Adiposity, Adult Weight Change, and Postmenopausal Breast Cancer Risk FREE

Obesity is a well-established risk factor for postmenopausal breast cancer.^1- 2 Risk associated with adiposity has been explained predominantly by increased production of available endogenous estrogens in the adipose tissue, potentially initiating and promoting breast carcinogenesis.^3- 5 Adulthood weight gain has also been associated with an increased risk of postmenopausal breast cancer.^6- 11 However, it is unclear whether the risk of breast cancer associated with weight gain is independent of the timing of weight gain or whether weight that is gained during potentially susceptible life stages (ie, perimenopause) is the relevant determinant of risk. Individual periods of hormonal change have distinct biological effects that may differentially affect the body size and breast cancer relation.¹²

Associations of adiposity and adult weight gain with postmenopausal breast cancer have been stronger in women who do not use menopausal hormone therapy (MHT).^2,6- 11 However, little is known about whether hormonal and reproductive factors reflecting early exposure to female hormones, such as age at menarche, modify the relations of adiposity and weight change to breast cancer risk.

In a large prospective study, we examined body mass index (BMI) (calculated as weight in kilograms divided by height in meters squared) at ages 18, 35, and 50 years and at the current age as well as weight change during those 4 periods in relation to postmenopausal breast cancer risk. We also investigated waist circumference, hip circumference, and waist-hip ratio in relation to breast cancer risk. We explored whether the associations of adiposity and weight change with breast cancer risk were modified by specific hormonal and reproductive factors, including age at menarche and MHT. We also examined whether associations with adiposity and weight change differed according to hormone receptor status and tumor stage at diagnosis.

The National Institutes of Health–AARP Diet and Health Study was established in 1995 when a baseline screening questionnaire that elicited information on medical history and lifestyle characteristics was returned by 566 407 AARP (formerly known as the American Association of Retired Persons) members aged 50 to 71 years who resided in 1 of 6 US states (California, Florida, Louisiana, New Jersey, North Carolina, and Pennsylvania) or 2 metropolitan areas (Atlanta, Georgia, and Detroit, Michigan) known to have high-quality cancer registries and large AARP memberships.¹³ A total of 132 943 women and 188 118 men satisfactorily completed the baseline questionnaire and an additional detailed risk factor questionnaire mailed in 1996. After excluding men, women with a history of cancer (except nonmelanoma skin cancer) (n = 9039), premenopausal women (n = 4445), and women with missing or extreme values of height or weight at ages 18, 35, or 50 years or at study baseline (n = 20 420), the analytic cohort included 99 039 postmenopausal women. Extreme values were defined based on the Box-Cox transformation, which minimizes the Kolmogorov-Smirnov test statistic for normality. Postmenopausal women were defined as women who reported having natural menopause (60.4%) or a bilateral oophorectomy (38.9%) or who were at least 57 years old (the 75th percentile of age at natural menopause in previous research [0.7%]).¹⁴

Members of the National Institutes of Health–AARP cohort are followed up annually for change of address by matching the cohort database to that of the National Change of Address maintained by the US Postal Service. Information on address changes is also obtained through receipt of US Postal Service processing of undeliverable mail, from other address change update services, and directly from participants who report address changes in response to study mailings, such as questionnaires, newsletters, and sample kits. Vital status is ascertained by annual linkage of the cohort to the Social Security Administration Death Master File on deaths in the United States, follow-up searches of the National Death Index for individuals who match to the Social Security Administration Death Master File, cancer registry linkage, questionnaire responses, and responses to other mailings.

All cases of breast cancer were identified by linkage to the 8 state cancer registries. For matching purposes, we have almost complete data on first and last name, address history, sex, and date of birth. All suspected matches underwent a review to reject potential matches that were unlikely to be true (approximately 4%), and uncertain matches underwent a final manual review. In a validation substudy, we previously linked a randomly selected subset of the cohort (n = 12 000) to all 8 cancer registries and compared those data with self-reports and subsequent medical record confirmation of incident cancer in this subcohort and found that 90% of all cancer cases are validly identified using cancer registries.¹⁵ The Special Studies Institutional Review Board of the US National Cancer Institute approved this study. Completion of the self-administered baseline questionnaire was considered to imply informed consent.

Information on weight at different ages (ages 18, 35, and 50 years and the current age), current height, and waist and hip circumferences was collected using the detailed risk factor questionnaire. Height and weight were used to calculate BMI, which was divided into 8 categories (15.0-18.4, 18.5-22.4, 22.5-24.9, 25.0-27.4, 27.5-29.9, 30.0-34.9, 35.0-39.9, and ≥40.0) that incorporated the definitions of underweight (<18.5), normal weight (18.5-24.9), overweight (25.0-29.9), and obesity (≥30.0) proposed by the World Health Organization.¹⁶ Current height was used to calculate BMI for each period.

Weight change was calculated as the difference in weight during 4 periods: from age 18 to 35 years, to approximate change in the early reproductive years; from age 35 to 50 years, to approximate change in the late reproductive years; from age 50 years to the current age, to approximate change during the perimenopausal and postmenopausal years; and from age 18 years to the current age, to approximate change during total adulthood. We chose cutoff points for weight change categories that are comparable with those used in the literature.^6,8,10,17

Information on hormone receptor status was obtained from cancer registries. The threshold for a positive hormone receptor status was at least 10 fmol of receptor per milligram of total protein. Because the Florida, Pennsylvania, and Michigan state cancer registries do not collect hormone receptor information, such data were unavailable for 847 cases originating from those state cancer registries. Thus, analyses regarding hormone receptor status excluded women from those 3 states. Women from states where hormone receptor information was available did not differ with respect to age, educational level, and BMI from women from states where such information was not available (data not shown).

We calculated person-years from the return date of the questionnaire (in 1996) to the first date of diagnosis of breast or other cancer (except nonmelanoma skin cancer), death, or December 31, 2000. Cox proportional hazards regression was used to estimate relative risks (RRs) and 95% confidence intervals (CIs) for breast cancer, with adjustment for the following potential confounders: age (continuous), race or ethnic group (white, African American, Hispanic, Asian, or other), family history of breast cancer (yes or no), level of education (<12 years of school or high school equivalent, 12 years of school or high school equivalent, post–high school vocational or technical training, some college or college graduate, or postgraduate), age at menarche (<12, 13-14, or ≥15 years), age at menopause (<40, 40-44, 45-49, 50-54, or ≥55 years), age at first birth (<20, 20-24, 25-29, or ≥30 years or nulliparous), parity (0, 1, 2, 3-4, or ≥5 live births), smoking status (never smoker, past smoker with categories of time since quitting of <5 and ≥5 years, or current smoker with categories of <20 and ≥20 cigarettes per day), physical activity (never, 1-3 times per month, 1-2 times per week, 3-4 times per week, or ≥5 times per week), fat intake (quintiles), alcohol consumption (quintiles), oophorectomy (yes or no), height (<155, 155-159, 160-164, 165-169, 170-174, 175-180, or >180 cm), and MHT use (current, past, or never). In weight change analyses, models were adjusted for weight at the beginning of the change interval to account for differences in body size at the beginning of the interval. We also ran models that mutually adjusted for weight change during all intervals. Tests for linear trend were conducted by treating the median values of each exposure category as a single continuous variable in the model. We also applied nonparametric regression using cubic splines¹⁸ to examine the association between weight change and breast cancer risk. To explore the effect of patterns of body size change throughout the lifetime on postmenopausal breast cancer risk, we classified participants as overweight or obese (BMI ≥25) or normal weight (BMI <25) with respect to BMI at ages 18, 35, and 50 years and the current age.

Relationships were evaluated in the entire data set and separately according to MHT use because MHT has been found to modify the associations of adiposity and weight change with breast cancer risk.¹ In addition, stratified analyses were performed based on age at menarche. We formally tested for interactions using log-likelihood ratio tests. In a separate analysis, we tested for heterogeneity according to estrogen and progesterone receptor status (ER/PR) and tumor stage using polytomous logistic regression with 3 distinct end points for each of these analyses: (1) ER+/PR+, (2) ER–/PR–, and (3) no breast cancer; (1) regional or distant metastatic disease, (2) in situ or localized disease, and (3) no breast cancer. We tested for collinearity between selected exposure variables by calculating the variance inflation factor. All the analyses were conducted using a software program (SAS version 9.1; SAS Institute Inc, Cary, North Carolina). All statistical tests were 2 sided.

During 383 447 person-years of follow-up of 99 039 postmenopausal women, 2111 women were diagnosed as having breast cancer (1740 invasive and 371 in situ). Women who had greater weight gain since age 18 years had an earlier age at first birth, were less likely to use MHT and to smoke, and were less physically active but consumed more total fat than women who had smaller weight gain (Table 1). The correlation coefficients between current BMI and BMI at ages 50, 35, and 18 years were 0.84, 0.64, and 0.33, respectively. Current BMI was positively correlated with weight gain since age 18 years (r = 0.82), hip circumference (r = 0.80), and waist circumference (r = 0.79). A total of 85.3% of women gained more than 2 kg since age 18 years, and 7.8% remained at a stable weight during that time (less than a ±2-kg change). Women gained an average of 15.6 kg since age 18 years. On average, women gained 5.0, 5.9, and 4.7 kg between ages 18 and 35 years, ages 35 and 50 years, and age 50 years and the current age, respectively.

In the analytic cohort as a whole, the multivariate RRs of breast cancer according to increasing current BMI categories (15.0-18.4, 18.5-22.4 [reference], 22.5-24.9, 25.0-27.4, 27.5-29.9, 30.0-34.9, 35.0-39.9, and ≥40) were 0.70, 1 [reference], 1.12, 1.19, 1.16, 1.22, 1.37, and 1.45, respectively (95% CI, 1.08-1.93) (P < .001 for trend). Estimates remained unchanged when in situ cases were excluded (data not shown).

Use of MHT modified the relations of BMI to breast cancer risk (P = .005 and .02 for interaction for current BMI and BMI at age 50 years, respectively). Hence, all subsequent analyses were stratified according to MHT use. Because associations were similar in never and former MHT users, we created a combined group henceforth referred to as MHT nonusers. Among MHT nonusers, current BMI and BMI at age 50 years showed strong positive relations with breast cancer risk (Table 2). Current BMI of 40 or greater showed a more than doubling of risk compared with women with current BMI of 18.5 to 22.4. In current MHT users, no associations were observed for current BMI or BMI at ages 50 and 35 years. In addition, in current MHT users, associations with BMI were similarly null for the subgroup of women with estrogen-only use and the group with estrogen plus progestin use (data not shown). In contrast, BMI at age 18 years was inversely associated with breast cancer risk regardless of MHT use.

Waist and hip circumferences and waist to hip ratio were each positively associated with breast cancer in MHT nonusers, whereas associations were null in MHT users. The association with waist circumference was slightly weakened after adjustment for current BMI (RR, 1.42; 95% CI, 0.79-2.57); however, results of the test for trend remained significant (P = .02 for trend).

Weight gains during the early reproductive years (ages 18-35 years), the late reproductive years (ages 35-50 years), the perimenopausal and postmenopausal years (age 50 years to the current age), and total adulthood (age 18 years to the current age) were each consistently associated with increased breast cancer risk in MHT nonusers compared with stable weight during those periods (Table 3). The associations with weight gain from age 50 years to the current age and weight gain from age 18 years to the current age were somewhat stronger in older women (>60 years) than younger women (≤60 years), although interaction with age was not significant (P = .28 and .32 for interaction, respectively). The association with weight gain from age 50 years to the current age was somewhat stronger in less overweight women at the beginning of the interval than in overweight women (for a ≥20-kg change: RR, 1.87 [95% CI, 1.24-2.81] for BMI <25 at age 50 years vs RR, 1.35 [95% CI, 0.90-2.03] for BMI ≥25 at age 50 years). However, none of the interactions between weight gain in a given interval and BMI at the beginning of that interval were significant (between BMI at age 18 years and weight gain from age 18 years to the current age: P = .55 for interaction; between BMI at age 18 years and weight gain from ages 18-35 years: P = .20 for interaction; between BMI at age 35 years and weight gain from ages 35-50 years: P = .97 for interaction; and between BMI at age 50 years and weight gain from age 50 years to the current age: P = .14 for interaction). No associations with weight gain were observed in current MHT users. The nonparametric regression curve showed a pattern similar to the categorical analyses; the multivariate RR increased with increasing weight change since age 18 years in MHT nonusers but not in current MHT users (Figure).

Adult weight loss was unrelated to breast cancer compared with stable weight (Table 3). The lack of association of weight loss with breast cancer risk was consistent across the entire lifetime. Associations between weight loss and breast cancer did not differ based on MHT use. Estimates remained unchanged when in situ cases were excluded (data not shown). To confirm that weight loss due to preclinical disease did not account for the null associations observed, we repeated all the analyses after excluding women who were diagnosed as having cancers of any type during the first 2 years of follow-up. Risk estimates remained unchanged (data not shown).

An exploratory analysis of body size patterns throughout life and breast cancer in MHT nonusers is given in Table 4. Compared with women who had a consistently normal weight throughout life, women who were overweight or obese (BMI ≥25) at ages 35 and 50 years and at study baseline had an approximately 1.4 times higher risk of breast cancer. In contrast, women who were overweight or obese at age 18 years in addition to being consistently overweight or obese at ages 35 and 50 years and at study baseline had no increased breast cancer risk compared with those who consistently maintained a normal weight. Similar relations were observed when using a BMI of at least 30 as the cutoff value (data not shown). In addition, when weight changes from age 50 years to the current age, from ages 35 to 50 years, and from ages 18 to 35 years were included simultaneously in the multivariate model, a positive relation between weight gain and breast cancer risk was observed for each stage of life (Table 4).

When weight gain in total adulthood and current BMI were simultaneously included in the multivariate model, a positive trend remained for weight change, but current BMI was no longer significantly associated with risk. We tested for collinearity of adult weight gain and current BMI by calculating the variance inflation factor. The values were 3.37 for weight gain and 3.34 for current BMI, suggesting no collinearity. Similarly, when adult weight gain and waist circumference were simultaneously included in the model, a positive trend remained for weight change but not for waist circumference (data not shown) (variance inflation factors: 2.03 for weight gain and 1.97 for waist circumference).

The associations among current BMI, BMI at age 18 years, and weight change in relation to breast cancer risk were modified by age at menarche (Table 5). More pronounced associations were seen in women with later age at menarche; relations were less pronounced in women with earlier age at menarche, although the interaction was significant only for weight change (P = .15, .83, and .007 for interaction, respectively). No interactions were observed with age at first birth, parity, or age at menopause. In current MHT users, no interactions between current BMI or adult weight gain and hormonal or reproductive factors in relation to breast cancer risk were observed (data not shown).

We subdivided breast cancer cases according to tumor stage at diagnosis and hormone receptor status in MHT nonusers (Table 6). Associations with current BMI were stronger for advanced disease (regional or distant metastases) than nonadvanced disease (in situ or localized) (P = .009 for heterogeneity). Estimates remained unchanged when in situ cases were excluded (data not shown). The associations of current BMI and weight gain with breast cancer according to tumor stage were suggestively more pronounced in women who had 1 or fewer mammograms during the past 3 years (P = .06 for heterogeneity) compared with women who underwent mammography more than once (P = .10 for heterogeneity).

In MHT nonusers, the positive associations between current BMI and adult weight gain and breast cancer risk were most evident for ER+PR+ tumors (P < .001 and P = .006 for heterogeneity, respectively). In contrast, associations with ER+PR−, ER−PR+, and ER−PR− tumors were nonsignificant, although statistical power to detect relations was weak because of the small numbers of cases. In current MHT users, no relation between current BMI or adult weight gain and breast cancer risk was observed, regardless of hormone receptor status (data not shown).

In this large prospective study of postmenopausal women, adulthood adiposity and weight gain were associated with increased breast cancer risk, particularly in MHT nonusers. Relations with weight gain were evident throughout adulthood and were not limited to specific periods in life. Risk associated with weight gain from age 18 years to the current age was stronger in women with later age at menarche than in those with earlier age at menarche. Women who gained weight or were overweight or obese were more likely to develop advanced disease or hormone receptor–positive tumors. These relations were all weaker or absent in current MHT users.

Most previous cohort studies^9,19- 24 have found a modest increase in breast cancer risk with increasing BMI, with risk estimates not exceeding 1.6 comparing extreme BMI categories. However, comparisons have been based on limited variation in BMI, with cutoff values generally not exceeding 30.0. For BMI values of 25.0 to 30.0, risk estimates have ranged from 1.3 to 1.6.^9,19- 24 The Pooling Project² reported a monotonic increase in breast cancer risk up to a BMI of 28.0, beyond which risk did not increase further. The present findings are consistent with those of the Women's Health Initiative, which found an RR of 2.5 for a BMI greater than 31.0.¹¹ No previous studies have included a BMI category of 40.0 or greater. Consistent with the present data, risk has been more pronounced in women not currently^8,21 or previously^9,11,24 using MHT. The present finding of an inverse association with BMI at age 18 years is also consistent with previous studies^11,23 and may be explained by increased frequency of anovulatory cycles and lower serum estradiol and progesterone levels in obese young women.^25- 26

At least 5 prospective studies have examined waist circumference in relation to breast cancer. Findings are mixed, with some studies showing a positive association^11,17,23 and others no relation.^21,27 Possible reasons for this inconsistency include small sample sizes, lack of stratification by MHT use, or variable adjustment for height or BMI.

The present findings indicate that the relations of adult weight gain to breast cancer are evident throughout the entire adulthood lifespan rather than being limited to a specific time in life. To our knowledge, this is the first prospective study to show significant increased risk in relation to multiple adult stages of life across a broad range of weight change. A case-control study⁷ from Long Island, New York, reported that weight gain after menopause showed an increased breast cancer risk, whereas weight gain from ages 20 to 30 years did not enhance risk, but the latter analysis did not evaluate potential effect modification by MHT use. In the Iowa Women's Health Study,²⁸ an increased risk of postmenopausal breast cancer was observed in women who progressively gained weight throughout adulthood, although that study used only 3 broad categories of weight change: stable, gain, and loss. The Nurses' Health Study⁶ examined weight change from age 18 years to menopause and weight change since menopause and found an increased risk of postmenopausal breast cancer in women who gained weight during those periods. Other studies^8,10- 11 have not addressed multiple life stages but have assessed adult weight change from age 18 years to the current age or have focused on postmenopausal weight change only.

Although substantial weight gain has been associated with enhanced breast cancer risk,²⁹ whether modest weight gains increase breast cancer risk is unclear. In the present study, the magnitude of the association between adult weight gain and breast cancer risk is consistent with previous prospective studies^6,8- 10 that reported a 50% to 90% increased risk for a 20- to 30-kg adult weight gain. With substantial statistical power, we evaluated even less pronounced weight gains. For example, we found a statistically significant 30% increased breast cancer risk with weight gain between as little as 2 to 10 kg from age 50 years to the current age, representing the perimenopausal to postmenopausal period for most women.

We found no significant association with weight loss, which is consistent with most studies in this area.^9- 10,30 In contrast, the Nurses' Health Study⁶ recently reported a significantly decreased breast cancer risk with sustained weight loss after menopause. Success in detecting an association with weight loss in that study could be due to substantially longer follow-up, a more rigorous definition of stable weight, and repeated assessments of weight change across successive questionnaire cycles. Similarly, the Iowa Women's Health Study²⁸ reported a decreased risk of breast cancer with adult weight loss.

The present data do not permit us to definitively conclude that both current BMI and adult weight gain are independently associated with increased breast cancer risk because current BMI represents the end result of adult weight gain up to that time. Although adult weight gain and current BMI were positively correlated in these data, this analysis indicated no statistically relevant collinearity between the two. The finding that current BMI was not significantly associated with risk, whereas a positive trend remained for weight change when those variables were modeled simultaneously, suggests that weight gain is an important risk factor for breast cancer. Because weight gain during adulthood mainly reflects the deposition of fat mass rather than lean body mass, weight gain potentially represents age-related metabolic change that may be important in breast cancer development.³¹ Findings of no association in women who were consistently overweight or obese during adulthood also suggest the importance of weight gain, although the adverse effect of overweight or obesity at ages 35 and 50 years and the current age may have been offset by the baseline effect of overweight or obesity at age 18 years.

The positive relations of adult weight gain and adiposity to postmenopausal breast cancer risk are biologically plausible because adipose tissue is a major source of female hormones.^3- 5 Serum hormone levels are positively associated with BMI even for values exceeding 30.0,³² and the association between BMI and postmenopausal breast cancer was found to be attenuated after adjustment for free estrogen.³ The present findings, being consistent with other study results,^{6- 7,33- 35} of stronger associations with hormone receptor–positive tumors support an estrogen mechanism. Because current MHT use increases circulating estrogen levels, use of MHT may obscure any effect of adiposity on breast cancer; neither adiposity nor weight gain affected breast cancer risk in current MHT users.

To our knowledge, this is the first study to present a significant interaction between adult weight gain and age at menarche in MHT nonusers, with stronger associations in women who had a later age at menarche. Although the exact mechanisms remain unclear, this finding suggests that the effect of weight gain is obscured in women with early age at menarche, who have a greater cumulative exposure to estrogen. The finding of a strong inverse association with BMI at age 18 years in women with late age at menarche also supports an estrogen mechanism. Further studies on weight gain and breast cancer risk according to age at menarche are warranted.

We found that obese women were more likely to present with advanced-stage breast cancer at diagnosis. Adverse outcome could reflect more rapid tumor growth because estrogen promotes cell proliferation in mammary tissue.³⁶ Alternatively, it could be due to less breast cancer detection in obese women. Because obesity is associated with low socioeconomic status,³⁷ women with low socioeconomic status may be less likely to undergo mammography or to seek medical advice.³⁸ Supporting this, associations of current BMI with breast cancer according to tumor stage were suggestively more pronounced in women without regular screening mammograms. Results are consistent with 3 previous case-control studies.^39- 41 One prospective study³⁴ reported a stronger association of adult weight gain with advanced breast cancer but did not examine current BMI.

This study has several limitations. Because the cohort comprised mainly white women, study findings may not apply to all women. Because anthropometry was assessed by self-report, results could be affected by imprecise measurements. However, the correlations between measured and self-reported weight and height (including recalled weight across several decades) are typically high, ranging from 0.80 to 0.95.⁴² Women tend to overreport their height and underestimate their weight slightly, with a tendency for heavier women to underestimate weight more than lean women.⁴³ Such misclassifications of weight and height would tend to overstate risk estimates. Hormone receptor status information was incomplete. However, such missingness was unrelated to the risk factor characteristics we measured, which may have introduced bias into the study. In addition, analyses including hormone receptor status were based on relatively few cases.

Advantages of this study include its large size, prospective design, and detailed anthropometry data with a substantial range in BMI levels. This enabled us to examine breast cancer risk according to narrow BMI categories with great precision across a variety of potentially important effect modifiers and to discern not only an enhanced risk of overweight but also a substantially elevated risk of obesity.

In summary, we found that weight gain throughout adulthood was associated with increased breast cancer risk. Relations with weight gain were not limited to specific periods in life. Overweight or obese women were more likely to develop advanced-stage breast cancer, which is associated with poor prognosis.⁴⁴ These findings may reinforce public health recommendations for the maintenance of a healthy weight throughout adulthood as a means of breast cancer prevention.

Correspondence: Jiyoung Ahn, PhD, Nutritional Epidemiology Branch, Division of Cancer Epidemiology and Genetics, National Cancer Institute, National Institutes of Health, EPS, 6120 Executive Blvd, Bethesda, MD 20892 (ahnj@mail.nih.gov).

Accepted for Publication: June 3, 2007.

Author Contributions:Study concept and design: Ahn, Schatzkin, Ballard-Barbash, Kipnis, Mouw, and Leitzmann. Acquisition of data: Schatzkin and Hollenbeck. Analysis and interpretation of data: Ahn, Schatzkin, Lacey, Albanes, Ballard-Barbash, Adams, Kipnis, Mouw, and Leitzmann. Drafting of the manuscript: Ahn and Leitzmann. Critical revision of the manuscript for important intellectual content: Ahn, Schatzkin, Lacey, Albanes, Ballard-Barbash, Adams, Kipnis, Mouw, Hollenbeck, and Leitzmann. Statistical analysis: Ahn, Lacey, Adams, Kipnis, and Leitzmann. Obtained funding: Schatzkin. Administrative, technical, and material support: Mouw and Hollenbeck. Study supervision: Schatzkin, Albanes, and Leitzmann.

Financial Disclosure: None reported.

Funding/Support: This research was supported by the Intramural Research Program of the National Institutes of Health, National Cancer Institute.

Disclaimer: The views expressed herein are solely those of the authors and do not necessarily reflect those of the Florida contractor or Department of Health. The Pennsylvania Department of Health disclaims responsibility for any analyses, interpretations, or conclusions.

Additional Contributions: Cancer incidence data from the Atlanta metropolitan area were collected by the Georgia Center for Cancer Statistics, Department of Epidemiology, Rollins School of Public Health, Emory University. Cancer incidence data from California were collected by the California Department of Health Services, Cancer Surveillance Section. Cancer incidence data from the Detroit metropolitan area were collected by the Michigan Cancer Surveillance Program, Community Health Administration, State of Michigan. The Florida cancer incidence data were collected by the Florida Cancer Data System under contract with the Department of Health. Cancer incidence data from Louisiana were collected by the Louisiana Tumor Registry, Louisiana State University Medical Center, New Orleans. Cancer incidence data from New Jersey were collected by the New Jersey State Cancer Registry, Cancer Epidemiology Services, New Jersey State Department of Health and Senior Services. Cancer incidence data from North Carolina were collected by the North Carolina Central Cancer Registry. Cancer incidence data from Pennsylvania were supplied by the Division of Health Statistics and Research, Pennsylvania Department of Health. Robert N. Hoover, MD, ScD, and Anne Thiebaut, PhD, from the National Cancer Institute, provided insightful input.