and history of biopsy were associated, and smoking was not associated with density scores. Aleut and Indian women were less likely to have high-density mammograms than were Eskimo women (P = 0.0448). No significant differences were found between ethnic group for conventional breast cancer risk factors. Mammogram density was associated with age at screening, parity, hormone replacement therapy, hysterectomy, history of biopsy, and ethnicity but not smoking status. Eskimo women had higher mammogram density than Aleuts or Indians.

Among Alaska native women the risk of breast cancer has quadrupled compared with 30 years ago (1). Alaska Native women previously had breast cancer rates half that of United States’ rates, but breast cancer incidence is now as high as that of United States white women (1). The reasons for the dramatic increase in breast cancer rates in Alaska Native women have not been studied. Contributing factors may be increased screening or increased longevity in this population.

Ethnic minorities have not been well represented in epidemiological research in breast cancer in the Unites States (2). Few data exist relative to the extent to which differences in breast cancer risk between race and ethnicity are related to differences in distributions of risk factors (2). There is sparse data on breast cancer risk factors among Native Americans and no data reported in Alaska Native women.

Mammogram density has also not been studied in Alaska Native women. The breast tissue parenchymal pattern as depicted on a mammogram, also known as mammogram density, is a factor that is influenced by hormonal breast cancer risk factors such as age at first birth, parity, menopause, and HRT3(3). In addition, density is a strong independent indicator of breast cancer risk (4, 5). The vast majority of these studies of mammogram density were performed among white women, with a few reports in other ethnic groups (2, 6, 7, 8, 9, 10). Mammogram density studies have not been done in Alaska Native women. A few studies also report that associations may exist between smoking and mammogram density and breast cancer risk (11, 12, 13).

Furthermore, differences in breast cancer risk have been reported between ethnic groups in Alaska.4 The objectives of this pilot study were to explore breast cancer risk factors, smoking, and the relationship of these to mammogram density in Alaska Native women who undergo mammography.

Subjects.

A retrospective review was performed of a sample of all sequential screening mammography examinations that were performed at the Alaska Native Medical Center between January and August of 1998 (n = 662 women). Alaska Native Medical Center is located in Anchorage and is a tertiary referral center providing healthcare to Alaska Native and Native Americans. Alaskan Natives include Indians, Eskimos, and Aleuts. Classification as Indian, Eskimo, or Aleut is part of the medical record and based on self-designation. Screening mammograms are defined as mammograms performed in patients who are without breast physical examination abnormalities, including nipple discharge, lumps, or thickening. Diagnostic mammograms, performed to evaluate such abnormalities, were excluded from this study. Consents from the Institutional Review Board of the Alaska Native Medical Center and from the North Central Cancer Treatment Group committee were obtained. A final report was made available to the Alaska Native Medical Center on August 13, 2000.

Mammogram Review and Data Collection.

Mammograms consisted of original standard mediolateral oblique and craniocaudal views obtained on dedicated mammographic equipment at the Alaska Native Medical Center. The mammograms were reviewed by a Food and Drug Administration-certified radiologist (M. A. R.) with 10 years of mammography experience in an ethnically diverse but non-Native American population at an academic medical center. Mammogram density was measured in two ways: (a) using the ACR (BIRADS) density patterns (14), a four category ordinal scale from 1 to 4 (fatty tissue to dense. The standard ACR BIRADS overall breast composition patterns are (1) almost entirely fat, (2) predominantly fat with scattered fibroglandular densities, (3) heterogeneously dense, and (4) extremely dense; (b) using percent density, which is a subjective assessment of the proportion of the area of the mammogram occupied by dense tissue. The radiologist was blinded to all clinical information. The BIRADS system and percent density scores have been used in mammography research and BIRADS is the standard adopted by the ACR, universally used in clinical practice (2, 6, 9, 15, 16).

As has been done in other studies (6), clinical data contained in the mammogram jacket in the mammogram questionnaire completed at time of mammography were recorded, including breast cancer risk factors (age, age at first birth, family history of breast cancer, HRT, lactation, hysterectomy, parity, HRT, prior breast biopsy, age at first birth) and other factors (current smoking status, use of birth control pills at screening, ethnic group). Age of menarche was not available. The clinical data were recorded by a separate observer (J. S. K.). Other medical data such as type of HRT or body weight were not available for this study per Institutional Review Board restrictions.

The distribution of ethnicity in this sample is as follows: 164 women reported as Indian, 155 Aleut, 209 Eskimo, 1 woman reported as both Indian and Eskimo, 1 woman reported as Hawaiian, 1 woman reported as “mixed, other,” and 131 women have missing information for ethnicity. Only data from those women classified as Indian, Aleut, or Eskimo were analyzed. The sample size was 528 women.

The subjects’ characteristics are summarized in Table 1 and are subcategorized by age group in Table 2, using age 50 years as a proxy for menopause. The overall rate of smoking was 44%. Smoking status percentage was similar across ethnic groups for women ages ≥ 50 years; however, for women < 50 years of age, Eskimos reported a higher proportion of current smokers (62.2%) than either Indians (45.3%) or Aleuts (40.8%; Fisher’s exact test P = 0.019). For all other characteristics, women were similar across ethnicity when considering either age group (age <50 or >50 years). The percentage of women reporting a first-degree family history was higher in Indians than Eskimos, but this did not achieve statistical significance. Information on parity was lacking for 195 (36.2%) of the women. Among the 333 women for whom parity information was available, 305 had given birth (91.6%) and 28 were nulliparous (8.4%).

Mammogram density as measured by BIRADS scores (Table 3) suggest a significant difference among ethnicities, without adjusting for potential confounding variables. In particular, Eskimo women were more likely to have higher dense mammograms with BIRADS scores of 3 or 4 than either Aleut or Aleut women, P (c2df=6 = 13.90) = 0.0308. Similarly, when comparing the mammogram density when classified as percent dense tissue (Table 3), Eskimo women have a significantly higher percent dense area (mean = 29.8, SD = 23.7) versus either Indian (mean = 24.4, SD = 20.8) or Aleut (mean = 23.8, SD = 20.9) women when comparing groups using the two-sample t test with Turkey’s multiple comparison correction.

When comparing BIRADS scores across ethnicities by age group (Table 4), there are more Eskimo women with mammograms scored as 4 (BIRADS score). Eskimo women also had higher mean percent density than the other ethnic groups, but this was not statistically significant. When comparing the age groups together, the larger sample size increase the power of the statistical test.

Although all three groups of women appeared to be similar for breast cancer risk factors, studied herein, across ethnic groups (with the potential exception of current smoking status), multivariate analyses were used to assess the independent effect of ethnicity on breast density. Using the percent area of the breast classified as dense as the outcome, standard least-squares regression techniques were used. As missing data were a concern [primarily for the parity index where 195 of 528 (36.9%) have missing information], both complete case methods and multiple imputation techniques were used. Using BIRADS breast density as the outcome, a cumulative logit, multinomial model was used for analysis. This model cumulates the logit scores or log odds for the higher BIRADS categories over the lower categories, thus estimating the odds of a subject having a higher BIRADS score. Regardless of outcome and the model type, all covariates of interest were offered to the model in a forward stepwise fashion, with P < 0.1 the criterion for a covariate to remain in the model. The following covariates were explored for their association: ethnicity; age of the women at screening; parity; age at first birth; history of nursing; birth control use at screening; HRT use at screening; family or personal history of breast cancer; history of breast biopsy; and current smoking status; factors are summarized in Tables 1and 2.

For the outcome, percent area of the breast considered dense breast density was significantly associated with parity, HRT, and history of biopsy (Table 5). Ethnicity was marginally associated when comparing Eskimo women to Indian women (P = 0.0923, model no. 1) using the complete case analysis in which the sample size was reduced by subjects missing information for parity. Similarly, age at screening was marginally significant (P = 0.0599) for this model. Modeling the data excluding parity (model no. 3) or when imputing parity for those subjects with missing information (model no. 2), ethnicity is significantly associated with breast density. The magnitude of effect is similar regardless of the modeling method used, with Indian women having breast density on average 4–6 percentage points less dense than Eskimo women. Aleut women were similar in average density to Indian women but were estimated to have 4–5.5 percentage points less dense breast on average than Eskimo women. Parity continued to be strongly related to breast density whether using complete-case methods (model no. 1) or when its value was imputed (model no. 2). The magnitude of effect was similar between models with women that have less dense breast by ∼2 percentage points/child.

A very similar pattern emerged for the analysis of the BIRADS scores (Table 6). Using the cumulative odds framework, Indian and Aleut women were less likely to have high BIRADS breast density scores than Eskimo women (odds ratio = 0.6 and 0.71, respectively, from model no. 1). Age at screening, parity, HRT, hysterectomy, and history of biopsy are all significantly associated with BIRADS scores. Smoking was not independently associated with density by either BIRADS or percent density scores.

Significant interactions between the significant breast cancer risk factors (age, parity, HRT, hysterectomy, and history of biopsy) and ethnicity were not seen in this sample, suggesting that these risk factors do not manifest themselves differently between ethnic groups.

Although breast cancer was previously believed to be a low-risk disease for Alaska Native women, rates have increased dramatically in the last 30 years (1). Breast cancer rates have also dramatically increased in other historically very low-risk ethnic groups, including the Native Americans in the Northern Plains, Native Hawaiians, and Japanese Americans in Hawaii (2, 17, 18). Ethnic minorities have been underrepresented in epidemiology research in cancer, yet such research may lead to insights to benefit these and other groups, and investigation of reasons for the increased rates of breast cancer in such groups is needed.

Mammogram density is an estimate of the proportion of fibroglandular tissue to fat in the breast. Density is known to vary with reproductive factors related to breast cancer risk such as, age at first birth, parity, and HRT and may also vary with diet, weight, alcohol intake, body mass index, and smoking (3, 11, 19, 20). However, mammogram density is also a strong independent risk factor (4, 20). There is a 6-fold variation in breast cancer incidence and mortality rates among Native American women, with Southwestern Native Americans having the lowest rates and Alaska and the Northern Plains regions having the highest rates (17, 18). Some researchers have used density patterns to examine populations with disparate rates of breast cancer. Previous studies of mammogram density among groups having disparate breast cancer risk have been performed between Asian and Caucasian women in the United Kingdom, Caucasian, Hispanic, and Native American women in New Mexico, and multiethnic groups in Seattle (2, 3, 6, 7, 8, 9, 10). Lower mammogram density has been described in Native American women in New Mexico, Seattle, and South Dakota as compared with other ethnic groups (6, 10, 21). A few studies have compared the relationship between mammogram density to breast cancer risk factors in various ethnic groups and to smoking status in white women, but such studies have not been done in Alaska Native women (6, 10, 11, 12).

Just as breast cancer patterns appear to differ nationwide among Native American tribes (18), differences were reported between ethnic groups and geographic locations in Alaska with data from 1969 to 1993, indicating that Eskimos had less risk than Indians4(22, 23). Age-adjusted breast cancer incidence rates were 46.7/100,000 among Eskimo and 95.2/100,000 among Indian women (14). Alaska Indians include Athabascans, who occupy the interior, and Tlingit, Haida, and Tsimpshian tribes, who reside mainly in the area of Alaska adjacent to British Columbia. Alaskan Indians have been considered linguistically related to Indians of the southwestern United States. Alaskan Eskimos constitute four recognized subgroups: Inupiaq in northern Alaska; Central Yupik in southwestern Alaska; Siberian Yupik on St. Lawrence Island; and Sugpiag on the North Pacific Rim. They are related to the Eskimos of Canada and Greenland. The Aleut group are those who occupied the Aleutian Island Chain and the Alaska Peninsula and possess some cultural and linguistic traits in common with Eskimos (23).

In our study group, the conventional breast cancer risk factors were similar among the three ethnic groups who obtained their mammograms at the Alaska Native Medical Center in Anchorage. Thus any differences, if they still exist, in breast cancer incidence rates by ethnicity are not explained by conventional risk factors. As compared with other studies on breast cancer risk factors in white women who undergo mammography, rates of family history positivity and HRT use among the entire study group were similar, and parity was higher (6, 24, 25, 26). Mean age at first birth for the women > 50 years (19.7–21.2 years) was slightly lower than data in white women of same age groups, and similarly in this group, mean number of births (4.3–4.7) is higher than that reported in white women (26, 27). Although the percentage of women reporting a first-degree family history was higher in Indians than in Eskimos, which would correlate to their reported higher risk 20 years ago, this difference did not achieve statistical significance. Parous women were 91.6% of the group, which is similar to Native American women (88.6%) and higher than white women in Seattle (79.9%; Ref. 6). Among the parous women in the group, 53.7% had more than or equal to four children, which is similar to Native Hawaiians (54.8%) and much higher than white (24%) women (2). The finding of somewhat more favorable parity data relative to breast cancer risk is paradoxical in light of the now equivalent risk in Alaska Native women as compared with white women. A similar paradox is seen in Native Hawaiian women, who have very high breast cancer incidence rates, despite high parity (2). Number of births varied by age with older women having higher mean values than younger women, a finding also seen in white women (27). Fewer births in the younger generation of Alaska Native women might be another contributing factor toward the increasing incidence of breast cancer in addition to other possible factors such as increasing screening and increasing longevity.

We found that breast cancer risk factors of age at screening, parity, HRT, hysterectomy, and history of biopsy were all significantly associated with BIRADS and/or percent density scores. These associations are consistent with other mammogram density studies (3, 20). Density did not correlate to smoking status; however, density did vary by ethnicity with Eskimo women having higher density than Indian women. Eskimos had higher mammogram density, which is opposite to that which would be expected based on the incidence data.4 There are several explanations for this paradox. First, density can vary with body weight, with diet, and with alcohol consumption (data not available in our study group; Refs. 27, 28, 29). Secondly, although mammogram density has been shown in many studies to predict breast cancer risk in women who are white, a previous study in Seattle found that Asian women were more likely and Native American women less likely to have dense mammograms and that density did not correlate to breast cancer risk in Asian women (6). Our findings are similar, indicating that Eskimo women have higher density, despite a historically lower risk of breast cancer. The Seattle study reported that variations in mammogram density by race may not conform to variations in breast cancer risk by race. This study concluded that density may not be a biomarker of breast cancer risk among every racial group, and our findings would be consistent with this hypothesis (6). Another possible reason for this lack of correlation between density and risk may be unknown genetic factors affecting density among racial groups.

A limitation of our study is that it was not performed in a fully multiethnic group to include all ethnic groups in Anchorage. Therefore, this study cannot determine whether density is different in Alaska Native women by group or subgroup, compared with other groups. However, because density interpretation was performed by an observer blinded to all clinical and ethnic information, any bias or misclassification in density interpretation within the study group would be uniform across the entire group and not result in systematic error or bias in subgroup results.

Smoking has been inversely associated with percent density in some studies and among premenopausal but not postmenopausal women in other studies (11, 28, 29). Some studies have even suggested a slight decreased risk for breast cancer among smokers based on mammogram pattern (29). In our study, smoking was not associated with mammogram density, and we did not find a correlation with differences in density or historical difference in breast cancer risk by ethnic group. This finding is may be limited by small subgroup sizes, however.

Breast cancer rates among the Alaska Native women now equal or exceed the United States’ rates but, if adjusted for risk factors, may be even higher. In Hawaii, for example, it was found that after adjustment for risk factors, the underlying breast cancer rate in Native Hawaiian women was actually 65% greater than that of white women (2). Furthermore, the increased incidence of breast cancer in Alaska Native women is not explained by known risk factor analysis. Comprehensive studies would be useful to more precisely correlate breast cancer incidence rates, breast cancer risk factors, and mammogram density in a larger multiethnic Alaskan population as has been done in Hawaii and Seattle (2, 6).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

1

Supported by National Cancer Institute Grant U10CA25224 to the North Central Cancer Treatment Group and National Cancer Institute Grant CA86098.

3

The abbreviations used are: HRT, hormone replacement therapy; ACR, American College of Radiology; BIRADS, Breast Imaging Reporting and Data System.

4

Internet address: http://www.cancercontrol.cancer.gov/womenofcolor/pdfs/alaska_native-tables.pdf.

Table 1

Subjects’ characteristics by ethnicity

IndianAleutEskimo
n 164 155 209 
Age in years at time of study mean (SD) 48.6 (11.2) 54.7 (10.4) 52.8 (10.5) 
Parity, n (%)    
 Nulliparous 7 (7.1) 7 (7.0) 14 (12.1) 
 Parous 91 (92.9) 99 (93.0) 115 (87.1) 
  Parity, mean, median (range) 3.9, 3 (1–12) 4.5, 4 (1–17) 4.1, 3 (1–12) 
  Age at first birth (yr), mean, median (range) 20.9, 20 (15–35) 20.2, 20 (14–32) 21.5, 20 (13–39) 
  First birth when age < 30 yrs, n (%) 80 (81.6) 92 (86.8) 105 (81.3) 
  Nursed, n (%) 54 (59.3) 55 (55.6) 74 (64.4) 
  ≥4 births 43 (47.2) 63 (63.6) 58 (50.4) 
  Unknown 66 (40.2) 49 (31.6) 80 (38.3) 
Hormone use at screening, n (%)    
 Birth control 7 (4.3) 5 (3.2) 10 (4.8) 
 HRT 69 (42.1) 69 (44.5) 84 (40.2) 
 Neither 84 (51.2) 81 (52.3) 114 (54.5) 
 Unknown 4 (2.4) 1 (0.5) 
Menopausal status, n (%)    
 Postmenopausal 104 (63.4) 116 (74.8) 135 (64.6) 
 Premenopausal 54 (32.9) 38 (24.5) 70 (33.5) 
 Unknown 6 (3.7) 0 (0.7) 4 (1.9) 
Hysterectomy, n (%)    
 Yes 43 (26.2) 50 (32.3) 65 (31.1) 
 No 118 (72.0) 105 (67.7) 141 (67.5) 
 Unknown 3 (1.8) 3 (1.4) 
Family history of breast cancer    
 Personal history 6 (3.7) 6 (3.9) 5 (2.4) 
 First degree relative 29 (17.7) 17 (11.0) 17 (8.1) 
 Second degree relative 8 (4.9) 10 (6.5) 11 (5.3) 
 No history 120 (73.2) 122 (78.7) 176 (84.2) 
 Unknown 1 (0.6) 
History of breast biopsy, n (%)    
 Yes 26 (15.9) 28 (18.1) 31 (14.8) 
 No 134 (81.7) 127 (81.9) 177 (84.7) 
 Unknown 4 (2.4) 0 (0.5) 
Time since last mammogram (yrs), mean, median (range) 1.5, 1.0 (0–7) 1.7, 1.0 (0–11) 1.5, 1.0 (0–15) 
Smoking history (current), n (%)    
 Yes 71 (43.3) 58 (37.4) 106 (50.7) 
 No 90 (54.9) 97 (62.6) 102 (48.8) 
 Unknown 3 (1.8) 1 (0.5) 
IndianAleutEskimo
n 164 155 209 
Age in years at time of study mean (SD) 48.6 (11.2) 54.7 (10.4) 52.8 (10.5) 
Parity, n (%)    
 Nulliparous 7 (7.1) 7 (7.0) 14 (12.1) 
 Parous 91 (92.9) 99 (93.0) 115 (87.1) 
  Parity, mean, median (range) 3.9, 3 (1–12) 4.5, 4 (1–17) 4.1, 3 (1–12) 
  Age at first birth (yr), mean, median (range) 20.9, 20 (15–35) 20.2, 20 (14–32) 21.5, 20 (13–39) 
  First birth when age < 30 yrs, n (%) 80 (81.6) 92 (86.8) 105 (81.3) 
  Nursed, n (%) 54 (59.3) 55 (55.6) 74 (64.4) 
  ≥4 births 43 (47.2) 63 (63.6) 58 (50.4) 
  Unknown 66 (40.2) 49 (31.6) 80 (38.3) 
Hormone use at screening, n (%)    
 Birth control 7 (4.3) 5 (3.2) 10 (4.8) 
 HRT 69 (42.1) 69 (44.5) 84 (40.2) 
 Neither 84 (51.2) 81 (52.3) 114 (54.5) 
 Unknown 4 (2.4) 1 (0.5) 
Menopausal status, n (%)    
 Postmenopausal 104 (63.4) 116 (74.8) 135 (64.6) 
 Premenopausal 54 (32.9) 38 (24.5) 70 (33.5) 
 Unknown 6 (3.7) 0 (0.7) 4 (1.9) 
Hysterectomy, n (%)    
 Yes 43 (26.2) 50 (32.3) 65 (31.1) 
 No 118 (72.0) 105 (67.7) 141 (67.5) 
 Unknown 3 (1.8) 3 (1.4) 
Family history of breast cancer    
 Personal history 6 (3.7) 6 (3.9) 5 (2.4) 
 First degree relative 29 (17.7) 17 (11.0) 17 (8.1) 
 Second degree relative 8 (4.9) 10 (6.5) 11 (5.3) 
 No history 120 (73.2) 122 (78.7) 176 (84.2) 
 Unknown 1 (0.6) 
History of breast biopsy, n (%)    
 Yes 26 (15.9) 28 (18.1) 31 (14.8) 
 No 134 (81.7) 127 (81.9) 177 (84.7) 
 Unknown 4 (2.4) 0 (0.5) 
Time since last mammogram (yrs), mean, median (range) 1.5, 1.0 (0–7) 1.7, 1.0 (0–11) 1.5, 1.0 (0–15) 
Smoking history (current), n (%)    
 Yes 71 (43.3) 58 (37.4) 106 (50.7) 
 No 90 (54.9) 97 (62.6) 102 (48.8) 
 Unknown 3 (1.8) 1 (0.5) 
Table 2

Subjects’ characteristics by ethnicity and age group

IndianAleutEskimo
Age < 50 yrAge ≥ 50 yrAge < 50 yrAge ≥ 50 yrAge < 50 yrAge ≥ 50 yr
n 64 100 49 106 90 119 
Age in years at time of study mean (SD) 42.8 (4.1) 58.6 (6.6) 43.6 (3.8) 59.9 (8.2) 43.1 (3.8) 60.1 (7.7) 
Parity, n (%)       
 Nulliparous 3 (4.7) 4 (4.0) 3 (6.1) 4 (3.8) 6 (6.7) 8 (6.7) 
 Parous 24 (37.5) 67 (67.0) 17 (34.7) 82 (77.4) 35 (38.9) 80 (67.2) 
  Parity, mean, median (range) 3.0, 3 (1–5) 4.3, 4 (1–12) 3.4, 3 (1–7) 4.7, 4 (1–17) 2.9, 3 (1–9) 4.7, 4 (1–12) 
  Age at first birth (yr), mean, median (range) 21.3, 20 (16–29) 20.7, 20 (15–35) 21.4, 21 (16–32) 19.9, 19 (14–30) 22.3, 21 (15–37) 21.2, 20 (13–39) 
  First birth when age < 30 yrs, n (%) 21 (87.5) 59 (88.1) 15 (88.2) 77 (93.9) 33 (94.3) 72 (90.0) 
  Nursed, n (%) 16 (66.7) 38 (56.7) 7 (41.2) 48 (58.5) 28 (80.0) 46 (57.5) 
  Unknown 37 (57.8) 29 (29.0) 29 (59.2) 20 (18.9) 49 (54.4) 31 (26.1) 
Hormone use at screening, n (%)       
 Birth control 6 (9.4) 1 (1.0) 5 (10.2) 10 (11.1) 
 HRT 9 (14.1) 60 (60.0) 8 (16.3) 61 (57.5) 14 (15.6) 70 (58.9) 
 Neither 46 (71.9) 38 (38.0) 36 (73.5) 45 (42.5) 66 (73.3) 48 (40.3) 
 Unknown 3 (4.6) 1 (1.0) 1 (0.8) 
Family history of breast cancer       
 Personal history 3 (3.1) 4 (4.0) 2 (4.1) 4 (3.8) 2 (2.2) 3 (2.5) 
 First degree relative 12 (18.7) 15 (15.0) 7 (14.3) 10 (9.4) 5 (5.6) 11 (9.3) 
 Second degree relative 7 (10.9) 1 (1.0) 3 (6.1) 7 (6.6) 10 (11.1) 1 (0.84) 
 First degree relative + personal history 1 (1.6) 1 (1.0)    1 (0.84) 
 No history 41 (64.1) 79 (79.0) 37 (75.5) 85 (80.2) 73 (81.1) 103 (86.6) 
 Unknown 1 (1.6) 
History of breast biopsy, n (%)       
 Yes 9 (14.1) 17 (17.0) 7 (14.3) 21 (19.8) 7 (7.8) 24 (20.2) 
 No 52 (81.3) 82 (82.0) 42 (85.7) 85 (19.8) 83 (92.2) 94 (79.0) 
 Unknown 3 (4.7) 1 (1.0) 1 (0.8) 
Time since last mammogram (yrs), mean, median (range) 1.79, 1.0 (0–7) 1.39, 1.0 (0–7) 1.76, 1.0 (0–9) 1.73, 1.0 (0–11.0) 1.41, 1.0 (0–7) 1.57, 1.0 (0–15) 
Smoking history (current), n (%)       
 Yes 29 (45.3) 42 (42.0) 20 (40.8) 38 (35.9) 56 (62.2) 50 (42.0) 
 No 33 (51.6) 57 (57.0) 29 (59.2) 68 (64.2) 34 (37.8) 68 (57.1) 
 Unknown 2 (3.1) 1 (1.0) 1 (0.8) 
IndianAleutEskimo
Age < 50 yrAge ≥ 50 yrAge < 50 yrAge ≥ 50 yrAge < 50 yrAge ≥ 50 yr
n 64 100 49 106 90 119 
Age in years at time of study mean (SD) 42.8 (4.1) 58.6 (6.6) 43.6 (3.8) 59.9 (8.2) 43.1 (3.8) 60.1 (7.7) 
Parity, n (%)       
 Nulliparous 3 (4.7) 4 (4.0) 3 (6.1) 4 (3.8) 6 (6.7) 8 (6.7) 
 Parous 24 (37.5) 67 (67.0) 17 (34.7) 82 (77.4) 35 (38.9) 80 (67.2) 
  Parity, mean, median (range) 3.0, 3 (1–5) 4.3, 4 (1–12) 3.4, 3 (1–7) 4.7, 4 (1–17) 2.9, 3 (1–9) 4.7, 4 (1–12) 
  Age at first birth (yr), mean, median (range) 21.3, 20 (16–29) 20.7, 20 (15–35) 21.4, 21 (16–32) 19.9, 19 (14–30) 22.3, 21 (15–37) 21.2, 20 (13–39) 
  First birth when age < 30 yrs, n (%) 21 (87.5) 59 (88.1) 15 (88.2) 77 (93.9) 33 (94.3) 72 (90.0) 
  Nursed, n (%) 16 (66.7) 38 (56.7) 7 (41.2) 48 (58.5) 28 (80.0) 46 (57.5) 
  Unknown 37 (57.8) 29 (29.0) 29 (59.2) 20 (18.9) 49 (54.4) 31 (26.1) 
Hormone use at screening, n (%)       
 Birth control 6 (9.4) 1 (1.0) 5 (10.2) 10 (11.1) 
 HRT 9 (14.1) 60 (60.0) 8 (16.3) 61 (57.5) 14 (15.6) 70 (58.9) 
 Neither 46 (71.9) 38 (38.0) 36 (73.5) 45 (42.5) 66 (73.3) 48 (40.3) 
 Unknown 3 (4.6) 1 (1.0) 1 (0.8) 
Family history of breast cancer       
 Personal history 3 (3.1) 4 (4.0) 2 (4.1) 4 (3.8) 2 (2.2) 3 (2.5) 
 First degree relative 12 (18.7) 15 (15.0) 7 (14.3) 10 (9.4) 5 (5.6) 11 (9.3) 
 Second degree relative 7 (10.9) 1 (1.0) 3 (6.1) 7 (6.6) 10 (11.1) 1 (0.84) 
 First degree relative + personal history 1 (1.6) 1 (1.0)    1 (0.84) 
 No history 41 (64.1) 79 (79.0) 37 (75.5) 85 (80.2) 73 (81.1) 103 (86.6) 
 Unknown 1 (1.6) 
History of breast biopsy, n (%)       
 Yes 9 (14.1) 17 (17.0) 7 (14.3) 21 (19.8) 7 (7.8) 24 (20.2) 
 No 52 (81.3) 82 (82.0) 42 (85.7) 85 (19.8) 83 (92.2) 94 (79.0) 
 Unknown 3 (4.7) 1 (1.0) 1 (0.8) 
Time since last mammogram (yrs), mean, median (range) 1.79, 1.0 (0–7) 1.39, 1.0 (0–7) 1.76, 1.0 (0–9) 1.73, 1.0 (0–11.0) 1.41, 1.0 (0–7) 1.57, 1.0 (0–15) 
Smoking history (current), n (%)       
 Yes 29 (45.3) 42 (42.0) 20 (40.8) 38 (35.9) 56 (62.2) 50 (42.0) 
 No 33 (51.6) 57 (57.0) 29 (59.2) 68 (64.2) 34 (37.8) 68 (57.1) 
 Unknown 2 (3.1) 1 (1.0) 1 (0.8) 
Table 3

Breast density by ethnicity

FactorIndianAleutEskimo
BIRADS, n (%)    
 1 50 (30.5) 54 (34.8) 50 (23.9) 
 2 46 (28.1) 40 (25.8) 55 (26.3) 
 3 67 (40.9) 59 (38.1) 93 (44.5) 
 4 1 (0.6) 2 (1.3) 11 (5.3) 
Percent breast density, mean, median (range) 24.4, 20 (0–80) 23.8, 20 (0–80) 29.9, 25 (0–100) 
FactorIndianAleutEskimo
BIRADS, n (%)    
 1 50 (30.5) 54 (34.8) 50 (23.9) 
 2 46 (28.1) 40 (25.8) 55 (26.3) 
 3 67 (40.9) 59 (38.1) 93 (44.5) 
 4 1 (0.6) 2 (1.3) 11 (5.3) 
Percent breast density, mean, median (range) 24.4, 20 (0–80) 23.8, 20 (0–80) 29.9, 25 (0–100) 
Table 4

Subjects’ breast density by ethnicity and age group

IndianAleutEskimo
Age < 50 yrAge ≥ 50 yrAge < 50 yrAge ≥ 50 yrAge < 50 yrAge ≥ 50 yr
BIRADS, n (%)       
 1 11 (17.2) 39 (39.0) 10 (20.4) 44 (41.5) 12 (13.3) 38 (31.9) 
 2 21 (32.8) 25 (25.0) 19 (38.8) 21 (19.8) 25 (27.8) 30 (25.2) 
 3 31 (48.4) 36 (36.0) 19 (38.8) 40 (37.7) 49 (54.4) 44 (37.0) 
 4 1 (1.6) 1 (2.0) 1 (0.9) 4 (4.4) 7 (5.9) 
Percent breast density, mean, median (range) 30.7, 25 (0–80) 20.4, 15 (0–70) 27.6, 25 (0–80) 22.1, 15 (0–80) 35.1, 30 (0–100) 25.9, 20 (0–80) 
IndianAleutEskimo
Age < 50 yrAge ≥ 50 yrAge < 50 yrAge ≥ 50 yrAge < 50 yrAge ≥ 50 yr
BIRADS, n (%)       
 1 11 (17.2) 39 (39.0) 10 (20.4) 44 (41.5) 12 (13.3) 38 (31.9) 
 2 21 (32.8) 25 (25.0) 19 (38.8) 21 (19.8) 25 (27.8) 30 (25.2) 
 3 31 (48.4) 36 (36.0) 19 (38.8) 40 (37.7) 49 (54.4) 44 (37.0) 
 4 1 (1.6) 1 (2.0) 1 (0.9) 4 (4.4) 7 (5.9) 
Percent breast density, mean, median (range) 30.7, 25 (0–80) 20.4, 15 (0–70) 27.6, 25 (0–80) 22.1, 15 (0–80) 35.1, 30 (0–100) 25.9, 20 (0–80) 
Table 5

Least-squares regression models for the outcome: percentage of breast area considered dense

Factor/estimate(SE)Model no. 1 Complete-case analysis with parity (n = 330)Model no. 2 Multiple imputation (n = 521)Model no. 3 Complete-case analysis without parity (n = 521)Ps Model #1 (model no. 3)
Intercept 34.7 (2.7) 33.3 (2.6) 28.5 (1.7) <0.0001 (<0.0001) 
Ethnicity     
 Indian −4.5 (2.7) −5.9 (2.2) −6.2 (2.2) 0.0923 (0.0049) 
 Aleut −4.0 (2.6) −5.1 (2.2) −5.5 (2.2) 0.1299 (0.0134) 
 Eskimo (referent)     
Age (per yr, centered) −0.3 (0.1) −0.4 (0.1) −0.5 (0.1) 0.0599 (<0.0001) 
Parity (per child) −2.1 (0.5) −1.4 (0.6) Omitted by design <0.0001 (n.a.)a 
HRT 4.6 (2.3) 5.2 (2.0) 5.0 (2.0) 0.0472 (0.0131) 
Hysterectomy −6.6 (2.5) −6.9 (2.2) −6.7 (2.2) 0.0082 (0.0022) 
Biopsy History 8.1 (2.8) 8.8 (2.5) 8.6 (2.5) 0.0040 (0.0006) 
Factor/estimate(SE)Model no. 1 Complete-case analysis with parity (n = 330)Model no. 2 Multiple imputation (n = 521)Model no. 3 Complete-case analysis without parity (n = 521)Ps Model #1 (model no. 3)
Intercept 34.7 (2.7) 33.3 (2.6) 28.5 (1.7) <0.0001 (<0.0001) 
Ethnicity     
 Indian −4.5 (2.7) −5.9 (2.2) −6.2 (2.2) 0.0923 (0.0049) 
 Aleut −4.0 (2.6) −5.1 (2.2) −5.5 (2.2) 0.1299 (0.0134) 
 Eskimo (referent)     
Age (per yr, centered) −0.3 (0.1) −0.4 (0.1) −0.5 (0.1) 0.0599 (<0.0001) 
Parity (per child) −2.1 (0.5) −1.4 (0.6) Omitted by design <0.0001 (n.a.)a 
HRT 4.6 (2.3) 5.2 (2.0) 5.0 (2.0) 0.0472 (0.0131) 
Hysterectomy −6.6 (2.5) −6.9 (2.2) −6.7 (2.2) 0.0082 (0.0022) 
Biopsy History 8.1 (2.8) 8.8 (2.5) 8.6 (2.5) 0.0040 (0.0006) 
a

n.a., not available.

Table 6

Cumulative logit estimates for the outcome: BIRADS breast density category

Factor/estimate(SE) odds ratioModel no. 1 Complete-case analysis with parity (n = 330)Model no. 2 Multiple imputation (n = 521)Model no. 3 Complete-case analysis without parity (n = 521)P Model no.1 (model no.3)
Ethnicity     
 Indian −0.52 (0.26) 0.60 −0.48 (0.20) 0.62 −0.51 (0.20) 0.60 0.0448 (0.0106) 
 Aleut −0.35 (0.25) 0.71 −0.46 (0.20) 0.63 −0.49 (0.20) 0.61 0.1678 (0.0148) 
 Eskimo (referent)     
Age (per yr) −0.03 (0.01) 0.97 −0.04 (0.1) 0.96 −0.05 (0.01) 0.95 0.0376 (<0.0001) 
Parity (per child) −0.21 (0.05) 0.81 −0.13 (0.05) 0.87 omitted by design <0.0001 (n.a.)a 
HRT 0.70 (0.23) 2.0 0.80 (0.19) 2.23 0.77 (0.19) 2.16 0.0021 (<0.0001) 
Hysterectomy −0.64 (0.24) 0.53 −0.72 (0.20) 0.49 −0.69 (0.20) 0.50 0.0084 (0.0005) 
Biopsy History 0.68 (0.27) 2.0 0.68 (0.23) 1.98 0.65 (0.23) 1.91 0.0137 (0.0048) 
Factor/estimate(SE) odds ratioModel no. 1 Complete-case analysis with parity (n = 330)Model no. 2 Multiple imputation (n = 521)Model no. 3 Complete-case analysis without parity (n = 521)P Model no.1 (model no.3)
Ethnicity     
 Indian −0.52 (0.26) 0.60 −0.48 (0.20) 0.62 −0.51 (0.20) 0.60 0.0448 (0.0106) 
 Aleut −0.35 (0.25) 0.71 −0.46 (0.20) 0.63 −0.49 (0.20) 0.61 0.1678 (0.0148) 
 Eskimo (referent)     
Age (per yr) −0.03 (0.01) 0.97 −0.04 (0.1) 0.96 −0.05 (0.01) 0.95 0.0376 (<0.0001) 
Parity (per child) −0.21 (0.05) 0.81 −0.13 (0.05) 0.87 omitted by design <0.0001 (n.a.)a 
HRT 0.70 (0.23) 2.0 0.80 (0.19) 2.23 0.77 (0.19) 2.16 0.0021 (<0.0001) 
Hysterectomy −0.64 (0.24) 0.53 −0.72 (0.20) 0.49 −0.69 (0.20) 0.50 0.0084 (0.0005) 
Biopsy History 0.68 (0.27) 2.0 0.68 (0.23) 1.98 0.65 (0.23) 1.91 0.0137 (0.0048) 
a

n.a., not available.

1
Lanier A. P., Kelly J. J., Holck P., Smith B., McEvoy T., Sandidge J. Cancer incidence in Alaska Natives thirty-year report 1969–1998.
Alaska Med.
,
43
:
87
-115,  
2001
.
2
Pike M. C., Kolonel L. N., Hendersen B. E., Wilkens L. R., Hankin J. H., Feigelson H. S., Wan P. C., Stram D. O., Nomura A. M. Y. Breast cancer in a multiethnic cohort in Hawaii and Los Angeles.
Cancer Epidemiol. Biomark. Prev.
,
11
:
795
-800,  
2002
.
3
Byrne C. Mammographic density: a breast cancer risk factor of diagnostic indicator?.
AUR
,
9
:
253
-255,  
2002
.
4
Byrne C., Schairer C., Wolfe J., Parekh N., Salane M., Brinton L. A., Hoover R., Haile R., et al Mammographic features and breast cancer risk: effects with time, age, and menopause status.
J. Natl. Cancer Inst. (Bethesda)
,
87
:
1622
-1629,  
1995
.
5
Warner E., Lockwood G., Tritchler D., Boyd N. F. The risk of breast cancer associated with mammographic parenchymal patterns: a meta-analysis of the published literature to examine the effect of method classification.
Cancer Detect. Prev.
,
69
:
67
-72,  
1992
.
6
El-Bastawissi A. Y., White E., Mandelson M. T., Taplin S. Variation in mammographic breast density by race.
Ann. Epidemiol.
,
4
:
257
-263,  
2001
.
7
Maskarinec G., Meng L., Shimozuma K. A pilot study of mammographic density patterns among Japanese women.
J. Epidemiol.
,
9
:
73
-77,  
1999
.
8
Hart B. L., Steinbock R. T., Mettler F. A., Pathak D. R., Bartow S. A. Age and race related changes in mammographic parenchymal patterns.
Cancer (Phila.)
,
63
:
2537
-2539,  
1989
.
9
Turnbull A. E., Kapera L., Cohen M. E. L. Mammographic parenchymal patterns in Asian and Caucasian women attending for screening.
Clin. Radiol.
,
48
:
38
-40,  
1993
.
10
Bartow S. A., Pathak D. R., Mettler F. A., Key C. R., Pike M. C. Breast mammographic pattern: a concatenation of confounding and breast cancer risk factors.
Am. J. Epidemiol.
,
142
:
813
-819,  
1995
.
11
Evis S., Warren R., McCann J., Duffy S., Luben R., Day N. High risk mammographic parenchymal patterns, hormone replacement therapy and other risk factors: a case-control study.
Int. J. Epidemiol.
,
29
:
629
-636,  
2000
.
12
Terry P. D., and Rohan T. E. Cigarette smoking and the risk of breast cancer in women: a review of the literature.
13
Braga C., Negri E., La Vecchia C., Filiberti R., Franceschi S. Cigarette smoking and the risk of breast cancer.
Eur. J. Cancer Prev.
,
5
:
159
-164,  
1996
.
14
Kopans D. B., Orsi C. J., Adler D. D. .
Breast imaging reporting and data system
, Ed.2 American College of Radiology Reston, VA  
1995
.
15
Heine J. J., Poonam M. Mammographic tissue, breast cancer risk, serial image analysis, and digital mammography, Part 1.
Acad. Radiol.
,
9
:
298
-316,  
2002
.
16
Jong R., Fishell E., Little L., Lockwood G., Boyd N. F. Mammographic signs of potential relevance to breast cancer risk: the agreement of radiologists’ classification.
Eur. J. Cancer Prev.
,
5
:
281
-286,  
1996
.
17
United States Department of Health and Human Services, Indian Health Service. Regional Differences in Indian Health 1998–1999, www.ihs.gov/ PublicInfo/Publications/trends98/region98.asp, 1999.
18
Kaur J. S. Migration patterns and breast carcinoma.
Cancer (Phila.)
,
88(5Suppl.)
:
1203
-1206,  
2000
.
19
Heine J. J., Malhotra P. Mammographic tissue, breast cancer risk, serial analysis, and digital mammography, Part 2.
Acad. Radiol.
,
9
:
317
-335,  
2002
.
20
De Waard F., Rombach J. J., Collette H. J. A., Slotboom B. Breast cancer risk associated with reproductive factors and breast parenchymal patterns.
J. Natl. Cancer Inst. (Bethesda)
,
72
:
1277
-1282,  
1984
.
21
Roubidoux M. A., Kaur J. S., Giroux J. Mammographic findings and family history risk for breast cancer in American Indian women.
Cancer (Phila.)
,
83
:
1830
-1832,  
1998
.
22
Lanier A. P., Blot W. J., Bender T. R., Fraumeni J. F. Cancer in Alaskan Indians, Eskimos, and Aleuts.
J. Natl. Cancer Inst. (Bethesda)
,
65
:
1157
-1159,  
1980
.
23
Lanier A. P., Bulkow L. R., Ireland B. Cancer in Alaskan Indians, Eskimos, and Aleuts, 1969–1983: implications for etiology and control.
Public Health Reports
,
104
:
658
-664,  
1989
.
24
Sickles E. A. Quality assurance: how to audit your own mammography practice.
Radiol. Clin. N. Am.
,
30
:
265
-274,  
1992
.
25
Stomper P. C., D’Souza D. J., DiNitto P. A., Arredondo M. A. Analysis of parenchymal density on mammograms in 1353 women 25–79 years old.
AJR
,
167
:
1261
-1265,  
1996
.
26
Simon M. S., Tang M-T. C., Bernstein L., Norman S. A., Weiss L., Burkman R. T., Daling J. R., Deapen D., Folger S. G., Malone K., Marchbanks P. A., McDonald J. A., Strom B. L., Wilson H. G., Spirtas R. Do thyroid disorders increase the risk of breast cancer?.
Cancer Epidemiol. Biomark. Prev.
,
11
:
1574
-1578,  
2002
.
27
Vachon C. M., Kushi L. H., Cerhan J. R., Kuni C. C., Sellers T. A. Association of diet and mammographic breast density in the Minnesota Breast Cancer Family Cohort.
Cancer Epidemiol. Biomark. Prev.
,
9
:
151
-160,  
2000
.
28
Sala E., Warren R., McCann J., Duffy S., Luben R., Day N. Smoking and high-risk mammographic parenchymal patterns: a case control study.
Breast Cancer Res.
,
2
:
59
-63,  
2000
.
29
Vachon C. M., Kuni C. C., Anderson K., Sellers T. A. Association of mammographically defined percent breast density with epidemiologic risk factors for breast cancer (United States).
Cancer Causes Control
,
11
:
653
-662,  
2000
.
30
Ehrsam G., Lanier A., Holck P., Sandidge J. Cancer mortality among Alaska Natives, 1994–1998.
Alaska Med.
,
43
:
50
-69,  
2001
.