Body mass and stage of breast cancer at diagnosis

Abstract

Obesity is a well-known risk factor for postmenopausal breast cancer. In contrast, the relationship between obesity and stage of breast cancer at diagnosis is less clear. We hypothesized that increased breast size in obese women may delay discovery of breast tumors. Thus, the purpose of our study was to examine whether there is an association between body mass and stage of breast cancer at diagnosis using hospital medical records. Newly diagnosed breast cancer cases (n = 966) in the Baltimore metropolitan area from 1991 to 1997 were included in our study. Patient information including age, ethnicity, weight, height and pathology data were obtained from hospital medical records. High body mass was significantly associated with late stage of breast cancer at diagnosis. Women who were obese (body mass index [BMI] ≥ 27.3) were more likely to be at an advanced stage at diagnosis compared with women with a BMI of < 27.3 (multivariate-adjusted odds ratio [OR] 1.57, 95% confidence interval [CI] 1.15–2.14). The association between body mass and stage at diagnosis was stronger among women younger than 50 years (OR 2.34, 95% CI 1.34–4.08) compared with women 50 years or older (OR 1.30, 95% CI 0.89–1.91). Our study suggests that higher body mass is associated with advanced stage of breast cancer at diagnosis. This finding may be of considerable concern, given the increasing prevalence of obesity in women in the United States and the poor prognosis associated with late-stage tumors. © 2001 Wiley-Liss, Inc.

Obesity is a well-known risk factor for breast cancer, with the effect being limited to postmenopausal women.1–9 In premenopausal women, there is an inverse association between body mass and the risk of breast cancer, i.e., obese women are less likely to get breast cancer than nonobese women. In contrast, the relationship between obesity and stage at diagnosis in either premenopausal or postmenopausal women is less clear. This is partially because this relationship has not been studied in detail and the results from existing studies are equivocal. For example, several10–20 but not all21, 22 studies report that being overweight is associated with late stage at diagnosis. In addition, some studies report that both large tumor size and lymph node involvement are associated with higher body weight,16, 23, 24 whereas other studies report that tumor size, but not lymph node involvement are associated with higher body weight.17, 25–28 Furthermore, 1 study has reported that lymph node involvement, but not tumor size, is associated with obesity.29

In obese women with larger breasts, palpation for lumps may be more difficult, resulting in delay in detection and diagnosis of breast cancer. This idea is supported by studies showing that women with large breasts (or bra size as a surrogate) are more likely to be diagnosed at a later stage of breast cancer than women with small breasts.18, 28, 30 A study by Reeves et al.11 reported that the positive association between body mass index (BMI) and stage was confined to tumors that were detected by women themselves. In a recent study, however, Hall et al.18 reported that the observed positive association was not modified by method of detection.

The purpose of our study was to help clarify the relationship between body mass and breast cancer by determining whether there is an association between obesity and stage of breast cancer at diagnosis. We also wished to determine whether there is a positive association between obesity and each of the 3 components for tumor staging (tumor size, node status and distant metastasis) and to explore whether the association between obesity and tumor stage was modified by age.

MATERIAL AND METHODS

Study subjects

This is a case series study of breast cancer using data from hospital medical records. Breast cancer cases were obtained from 2 hospitals in the Baltimore metropolitan area. Although all recorded cases of breast cancer were obtained by the investigators, the study subjects were limited to all newly diagnosed breast cancer patients aged 20–85 years, with diagnosis between 1991 and 1997. Date and place of final diagnosis were ascertained from medical charts and used to determine eligible cases. For newly diagnosed patients with breast cancer, exclusion criteria included: (i) previous malignancy (cancer at another anatomic site or breast cancer in the contralateral breast); (ii) age outside the 20–85-year range; and (iii) ethnic groups other than white or black because of too few numbers. Subjects with names on the hospital record list but whose charts were unavailable were also excluded from the study (n = 29). Subjects with bilateral breast cancer at the same time were counted as 1 subject. The study protocol was approved by the University of Maryland Interview Review Board (IRB) and the IRB of the collaborating hospital.

Variables

Information about the variables of interest was abstracted from the patients' hospital records. Body weight and height recorded at the time of diagnosis were used to compute BMI (weight [kg]/height [m²]). If there was more than 1 record of body weight and height, the earliest record was abstracted from the chart. In order to explore the dose-response relationship between body mass and breast cancer stage at diagnosis, BMI was treated as a continuous variable. BMI was further categorized in stratified analysis in which overweight was defined as BMI of 25–29.9 and obesity was defined as BMI of 30 and above. These definitions were based on the first federal guidelines that were released by National Institutes of Health in 1998 for the identification, evaluation and treatment of overweight and obese adults.31 These definitions are comparable with those used by the World Health Organization. In some stratified analysis, obesity also was defined as a BMI ≥ 27.3. This value corresponds to the 85^th percentile of the body mass distribution in U.S. women aged 20–29 years and is used by the National Center for Health Statistics to classify “overweight” adult females.32

We used the tumor-node-metastasis (TNM) classification of stage of breast cancer at diagnosis as established by the American Joint Committee on Breast Cancer as the outcome of primary interest.33–35 This staging system consists of 3 components: (i) tumor size (T); (ii) absence or presence and extent of regional lymph node metastasis (N); and (iii) absence or presence of distant metastasis (M). There are 5 TNM stages ranging from stage 0 (in situ), to stage IV.

In most cases (94%), information about tumor characteristics such as tumor size, lymph node status and summary TNM staging was available in pathology reports from biopsy, lumpectomy or mastectomy specimens. Pathologic findings as well as results from physical and radiologic examination for determination of distant metastasis were cited and summarized by the physicians.

Only small proportions of women were diagnosed at either stage III (10%) or stage IV (6%). Therefore, a dichotomous variable of TNM stage was used in analysis: stage 0/I versus stage II and above. An ordinal variable for stage (from 0 to IV, with III and IV combined due to small numbers) was also used as the outcome variable in polychotomous ordinal logistic regression (similar results obtained, data not shown).

Other variables in the analysis included age at time of diagnosis, ethnicity, marital status and health insurance coverage. Income and education were determined by using information on residence and socioeconomic status (income and education levels) from the census tracts or the ZIP code of areas in which patients resided according to the U.S. Bureau of the Census in 1990.36–41 The use of proxy socioeconomic information based on census data at the census tract level has been demonstrated to be a valid approach for categorizing socioeconomic status of research subjects.42 For about 15% of the study population, residence was either incorrect or outside the state of Maryland. For these subjects, ZIP codes instead of residence addresses were used to obtain ZIP code-based socioeconomic status information from the 1990 U.S. census data.

Information on other potential risk factors for breast cancer such as menstrual history, reproductive history, hormone use history and lifestyle factors (e.g., physical activity) were not routinely recorded in medical charts. Therefore, this information could not be included in the data analysis.

Data collection procedures

All medical records were reviewed and abstracted by 1 investigator (Y.C.). To test for reliability, another investigator performed the abstraction on a sample of medical records (10%). Data on main variables from these 2 sources were found to be consistent. In addition, data for main variables (with the exception of body weight and height) were obtained from the hospital tumor registry office for 1 hospital. High consistency was found when comparing tumor characteristics and treatment procedures from this source with our data. In the few instances in which inconsistencies were found, consultation was made with an oncologist for correct diagnosis and tumor staging.

Data analysis

The relationship of BMI to stage at diagnosis was evaluated with: (i) Student's t-tests for comparison of BMI between women with early stages and women with advanced stages of breast cancer; (ii) Chi square tests for frequency tables of BMI by stage, with BMI either having 3 categories (<25, 25–29.9, ≥30) or 2 categories (<27.3, ≥27.3); and (iii) logistic regression analysis using the maximum likelihood method. Odds ratios (ORs) and 95% confidence intervals (95% CIs) adjusted for age and ethnicity only or adjusted for age, ethnicity, income, education and marital status were calculated and reported.

To examine whether the association between body mass and stage of breast cancer at diagnosis differed by age, stratified analyses were carried out among women <50 years and women ≥50 years. Additional analyses were performed to assess whether the body mass-stage association differed by ethnicity, income, education, marital status or type of health insurance. If these variables were not found to be effect modifiers, they were examined as potential confounders. All analyses were repeated with each of the additional components of TNM (tumor size, lymph node status and distant metastasis) as the outcome variable.

RESULTS

Descriptive data on the study subjects are shown in Table I. A total of 966 eligible cases were included in the data analysis. Of the entire study sample, the mean age was 56 years (age range 28–85 years). A total of 585 subjects (60%) were white women, and 381 subjects (40%) were black women. Half of the study subjects were not married. About one-third of the subjects had Medicare insurance, and a few subjects (10%) had Medicaid or had no insurance.

Table I. Characteristics of Study Subjects (n = 966)

Characteristic	No.1	%
1 Numbers for each characteristic may not sum to total because of missing data.
Age (yr), mean (SD)	56.5 (13.4)
<50	344	25.6
≥50	622	74.4
Ethnicity
White	585	60.6
Black	381	39.4
Marital status
Married	434	49.7
Not married	439	50.3
Health insurance
Medicare	270	31.8
Medicaid	71	8.4
Commercial	496	58.4
No insurance	13	1.5
Body weight (kg), mean (SD)	74.5 (17.7)
Body mass index (kg/m^2), mean (SD)	28.2 (6.4)
<27.3	423	50.7
≥27.3	411	49.3
TNM stage
0	129	14.2
I	330	36.3
II	304	33.4
III	90	9.9
IV	56	6.2

Information on body weight was available for 92% of study subjects. The average body weight was 74.5 kg. Information on BMI (weight and height) was available for 86% of study subjects. The average BMI was 28.2, and the median was 27.1. Almost half of the subjects had a BMI ≥ 27.3, and about one-third had a BMI ≥ 30.

With regard to tumor characteristics of our study population, about 14% of the subjects were diagnosed at stage 0 (carcinoma in situ), 36% were diagnosed at stage I, and almost 50% were diagnosed at stage II and above. More tumors were diagnosed at an early stage in women over 50 years compared with women younger than 50 years. In women 50 years or older, more than half had stage 0/I at diagnosis, whereas in women younger than 40 years, only 29% were diagnosed at stage 0/I.

Women with the highest BMI levels were more likely to be at a later stage of breast cancer at diagnosis than women with a lower BMI. Table II shows that the percentage of tumors diagnosed at stage II and later stages was greater for overweight women (BMI of 25–29.9) and obese women (BMI ≥ 30) than for women with a normal BMI (i.e., 51% in obese women, 50% in overweight women and 41% in normal weight women). Both overweight (BMI 25–29.9) and obese (BMI ≥ 30) women had a greater odds of diagnosis with cancer at TNM stage II or higher compared with those with BMI < 25. For instance, women with a BMI of 25–29.9 had an OR of 1.44 (95% CI 1.02–2.03), and women with a BMI of ≥30 had an OR of 1.54 (95% CI 1.10–2.16). When the cut point of 27.3 was chosen to define obesity, obese women were 1.5 times more likely to have a later stage of diagnosis than women with a BMI < 27.3 (95% CI 1.13–1.98; Table II). This trend remained after adjustment for age, ethnicity, income, education and marital status.

Table II. OR and 95% CI for Being at Stage II or Above At Diagnosis in Overweight/Obese Women Compared With Normal Weight Women

Body mass index	Stage 0/I		Stage ≥ II		Unadjusted OR (95% CI)	Adjusted OR (95% CI)1
	No.	%	No.	%
1 ORs comparing late stage (TNM ≥ II) with early stage (TNM 0/I), adjusted for age, ethnicity, income, education and marital status. 2 p = 0.03, from Student's t-test.
Mean (SD)	27.5 (6.1)2		28.8 (6.5)2
<27.3	236	57.8	172	42.2	1.0	1.0
≥27.3	184	47.8	201	52.2	1.50 (1.13–1.98)	1.57 (1.15–2.14)
<25	165	59.4	113	40.6	1.0	1.0
25–29.9	124	50.4	122	49.6	1.44 (1.02–2.03)	1.44 (0.98–2.10)
≥30	131	48.7	138	51.3	1.54 (1.10–2.16)	1.53 (1.05–2.23)

In further analyses, a high body mass was positively associated with tumor size. Women with a BMI of ≥27.3 had 1.5 times the odds of having a tumor > 2 cm compared with women with a lower BMI (adjusted OR 1.57, 95% CI, 1.12–2.17; Table III). In contrast, the associations between BMI and the 2 other TNM components, axillary node involvement (adjusted OR 1.28, 95% CI 0.90–1.81) and distant metastasis (adjusted OR 1.60, 95% CI 0.72–3.48), were not statistically significant.

Table III. Association Between Body Mass Index (BMI) and Tumor Characteristics

	BMI < 27.3		BMI ≥ 27.3		Unadjusted OR1 (95% CI)	Adjusted OR (95% CI)^1,2
	No.	%	No.	%
1 ORs > 1 indicates obese women were more likely to have less favorable tumor characteristics compared with nonobese women. ^2ORs adjusted for age, ethnicity, income, education and marital status.
Tumor size
Ca in situ, ≤2 cm	273	67.2	225	58.9	1.0	1.0
>2 cm	133	32.8	157	41.1	1.43 (1.07–1.92)	1.57 (1.12–2.17)
Lymph node status
Negative	262	70.2	233	65.6	1.0	1.0
Positive	111	29.8	122	34.4	1.24 (0.90–1.69)	1.28 (0.90–1.81)
Distant metastasis
No	398	96.8	370	94.4	1.0	1.0
Yes	13	3.2	22	5.6	1.82 (0.90–3.67)	1.60 (0.73–3.48)

The association of body mass with advanced stages was examined by age. When the BMI cut point of 27.3 for obesity was used, the positive association was evident in women <50 years (OR 2.34, 95% CI 1.34–4.08). In women aged 50 years or older, however, this positive association was not significant (OR 1.30, 95% CI 0.89–1.91; Table IV). Furthermore, statistical testing showed that the strength of body mass-stage association differed by age (p = 0.03).

Table IV. Association Between Body Mass Index (BMI) and Stage of Breast Cancer at Diagnosis by Age

BMI	Stage 0/I		Stage ≥ II		OR (95% CI)1
	No.	%	No.	%
1 Adjusted for age, ethnicity, income, education and marital status.
Age: <50 yr
<27.3	84	51.5	79	48.5	1.0
≥27.3	37	30.6	84	69.4	2.34 (1.34–4.08)
Age: ≥50 yr
<27.3	152	62.0	93	38.0	1.0
≥27.3	147	55.7	117	44.3	1.30 (0.89–1.91)

Further analysis was done to examine the body mass-stage association by ethnicity in each age group (<50 years or 50 years and above). No significant interaction between ethnicity and BMI was detected (p > 0.1). In addition, the positive association between BMI and stage of breast cancer at diagnosis was not modified by income, education, marital status or type of health insurance.

DISCUSSION

In the current study, high body mass was significantly associated with stage of breast cancer at diagnosis. Adjustment for age, ethnicity, income, education and marital status did not substantially change the estimates of the association. This finding is consistent with those of several previous studies.10–20 For example, in the study by Jones et al.16 of women aged 26–79 years, severe obesity was significantly associated with TNM stage at diagnosis. In a recent study among young women (aged 20–44 years), a positive association between high BMI and advanced stage at diagnosis was found.18 Our results also indicate that high body mass is significantly associated with tumor size, but not node status. This finding is consistent with those of some previous studies.17, 25–28 However, it is inconsistent with findings in other previous studies, which show that obesity is associated with both tumor size and node status.16, 23, 24

There may be several reasons for the observed association between body mass and stage of breast cancer at diagnosis. First, this association could be due to a delay in diagnosis among obese women. Overweight/obese women have larger breasts, and thus tumor detection may be more difficult in these women simply because tumors are more difficult to palpate in larger breasts. This hypothesis is supported by several studies that show a positive relationship between breast size and stage of breast cancer.11, 18, 28, 30

Second, obesity may be associated with a more advanced stage at diagnosis because of an underlying endocrinologic abnormality related to tumor progression.43–46 Some studies suggest that locally increased estrogen levels promote tumor growth.47–49 Obesity causes an increased production of the estrogen known as estrone via the aromatization of androstenedione in peripheral adipose tissue.46, 50 In addition, obesity is associated with low levels of sex hormone-binding globulin, which results in a significantly higher level of the biologically active, unbound form of estrogen known as estradiol.46, 51, 52 Therefore, it is possible that obesity leads to an overall increase in the active levels of estrone and estradiol and that the high levels of these hormones promote the growth of breast tumors in obese women.

We also examined whether age influenced the positive association between body mass and stage at diagnosis. In our study, the association between body mass and stage at diagnosis was stronger in young women (<50 years) than in older women. This finding was unexpected because other studies suggest that obesity is a risk factor for breast cancer in postmenopausal women, but not in premenopausal women.1, 2, 6–9 Thus, we anticipated that the association between obesity and stage of breast cancer would be stronger in older women because they are more likely to be postmenopausal than younger women, or at least equally strong in younger and older women.

The reason for the apparent stronger association between body mass and stage at diagnosis in younger women is unclear. One possible explanation may be that the baseline incidence of breast cancer (both early and late stage) is higher in older women than in younger women. A higher rate of breast cancer in the comparison group (those with normal BMI) in older women may yield a lower odds ratio, even though the breast cancer risk attributable to obesity may be similar in younger and older women.

A second explanation may be that obesity is protective against an early or local stage of breast cancer, but not against tumors at advanced stages in young women. This hypothesis is supported by a recent case-control study of obesity and breast cancer in women aged 20–44 years.18 In this study, the investigators found that comparing women with local stage tumors with controls yielded an OR of 0.7 for the association between BMI and breast cancer, whereas comparing women with advanced (regional/distant) stage tumors to controls yielded an OR of 1.0. When comparisons were made between early and advanced stage tumors, the association with BMI was positive. This hypothesis seems to be reasonable in that it explains 2 findings simultaneously, i.e., a negative association between BMI and breast cancer risk in young women and a positive association between a high BMI and advanced stage of breast cancer at diagnosis.

Although our study has several strengths (e.g., a large sample size, reliable information on BMI, reliable information on date of diagnosis and tumor stage), it also has some limitations due to the insufficiency of data obtained from medical charts. First, menopausal status was not consistently (or routinely) recorded in medical charts, and thus we were unable to explore the association between body mass and stage at diagnosis in premenopausal and postmenopausal women separately. We did explore the relationship among age (as a proxy for menopausal status), body mass and stage at diagnosis. However, we acknowledge that this approach is limited because women may become menopausal at different ages.

Second, premorbid body weight and early adulthood body weight were not routinely recorded, and therefore we were unable to examine these variables in our study. These measures might be more informative than current weight. Third, information about potential risk factors for breast cancer such as menstrual history, reproductive history, hormone use history and lifestyle factors were unavailable for analysis. It is possible that these factors are also related to stage of breast cancer at diagnosis and that our results are biased because we were unable to adjust for these potential risk factors. For instance, the positive association between body mass and stage of breast cancer found in our study may be either an underestimate or an overestimate if menstrual history, reproductive history, hormone use history and/or lifestyle factors have confounding effects. Finally, detection methods were not available in a relatively high proportion of study subjects. This hindered the exploration of potential mechanisms for our findings of a positive association between body mass and stage at diagnosis.

In summary, our study found that high BMI is associated with a later stage of breast cancer at diagnosis. This association was more apparent in young women than in older women. In addition, the study found that high BMI was associated with larger tumor size, but not node involvement. Future studies are needed to confirm these findings and to explore the potential mechanisms involved in the association between BMI and stage at diagnosis.

Acknowledgements

This work is dedicated to the memory of Dr. Trudy Bush. The authors are grateful to Ms. L. Van Ruiten for her help with manuscript preparation.