Adult weight gain and histopathologic characteristics of breast cancer among postmenopausal women

Authors


Abstract

BACKGROUND

Although the link between postmenopausal breast cancer and adiposity is well established, the association between weight gain and specific histopathologic characteristics of breast carcinoma has not been studied carefully.

METHODS

Using 1200 incident invasive breast cancers among 44,161 postmenopausal women who were not taking hormone therapy in the American Cancer Society's Cancer Prevention Study II Nutrition Cohort, the authors computed age-adjusted rates and rate ratios (RR) for breast cancer by histology, stage, grade, and estrogen receptor (ER) and progesterone receptor (PR) status by categories of adult weight gain.

RESULTS

Age-adjusted rates of breast cancer were highest for women who reported the most weight gain, regardless of histologic type. For weight gain >60 pounds, compared with weight gain ≤20 pounds the RR for ductal carcinoma was 1.89 (95% confidence interval [95%CI], 1.53-2.34), and the RR for lobular carcinoma was 1.54 (95%CI. 1.01-2.33). Weight gain was associated with increased risk at every tumor stage and grade. The risk for regional or distant stage was elevated significantly in every category of weight gain and was 3 times higher among women who had the greatest weight gain (RR, 3.15; 95%CI, 2.21-4.48). Weight gain was associated with increased risk of ER-positive/PR-positive tumors (P for trend <.0001) but not ER-negative/PR-negative tumors (P for trend = .09). The results essentially remained unchanged when the analysis was restricted to women who had regular screening mammograms.

CONCLUSIONS

Excess adiposity is an important contributor to breast cancer risk among postmenopausal women, regardless of histologic type, and especially for tumors of advanced stage and high grade. Cancer 2006. © 2006 American Cancer Society.

Obesity is a well known risk factor for postmenopausal breast cancer.1 It is believed that the increase in risk results predominantly from circulating estrogens derived from aromatization of plasma androstenedione to estrone in adipose tissue,2, 3 although a number of peptide hormones also may play a role. Despite the growing body of literature examining the increased risk of breast cancer from obesity and weight gain, the relation between body weight and the histopathologic characteristics of breast carcinomas has received less attention. Although several studies have investigated at estrogen receptor (ER) and progesterone receptor (PR) status,4–12 few investigators have examined the association between adiposity and tumor stage, grade, or histologic type.12–16

Previous studies have shown that the relationship between obesity and breast cancer depends on whether a woman is taking postmenopausal hormone therapy.1, 17, 18 For example, in the American Cancer Society Cancer Prevention Study II (CPS-II) Nutrition Cohort, we found that adiposity was associated strongly with an increased relative risk of incident breast cancer only among women who currently were not using postmenopausal hormone therapy,18 and adult weight gain (defined as weight gain since age 18 years) was a stronger predictor of breast cancer risk than body mass index (BMI). In the current analysis, we examined the relation between several histopathologic characteristics of invasive breast cancer and adult weight gain among postmenopausal women who were not using postmenopausal hormone therapy.

MATERIALS AND METHODS

Study Cohort

Women in this analysis were drawn from the 97,786 female participants in the CPS-II Nutrition Cohort, a prospective study of cancer incidence and mortality among men and women in the United States that was established in 1992. The Nutrition Cohort is a subgroup of the approximately 1.2 million participants in CPS-II, a prospective study of cancer mortality that was established by the American Cancer Society in 1982, as described in detail elsewhere.19 Briefly, Nutrition Cohort participants were recruited from original members of the CPS-II cohort who resided in 21 states and were ages 50 to 74 years in 1992. Participants completed a mailed, self-administered questionnaire in 1992 or 1993 that included a food frequency diet assessment and information on demographic, medical, behavioral, environmental, and occupational factors. Beginning in 1997, follow-up questionnaires have been sent to cohort members every 2 years to update exposure information and to ascertain newly diagnosed cancers; response rates for all follow-up questionnaires have been at least 90%. This study is approved by the Emory University Institutional Review Board, and informed consent was obtained from all participants to gather information on cancer diagnosis and to retrieve medical records for confirmation of diagnoses.

The current analysis was limited to women who were postmenopausal at baseline in 1992 and were not currently using postmenopausal hormone therapy (n = 60,865 women). Women were considered menopausal if they reported that their periods had stopped permanently; women with missing or unclear menopausal data were classified as menopausal if they were older than age 60 years. We excluded women who were lost to follow-up from baseline in 1992 through August 31, 2001 (n = 2428 women); women who reported prevalent breast or other cancer at baseline, except nonmelanoma skin cancer (n = 9495 women); participants with extreme values of weight (those below the 0.1 percentile and above the 99.9 percentile) or with unknown weight (n = 2315 women). We also excluded women who reported a net weight loss >5 pounds after age 18 years (n = 2466 women), because the relationship between breast cancer and weight loss may be very different than its relationship with weight gain. After all exclusions, the final analytic cohort consisted of 44,161 postmenopausal women who were cancer-free at baseline in 1992.

Patient Ascertainment

We documented 1368 incident cases of breast cancer that were diagnosed between enrollment in 1992 and August 31, 2001 among the 44,161 women who were eligible for this analysis. Of these, 1289 women were identified by self-report on the follow-up questionnaires in either 1997 to 1999 or 1999 to 2001 and subsequently were verified by obtaining medical records or through linkage with state registries when complete medical records could not be obtained.19 Previous work linking our cohort members to state cancer registries has shown that the ability of our respondents to report a past diagnosis of cancer accurately is high (sensitivity, 0.93; specificity, >0.99 for report of any cancer).20

The current analysis was limited to invasive breast cancer; thus, women with in situ breast cancers were excluded if they were diagnosed in the interval between 1992 and 1997 (n = 130 women) or if they were censored at the time of diagnosis in subsequent intervals. We also excluded 3 women who had an unknown date of diagnosis and 35 women for whom records were not available to determine tumor type, leaving a total of 1200 women with breast cancer who had information on histopathologic characteristics.

Histopathologic Classification of Breast Tumors

We examined the influence of adult weight gain on the following tumor characteristics: histologic type; Surveillance, Epidemiology, and End Results (SEER) summary stage (localized and regional or distant); Grade (1, 2, 3); and ER/PR status. Data were obtained from state registries or abstracted from medical records by Certified Tumor Registrars using established registry guidelines. A subset of records also was reviewed by a pathologist (T.G.) to assess coding accuracy. We used SEER summary stage rather than other staging criteria, because it is the most common staging protocol used by central cancer registries. Histologic tumor types were grouped as ductal (International Classification of Diseases for Oncology [ICD-O] code 8500), lobular and mixed lobular/ductal (ICD-O codes 8520 and 8522), or “other” histology. Thus, “other” histology is a heterogeneous category comprised of all other breast carcinomas that are not coded as ductal, lobular, or mixed lobular ductal and included some poorly differentiated carcinomas of unspecified histology and a smaller number of well differentiated tubular, mucinous, and papillary carcinomas. This category was not subdivided in analyses because of the small number of cases. ER status was available for 699 women (58%), and PR status was available for 680 women (57%).

Statistical Analysis

Age-adjusted breast cancer incidence rates were calculated for categories of adult weight gain standardized to the age distribution of women in the CPS-II Nutrition Cohort at baseline. Cox proportional hazards modeling was used to examine the association of weight gain with histopathologic characteristics of incident breast cancer (described above) while adjusting for other potential risk factors. Weight gain was modeled as a time-dependent variable in this analysis using data from the baseline survey (1992) and from the 1997 and 1999 follow-up surveys. In 1992, current weight and weight at age 18 years were reported. Current weight also was recorded on each subsequent follow-up survey. Total adult weight change was calculated from reported weight at age 18 years and the current weight on each questionnaire. The referent category included women who reported ≤5 pounds lost (2.3 kg) or ≤20 pounds gained (9.1 kg) since age 18 years: Categories of weight gain were defined as follows: from 21 to 40 pounds (9.5-18.2 kg), from 41 to 60 pounds (18.6-27.3 kg), and >60 pounds (27.3 kg) gained since age 18 years.

All models were stratified by single year of age at enrollment. Potential confounders that were included in multivariate models were age at menarche (younger than 12 years, 12 years, 13 years, 14 years or older, or unknown), age at menopause (younger than 45 years, from 45 years to younger than 50 years, from 50 years to younger than 54 years, 54 years or older, or unknown), number of live births (nulliparous, 1-2 births, ≥3 births, or unknown), age at first live birth (nulliparous, younger than 20 years, 20-24 years, 25-29 years, 30 or older years, or unknown), oral contraceptive use (ever, never, or unknown), family history of breast cancer in mother or sister (yes or no), personal history of breast cysts (yes or no), screening mammography (none, recent [within the last 2-3 years], not recent, or unknown) as a time-dependent covariate, height (<63 inches, from 63 to <65 inches, from 65 to <67 inches, or ≥67 inches) education (less than high school graduate, high school graduate or unknown, some college, college graduate), recreational physical activity (at baseline) as assessed by metabolic equivalents of energy expenditure units (METS) for exercise (<7 MET-hours per week, from 7 to <15 MET-hours per week, from 15 to <25.5 MET-hours per week, ≥25.5 MET-hours per week, or unknown),21 alcohol use (none, <1 drink per day, 1 drink per day, >1 drink per day, or unknown), aspirin use (not currently using, current use <15 days per month, current use ≥15 days per month, current use of unknown frequency, or unknown), and race (white, black, or other). Baseline MET-hours per week of recreational physical activity were calculated for each participant by multiplying the hours spent engaged in various activities (walking, jogging/running, swimming, tennis or racquetball, bicycling or stationary biking, aerobics, calisthenics, and dancing) times the MET score estimated for each activity according to the Compendium of Physical Activities.22 Tests of linear trend for increasing weight gain were computed; all reported P values are 2-sided.

Women were censored during follow-up for death, report of breast cancer that could not be verified or verified diagnosis of in situ breast cancer, report of another cancer, interval postmenopausal hormone therapy use, weight loss (>5 pounds since age 18 years), or extreme weight gain (>70 pounds in a 2-year interval.)

Finally, to assess whether the frequency of mammographic screening influenced our results, we conducted a sensitivity analysis in which we reran all models described above for histology, stage, grade, and receptor status and included only women who reported regular mammographic screening. We excluded women who had not had a mammogram within 3 years prior to baseline (n = 7477 women; 330 diagnoses of breast cancer) and censored women who did not report having mammograms every 2 to 3 years throughout follow-up (n = 4832 women [74 diagnoses] in 1997 and n = 4401 women [92 diagnoses] in 1999).

RESULTS

This analysis included 1200 cases of incident breast cancer among 44,161 postmenopausal women who were cancer-free at baseline in 1992 and who were not taking postmenopausal hormone therapy. Most participants were white (97%), middle-aged or elderly (mean age, 64 years), and educated (28% college graduates). Seventy-nine percent reported having a natural menopause, and 18%, including 199 women who were diagnosed with breast cancer, were former users of hormone therapy at baseline. The median duration of use among former hormone therapy users was 2 years. Other lifestyle and demographic characteristics of the CPS-II Nutrition Cohort are provided elsewhere.18, 19

Table 1 shows the percent of patients who were diagnosed with breast cancer in each histopathologic category of interest. Regardless of histologic type, most tumors were localized and were ER-positive/PR-positive. Seventy-five percent of ductal breast cancers were localized, 45% were Grade 2, and 67% (of those with known receptor status) were ER-positive/PR-positive; 75% of lobular and mixed tumors were localized, 51% were Grade 2, and 77% were ER-positive/PR-positive. Tumors in the “other” histology category were more likely to be Grade 3 (45%). ER/PR status was missing or was classified as “borderline” or “equivocal” for 46% of breast cancers. Availability of ER or PR information was not associated with socioeconomic or demographic variables (data not shown).

Table 1. Distribution of Tumor Classifications by Histologic Type
 No. of Cases (%)*
VariableTotal No.Ductal (n = 865)Lobular (n = 208)Other Histology (n = 127)
  • ER indicates estrogen receptor; +, positive; PR, progesterone receptor; -, negative.

  • *

    Percentages do not include missing data.

  • Lobular includes lobular carcinoma (International Classification of Diseases for Oncology [ICD-O] code 8520;, n = 141 women) and mixed lobular and ductal carcinoma (ICD-O code 8522; n = 67 women).

  • ER+PR-, n = 95 women; ER-PR+, n = 13 women.

Disease stage    
 Localized877635 (74.8%)155 (75.2)87 (73.7)
 Regional/distant296214 (25.2%)51 (24.8)31 (26.3)
 Stage unknown271629
Hormone receptor status    
 ER+PR+445311 (66.7)100 (76.9)34 (61.8)
 ER-PR-9881 (17.4)5 (3.9)12 (21.8)
 ER+PR- or ER-PR+10874 (15.9)25 (19.2)9 (16.4)
 ER or PR unknown5493997872
Tumor grade    
 Grade 1184147 (21.1)19 (17.9)18 (27.7)
 Grade 2387315 (45.1)54 (50.9)18 (27.7)
 Grade 3298236 (33.8)33 (31.1)29 (44.6)
 Grade unknown33116710262

Table 2 shows age-standardized rates and rate ratios (RR) for incident breast cancer by adult weight gain stratified by histology. Age-standardized rates of breast cancer were highest for women who reported the most weight gain compared with women who had the least weigh gain, regardless of histologic type. The RR of breast cancer increased with a statistically significant trend for ductal and “other” histologic types. For lobular and mixed carcinomas, the increase in risk was limited to women who gained >60 pounds since age 18 years and was of borderline statistical significance (RR, 1.54; 95% confidence interval [95%CI], 1.01-2.33) for weight gain >60 pounds compared with weight gain ≤20 pounds (P for trend = .06). These results did not change appreciably when tumors with lobular (ICD code 8520) and mixed lobular and ductal (ICD code 8522) histology were analyzed separately, but the estimates of risk were less precise as a result of the smaller sample size. The association with increasing weight gain was strongest among tumors in the “other” histology group, although this estimate was based on relatively few women.

Table 2. Rate Ratios of Breast Cancer Incidence Stratified by Histology According to Weight Gain in the Cancer Prevention Study-II Nutrition Cohort, 1992-2001
 Adult Weight Gain from Age 18 Years to 1992
Histology>5 to 20 Pounds (n = 327)21-40 Pounds (n = 385)41-60 Pounds (n = 269)≥61 Pounds (n = 219)P
  • RR indicates rate ratio; 95% CI, 95% confidence interval.

  • *

    Rates (per 100,000) were standardized to the age distribution of the Nutrition Cohort.

  • RRs were adjusted for age at interview.

  • RRs were adjusted for age at interview, age at menarche, age at menopause, number of live births, age at first live birth, oral contraceptive use, family history of breast cancer, history of breast cysts, history of mammography, height, education, physical activity, alcohol use, aspirin use, and race.

Ductal     
 No. of cases234278205148 
 Age-standardized rate*222.4273.6334.9348.6 
 Age-adjusted RR (95% CI)1.001.27 (1.07-1.51)1.61 (1.33-1.94)1.81 (1.47-2.22)<.0001
 Multivariate RR (95% CI)1.001.28 (1.08-1.53)1.65 (1.37-2.00)1.89 (1.53-2.34)<.0001
Lobular and mixed lobular/ductal     
 No. of cases69634036 
 Age-standardized rate*63.361.663.688.2 
 Age-adjusted RR (95% CI)1.000.98 (0.69-1.37)1.06 (0.72-1.57)1.50 (1.00-2.24).07
 Multivariate RR (95% CI)1.000.99 (0.70-1.39)1.08 (0.73-1.61)1.54 (1.01-2.33).06
Other histology     
 No. of cases24442435 
 Age-standardized rate*22.743.440.284.6 
 Age-adjusted RR (95% CI)1.001.97 (1.20-3.23)1.85 (1.05-3.25)4.25 (2.52-7.17)<.0001
 Multivariate RR (95% CI)1.002.09 (1.27-3.45)2.00 (1.13-3.55)4.67 (2.72-8.01)<.0001

Table 3 shows the age-standardized rates and RRs for breast cancer by total adult weight gain stratified separately by stage and grade. Like histologic type, increasing adult weight gain was associated with increased risk for each stage and grade. There was a statistically significant, elevated risk for regional or distant stage in every category of weight gain, and the risk was 3 times higher among women who gained >60 pounds since age 18 years compared with women who gained ≤20 pounds (RR, 3.15; 95%CI, 2.21-4.48; P for trend, <.0001). The increased risk associated with weight gain was less pronounced for localized tumors (RR, 1.68; 95%CI, 1.36-2.08 for weight gain >60 pounds compared with ≤20 pounds; P for trend, <.0001). The association with weight gain was similar across tumor grade, although it was stronger for Grade 3 tumors (RR, 2.84; 95%CI, 1.99-4.06; P for trend, <.0001) than for Grade 1 tumors (RR, 2.17; 95%CI, 1.37-3.44; P for trend = .0006) or Grade 2 tumors (RR, 1.67; 95%CI, 1.22-2.29; P for trend, <.0001) when women who had the highest weight gain were compared with women who had the lowest weight gain.

Table 3. Rate Ratios of Breast Cancer Incidence Stratified by Stage and Grade at Diagnosis According to Weight Gain in the Cancer Prevention Study-II Nutrition Cohort, 1992-2001
 Adult Weight Gain from Age 18 Years to 1992*
Stage/Grade>5 to 20 Pounds (n = 327)21-40 Pounds (n = 385)41-60 Pounds (n = 269)≥61 Pounds (n = 219)P
  • RR indicates rate ratio; 95% CI, 95% confidence interval.

  • *

    Columns that do not add up to the total reflect missing data.

  • Rates (per 100,000) were standardized to the age distribution of the Nutrition Cohort.

  • RRs were adjusted for age at interview.

  • §

    RRs were adjusted for age at interview, age at menarche, age at menopause, number of live births, age at first live birth, oral contraceptive use, family history of breast cancer, history of breast cysts, history of mammography, height, education, physical activity, alcohol use, aspirin use, and race.

Localized disease     
 No. of cases260275199143 
 Age-standardized rate244.0267.3324.6341.9 
 Age-adjusted RR (95% CI)1.001.13 (0.96-1.34)1.41 (1.17-1.70)1.59 (1.30-1.95)<.0001
 Multivariate RR (95% CI)§1.001.15 (0.97-1.37)1.46 (1.21-1.76)1.68 (1.36-2.08)<.0001
Regional/distant disease     
 No. of cases63996470 
 Age-standardized rate60.198.7104.6165.2 
 Age-adjusted RR (95% CI)1.001.66 (1.21-2.28)1.84 (1.30-2.61)3.10 (2.20-4.36)<.0001
 Multivariate RR (95% CI)§1.001.68 (1.22-2.31)1.89 (1.33-2.69)3.15 (2.21-4.48)<.0001
Grade 1     
 No. of cases49634032 
 Age-standardized rate45.161.265.380.7 
 Age-adjusted RR (95% CI)1.001.40 (0.96-2.03)1.56 (1.02-2.36)1.94 (1.24-3.04).002
 Multivariate RR (95% CI)§1.001.45 (0.99-2.10)1.67 (1.09-2.55)2.17 (1.37-3.44).0006
Grade 2     
 No. of cases1091179566 
 Age-standardized rate103.6117.3151.1159.3 
 Age-adjusted RR (95% CI)1.001.14 (0.88-1.48)1.58 (1.20-2.08)1.70 (1.25-2.31)<.0001
 Multivariate RR (95% CI)§1.001.13 (0.87-1.47)1.56 (1.18-2.06)1.67 (1.22-2.29)<.0001
Grade 3     
 No. of cases681056263 
 Age-standardized rate62.4103.7104.9147.0 
 Age-adjusted RR (95% CI)1.001.64 (1.21-2.22)1.64 (1.16-2.31)2.58 (1.83-3.64)<.0001
 Multivariate RR (95% CI)§1.001.69 (1.25-2.30)1.77 (1.25-2.51)2.84 (1.99-4.06)<.0001

The age-standardized rates and RRs for ER and PR status by total adult weight gain are shown in Table 4. Among women with ER-positive/PR-positive tumors, increasing weight gain was associated with a statistically significant, increased risk of breast cancer (P for trend, <.0001). Even women who gained from 21 to 40 pounds after age 18 years were at increased risk (RR, 1.32; 95%CI, 1.03-1.71). This risk more than doubled among women who reported weight gains of >60 pounds (RR, 2.42; 95%CI, 1.82-3.23). The risk of ER-negative/PR-negative tumors and of ER-positive/PR-negative or ER-negative/PR-positive tumors was elevated among women who had the highest weight gain, but these increases were not statistically significant. Because data on ER and/or PR status were not available for approximately 50% of the women who were diagnosed with breast cancer, the results for women with missing ER/PR status are shown and resemble those for the women who had known ER-positive/PR-positive status. This likely reflects the fact that >75% of the tumors detected among white women in the United States are ER-positive at diagnosis.23, 24 Our results were similar when we considered ER and PR status separately, although there was some suggestion that ER-positive status had a stronger association with weight gain than PR-positive status (data not shown). Most of our PR-positive tumors also were ER-positive (97%).

Table 4. Rate Ratios of Breast Cancer Incidence Stratified by Estrogen Receptor/Progesterone Receptor Status According to Weight Gain in the Cancer Prevention Study-II Nutrition Cohort, 1992-2001
 Adult Weight Gain from Age 18 Years to 1992*
ER/PR Status>5 to 20 Pounds (n = 327)21-40 Pounds (n = 385)41-60 Pounds (n = 269)≥61 Pounds (n = 219)P
  • ER indicates estrogen receptor; PR, progesterone receptor; +, positive; -, negative; RR, rate ratio; 95% CI, 95% confidence interval.

  • *

    Rates (per 100,000) were standardized to the age distribution of the Nutrition Cohort.

  • RRs were adjusted for age at interview.

  • RRs were adjusted for age at interview, age at menarche, age at menopause, number of live births, age at first live birth, oral contraceptive use, family history of breast cancer, history of breast cysts, history of mammography, height, education, physical activity, alcohol use, aspirin use, and race.

ER+PR+     
 No. of cases11113511089 
 Age-standardized rate*104.2132.4181.2214.0 
 Age-adjusted RR (95% CI)1.001.30 (1.01-1.67)1.82 (1.40-2.37)2.31 (1.74-3.05)<.0001
 Multivariate RR (95% CI)1.001.32 (1.03-1.71)1.88 (1.44-2.45)2.42 (1.82-3.23)<.0001
ER-PR-     
 No. cases31311818 
 Age-standardized rate*28.530.431.242.7 
 Age-adjusted RR (95% CI)1.001.06 (0.64-1.74)1.04 (0.58-1.85)1.55 (0.86-2.78).20
 Multivariate RR (95% CI)1.001.10 (0.67-1.81)1.14 (0.63-2.06)1.78 (0.98-3.23).09
ER+PR- or ER-PR+     
 No. of cases32421915 
 Age-standardized rate*28.240.330.238.6 
 Age-adjusted RR (95% CI)1.001.39 (0.88-2.21)1.07 (0.61-1.90)1.31 (0.71-2.43).52
 Multivariate RR (95% CI)1.001.39 (0.87-2.20)1.08 (0.61-1.93)1.32 (0.70-2.49).52
ER or PR unknown     
 No. of cases15317712297 
 Age-standardized rate*148.2176.9197.8227.1 
 Age-adjusted RR (95% CI)1.001.24 (1.00-1.54)1.48 (1.17-1.88)1.84 (1.43-2.38)<.0001
 Multivariate RR (95% CI)1.001.26 (1.01-1.57)1.53 (1.20-1.94)1.91 (1.47-2.48)<.0001

Results from the sensitivity analysis revealed no appreciable changes by grade, histology, or ER/PR status between the subcohort of women who reported having a screening mammogram at regular intervals compared with the entire study population. For example, among women in the highest category of weight gain, the RR for ductal tumors was 1.88 (95%CI, 1.45-2.45) in the screened subcohort compared with 1.89 (95%CI, 1.53-2.34) in the full cohort. For Grade 1 tumors, the RR was 2.01 (95%CI, 1.16-3.51) in the screened subcohort compared with 2.17 (95%CI, 1.37-3.44) in the full cohort for ≥61 pounds of weight gain. Similarly, for Grade 3 tumors, the RR was 3.01 (95%CI, 1.96-4.61) in the screened subcohort compared with 2.84 (95%CI, 1.99-4.06) in the full cohort for ≥61 pounds of weight gain. In the screened subcohort, among women with ≥61 pounds of weight gain, the RR for ER-positive/PR-positive tumors was 2.50 (95%CI, 1.76-3.54) compared with 2.42 (95%CI, 1.82-3.23) in the full cohort, and the RRs were identical in this category for ER-negative/PR-negative tumors.

Results from the sensitivity analysis for disease stage are shown in Table 5. Although there was no difference in the distribution of tumors by stage (25% of tumors were regional or distant in the entire cohort, compared with 22% in the regularly screened subcohort), the association with adult weight gain and regional or distant stage was somewhat stronger in each category of weight gain among the regularly screened group. In the highest category of weight gain (>60 pounds), the RR was 3.92 (95%CI, 2.49-6.17) compared with women who gained ≤20 pounds. The results for localized disease did not change among those who had regular mammographic examinations.

Table 5. Rate Ratios of Breast Cancer Incidence among Women with Regular Mammographic Screening by Stage at Diagnosis according to Weight Gain in the Cancer Prevention Study-II Nutrition Cohort, 1992-2001
 Adult Weight Gain from Age 18 Years to 1992*
 Stage at Diagnosis>5 to 20 Pounds (n = 218)21-40 Pounds (n = 265)41-60 Pounds (n= 186)≥61 Pounds (n = 147)P
  • RR indicates rate ratio; 95% CI, 95% confidence interval.

  • *

    Columns that do not add up to the total reflect missing data.

  • RRs were adjusted for age at interview.

  • RRs were adjusted for age at interview, age at menarche, age at menopause, number of live births, age at first live birth, oral contraceptive use, family history of breast cancer, history of breast cysts, history of mammography, height, education, physical activity, alcohol use, and race.

Localized     
 No. of cases18219714696 
 Age-adjusted RR (95% CI)1.001.17 (0.95-1.43)1.52 (1.22-1.89)1.61 (1.26-2.07)<.001
 Multivariate RR (95% CI)1.001.17 (0.96-1.44)1.54 (1.23-1.92)1.64 (1.27-2.11)<.001
Regional/distant disease     
 No. of cases35643747 
 Age-adjusted RR (95% CI)1.001.94 (1.28-2.93)1.96 (1.23-3.11)3.91 (2.52-6.07)<.001
 Multivariate RR (95% CI)1.001.95 (1.29-2.94)1.95 (1.22-3.12)3.92 (2.49-6.17)<.001

DISCUSSION

To our knowledge, this is the first study to examine the association between adult weight gain and histopathologic characteristics of breast tumors among postmenopausal women who were not taking postmenopausal hormone therapy. In the current analysis, we show that excess adiposity is an important determinant of breast cancer risk regardless of stage, grade, or histologic type, but the magnitude of the effect varies according to these factors. Our findings are especially relevant to current medical practice and public health issues given the recent decline in postmenopausal hormone use25, 26 and the increasing prevalence of overweight and obesity in the United States and other Westernized countries.27, 28 The early termination of the Women's Health Initiative estrogen plus progestin trial led to a sharp decline in postmenopausal hormone use.25, 26 During 1999 to 2002, 68% of U.S. women age 60 or older were overweight or obese (BMI≥25 kg/m2).28 Thus, our results pertain directly to what is likely to be an increasing segment of the population.

Weight gain was associated with all histologic types of breast cancer, although the association of lobular and mixed tumors with weight gain was weaker (RR, 1.54; 95%CI, 1.01-2.33 for weight gain >60 pounds vs. weight gain ≤20 pounds) than was seen with ductal or “other” histologies. Few studies specifically have examined risk factors for lobular breast carcinomas,13, 29, 30 and only 1 study included any measure of weight or adiposity.13 Li et al.13 reported a nonsignificant elevation in the risk of lobular breast cancer for a weight ≥175 pounds (odds ratio, 1.3; 95%CI, 0.8-2.0) but reported no association with BMI or maximum weight. Both the incidence of lobular breast carcinomas31 and the prevalence of obesity32 have been increasing in the United States in recent years. The relation between weight gain and lobular carcinomas deserves attention in future studies.

High weight gain was associated more strongly with regional/distant stage than with localized tumors and more strongly with Grade 3 tumors than with Grade 1 or 2 tumors. Women who gained >60 pounds during adulthood were 3 times more likely to have regional or distant metastatic disease compared with women who gained ≤20 pounds during adulthood (RR, 3.15; 95%CI, 2.21-4.48). Although no previous study has examined adult weight gain specifically, our results are consistent with 3 of the 4 previous studies that have examined stage by other measures of adiposity, including BMI and waist:hip measurements.14–16 All of those studies found a statistically significant, increased risk of American Joint Committee on Cancer Stage ≥II compared with Stage 0 or I among women in the highest category of BMI,14–16 with high waist:hip ratio in 1 study16 and with large breast size measured by bra cup size (D vs. A) in another.15 Our results on stage differ from those recently published by MacInnis et al.,12 who examined several measures of adiposity among 357 women with breast cancer and observed no overall differences by stage; however, in a subgroup of women who were ≥15 years past menopause, those authors observed that the association with body size was stronger for Grade 2 tumors (n = 59 women) and Grade 3 tumors (n = 44 women) compared with Grade 1 tumors (n = 36 women). Although MacInnis et al.12 did not limit their analysis to women who were not using postmenopausal hormone therapy, they did report stronger associations with measures of adiposity among women who were not using postmenopausal hormone therapy. It is possible that the use of postmenopausal hormone therapy may have masked the association with stage in their small study.

It has been suggested that heavier women may be diagnosed with breast cancer at a later stage, either because tumors are more difficult to detect in a large breast or because heavier women are screened less frequently than thin women.33 However, when we restricted our analysis to women who had screening mammograms at regular intervals, we found no appreciable differences in our risk estimates. Our findings are in agreement with more recent studies,15, 16, 34 including a study that reviewed >100,000 mammography examinations and found no differences in mammography sensitivity by categories of BMI.34 This suggests that biologic differences, rather than delayed detection, likely are responsible for the association of more advanced stage with increasing weight gain.

Numerous studies have examined whether the risk of postmenopausal breast cancer associated with obesity differs according to ER and PR status.4–12 In most of those studies,4, 6–8, 10–12 excess adiposity, whether measured by BMI or other measures of body fat, was associated more clearly with ER-positive or ER-positive/PR-positive tumors rather than with ER-negative or PR-negative tumors. Our results were consistent with those observations. We observed a statistically significant increased risk of ER-positive/PR-positive tumors at each category of adult weight gain, beginning with 21 to 40 pounds (RR, 1.32; 95%CI, 1.03-1.71; P for trend, <.0001). The risk more than doubled for ER-positive/PR-positive tumors among women with >60 pounds of weight gain (RR, 2.42; 95%CI, 1.82-3.23). Only women in this highest category showed an increased risk of ER-negative/PR-negative breast cancer, and the association was not statistically significant (RR, 1.78; 95%CI, 0.98-3.23; P for trend = .09). Colditz et al.11 suggested that PR expression, but not ER expression, may be associated independently with BMI. We did not observe this association in the current study.

The primary strength of this study is its large, prospective design, which gives us the ability to examine, with adequate statistical power, the association between weight gain and several histopathologic characteristics of breast tumors among women who are not using postmenopausal hormone therapy. Several limitations of this study should be acknowledged. One limitation of this study, as mentioned above, is incomplete data on receptor status. However, these missing data do not appear to be related to any sociodemographic or risk factor characteristics that would introduce bias into this study. Another limitation of our study is the potential for measurement error from self-reported weight and recall of weight at age 18 years. These errors should have little impact on our results, because correlations between measured and self-reported weight typically are >0.95,1 and correlations between recalled weight over several decades and measured weight have been between 0.80 and 0.87 in several studies.35 A recent article by Spencer et al.36 assessed the validity of self-reported height and weight among 4808 participants in the European Prospective Investigation into Cancer (EPIC) study and found that women tended to underestimate weight by an average of 1.40 kg (range, 1.31-1.49 kg), or approximately 3.1 pounds, with a tendency for heavier women to underestimate weight more than lean women. Another limitation is our lack of serial measures of weight from several decades in adulthood, which would have allowed us to examine other patterns of weight change and would have strengthened this study. Finally, it should be noted that these results are from a cohort of predominantly white, middle-aged or elderly, educated women, and they may not be generalizable to other populations of interest.

In conclusion, our data indicate that adult weight gain influences most histopathologic characteristics of breast tumors among women who are not using postmenopausal hormone therapy. The risk of breast cancer increased with increasing weight gain regardless of tumor stage, grade, or histologic type but was stronger for higher stage and grade. Women with the highest amounts of adult weight gain were 3 times more likely to have regional or distant metastatic disease than women who gained ≤20 pounds in adulthood. Weight gain does not appear to be associated strongly with ER-negative/PR-negative tumors, as shown previously. These associations persist even when the analyses are restricted to women who have screening mammograms at regular intervals, suggesting that the increased risk among heavy postmenopausal women is not due to detection bias. Adult weight gain is one of the few well established risk factors for breast cancer that is modifiable. These data further illustrate the relation between adult weight gain and breast cancer and the importance of maintaining a healthy body weight throughout adulthood.

Ancillary