906

Can the Gail Model be Useful in American Indian and Alaska Native Populations? Judith Salmon Kaur, M.D.1 Marilyn A. Roubidoux, M.D.2 Jeff Sloan, Ph.D.3 Paul Novotny, M.S.3 1

College of Medicine, The Mayo Clinic, Rochester, Minnesota.

2

Department of Radiology, University of Michigan Medical Center, Ann Arbor, Michigan.

3

The Mayo Foundation, Rochester, Minnesota.

BACKGROUND. Very little is known about breast carcinoma risk factors for American Indian/Alaska Native (AI/AN) women undergoing screening. The Gail model has been a useful tool for predicting the risk of breast carcinoma in several populations. It has not been applied systematically to AI/AN women. METHODS. The current study was a retrospective review of 1458 screening mammograms performed for AI/AN women. The authors applied the Gail model to estimate both absolute risk and relative risk for breast carcinoma for AI/AN women screened in South Dakota, Arizona, and Alaska. RESULTS. The mean age of the women was 52.4 years. The onset of menses was not significantly different than expected. The average age at first birth was 20 years, very few women were nulliparous, and few women were age ⬎ 30 years at first live birth. The proportion of women reporting a first- or second-degree relative with breast carcinoma was similar to the proportion in the general population. The results of the model indicated an overall average relative risk that ranged from 1.42 to 2.69 compared with white American women, depending on the model assumptions used. Using a modified Gail model and calculating an imputed absolute risk, the expected incidence of breast carcinoma in this population increased to rates of 170 –180 per 100,000 in the next 10 years, a significant increase over the Surveillance, Epidemiology and End Results– derived incidence rates from 1988 to 1992 of 31.6 per 100,000 for AI women in New Mexico and 78.9 per 100,000 for AN women. CONCLUSIONS. The model indicated a likelihood of increasing rates of breast carcinoma in the study population. The data obtained were useful in generating preliminary estimates of breast carcinoma risk in the study population, for which no prospective population survey has been completed. The inherent weaknesses in the current retrospective study indicated the need for a large-scale prospective data collection to confirm these exploratory findings. Cancer 2004;100:906 –12. © 2004 American Cancer Society.

KEYWORDS: breast carcinoma, American Indian, Alaska Native, Gail model.

Presented in part at the Fifth Biennial Conference on Cancer in Indian Country, Scottsdale, Arizona, November 11–14, 2001. Supported by the North Central Cancer Treatment Group (Rochester, MN). Address for reprints: Judith Salmon Kaur, M.D., Mayo Foundation, 200 First Street SW, Rochester, MN 55905; Fax: (507) 538-0823; E-mail: [email protected] Received April 1, 2003; revision received November 28, 2003; accepted December 3, 2003. © 2004 American Cancer Society DOI 10.1002/cncr.20047

A

ccess to annual screening mammography among American Indians and Alaska Native (AI/AN) women has been quite limited until very recently. Although average breast carcinoma mortality rates are lower than those of all races in the Surveillance, Epidemiology, and End Results (SEER) database, rates vary widely between tribes and Indian Health Service (IHS) regions.1–3 There is a general consensus that the rates of breast carcinoma in the AI/AN population have been increasing over the past 25 years. This is supported by the Alaska Tumor Registry, which has a 25-year record.4,5 Five-year survival rates of AI/AN women with breast carcinoma are the lowest of any group.6,7 Therefore, mammographic screening may be very important for changing this pattern. Previous data on the usefulness of mammography and risk factor

Breast Carcinoma Risk Profiles in Native Women/Kaur et al.

assessment were based on the general white and black population.8 –13 In these populations, screening mammography has been advocated to reduce breast carcinoma mortality rates among women age ⬎ 50 years. Risk factors for breast carcinoma that were verified from these larger populations included older age, young age of onset of menses, late onset of menopause, family history of breast carcinoma in a firstdegree relative, a personal history of breast carcinoma, nulliparity, age at first live birth ⬎ 30 years, higher breast density, and history of atypical hyperplasia on previous breast biopsies. The original Gail model was developed from data using ⬎ 4000 matched pairs of breast carcinoma cases and controls from the Breast Cancer Detection and Demonstration Project.14 That study was conducted from 1973 to 1980 and involved white American women. The generalizability of these data to other racial and ethnic groups is uncertain.14 –16 The National Cancer Institute has recently included risk assessment data applicable to Hispanic women on the ‘risk disk’, the tool used by physicians and other clinicians to help determine an individual woman’s 5-year and lifetime risk of developing invasive breast carcinoma.17 Other imputed risk factors include alcohol and obesity, but these have not been studied systematically as well as the Gail model.18 –20 Very little is known about the breast carcinoma risk factors or mammographic characteristics among AI/AN women undergoing screening. The Gail model, despite its limitations, has been widely used to calculate a woman’s risk of developing breast carcinoma. It is the essential qualifier for women participating in breast carcinoma prevention trials. This model allows one to estimate the risk of developing breast carcinoma over a specified time interval for a woman of a certain age with accepted risk factors. The elements in the model are age, age at menarche, number of breast biopsies, age at first live birth, and number of first -degree relatives with breast carcinoma. In the first National Surgical Adjuvant Breast and Bowel Project Breast Cancer Prevention Trial there were very few AI/AN women screened for risk factors, and even fewer were placed on protocol. Efforts are ongoing for AI/AN women to be screened and enrolled in the new P2 STAR protocol.21 We developed a protocol to evaluate these crucial issues as they relate to use, sensitivity, and appropriateness of screening in AI/AN women. The current study is sponsored by the North Central Cancer Treatment Group, using the Mayo Clinic (Rochester, MN) as its research base. The protocol was developed with tribal input and approved through tribal health boards, IHS area offices, all institutional review boards (IRB), and the national IHS IRB. This represents the

907

most accurate data on AI/AN women regarding breast carcinoma risk profiling. The current study covers the risk factor analysis of AI/AN women referred for mammographic screening.

MATERIALS AND METHODS The current study was a retrospective review of mammograms and accompanying standardized IHS mammography questionnaires for 1458 AI/AN women referred for screening mammography. The study was designed to accrue the mammography records of 1000 women. This sample size would provide 95% confidence intervals that were roughly 9 percentage points wide. Study sites included South Dakota, Arizona, and Alaska. All information was blinded for individual identifiers. Age, family history of breast carcinoma, age at first pregnancy, parity, age of menses, age of menopause, mammographic abnormalities, and density and biopsy results were recorded. Mammographic density was assessed visually by an experienced mammographer accredited by the American College of Radiology and the Food and Drug Administration (M.A.R.). The data regarding breast density for this cohort were published previously.22 The risk factors for each subject first were used to calculate a relative risk of developing breast carcinoma. These relative risks then were used to estimate the probability of developing breast carcinoma in the next 10 years (absolute risk). The relative risk and absolute risk according to the Gail model were calculated for each subject using the instructions set forth by Vogel.23 The mean of these individual probabilities was then multiplied by 100,000 to estimate an absolute risk for breast carcinoma cases in the population in the next 10 years per 100,000 women. The Gail model elements including woman’s age, age at menarche, number of breast biopsies, age at first live birth, and the number of first-degree relatives with breast carcinoma were accessed from mammography records. After reviewing the medical records from the AI/AN women, data on some of the necessary variables for the Gail model were unavailable. In some cases, data were unavailable because patients did not provide them or because there were administrative oversights in their collection. In other cases, only data on whether a breast biopsy was performed (and not on how many biopsies were performed) were available. Table 1 shows the variables used in the Gail model and the variables on which data were available in the AI mammography data set. To evaluate breast carcinoma risk, three values were estimated for each patient, based on the Gail model: minimum possible risk, maximum possible

908

CANCER March 1, 2004 / Volume 100 / Number 5

TABLE 1 Variables Affecting Breast Carcinoma Risk Variable Age No. reported No. unavailable Mean (range) in years Age at menarchea No. reported No. unavailable Mean (range) in years Age at first pregnancy No. reported No. unavailable Mean (range) in years Percentage of women with a first-degree relative with breast carcinoma No. reported No. unavailable Percentage Percentage of women with a previous breast biopsy No. reported No. unavailable Percentage a

Alaska

Arizona

Rural South Dakota

Urban South Dakota

653 0 52.3 (28–85)

455 0 52.9 (19–85)

175 0 49.0 (19–82)

175 0 54.4 (27–83)

0 653 — (—)

432 23 12.5 (7–18)

165 10 13.1 (9–19)

172 3 12.8 (9–18)

325 328 20.8 (13–39)

404 51 20.3 (13–35)

163 12 19.8 (14–38)

164 11 20.8 (15–38)

651 2 8.1

453 2 8.8

171 4 9.4

175 0 11.4

645 8 14.6

455 0 8.1

175 0 9.1

175 0 14.3

Unavailable data from Alaska accounted for 88% of the missing values.

risk, and imputed risk based on the distributions of the existing data.24 These three relative risks are equal to the calculated relative risk from the Gail model if there were no missing data. In calculating the minimum estimated breast carcinoma risk, missing values were assumed to equal the value that would provide the lowest relative risk. For example, if age at menarche was unavailable, it was assumed that the age at menarche was ⱖ 14 years. This resulted in a relative risk of 1.000. The calculation of maximum estimated breast carcinoma risk assumed that each missing value equaled the value that would yield the greatest relative risk. For example, missing ages at menarche were assumed to be ⬍ 12 years, resulting in a relative risk of 1.207. The third estimate of breast carcinoma risk was obtained by assuming that unavailable values would follow the same distribution as the available values. In the current data set, 19% of women had an age of menarche ⬍ 12 years, 55% had an age of menarche of 12–13 years, and 26% had an age of menarche ⱖ 14 years. Therefore, unavailable values for age of menarche were randomly assigned, with a 19% chance of being ⬍ 12 years, a 55% chance of being 12–13 years, and a 26% chance of being ⱖ 14 years.

RESULTS The mean age of women screened was 52.4 years (median, 51.0 years). There was no evidence of very

young onset, and the average onset of menses occurred at age 13 years in records in which onset of menses was recorded. The average age at first birth was 20 years. Very few women were nulliparous, and very few women had their first live birth at age ⬎ 30 years. The number of women in the current cohort who reported a first- or second-degree relative with breast carcinoma ranged from 17% in Arizona to 25% in South Dakota. There were 129 women with a personal history of malignancy. These malignancies included breast, cervical, and colon carcinoma, as well as melanoma.

Risk of Breast Carcinoma Table 2 shows the distribution of estimated relative breast carcinoma risk (overall and by site). The three analytic approaches to imputing unavailable data (maximum risk, minimum risk, and imputed risk) are presented separately. These risk rates represent the worst case, best case, and representative case, respectively, for the risk rates that one would obtain from the Gail model if complete data were available. The overall average relative risk ranges from 1.42 to 2.69 compared with white American women, depending on the model assumptions used. Rates across sites did not change substantially if the minimum risk approach was used (range, 1.36 –1.54). If the maximum or imputed risk approach was used, the risk for AN women

Breast Carcinoma Risk Profiles in Native Women/Kaur et al.

909

TABLE 2 Estimated Breast Relative Risk of Developing Breast Carcinoma Characteristic Gail model maximum risk Mean Median Standard deviation Gail model imputed risk Mean Median Standard deviation Gail Model minimum risk Mean Median Standard deviation

Overall (n ⴝ 1458)

Alaska (n ⴝ 653)

Arizona (n ⴝ 455)

Rural South Dakota (n ⴝ 175)

Urban South Dakota (n ⴝ 175)

2.69 1.55 3.65

3.39 2.33 4.62

2.04 1.37 2.16

2.26 1.24 3.10

2.24 1.37 2.49

2.45 1.55 3.12

2.86 1.93 3.76

2.03 1.37 2.14

2.21 1.24 3.01

2.23 1.37 2.47

1.42 1.21 0.69

1.36 1.00 0.76

1.46 1.24 0.58

1.45 1.24 0.69

1.54 1.37 0.65

was substantially higher compared with the risk among women at the other sites. This was due to the greater number of missing values in the Alaska dataset, particularly age of menarche. Using a modified Gail model and calculating an absolute risk rate, the expected incidence of breast carcinoma in this population over a 10-year span increases to 170 –180 per 100,000. This is a significant increase over the published SEER incidence rates from 1988 to 1992 of 31.6 per 100,000 for AI women in New Mexico and 78.9 per 100,000 for AN women. The patients evaluated in the current protocol outside of Alaska are not included in SEER databases, because it does not cover the AI in the Phoenix area or the Dakotas. Therefore, it is possible that the different populations may have different risk profiles.

DISCUSSION Breast carcinoma in past generations was reported to be very rare among AI/AN women.25–28 Data are fragmented currently because of the distribution of health services and data collection within the IHS.29 –35 Nonetheless, IHS data demonstrate significant increases in breast carcinoma–specific mortality. In the current study, the imputed risk from the Gail model application to our population predicts significant increases in absolute risk for the development of breast carcinoma over the next decade. The relative risk ranges from 1.42 to 2.69, indicating a population at increased risk. This amounts to an expected number of additional breast carcinoma cases over the next 10 years that will, at best, range from a low of an approximately 50% increase to a possible near-tripling relative to the most recently published SEER data. Screening for breast carcinoma is a new service to most AI/AN communities. Until recently, there were

no data on the mammographic characteristics of AI/AN women. Roubidoux et al.22 and Boyd et al.36 described significantly decreased breast density compared with a normative Caucasian population, which suggests screening mammography should be very valuable in this population. The lower breast density may mitigate risk attribution calculated from the Gail model.9,37– 47 There have been limited data on breast carcinoma risk factors among AI/AN women. Only Welty et al.48 have reported rates of first-degree relatives with breast carcinoma in women screened through the Sioux Cancer Study.48 There are no data on the incidence across all of AI populations. The SEER data are not generalizable across all tribes or regions. The most reliable data on incidence were obtained from the Alaska Tumor Registry, which is now a SEER special project site. Over the 25 years of recording cancer data in Alaska, the incidence of breast carcinoma has increased from 27 per 100,000 to 77 per 100,000.4,49 In addition, a recent Alaska Tumor Registry report titled “Alaska Native Cancer Survival Report” shows that from 1984 to 1994, the observed 5-year survival rate was 76.6%, which is significantly lower than the 84% for white women in Alaska.50 So it is unclear whether this reduced survival is due to advanced stage at presentation, differences in treatment for women with the same stage of disease, or other undetermined factors.51–57 To date, no study has systematically investigated the risk factors for breast carcinoma in the AI/AN population.58 In the general white population, women have a longer life expectancy than men. This also is true of AI/AN women. As women age, their risk of breast carcinoma increases. To the extent that other health concerns are addressed so that people live longer, this may also be playing a role in the increase

910

CANCER March 1, 2004 / Volume 100 / Number 5

in breast carcinoma cases observed recently in AI/AN communities. The SEER registry data from 1988 to 1992 show that the rate of breast carcinoma incidence among AI women ages 30 –54 years was similar to the corresponding rates among white and black women.59 As the life expectancy of AI/AN women increases, their risk of developing breast carcinoma also is likely to increase. The Gail model previously has been used to calculate risk for black women, white women, and, more recently, Hispanic women.60 The data presented in the current study represent the first attempt to investigate risk profiles in this special population. There are many caveats in understanding these data and it is anticipated this will lead to further data collection to validate or modify the information presented by us. This model can only be applied to the population of women under study who received screening mammograms and is not readily applicable to all AI/AN women. One potential problem concerns whether the women examined in the current project were representative of all AI/AN women. That is doubtful. In any previously underscreened population, it is typical that higher-risk women will be the first to be referred. This phenomenon is evidenced in our data by the number of women with a previous history of any malignancy, as well as by the number of women with a previous personal history of breast carcinoma. Women who are being followed for a personal history of malignancy probably are being followed by an oncologist, who would reinforce the need for regular screening. Nonetheless, the data on first-degree relatives is consistent with the data from the Sioux Cancer Study, which was a population-based study. Another issue is that the original Gail model was modeled on white American women. The relative risks associated with each of the Gail model variables could be different between white Americans and AI. These differences would impact all calculations. These are inherent limitations of using the Gail model to support predictions of breast carcinoma incidence in this population. Some have found that the model overpredicts absolute breast carcinoma risk in younger women who did not receive annual mammographic screening.61 These data suggest that for this cohort of women, the rate of breast carcinoma cases in the next decade probably will be significantly greater than was previously recognized. Even when all the model factors were set to yield the lowest possible relative risk for each participant, the resulting mean relative risk estimates were still greater than the baseline risk based on white American women in the Gail model (relative risk, 1.36 –1.54). The success of the Centers for Disease Control

(CDC)-sponsored breast and cervical carcinoma screening programs should increase the number of women having access to mammography. Researchers have already observed an increase in New Mexico of the number of cases of ductal carcinoma in situ in all ethnic groups due to access to screening (unpublished data). In contrast to population-based incidence rates, cancer detection rates are calculated only for the population of women who are screened, thereby reducing the impact of differential access to and use of screening services as an explanation for racial and ethnic differences. Commemorating Breast Cancer Awareness Month in April 2000, the CDC released data showing detection of invasive breast carcinoma at a rate of 4.9 per 1000 AI/AN women screened, compared with 7.7 per 1000 white women, 6.4 per 1000 African American women, 6.2 per 1000 Asian–Pacific Islander women, and 4.9 per 1000 Hispanic women. AI/AN women had the highest percentage (68%) of firstround cancers detected in Stage I.62 Ultimately, regular screening should improve the survival rates. Unfortunately, the CDC minimum datasets required of participating programs are not collecting Gail model information on women screened. It is entirely possible that the women screened through the CDC program may have a different risk profile than the women in the current study. Clinically, only 25% of women with breast carcinoma have identifiable risk factors. Therefore, the primary goal in a clinical setting should still be to offer annual screening mammograms to all women within a specific age group. The current IHS policy is to recommend annual screening to all women at age 40. AI/AN women have a younger age distribution compared with the general U.S. population. Therefore, proportionately more of the cases diagnosed may be in women age ⬍ 50 years. To conclude, the rate of breast carcinoma among AI/AN women will increase to meet or exceed that of U.S. white women. Since we completed the current study, the observed rate of breast carcinoma among AN women from 1987 to 1999 was 125.2 per 100,000, exceeding the rate among white women in the U.S.!63 We believe that screening will also produce a shift to earlier stage at diagnosis and improved survival. We would like to see comprehensive Gail model data collected on AI/AN women to advance our understanding of the risk factors in the AI and AN populations.

REFERENCES 1.

Valway S, editor. Cancer mortality among Native Americans in the United States: regional differences in Indian health, 1984-88 and trends over time, 1968-87. Rockville, MD: Indian Health Service, 1992.

Breast Carcinoma Risk Profiles in Native Women/Kaur et al. 2.

3.

4.

5.

6.

7.

8.

9.

10.

11.

12.

13.

14.

15. 16. 17. 18.

19. 20.

21.

22.

Hampton JW. The heterogeneity of cancer in Native American populations. In: Jones L, editor. Minorities and cancer. New York: Springer-Verlag, 1989:45–53. Partin M, Slater J, Korn J, et al. Cancer incidence among American Indians in Minnesota, 1988-1993. St. Paul: Minnesota Department of Health, 1996. Lanier AP. Cancer incidence in Alaska Natives: comparison of two time periods, 1989-93 vs. 1969-73. Cancer. 1998;83: 1815–1817. Lanier AP, Kelley JJ, Holck P, et al. Alaska Native cancer update 1985-97 by sex, age, service unit and year. Anchorage: Alaska Native Epidemiology Center, 2000. Baquet C, Ringen K. Surveillance, Epidemiology and End Results (SEER) report. Cancer among blacks and other minorities: statistical profile. Bethesda, MD: National Cancer Institute, Division of Cancer Prevention and Control, 1986. Swan J, Edwards H. Cancer rates among American Indians and Alaska Natives: is there a national perspective? Cancer. 2003;98:1262–1272. Caplan LS, Wells BL, Haynes S. Breast cancer screening among older racial/ethnic minorities and whites: barriers to early detection. J Gerontol. 1992;47:101–110. Byrne C, Schairer C, Wolfe J, et al. Mammographic features and breast cancer risk: effects with time, age, and menopause status. J Natl Cancer Inst. 1995;87:1622–1629. Byrne C, Colditz GA, Willett WC, et al. Plasma insulin-like growth factor (IGF) I, IGF-binding protein 3, and mammographic breast density. Cancer Res. 2000;60:3744 –3748. Fabian CJ, Kimler BF, Zalles CM, et al. Short-term breast cancer prediction by random periareolar fine-needle aspiration cytology and the Gail risk model. J Natl Cancer Inst. 2000;92:1217–1227. Abu-Rustum NR, Herbolsheimer H. Breast cancer risk assessment in indigent women at a public hospital. Gynecol Oncol. 2001;81:287–290. Stanford JL, Weiss NS, Voigt LF, et al. Combined estrogen and progestin hormone replacement therapy in relation to risk of breast cancer in middle-aged women. JAMA. 1995; 274:137–142. Gail MH, Brinton LA, Byar DP, et al. Projecting individualized probabilities of developing breast cancer for white females who are being examined annually. J Natl Cancer Inst. 1989;81:1879 –1886. Vogel VG. Assessing women’s potential risk of developing breast cancer. Oncology (Huntingt). 1996;10:1451–1458. Bjurstam N, Bjorneld L, Duffy SW, et al. The Gothenburg breast screening trial. Cancer. 1997;80:2091–2099. Theisen C. Updated breast cancer risk disk available [abstract]. J Natl Cancer Inst. 2001;93:581a. Peacock SL, White E, Daling JR, et al. Relation between obesity and breast cancer in young women. Am J Epidemiol. 1999;149:339 –346. Smith-Warner SA, Spiegelman D, Yan SS, et al. Alcohol and breast cancer in women. JAMA. 1998;279:535–540. Mannisto S, Virtanen M, Kataja V, et al. Lifetime alcohol consumption and breast cancer: a case-control study in Finland. Public Health Nutr. 2000;3:11–18. Fisher B, Costantino JP, Wickerham DL, et al. Tamoxifen for prevention of breast cancer: report of the National Surgical Adjuvant Breast and Bowel Project P-1 study. J Natl Cancer Inst. 1998;90:1371–1388. Roubidoux MA, Kaur JS, Giroux J. Mammographic findings and family history risk for breast cancer in American Indian women. Cancer. 1998;83:1830 –1832.

911

23. Vogel VG. Assessing women’s potential risk of developing breast cancer. Oncology. 1996;10:1451–1463. 24. Cohen J. Statistical power analysis for the behavioral sciences. Mahwah, NJ: Lawrence Erlbaum Associates, 1988. 25. Hampton JW, Keala J, Luce P. Overview of National Cancer Institute networks for cancer control research in Native American populations. Cancer. 1996;78:1545–1552. 26. Mahoney MC, Michalek AM. A bibliometric analysis of cancer among American Indians and Alaska Natives. Alaska Med. 1995;37:54 – 62. 27. Eidson M, Becker TM, Wiggins CL, Keyu CR, Samet JM. Breast cancer among Hispanics, American Indians, and non-Hispanic whites in New Mexico. Int J Epidemiol. 1994; 23:231–237. 28. Black WC, Bordin GM, Varsa EW, Herman D. Histologic comparison of mammary carcinomas among a population of Southwestern American Indian, Spanish American, and Anglo women. Am J Clin Pathol. 1979;71:142–145. 29. Michalek AM, Mahoney MC, Papas M, Tenney M, Burhansstipanov L. Tribal-based cancer control activities among Alaska Natives: services and perceptions. Alaska Med. 1996; 38:59 – 64. 30. Gordon P. Mammography and Pap smear screening of Yacqui Indian women. Public Health Rep. 1994;109:99 –103. 31. Coughlin SS, Uhler RJ, Blackman DK. Breast and cervical cancer screening practices among American Indian and Alaska Native women in the United States, 1992-1997. Prev Med. 1999;29:287–295. 32. Johnson C, Archer J. Accuracy of Pap smear and mammogram self-reports in a southwestern Native American tribe. Am J Prev Med. 1995;11:360 –363. 33. Lanier A, Mostow EN. Screening for cancer in remotely populated regions—lessons from mammography and breast cancer. Arctic Med Res. 1991;Suppl:462– 464. 34. Nutting PA, Calonge BN, Iverson DC, Green LA. The danger of applying uniform clinical policies across populations: the case of breast cancer in American Indians. Am J Public Health. 1994;84:1631–1636. 35. Falls Down D, Hammond N, Kuefler P, et al. Cancer control survey among Crow and Cheyenne reservations [abstract]. Proc Am Soc Clin Oncol. 1991;10:A230. 36. Boyd NF, Byng JW, Jong RA, et al. Quantitative classification of mammographic densities and breast cancer risk: results from the Canadian National Breast Screening Study. J Natl Cancer Inst. 1995;87:670 – 675. 37. MacKarem G, Roche CA, Hughes KS. The effectiveness of the Gail model in estimating risk for development of breast cancer in women under 40 years of age. Breast J. 2001;7:34 – 39. 38. Lam PB, Vacek PM, Geller BM, Muss HB. The association of increased weight, body mass index, and tissue density with the risk of breast carcinoma. Cancer. 2000;89:369 – 375. 39. Vachon CM, Kushi LH, Cerhan JR, Kuni CC, Sellers TA. Association of diet and mammographic breast density in the Minnesota breast cancer family cohort. Cancer Epidemiol Biomarkers Prev. 2000;9:151–160. 40. 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 –515. 41. Boyd NF, Lockwood GA, Martin LJ, et al. Mammographic densities and risk of breast cancer among subjects with a family history of this disease. J Natl Cancer Inst. 1999;91: 1404 –1408.

912

CANCER March 1, 2004 / Volume 100 / Number 5

42. Van Gils CH, Hendriks JH, Otten JD, Holland R, Verbeek AL. Parity and mammographic breast density in relation to breast cancer risk: indication of interaction. Eur J Cancer Prev. 2000;9:105–111. 43. Vachon CM, Kuni CC, Anderson K, Anderson VE, Sellers TA. Association of mammographically defined percent breast density with epidemiologic risk factors for breast cancer (United States). Cancer Causes Control. 2000;11:653– 662. 44. Mezzetti M, La Vecchia C, Decarli A, Boyle P, Talamini R, Franceschi S. Population attributable risk for breast cancer: diet, nutrition, and physical exercise. J Natl Cancer Inst. 1998;90:389 –394. 45. Schaierer C, Lubin J, Troisi R, Sturgeon S, Brinton L, Hoover R. Menopausal estrogen and estrogen-progestin replacement therapy and breast cancer risk. JAMA. 2000;283:485– 491. 46. Yaffe MJ, Boyd NF, Byng JW, et al. Breast cancer risk and measured mammographic density. Eur J Cancer Prev. 1998; 7:S47–S55. 47. Singletary KW, Gapstur SM. Alcohol and breast cancer. JAMA. 2001;286:2143–2151. 48. Welty TK, Zephier N, Schweigman K, Blakes B, Leonardson G. Cancer risk factors in three Sioux tribes. Use of the Indian-specific health risk appraisal for data collection and analysis. Alaska Med. 1993;35:265–272. 49. Hildes JA, Schaefer O. The changing picture of neoplastic disease in the western and central Canadian Arctic. Can Med Assoc J. 1984;130:25–32. 50. Lanier AP, Holck P, Kelly J, Smith B, McEvoy J. Alaska Native cancer survival report. Anchorage: Alaska Native Health Board and the Alaska Native Medical Center, April 1999. 51. Jones BA, Stanislav VK, Curnen MG, Owens PH, Dubrow R. Can mammography screening explain the race difference in stage at diagnosis of breast cancer? Cancer. 1995;75:2103– 2113.

52. Elledge RM, Clark GM, Chamness GC, Osborne CK. Tumor biologic factors and breast cancer prognosis among white, Hispanic, and black women in the United States. J Natl Cancer Inst. 1994;86:705–712. 53. Burhansstipanov L. National Cancer Institute’s Native American cancer research projects. Alaska Med. 1993;35: 248 –254. 54. Joe JR, Young RS. Overview: cancer in Indian Country—a national conference. Alaska Med. 1993;35:239 –242. 55. Kaur JS. Migration patterns and breast carcinoma. Cancer. 2000;88 Suppl 5:1203–1206. 56. Kaur JS. Native women and cancer. Health Care Women Int. 1999;20:445– 453. 57. Kaur JS. Helping the Native American cancer patient, their family and community. A panel presentation. Alaska Med. 1993;35:297–300. 58. Michalek AM, Mahoney MC. Cancer in native populations: lessons to be learned. J Cancer Educ. 1990;5:243–249. 59. Miller BA, Kolonel LN, Bernstein L, et al., editors. Racial/ ethnic patterns of cancer in the United States, 1988-1992. Bethesda, MD: National Cancer Institute, 1996. 60. Euhus DM. Understanding mathematical models for breast cancer risk assessment and counseling. Breast J. 2001;7:224 – 232. 61. Spiegelman D, Colditz GA, Hunter D, et al. Validation of the Gail et al. model for predicting individual breast cancer risk. J Natl Cancer Inst. 1994;86:600 – 607. 62. May DS, Lee NC, Richardson LC, Giustozzi AG, Bobo JK. Mammography and breast cancer detection by race and Hispanic ethnicity: results from a national program (United States). Cancer Causes Control. 2000,11:697–705. 63. Lanier A, Kelly JJ, McEvoy T, Sandidge J. Alaska Native cancer update 1987-99. Anchorage: Alaska Native Tribal Health Consortium, 2002.