American Journal of Epidemiology Vol. 151, No.

8
Copyright O 2000 by The Johns Hopkins University School of Hygiene and Public Health Printed In U.S.A.
All rights reserved

Body Size and Breast Cancer Risk in Black Women and White Women
The Carolina Breast Cancer Study

Ingrid J. Hall,1 Beth Newman,1 Robert C. Millikan,1 and Patricia G. Moorman2

The relation between body size and breast cancer risk was investigated in a population-based, case-control
study of Black women (350 cases, 353 controls) and White women (523 cases, 471 controls) from North
Carolina, aged 20-74 years in 1993-1996. Logistic regression analyses compared tertiles of each body size
variable, adjusting for age and breast cancer risk factors (results shown for highest relative to lowest fertile).
Among premenopausal women, body mass index (kg/m2) was inversely associated with breast cancer (odds
ratio (OR) = 0.46, 95% confidence interval (Cl): 0.26, 0.80) for Whites but not for Blacks. There was essentially

Downloaded from http://aje.oxfordjournals.org/ by guest on November 28, 2015

no association among postmenopausal women. Higher waist/hip ratio, adjusted for body mass index, increased
risk for all women. Odds ratios for Black and White premenopausal women were 2.50 (95% Cl: 1.10, 5.67) and
2.44 (95% Cl: 1.17, 5.09), respectively; odds ratios for Black and White postmenopausal women were 1.62 (95%
Cl: 0.70, 3.79) and 1.64 (95% Cl: 0.88, 3.07), respectively. Findings for body mass index differed among Black
women when stratified by age (<50 years) (OR = 0.50, 95% Cl: 0.25, 1.01) instead of menopausal status. Thus,
the associations of breast cancer with body mass index and waist/hip ratio among Black women are similar to
those documented for Whites, despite different body size profiles on average. Am J Epidemiol'2000; 151:754-64.

Blacks; breast neoplasms; obesity

Breast cancer incidence rates among women older
than 45 years of age are higher for Whites than for
Blacks. In contrast, among women younger than 45
years, Black women are more likely to develop breast
cancer than are White women (1). Studies examining
risk factors for breast cancer among Black women
have implicated those previously identified for White
women (2-9). However, the crossover from higher to
lower age-specific incidence rates in Black women at
age 45 cannot be explained by current data on risk fac-
tors (2, 10).
In a review of published data on the epidemiology of
breast cancer in Black women, Trock posits that per-
haps the "disparity in incidence reflects an ethnic dif-
ference in the prevalence of risk factors rather than in
the magnitude of their effects" (2, p. 16). Numerous
studies document differences in obesity between Black
Received for publication December 16, 1998, and accepted for
publication June 22, 1999.
Abbreviations: Cl, confidence interval; OR, odds ratio.
1
Department of Epidemiology, School of Public Health, and
Uneberger Comprehensive Cancer Center, University of North
Carolina, Chapel Hill, NC.
2
Department of Epidemiology and Public Health, Yale University,
New Haven, CT.
754
women and White women, asserting that the preva-
lence of obesity in Black women is twice that of White
women (11-16). This difference has been detected as
early as 10 years of age (16, 17). Obesity, therefore, is
particularly interesting as a risk factor for breast can-
cer among Black women as it contributes to an appar-
ent paradox. If increased obesity is associated with
increased postmenopausal breast cancer (18-24), one
might expect higher rates of postmenopausal breast
cancer among Blacks, since their prevalence of obesity
is higher. Moreover, since obesity is inversely related
to breast cancer risk among premenopausal women
(18-28), young Black women should have a lower
incidence of breast cancer than young White women
have. Neither of these expectations is upheld by the lit-
erature (1) although, clearly, body size is not the only
predictor for breast cancer risk.
There are few population-based analytical studies
that focus on risk factors for breast cancer in Black
women. In one (6), analyses of obesity were not strat-
ified by menopausal status, an important effect modi-
fier. In another (9), the relation between body mass
index and breast cancer risk among young Black
women was contrary to the inverse association docu-
mented extensively in the literature for White women.
The goal of this study was to determine whether body
mass index, height, and waist/hip ratio, as indices of

body size and shape, influence the risk of breast can-
cer in a population-based sample, comparing the
results between younger and older Black women and
White women. Race-specific similarities in odds ratios
obtained from separate analyses of Black women and
White women would suggest that factors affecting
breast cancer act independently of the very different
body size profiles observed between Black women and
White women.
MATERIALS AND METHODS
Study participants, design, and data collection
Data were used from the Carolina Breast Cancer
Study, a population-based, case-control study (29).
Eligible cases were women between the ages of 20 and
74 years who were diagnosed as having a first, pri-
mary, invasive breast cancer between May 1, 1993,
and May 30, 1996, and who lived in a 24-county area
of central and eastern North Carolina. The sampling
design ensured oversampling of Black and young
(20-49 years) women. Of the 1,196 eligible cases,
physicians refused contact for 73 cases (6 percent), 42
(3.5 percent) could not be located, and 191 (16 per-
cent) refused to participate. Thus, data for 890 women
were available for analysis, providing an overall
response rate among eligible and locatable cases of 77
percent.
Controls were selected from the same geographic
area as cases by using records from the North Carolina
Division of Motor Vehicles and US Health Care
Financing Administration to identify women <65 years
and women 65-74 years of age, respectively. Controls
were frequency matched to cases by race and 5-year
age group using a modification of randomized recruit-
ment (30, 31). Of the 1,593 eligible controls, 348 (22
percent) could not be located, and 404 (25 percent)
refused to participate, leaving 841 women for analysis.
The response rate among eligible and locatable con-
trols was 68 percent.
Participants were interviewed in person according to
a pretested, standardized questionnaire. Data collec-
tion included family history of cancer, reproductive
and menstrual history, hormone use, alcohol consump-
tion, occupational exposures, and sociodemographic
characteristics, among a number of other suspected
risk factors for breast cancer. Measurements of height,
weight, and waist and hip circumferences were taken
at the time of interview using standardized scales and
tape measures. Height and weight were measured
twice to the nearest 0.5 cm and 0.5 kg, respectively,
and the two measurements were averaged. The waist
circumference measurement was taken at the natural
indentation of the waist, and the hip circumference
Am J Epidemiol Vol. 151, No. 8, 2000
Body Size and Breast Cancer Risk 755
was taken at the greatest protrusion of the buttocks.
Both circumferences were measured two times and
averaged. A third measurement was taken if the first
two differed by more than 1.0 cm, in which case the
two closest values were averaged. Body measurement
data were not available for three women. Eighty per-
cent of interviews were completed within 5 months of
diagnosis for cases or of selection for controls.
The data set consisted of 994 (57 percent) White
women and 703 (41 percent) Black women (race deter-
mined by self-report). In addition, the study included
11 (0.6 percent) Hispanic, 9 (0.5 percent) Asian, and
11 (0.6 percent) Native American participants. Results
were not altered by the exclusion of the least repre-
sented minorities, and thus, the results reported here
are for Black women and White women only.
Downloaded from http://aje.oxfordjournals.org/ by guest on November 28, 2015
Statistical analysis
Tertiles were calculated for body mass index, height,
and waist/hip ratio using the distributions for all cases
and controls. In addition, race-specific analyses were
conducted to provide a direct comparison of results
with those of the published literature, most of which
were reports on studies conducted in predominately or
exclusively White samples. We also wanted to investi-
gate whether the vastly different distributions of body
mass index and waist/hip ratio among Black women
and White women differentially affected estimates of
risk. Therefore, race-specific tertiles were calculated
using separate distributions for all Black women and
all White women. We classified women according to
menopausal status using the definitions of Palmer et al.
(8). Women were categorized as postmenopausal if
they reported natural menopause, bilateral oophorec-
tomy, or a hysterectomy and were >55 years of age.
Women who reported still having menstrual cycles or
who had at least one remaining ovary and were <42
years were classified as premenopausal. Women who
had hysterectomy without bilateral oophorectomy and
who were aged 42-55 years were considered peri-
menopausal.
Logistic regression was used to control for several
potential confounders (32, 33), including years of edu-
cation (<12, 12, 13-15, >16), age at menarche (<11,
12-13, >14), breastfeeding (never, ever), family history
of breast cancer in afirst-degreerelative (no, yes), age at
menopause (<45, 46-49, >50, among postmenopausal
women), a cross-classification of parity and age at first
full-term pregnancy (nulliparous, 1-2 children, and age
<29 years; >3 children and age <29 years; 1-2 children
and age >29 years; >3 children and age >29 years), oral
contraceptive use (never, ever), hormone replacement
therapy (never, ever), alcohol consumption (never, ever),
cigarette smoking (never, ever), and weight change in
756 Hall et al.
the past year (pounds gained: 6-10, 11-19, >20; pounds
lost: 6-10, 11-19, >20) (1 pound = 0.45 kg). Age was
coded as a continuous variable.
The data were stratified by race and menopausal sta-
tus, excluding perimenopausal women, resulting in
four subgroups. As 28 percent of postmenopausal
women were under 50 years of age, we repeated analy-
ses stratifying on race and age at diagnosis for cases or
age at selection for controls (<50 vs. ^50 years). This
allowed us to include perimenopausal women in the
analyses. An offset term was incorporated in each
model to adjust for sampling fractions using Proc
Genmod in SAS version 6.12 software (SAS Institute,
Inc., Cary, North Carolina). In the initial logistic
regression, we examined all variables. No interaction
was detected, as point estimates between the body
mass index and risk of breast cancer did not vary sig-
nificantly, using the likelihood ratio test (a = 0.20),
among categories of any covariate. Covariates were
assessed individually for confounding by removing
each from the fully adjusted model. The variables for
oral contraceptive use, hormone replacement therapy,
alcohol, smoking, and weight change made little dif-
ference in results and, therefore, were omitted from the
analyses reported here. In subsequent runs, we retained
only age, education, and the variables related to repro-
ductive history. No covariate acted as a confounder
(^20 percent change in P) of the body size-breast can-
cer relations, but all were retained regardless of statis-
tical significance to provide fully adjusted estimates
for comparison with those in the published literature.
RESULTS
Characteristics of the study population are presented
in table 1. Unweighted percentages are reported, repre-
senting the sample rather than the underlying popula-
tion. Examination of differences between Black controls
and White controls reveals that Black women exhibited
a different distribution for age at menarche, with a
smaller proportion in the middle category and more in
the younger and older categories; younger age at
menopause; more children at younger ages; and fewer
years of education. Among Black women, cases were
more likely to report an earlier age at menarche, later age
at menopause, and breast cancer in a first-degree family
member and less likely to report five or more pregnan-
cies of 7 months or longer or breastfeeding compared
with controls. For White women, cases were more likely
than controls to report an earlier age at menarche, nulli-
parity, later age at first full-term pregnancy, a family his-
tory of breast cancer, and never having breastfed.
Table 2 provides the distributions of body size vari-
ables. The mean body mass index among controls was
31.7 kg/m2 among Black women and 26.2 kg/m2
among White women. Using tertiles based on cases
and controls from both races combined, the distribu-
tions of body mass index reveal different patterns:
among controls, greater than 50 percent of Black
women fall into the highest tertile of body mass index,
while 40 percent of White women are in the lowest ter-
tile. The means for height were 162.2 cm and 162.8
cm, respectively, for Black women and White women,
and their corresponding tertile distributions were sim-
ilar. The waist/hip ratio showed distributions similar to
those for the body mass index with means that were
somewhat higher among Black women and that
remained unchanged after adjustment for body mass
index (mean = 0.84 among Black women, mean =
0.79 among White women).
Premenopausal women
Downloaded from http://aje.oxfordjournals.org/ by guest on November 28, 2015
The relations between various body size measures
and breast cancer risk for premenopausal women,
adjusted for breast cancer risk factors, are presented in
table 3. Among White women, those in the third tertile
of body mass index were at about half the risk com-
pared with women in the first tertile. There was no
association among premenopausal Black women.
Adult height was associated with increased breast can-
cer risk in premenopausal Black women, while there
was no association observed among White women.
The waist/hip ratio, adjusted for body mass index, was
associated with increased breast cancer risk in Black
and White premenopausal women and was suggestive
of a dose-response relation, although the confidence
intervals largely overlapped.
A large proportion of younger women (<50 years)
was excluded from the prior analysis because they
were not premenopausal. We therefore repeated the
analyses among women younger than 50 years. Only
one difference in the pattern of the results reported
above was noteworthy: among younger Black women,
increasing body mass index was associated with a 50
percent lower risk of breast cancer comparing women
in the third tertile of body mass index with those in the
first (odds ratio (OR) = 0.50, 95 percent confidence
interval (CI): 0.25, 1.01), making these results more
comparable with what was observed for White
women.
Postmenopausal women
Table 4 presents estimates for the relations between
various body size measures and breast cancer risk for
postmenopausal women, adjusted for breast cancer
risk factors. Body mass index showed essentially no
association with breast cancer in postmenopausal
White women; in contrast, there was the suggestion of
Am J Epidemiol Vol. 151, No. 8, 2000
 Body Size and Breast Cancer Risk 757

TABLE 1. Distribution of known risk factors for breast cancer among Black women and WhKe women, Carolina Breast Cancer
Study, North Carolina, 1993-1996

Black women (n = 703) White women ( n = 994)

variable Cases(n = 350) Controls (n = 353) Cases (n = 523) Controls (n = 471)
No. % No. % No. % No. %

Age group (years)

20-34 29 8 18 7 36 7 25 5
35-44 84 24 89 25 151 29 110 23
45-54 100 28 114 32 179 34 145 31
55-64 67 19 71 20 70 13 86 18
265 70 20 61 17 87 17 105 22

Age at menarche (years)

11 94 27 83 23 111 21 86 18
12-13 177 50 167 47 300 57 271 58
214 78 22 102 28 111 21 112 24

Menopausal status

Downloaded from http://aje.oxfordjournals.org/ by guest on November 28, 2015

Pre 135 38 136 39 255 49 183 39
Peri 36 10 35 10 55 11 51 11
Post 179 51 182 52 213 41 237 50

Age at menopause (years)

Premenopausal 128 36 126 36 240 46 176 37
45 124 35 134 38 141 27 147 31
46-49 33 9 39 11 80 15 59 13
50-54 45 13 41 12 42 8 54 11
255 16 5 8 2 14 3 16 3

Parity
0 44 13 44 12 89 17 52 11
1-2 134 38 129 37 290 55 254 54
34 103 29 90 25 115 22 119 25
25 56 16 73 21 21 4 21 4

Age at first full-term

pregnancy (years)
Nulliparous 44 13 44 12 89 17 52 11
24 243 69 248 70 271 52 277 59
25-29 39 11 33 9 97 19 101 21
230 24 7 28 8 66 13 41 9

Education
<High school 108 31 118 33 59 11 59 13
High school graduate 102 29 94 27 144 27 126 27
Some college 75 21 78 22 144 28 142 30
^College 65 19 63 18 176 34 144 31

Family historyt
No 293 84 301 85 426 81 399 85
Yes 46 13 35 10 85 16 59 13

Breastfeeding
Never 232 66 208 59 335 64 260 55
Ever 103 29 124 35 178 34 181 38
May not total 100 because of missing values.
t Breast cancer in first-degree relative, that is, mother, father, or sibling.

an inverse association observed between body mass cer risk in postmenopausal Black women, among post-
index and breast cancer among postmenopausal Black menopausal White women, taller women had a non-
women. While height was not important to breast can- significant increased risk compared with shorter
Am J Epidemiol Vol. 151, No. 8, 2000
758 Hall et al.

TABLE 2. Distribution of body size Indices among Black women and White women, using population-derived tertiles, Carolina
Breast Cancer Study, North Carolina, 1993-1996*

Hack women (n = 703) White women (n = 994)

Cases (n = 350) Controls (n = 353) Cases(n = 523) Controls (r> = 471)
No. %t No. % No. % No. %

Mean body mass index

(kg/m*) 31.66 (7.4)* 26.19 (5.62)
14.62-24.61 53 15 46 13 255 49 195 41
24.62-30.12 114 33 114 32 169 32 171 36
30.13-59.26 182 52 191 54 98 18 104 22

Mean height (cm) 162.17'(6.71) 162.84 (6.41)

140.0-160.0 128 36 136 39 176 34 168 36
160.1-165.0 90 26 93 32 160 31 124 26
165.1-188.0 109 31 94 27 170 33 147 31

Mean waist/hip ratio 0.84 (0.08) 0.79 l(0.08)

0.60-0.77 68 19 74 21 210 40 218 46

Downloaded from http://aje.oxfordjournals.org/ by guest on November 28, 2015

0.78-0.85 110 31 122 35 194 37 131 28
0.86-1.34 153 44 130 37 102 20 91 19
Tertiles derived from all cases and controls.
t Totals may not equal 100 because of missing values.
t Numbers in parentheses, standard deviation.

TABLE 3. Odds ratios (ORs) and 95% confidence Intervals (CIs) for breast cancer according to indices
of body size in premenopausal women, Carolina Breast Cancer Study, North Carolina, 1993-1996t

Black women White women

No. of No. of
Variable cases/ cases/
OR 95% Cl OR 95% Cl
no. of no. of
controls controls

Body mass index (kg/m2)^

14.62-24.61 23/22 1.00 140/76 1.00
24.62-30.12 35/41 0.71 0.28, 1.79 71/61 0.74 0.45, 1.21
30.13-59.26 77/72 0.89 0.38, 2.07 43/46 0.46 0.26, 0.80
Continuous body mass
index (units/kg/m2) 1.00* 0.98, 1.02 0.98* 0.95,1.01

Height (cm)
140.0-160.0 40/51 1.00 77/50 1.00
160.1-165.0 36/35 1.36 0.65, 2.87 75/47 1.06 0.61, 1.82
165.1-188.0 50/35 2.93 1.44, 5.95 95/72 0.77 0.46, 1.29
Continuous height (units/cm) 1.05** 1.00, 1.10 0.99* 0.95,1.02

Waist/hip ratJo
0.60-0.77 30/40 1.00 124/98 1.00
0.78-0.85 51/50 1.69 0.81, 3.52 86/50 1.77 1.06,2.94
0.86-1.34 46/32 2.50 1.10, 5.67 33/22 2.44 1.17,5.09
Continuous waist/hip ratio
(units/0.01 unit) 1.05** 1.00, 1.10 1.04** 1.01,1.08

* p trend, not significant; p trend, <0.05.

t Tertiles derived from all cases and controls.
X Odds ratios were adjusted for age, age at menarche, parity/age at first full-term pregnancy, lactation, edu-
cation, and sampling probabilities.
Odds ratios were additionally adjusted for body mass index.

women. A higher waist/hip ratio was associated with We again stratified by age, as many postmenopausal
an increased risk of breast cancer, although not statis- women were younger. Analyses performed after strati-
tically significant, in all postmenopausal women. fying on age 50 years or older resulted in a similar pat-
Am J Epidemiol Vol. 151, No. 8, 2000
 Body Size and Breast Cancer Risk 759

TABLE 4. Odds ratios (ORs) and 95% confidence Intervals (CIs) for breast cancer and Indices of body
size In postmenopausal women, Carolina Breast Cancer Study, North Carolina, 1993-19961

Black women White women

No.oJ No. of
Variable cases/ cases/
OR 95% Cl OR 95% Cl
no. of no. of
controls controls

Body mass index (kg/m2)^

14.62-24.61 26/23 1.00 88/95 1.00
24.62-30.12 60/61 0.69 0.33, 1.47 80/98 0.98 0.60, 1.59
30.13-59.26 92/97 0.68 0.33, 1.42 45/43 1.08 0.58, 2.00
Continuous body mass
index (units/kg/m2) 1.00* 0.98, 1.02 1.02* 0.98, 1.05

Height (cm)
140.0-160.0 78/76 1.00 77/101 1.00
160.1-165.0 47/44 1.07 0.59,1.94 69/65 1.59 0.95, 2.67
165.1-188.0 42/49 1.00 0.55,1.83 63/55 1.63 0.96, 2.76

Downloaded from http://aje.oxfordjournals.org/ by guest on November 28, 2015

Continuous height (units/cm) 1.00* 0.97,1.04 1.03* 1.00, 1.07

Waist/hip ratjo
0.60-0.77 30/29 1.00 59/58 1.00
0.78-0.85 49/55 0.81 0.34, 1.94 88/72 1.92 1.14,3.23
0.86-1.34 91/86 1.62 0.70, 3.79 63/91 1.64 0.88, 3.07
Continuous waist/hip ratio
(unlts/0.01 unit) 1.03* 0.99, 1.07 1.03* 1.00, 1.06
* p trend, not significant; * p trend, <0.05.
t Tertjles derived from all cases and controls.
% Odds ratios were adjusted for age, age at menarche, parity/age at first full-term pregnancy, age at
menopause, lactation, education, and sampling probabilities.
Odds ratios were additionally adjusted for body mass index.

tern of results as described above, with one exception.
There was no association observed between body mass
index and breast cancer in Black women >50 years.
The inverse association observed for postmenopausal
Black women was lost when younger women were
removed from the postmenopausal group.
Race-specific tertiles
ously seen with body mass index was lost, and no asso-
ciation was observed in race-specific analyses. For
waist/hip ratio, there was the appearance of a pro-
nounced trend among White postmenopausal women
(OR = 1.66,95 percent Cl: 0.94,2.93; and OR = 2.15,
95 percent Cl: 1.17, 3.95, for the second and third ter-
tiles, respectively), although confidence intervals
overlapped.

Analyses were repeated using tertiles derived from DISCUSSION

the separate body size distributions of Black women
and White women to account for the differences in dis-
tributions by race. A trend of borderline significance
emerged for waist/hip ratio among White pre-
menopausal women (OR = 1.48, 95 percent Cl: 0.89,
2.44; and OR = 2.21, 95 percent Cl: 1.19,4.11, for the
second and third tertiles, respectively). In contrast, the
trends previously observed for premenopausal Black
women became flatter when using race-specific tertiles
of waist/hip ratio (OR = 2.09, 95 percent Cl: 1.04,
4.19; and OR = 2.11, 95 percent Cl: 0.84, 5.32, for the
second and third tertiles, respectively). Race-specific
cutpoints yielded similar results for postmenopausal
women, with the following differences. For Black
postmenopausal women, the inverse association previ-
In our data, body mass index was inversely related
to breast cancer risk among premenopausal White
women but not among premenopausal Black women.
There was little or no evidence for increased breast
cancer risk among postmenopausal women of either
race. Waist/hip ratio was associated with increased risk
of breast cancer among both premenopausal and post-
menopausal women of both races. Height was posi-
tively associated with breast cancer only among pre-
menopausal Black women and postmenopausal White
women. Analyses stratified by age altered the pattern
of results minimally for waist/hip ratio and height and,
for White women, body mass index. In contrast, asso-
ciations with body mass index differed among Black

Am J Epidemiol Vol. 151, No. 8, 2000

760 Hall et al.
women when age was substituted for menopausal sta-
tus as the defining subgroup characteristic. The lack of
association seen among premenopausal Black women
became an inverse association among younger Black
women, whereas the inverse association observed for
postmenopausal Black women was lost among older
Black women. For most analyses, the pattern of results
was essentially unchanged when race-specific tertiles
were used; however, the dose-response nature of the
association with waist/hip ratio, though somewhat ten-
uous, became much more apparent among White
women with race-specific cutpoints.
Obesity is an important public health issue for Black
women. A 1991 study (34) reported that, among
nonobese women, Blacks were 60 percent more likely
than Whites to become obese. The rates of obesity in
Black women have been shown to be twice those of
Whites (12), and the prevalence of overweight is more
than twofold (11), nearing 50 percent in non-Hispanic
Black women (35). In our population, the prevalence
of overweight, defined as body mass index >27.3
kg/m2 (14), among Black controls and White controls
was 71 percent and 39 percent, respectively.
Younger Black women exhibited similar inverse
associations between body mass index and breast can-
cer risk as did younger White women. Hence, obesity,
as measured by body mass index, does not account for
the increased breast cancer incidence observed in
Black women <50 years (1). The absence of the
inverse association between premenopausal breast
cancer and body mass index and the appearance of an
inverse association for postmenopausal breast cancer
in Black women may be related to the high rate of sur-
gical menopause in this population. In our data, a large
number of young Black women (27 percent) reported
hysterectomy with oophorectomy. The movement of
these younger women from the under age 50 subgroup
to the postmenopausal subgroup effectively altered the
pattern of results in Black women, making them less
consistent with the results seen in Whites in this and
other studies. Although confidence intervals were gen-
erally overlapping, the presence of inverse associa-
tions among women younger than 50 years (regardless
of menopausal status) suggests that the protective
effect of obesity may have more to do with age at diag-
nosis (or possibly menopausal status sometime prior to
diagnosis) than with menopausal status at the time of
breast cancer occurrence.
The absence of an association between body mass
index and postmenopausal breast cancer in Black
women may be related to their high prevalence of estro-
gen receptor-negative breast tumors (36). Such tumors
are likely unresponsive to the increased estrogen avail-
ability caused by obesity. The null finding among White
women goes against the positive association seen in
most literature. However, recent findings have detected
an association with body mass index only among post-
menopausal women who never used hormone replace-
ment therapy (21, 23, 37). When we restricted analyses
to women 50 years who had never used hormone
replacement therapy, there was still no association
observed among Black women; however, the body
mass index-breast cancer association was strengthened
substantially for White women, yielding an odds ratio
of 3.04 (95 percent CI: 1.00, 9.29) for the highest fertile
compared with the lowest. London et al. (38) and oth-
ers (21, 39) conducted analyses by decade of age at
diagnosis and saw the strongest associations among the
oldest postmenopausal women (>55 years). We also
observed an increase in odds ratios for all women when
analyses were restricted to women >55 years and a fur-
Downloaded from http://aje.oxfordjournals.org/ by guest on November 28, 2015
ther increase in effect estimates when analyses were
restricted to those >60 years (data not shown).
Biologic plausibility of the obesity-breast cancer
association can be inferred from extensive literature
documenting higher levels of non-protein-bound estra-
diol (40), the biologically active component, and lower
levels of sex hormone binding globulin (41) in breast
cancer cases compared with controls. Obese women
also have lower levels of sex hormone binding globu-
lin than do their leaner counterparts (42). The increase
in breast cancer risk associated with obesity in post-
menopausal women is presumed to be mediated by
conversion of androstenedione to estrone in adipose
tissue, which occurs at a greater rate in obese women
(43, 44). Consequently, the estrogen-sensitive tissues
of obese women are exposed to more stimulation than
are those of leaner women. The amount of additional
estrogen exposure from peripheral conversion may be
less consequential among premenopausal women in
whom cyclic hormone levels predominate. In addition,
premenopausal obesity is frequently associated with
impaired ovulation, possibly indicating an increase in
anovulatory cycles (45), which may decrease exposure
to both estradiol and progesterone and thereby reduce
risk of breast cancer in young women (46). Previous
work found that women with evidence of anovulatory
cycles (i.e., irregular cycles >36 days), as a result of
polycystic ovary disease, were more than 30 pounds
heavier, on average, than were women with no men-
strual abnormalities, after adjustment for height and
age (47), while more recent work has shown that
women with irregular cycles had decreased breast can-
cer risk compared with women with regular cycles
(48). Moreover, leptin levels, which increase with
increasing fat stores, inhibit ovarian estrogen produc-
tion (49) and may contribute to reduced risk of breast
cancer in heavy, younger women.
Am J Epidemiol Vol. 151, No. 8, 2000

Metabolism and estrogen binding also can be
affected by differences in fat distribution, as women
with upper body fat localization have lower levels of
sex hormone binding globulin, leaving free estrogen to
target sensitive tissues (50). Thus, abdominal or cen-
tral fat patterning has been associated with increased
breast cancer risk in the absence of a breast cancer-
body mass index association (24, 51-55). In addition
to increased levels of free estrogen, women with
abdominal obesity often show hyperinsulinemia (24).
Hyperinsulinemia is associated with increased levels
of insulin-like growth factor type I, and higher levels
of this growth factor have been detected in breast can-
cer cases when compared with controls (56-58).
Insulin-like growth factor type I has been demon-
strated to work synergistically with estrogen in stimu-
lating breast cancer growth (59). Thus, abdominal obe-
sity should be considered a risk factor for breast
cancer, separate from body mass, and our results sug-
gest that it may be most important for premenopausal
women of both races.
Adult height has been implicated as a predictor of
breast cancer risk in older women (38, 60-62). No
association with height was detected in other studies
(22, 54, 63-65), while a few reports show a positive
association with height and breast cancer risk in both
younger and older women (17, 19, 26). Our results
contribute to the body of literature by demonstrating
an association among older White women and no
effect among younger White women. Data on the rela-
tion between height and breast cancer risk among
Black women are scarce (8), so there is little frame of
reference for our findings. The reasons behind this
association require additional scrutiny.
Limitations of this study include its retrospective
nature. Anthropometric measurements of cases were
taken after diagnosis when 34 percent of cases, com-
pared with 22 percent of controls, had lost more than 5
pounds. This difference may reflect an effect of treat-
ment among cases. We attempted to minimize the
effect by completing most interviews within 5 months
of diagnosis. Hence, the relations between body mass
index and breast cancer observed in our study may
underestimate the true magnitude of positive associa-
tions. For inverse associations, the effect of weight
loss by cases serves to make the observed association
more pronounced. In addition, response rates often
varied by age, race, and case-control status. In general,
among younger women who could be located and were
eligible, there was little difference in response by race:
81-82 percent of cases and 72-73 percent of controls.
However, almost twice as many young Black controls
compared with young non-Black controls could not be
located. Response rates for older women were lower
Am J Epidemiol Vol. 151, No. 8, 2000
Body Size and Breast Cancer Risk 761
overall compared with younger women. Among Black
and non-Black older cases who could be located and
were eligible, the response rates were 68 percent and
74 percent, respectively. Rates were 59 percent and 67
percent among Black and White older controls, respec-
tively. Again, more than twice as many older Black
controls could not be located compared with non-
Black controls. If heavier women were underrepre-
sented to a greater degree in the control groups, we
would expect the true odds ratios for the highest tertile
of body mass index to reflect even stronger inverse
relations, whereas the true odds ratios for waist/hip
ratio would be attenuated relative to our results. Again,
if body size is associated with nonresponse, these
biases might operate more for Black women than for
White women. However, such selection bias may be
considered less likely, since the mean body mass index
Downloaded from http://aje.oxfordjournals.org/ by guest on November 28, 2015
did not differ for either cases or controls between
women fully participating in the study and those agree-
ing only to a brief telephone survey on breast cancer
risk factors (66). While we acknowledge that response
rates for older women were not optimal, the most
interesting results of this report were observed among
younger women for whom response rates were much
higher. It should be noted that the women in our study
are heavier, on average, than are women from other
parts of the country (16, 67, 68), potentially decreasing
generalizability of results.
We lacked data to adjust analyses for other factors
that may confound the observed results. Our adjust-
ment for socioeconomic status was minimal (69),
using only years of education as a proxy. Neither phys-
ical activity nor dietary considerations were taken into
account in our study. Both have been implicated as
possible risk factors for breast cancer (70, 71). A recent
publication by Michels et al. (72) asserts that studies of
body mass index, an index of body composition, must
be additionally adjusted for an indicator of body size,
such as height. In our data, additional adjustment for
either height or waist/hip ratio did not change the pat-
tern of results for body mass index previously
observed (data not shown).
Misclassification of fat distribution, based on the
waist/hip ratio, is also likely. Fat can be distributed
subcutaneously or viscerally within the abdominal
cavity, as measured by computed tomography or mag-
netic resonance imaging (73), and race-specific differ-
ences in localization of fat have been documented. A
1995 study (74) of visceral adipose tissue in Black
women and White women, matched for percentages of
body fat and body mass index, showed that not only
did large abdominal fat deposits exist in apparently
lean women but that Black women had 23 percent less
visceral adipose tissue, or metabolically active fat, and
762 Hall et al.
more subcutaneous fat. Lovejoy et al. (75) showed that
the visceral fat area was smaller in Black women com-
pared with White women of similar age, body mass
index, and waist/hip ratio and that the correlation
between the visceral fat area and the waist/hip ratio
was weaker among Black women (75). Conway et al.
(74) also indicated that Black women have greater
bone density, while others have suggested increased
muscle mass (76). Thus, the heaviness of Black
women may reflect different underlying physiologic
parameters and may therefore have different implica-
tions with respect to disease risk. Results from
research on diabetes and cardiovascular disease have
suggested that the adverse physiologic effects of over-
weight were not as strong among Blacks as they were
among Whites (77, 78).
In conclusion, studies of body mass index and
breast cancer are challenging to interpret, and among
Black women, results are sensitive to stratification by
age or menopausal status. Our results are consistent
with those of the literature, showing an inverse rela-
tion between risk and body mass index among
younger women of both races and a positive associa-
tion between breast cancer risk and waist/hip ratio.
Thus, central fat distribution, adjusted for body mass
index, may be an adverse risk factor for breast cancer
as positive dose-response relations were consistently
observed, although the use of race-specific tertiles
was necessary to discern the trend in postmenopausal
White women. Future studies of body size and breast
cancer would benefit from better classification of fat
distribution according to fat localization and from the
inclusion of information on additional socioeconomic
indicators, physical activity, and dietary factors.
ACKNOWLEDGMENTS
This study was supported by the Specialized Program of
Research Excellence (SPORE) in Breast Cancer (P50-
CA58223) and by training grant 5-T32-CA09330, both from
the National Cancer Institute.
The authors thank the nurses of the Carolina Breast
Cancer Study for their diligence in collecting questionnaire
data and body measurements and Dr. Beverly Rockhill and
Jessica Tse for help with statistical analysis.
REFERENCES
1. Ries LAG, Kosary CL, Hankcy BF, et al. SEER cancer statis-
tics review, 1973-1994: tables and graphs. Bethesda, MD:
National Cancer Institute, 1997. (NIH publication no. 97-
2789).
2. Trock BJ. Breast cancer in African-American women: epi-
demiology and tumor biology. Breast Cancer Res Treat 1996;
40:11-24.
3. Moormeier J. Breast cancer in Black women. Ann Intern Med
1996; 124:897-905.
4. Austin H, Cole P, Wynder E. Breast cancer in Black American
women. Br J Cancer 1979;24:541-4.
5. Schatzkin A, Palmer JR, Rosenberg L, et al. Risks for breast
cancer in Black women. J Natl Cancer Inst 1987;78:213-17.
6. Mayberry RM, Stoddard-Wright C. Breast cancer risk factors
among Black women and White women: similarities and dif-
ferences. Am J Epidemiol 1992;136:1445-56.
7. Laing AE, Demenais FM, Williams R. Breast cancer risk fac-
tors in African-American women: the Howard University
tumor registry experience. J Natl Med Assoc 1993;85:931-9.
8. Palmer JR, Rosenberg L, Harlap S. Adult height and risk of
breast cancer among US Black women. Am J Epidemiol 1995;
141:845-9.
9. Mayberry RM. Age-specific patterns of association between
breast cancer and risk factors in Black women, ages 20 to 39
and 40 to 54. Ann Epidemiol 1994;4:205-13.
10. Kelsey JL, Gammon MD, John EM. Reproductive factors and
Downloaded from http://aje.oxfordjournals.org/ by guest on November 28, 2015
breast cancer. Epidemiol Rev 1993;15:36-47.
11. Walcott-McQuigg JA, Sullivan J, Dan A, et al. The relation-
ship between stress and weight-control behavior in African-
American women. J Natl Med Assoc 1995;87:427-32.
12. Melnyk MG, Weinstein E. Preventing obesity in Black women
by targeting adolescents: a literature review. J Am Diet Assoc
1994;94:536-40.
13. Kumanyika SK. Special issues regarding obesity in minority
populations. Ann Intern Med 1993; 119:650-4.
14. Williamson DF. Descriptive epidemiology of body weight and
weight change in U.S. adults. Ann Intern Med 1993;119:
646-9.
15. Croft JB, Strogatz DS, James SA, et al. Socioeconomic and
behavioral correlates of body mass index in Black adults: the
Pitt County Study. Am J Public Health 1992;82:821-6.
16. Burke GL, Savage PJ, Manolio TA, et al. Correlates of obesity
in young Black and White women: the CARDIA Study. Am J
Public Health 1992,82:1621-5.
17. Campaigne DN, Morrison JA, Schumann BC. Indexes of obe-
sity and comparisons with previous national survey data in 9-
and 10-year-old Black and White girls: The National Heart,
Lung, and Blood Institute Growth and Health Study. J Pediatr
1994;125:675-80.
18. Trentham-Dietz A, Newcomb PA, Storer BE, et al. Body size
and risk of breast cancer. Am J Epidemiol 1997; 145:1011-19.
19. Yong L-C, Brown D, Schatzkin A, et al. Prospective study of
relative weight and risk of breast cancer the Breast Cancer
Detection Demonstration Project Follow-up Study, 1979
1989. Am J Epidemiol 1996;143:985-95.
20. Ziegler RG, Hoover RN, Nomura AM, et al. Relative weight,
weight change, height, and breast cancer risk in Asian-
American women. J Natl Cancer Inst 1996;88:650-60.
21. Franccschi S, Favero A, LaVeccia C, et al. Body size indices
and breast cancer risk before and after menopause. Int J Cancer
1996;67:181-6.
22. Taioli E, Barone J, Wynder EL. A case-control study on breast
cancer and body mass. Eur J Cancer 1995;31A:723-8.
23. Harris RE, Namboodiri KK, Wynder EL. Breast cancer risk:
effects of estrogen replacement therapy and body mass. J Natl
Cancer Inst 1992;84:1575-82.
24. Bruning PF, Bonfrer JMG, Hart AAM, et al. Body measure-
ment, estrogen availability, and the risk of human breast can-
cer a case-control study. Int J Cancer 1992;51:14-19.
25. Ursin G, Longnecker MP, Haile RW. A meta-analysis of body
mass index and risk of premenopausal breast cancer.
Epidemiology 1995;6:137^tl.
26. Brinton LA, Swanson CA. Height and weight at various ages
and risk of breast cancer. Ann Epidemiol 1992;2:597-609.
27. Vatten LJ, Kvinnsland S. Prospective study of height, body
mass index, and risk of breast cancer. Acta Oncol 1992;31:
Am J Epidemiol Vol. 151, No. 8, 2000

195-200.
28. Peacock SL, White E, Dating JR, et al. Relation between obe-
sity and breast cancer in young women. Am J Epidemiol 1999;
149:339-46.
29. Newman B, Moorman PG, Millikan R, et al. The Carolina
Breast Cancer Study: integrating population-based epidemiol-
ogy and molecular biology. Breast Cancer Res Treat 1995;
35:51-60.
30. Weinberg CR, Sandier DP. Randomized recruitment in case-
control studies. Am J Epidemiol 1991;134:421-32.
31. Weinberg CR, Wacholder S. The design and analysis of case-
control studies with biased sampling. Biometrics 1990;46:
963-75.
32. Hosmer DW, Lemeshow W. Applied logistic regression. New
York; John Wiley & Sons, Inc, 1989.
33. Kleinbaum DG, Kupper LL, Morgenstem H. Epidemiologic
research: principles and quantitative methods. New York: Van
Nostrand Reinhold Company, 1982.
34. Williamson DF, Kahn HS, Byers T. The 10-y incidence of obe-
sity and major weight gain in Black and White US women
aged 30-55 y. Am J Clin Nutr 1991;53(suppl):1515S-18S.
35. Kuczmarski RJ, Flegal KM, Campbell SM, et al. Increasing
prevalence of overweight among US adults. JAMA 1994;272:
205-11.
36. Kelsey JL. Breast cancer epidemiology: summary and future
directions. Epidemiol Rev 1993;15:256-63.
37. Huang Z, Hankinson SE, Colditz GA, et al. Dual effect of
weight and weight gain on breast cancer risk. JAMA
1997;278:1407-ll.
38. London SJ, Colditz GA, Stampfer MJ, et al. Prospective study
of relative weight, height, and risk of breast cancer. JAMA
1989;262:2853-8.
39. LaVecchia C, Negri E, Franceschi S, et al. Body mass index
and post-menopausal breast cancer: an age-specific analysis.
BrJ Cancer 1997;75:441^.
40. Bruning PF, Bonfrer JMG, Hart AAM. Non-protein bound
estradiol, sex hormone binding globulin, breast cancer, and
breast cancer risk. Br J Cancer 1985;51:479-84.
41. Moore JW, Key TJA, Bulbrook RD, et al. Concentrations of
sex hormone binding globulin (SHBG) in a population of nor-
mal women who had never used exogenous sex hormones.
Steroids 1988;52:391-2.
42. Ingram D, Nottage E, Ng S, et al. Obesity and breast disease.
The role of the female sex hormones. Cancer 1989;64:1049-
53.
43. Kirschner MA, Ertel N, Schneider G. Obesity, hormones, and
cancer. Cancer Res 1981;41:3711-17.
44. Hershcopf RJ, Bradlow HL. Obesity, diet, endogenous estro-
gens, and the risk of hormone-sensitive cancer. Am J Clin Nutr
1987;45:283-9.
45. Pike MC. Reducing cancer risk in women through lifestyle-
mediating changes in hormone levels. Cancer Detect Prev
1990; 14:595-607.
46. Henderson BE, Ross RK, Judd HL, et al. Do regular ovulatory
cycles increase breast cancer risk? Cancer 1985;56:1206-8.
47. Hartz AJ, Barboriak PN, Wong A, et al. The association of obe-
sity with infertility and related menstrual abnormalities in
women. Int J Obes 1979;3:57-73.
48. den Tonkelaar I, de Waard F. Regularity and length of men-
strual cycles in women ages 41^46 in relation to breast cancer
risk: results from the DOM project. Breast Cancer Res Treat
1996;38:253-8.
49. Spicer LJ, Francisco CC. The adipose gene product, leptin:
evidence of a direct inhibitory role in ovarian function.
Endocrinology 1997; 135:3374-9.
50. Kaye SA, Folsom AR, Soler JT, et al. Associations of body
mass and fat distribution with sex hormone concentrations in
postmenopausal women. Int J Epidemiol 1991 ;20:151-6.
51. Kaaks R, Van Noord PA, Den Tonkelaar L et al. Breast-cancer
incidence in relation to height, weight, and body-fat distribu-
tion in the Dutch "DOM" cohort. Int J Cancer 1998;79:
647-51.
Am J Epidemiol Vol. 151, No. 8, 2000
Body Size and Breast Cancer Risk 763
52. Schapira DV, Kumar NB, Lyman GH, et al. Abdominal obesity
and breast cancer risk. Ann Intern Med 1990;l 12:182-6.
53. Folsom AR, Kaye SA, Prineas RJ, et al. Increased incidence
of carcinoma of the breast associated with abdominal adipos-
ity in postmenopausal women. Am J Epidemiol 1990;131:
794-803.
54. Ballard-Barbash R, Schatzkin A, Carter CL, et al. Body fat dis-
tribution and breast cancer in the Framingham Study. J Natl
Cancer Inst 199O;82:286-9O.
55. Ng E-H, Gao F, Ji C-Y, et al. Risk factors for breast carcinoma
in Singaporean Chinese women: the role of central obesity.
Cancer 1997;80:725-31.
56. Peyrat JP, Bonneterre J, Hecquet B, et al. Plasma insulin-like
growth factor-1 (IGF-I) concentrations in human breast cancer.
Eur J Cancer 1993;29A:492-7.
57. Bruning PF, Van Doom J, Bonfrer JMG, et al. Insulin-like
growth-factor-binding protein 3 is decreased in early-stage
operable pre-menopausal breast cancer. Int J Cancer 1995;62:
266-70.
58. Hankinson S, Willett W, Colditz G, et al. Circulating concen-
trations of insulin-like growth factor-1 and risk of breast can-
cer. Lancet 1998;351:1393-6.
Downloaded from http://aje.oxfordjournals.org/ by guest on November 28, 2015
59. Thorsen T, Lahooti H, Rasmussen M, et al. Oestradiol treat-
ment increases the sensitivity of MCF-7 cells for the growth
stimulatory effect of IGF-1. J Steroid Biochem Mol Biol
1992;41:537-40.
60. Hsieh CC, Trichopoulos D, Katsouyanni K, et al. Age at
menarche, age at menopause, height, and obesity as risk fac-
tors for breast cancer: associations and interaction in an inter-
national case-control study. Int J Cancer 1992;46:796-800.
61. De Stavola BL, Wang DY, Allen DS, et al. The association of
height, weight, menstrual and reproductive events with breast
cancer results from two prospective studies on the island of
Guernsey (United Kingdom). Cancer Causes Control 1993;
4:331^10.
62. Wang DY, De Stavola BL, Allen DS, et al. Breast cancer risk
is positively associated with height. Breast Cancer Res Treat
1997 ;43:123-8.
63. Zhang Y, Rosenberg L, Colton T, et al. Adult height and risk of
breast cancer among White women in a case-control study. Am
J Epidemiol 1996;143:1123-8.
64. Parazzini F, LaVecchia C, Negri E, et al. Anthropometric vari-
ables and risk of breast cancer. Int J Cancer 199O;45:397^tO2.
65. Lund E, Adami H-O, Meirik O. Anthropometric measures and
breast cancer in young women. Cancer Causes Control 1990;
1:169-72.
66. Moorman PG, Newman B, Millikan RC, et al. Participation
rates in a case-control study: the impact of age, race, and race
of interviewer. Ann Epidemiol 1999;9:188-95.
67. Lackland DT, Orchard TJ, Keil JE, et al. Are race differences
in the prevalence of hypertension explained by body mass and
fat distribution? A survey in a biracial population. Int J
Epidemiol 1992;21:236^45.
68. Striegel-Moore RH, Wilfley DE, Caldwell MB, et al. Weight-
related attitudes and behaviors of women who diet to lose
weight: a comparison of Black dieters and White dieters. Obes
Res 1996;4:109-15.
69. Abramson JH, Gofin R, Habib J, et al. Indicators of social
class: a comparative appraisal of measures for use in epidemi-
ological studies. Soc Sci Med 1982; 16:1739-46.
70. Gammon MD, John EM, Britton JA, et al. Recreational and
occupational physical activities and risk of breast cancer. J
Natl Cancer Inst 1998;90:100-17.
71. Mansfield CM. A review of the etiology of breast cancer. J Natl
MedAssoc 1993;85:217-21.
72. Michels KB, Greenland S, Rosner BA. Does body mass
index adequately capture the relation of body composition
and body size to health outcomes? Am J Epidemiol 1998;
147:167-72.
73. Seidell JC, Bakker CJG, van der Kooy K. Imaging techniques
for measuring adipose tissue distributiona comparison
between computed tomography and I. S. T. magnetic reso-
764 Hall et al.
nance. Am J Clin Nutr 1990;51:953-7.
74. Conway JM, Yanovski SZ, Avila NA, et al. Visceral adipose
tissue differences in Black and White women. Am J Clin Nutr
1995:61:765-71.
75. Lovejoy JC, de la Bretonne JA, Klemperer M, et al.
Abdominal fat distribution and metabolic risk factors: effects
of race. Metabolism 1996;45:1119-24.
76. Ortiz O, Russell M, Daley TL, et al. Differences in skeletal
muscle and bone mineral mass between Black and White
females and their relevance to estimates of body composition.
Am J Clin Nutr 1990;52:45-51.
77. Dowling HJ, Pi-Sunyer FX. Race-dependent health risks of
upper body obesity. Diabetes 1993;42:537^3.
78. Stevens J, Kiel JE, Rust PF, et al. Body mass index and body
girths as predictors of mortality in Black and White women.
Arch Intern Med 1992;152:1257-62.
Downloaded from http://aje.oxfordjournals.org/ by guest on November 28, 2015
Am J Epidemiol Vol. 151, No. 8, 2000