- Email: [email protected]
Increased risk of ‘high-risk’ colorectal adenomas in overweight men
GASTROENTEROLOGY
1993:104:137-144
Increased Risk of ‘High-Risk’ Colorectal Adenomas Overweight Men EKKEHARD WOLFGANG
BAYERDoRFFER,* GERD ALEXANDER MANNES,* THOMAS OCHSENKtiHN,* KoPCKE,* BALDUR WIEBECKE,’ and GUSTAV PAUMGARTNER*
*Medical Department Germany
II, Klmikum Grol3hadern. *lnstltute
of Biometry and Epidemiology,
Background: Epidemiologic studies have suggested that the incidence of colorectal carcinoma may be related to overnutrition, but retrospective analysis of its relation to the body mass index (BMI: kg/m2) has produced conflicting data. Methods: To avoid as many sources of statistical bias as possible, the relation between BMI and the presence of colorectal adenomas was investigated in a cross-sectional study. Results: Two thousand twelve consecutive colonoscoped patients were investigated (532 patients with malignancies or other conditions associated with weight loss were excluded). The relation between BMI and observed colorectal adenomas was evaluated by a logistic model controlling for other prognostic factors such as age, sex, and serum cholesterol level. The subgroup of “high-risk” adenomas with an increased risk of malignant transformation was positively associated with the BMI in men of the age group 50.5-68.1 years (quintiles Ill and IV: odds ratio for the top quintile vs. the lowest quintile, 3.21; 95% confidence interval, 1.15-8.98). Conclusions: It was concluded that the risk of developing high-risk adenomas tends to be increased in men who are overweight and that this association is independent of the positive association with the serum cholesterol level recently described.
T
he frequency
of colorectal
cancer
in industrialized
countries is several times higher than in developing countries.lV3 Nutritional factors have been sug-
gested to be a major reason for this difference.4*5 Epidemiologic studies have indicated a strong correlation between lesterol,
in
the per capita consumption of animal fat, choand total caloric intake on the one hand and
the incidence of colorectal cancer on the other.4*7-‘0 In support of the theory that nutritional factors are important in the genesis of colorectal cancer, other investigations have found an increased body mass index (BMI) to be associated with an increased mortality from cancer including colorectal cancer.“-” Conflicting data have been obtained on the association between BMI and colorectal cancer. Some studies de-
scribed others
and “Institute
a positive found
of Pathology, University of Munich,
association
no association
with
BMI”-“,
between
while
BMT and colorec-
tal cancer.20-24 Many
studies
elevated
studies
vascular
colon
Because
higher
colorectal
studies
studies
rates
of an
or case-
risks of cardio-
cardiovascular
death
risks
diseases
in younger
have
patients
cancer
and because
they are also posi-
with
an increased
BM1,‘2,21,22 these
associated
may be a major cancer
as cohort
to assess the mortality
disease.
significantly tively
on the health
BMI were designed
control
than
reporting
is being
source studied
quately designed to examine and colon cancer. Another
of statistical and
influence
not
ade-
the relation between BMI problem with interpreta-
tion may derive from the possibility cancer undetected at the baseline have a metabolic
are thus
bias when
that an advanced investigation may
on the BMI itself.23*24
To control for potential confounding risk factors, we applied a different approach to the problem. There is a lot of evidence from adenomatous
that most colorectal cancers arise po1yps2’ (this is referred to as the
adenoma-dysplasia-carcinoma tempted
to correlate
BMI with
sequence’“), the different
and we atstages in
this sequence. Using this approach, Mannes et al. have recently shown that the serum cholesterol level is positively
correlated with the risk for colorectal adewith a report by nomas,23 which is in good agreement for Tijrnberg et al., who found the same association colorectal carcinomas in a case control study.27 While nutritional factors are considered important in the genesis of colorectal cancer,28 the role of total caloric intake, unlike food composition, has not been investigated so far. Because BMI is a good quantitative measure of total caloric intake, we investigated the risk of colorectal
adenomas
in relation
to BMI.
Abbreviations used in this paper: 95% Cl, 95% confidence interval; BMI, body mass index; CHD, coronary heart disease. cc‘ 1993 by the American Gastroenterological Association 0016-5085/93/$3.00
138
SAVERDORFFER
GASTROENTEROLOGV
ET AL.
Patients and Methods
serum
All patients
(n = 2012) who underwent
January
addition
to personal
weight
1987 and March
data and medical
were recorded,
blood
(patients
cepted),
stool
diarrhea, origin,
also performed polypectomy tients
histories,
and in all patients for colonoscopy
in the
with
constipation,
occult
acute
as follow-up
examination for colorectal
inflammatory
and
bowel
or overt
hemorrhage
abdominal neoplasia.
or resection
In
height
pain
of
moved
CF-IB, In
polyps
logically.
CF-ITI,
addition
Because
after
or biopsy
material were
tubulovillous,
with
quency
teria of the World
Health
sia of adenomas
were also examined
corehisto-
histologically
in accordance
0rganization3’
was graded
unit
with the cri-
Epithelial
mild, moderate,
as
body weight
Fasting
analyses, weight
initiated.
About
day after
admission
and
Total serum cholesterol matically
at the Department
commercial heim,
again
before
concentration
kit (CHOD-PAP
8 AM and was
preparation.
was determined
of Clinical method,
quintile.
Chemistry
using a
Boehringer,
Mann-
Statistics To control
for potential
with
diseases
with
weight,
such
as previous
chronic
inflammatory
confounding
an obvious or current
bowel disease,
ciated with weight
loss, were excluded,
partial
Patients
colectomy.
excluded, influence
variables,
all
influence
on body
malignant
disease,
or other diseases
asso-
as were patients
with
with familial
polyposis
were also
because genetic factors may have a predominant on the development of adenomas.
The reasons
for exclusion
were recorded
in the protocol.
The remaining patients were divided into quintiles body mass range. To control for potential confounding ables, all factors
known
to be associated
of the vari-
with an increased
frequency of colorectal adenomasage, sex, serum cholesterol level, and indication for colonoscopy - were included in the analyses. The presence of adenomas, “high-risk” adenomas, and “low-risk” adenomas in patients in the individual BMI quintiles
were compared
with
respect
numbers
subjects
were
between
was also per-
the BMI
quintiles
small,
we com-
colorectal subjects
as defined
and the quintiles
Ninety-five
were computed values
ined
percent
in accordance
adenoma
with
the fre-
by Bray et a1.32 For
2 and
3 were
taken
4 and 5 as overweight.
as All
confidence
intervals
with Woolf
(95% CI)
s method.33
All P
were two tailed.
to age, sex,
were excluded
ulcerative colitis, partial colectomy, ing
enzy-
Germany).3’
patients
adenomas
Results
were taken on the bowel
of the frequency
of the association
dyspla-
bowel preparation
10% of the blood samples
age dependency colorectal
pro-
rel. 1988).
data were analyzed for men and women separately and for both together. Odds ratios for crude and adjusted percentages were calculated for BMI as categorical data in the highest and lowest
or severe.30
were taken between before
package,
in normal-weight
Five hundred
blood samples
for these
the BMDP-LR
of the observed
in overweight
Laboratory Tests 9 AM on the day of admission
was adjusted
using
analysis
and when
the association
or familial
continuous
in the age quintiles.
pared
and of pa-
fiber-optic
adenomas,
Additionally,
cancer
evaluation,
classified
or villous
for colonoscopy.
of age and serum choles-
software
of the marked
of colorectal
disease
macroscopic
Statistical
formed
of patients
and CF-101
to
Adenomas
tubular,
(BMDP
unwas
Endoscopy and Histology with Olympus
by logistic
BMI and the observed
Colonoscopy
in our endoscopy
for the influence analysis
gram
No. 1
data, respectively.
parameters
these
lonoscopes.
indication
of adenomas
normal
was performed
and
terol level the presence
ex-
polyposis.
Colonoscopy
and frequency
To standardize
medicine
the BMI was calcu-
were
and suspected
with chronic
colonoscopy
of internal
1988 were evaluated.
lated.‘” Indications
known
variables
the cecum in our department
between
level,
The t test and the x2 test were used to examine
Patients and Study Design including
cholesterol
Vol. 104.
108 with
chronic
diseases
of 2012 patients examthe study: 59 patients with
thirty-two
from
71 with Crohn’s disease, 48 with 344 with malignant disease, includ-
colorectal associated
The characteristics
cancer,
and
with
weight
of the remaining
10 with
other
loss. 1480 patients,
such as mean age, mean serum cholesterol mean BMI are shown in Table 1. Patients
level, and with ade-
nomas were statistically significantly older and had a higher serum cholesterol level; however there was no difference in BMI (Table 1). With respect to the distributions
of the patients’
age and serum cholesterol
level
in the BMI quintiles, it was found that patients in the highest BMI quintile were younger and had lower serum cholesterol levels than those in the lower BMI quintiles. Because the frequency of adenomas is associated with age and serum cholesterol leve1,23 these parameters were considered in the statistical analysis. As Table 2 shows, the number of observed colorectal adenomas is clearly related to older age. The observed numbers of patients with colorectal adenomas were significantly higher in men than in women except in the oldest age quintile where the percentage of adenomas was similar for both sexes (Table 2). The relation
between
BMI and the presence
of colo-
January
Table
OVERWEIGHT
1993
of 1480
1. Characteristics
Patients
With (n = 288)
and Without
AND
(n = 1 192) Colorectal
COLORECTAL
With adenoma (n = 172) 60.5 237 24.57
Age W Cholesterol (mg/dL) BMI (kg/m2) NOTE. Values
are means
rectal adenomas
r 11.5 I!Z 60 -+ 3.23
in the age quintiles
no overall
the presence
of colorectal
With adenoma (n = 116)
f 15.3 + 67 f 3.33
64.5 242 23.22
ble 3). In the oldest
in Table
between
adenoma
in the younger
years and >39/<50.5
is shown
association
and >58.6/<68.1
association
51.3 218 24.70
BMI and
association showed
in the age quintiles
(Table
i 10.7 f 75 + 3.26
We found
no
with the high-risk
age quintiles,
i.e., under
39
0.14,
years neither
for men
nor for
(Table
for develop-
found
no association
analyzed
nomas,
for a slight
association
dysplasia,
adenomas
of 22 cm, and multiple
been
considered
adenomas
to harbor
malignant.
teria
were
and for women
evaluated
Next we analyzed
separately (Table
(26
an increased
All adenomas
men,
with
a cm)
risk of
that met these crifor all patients,
for
and low-risk
To
BMI quintile
of 3.17 for high-risk
13.70,
3). In the age quintile
adenoma
obtain
58.6 and 68.1 years of age, the odds ratio lated to be 2.92 (CI, 0.69-12.41,
Table
of high-risk positive
we
and BMI and or low-risk
ade-
but not significant
between
58.6 and
68.1
with
high-risk
information
between
low-risk
we calculated
for
adenomas.
adenoma
influence
adenomas
The
ratio
of low-risk
in the BMI quintiles
of age as indicated influence
and those
reveals
and an ob-
by the highest
of the BMI,
val-
which
is
indicated by the rapidity by which the ratio declines. As calculated by the log linear model, the influence seems
between
and
was calcu-
3). When
further
the ratio
ues and also a clear
(CI, 0.73-
of men
mentioned
3).
men
vious
adenomas.
For the age quintile 50.5-58.6 years in men we calculated an odds ratio between the lowest and the highest Table
years (Table
in men
in age quintiles,
adenoma
in the age group
high-risk
3).
high-risk
except
in the above
between
nomas
severe
was no such
adenomas
3).
In the case of women adenomas
there
with odds ratios of 0.27,
respectively,
subgroup
becoming
k 16.5 +- 57 + 3.82
to the BMI when compared
adenomas,
and 0.19,
age groups
relation
also in neither
have
49.8 223 23.97
age quintile
ing malignancy varies, depending on a number of different factors. Villous or tubulovillous adenomas, adediameter
Without adenoma (n = 671)
3). The low-risk
an inverse
years.
women (data not shown). The potential of colorectal
with
(n = 787)
f SD.
3. We found >50.5/<58.6
Women
Without adenoma (n = 521)
139
Adenomas
Men (n = 693)
Characteristic
ADENOMAS
we
to be strongest 68.1 years
next older
in the age group
and weaker
BMI quintiles
in the next
(Table
between
58.6
younger
and
4).
pooled the two age groups, i.e., men from 50.5 to 68.1 years, the odds ratio was significant at 3.21 and a CI of
The total of observed colorectal adenomas of all age groups and the subgroups of high-risk and low-risk
1.15-8.98
adenomas in the BMI quintiles of men and women are shown in Table 5. Statistical analysis applying a logis-
(Table
age quintiles revealed
3). Statistical
combined
no significant
analysis
employing association
of these
the logistic between
two
model
tic model
the overall
group of adenomas and BMI (P = 0.106), or between adenomas and cholesterol (P = 0.185) for men in age quintiles >50.5 and <58.6 and >58.6 and <68.1 (Ta-
Table 2. Numbers
Age <39 <50 <58.6 <68.1 >68.1
of Patients
All patients 300 294 296 298 292 1480
and Those With Adenomas Adenomas 10 34 61 77 106 288
(3.3) (11.6) (20.6) (25.8) (36.3) (19.5)
(%)
showed
a significant
association
between
the
frequency of adenomas sexes. The association
in all patients and age for both between cholesterol and ade-
nomas
in men.
was significant
BMI was significantly
in the Age Quintiles Men
Adenomas
119 142 154 142 136 693
8 26 39 48 51 172
(%)
(6.7) (18.3) (25.3) (33.8) (37.5) (24.8)
Women 181 152 142 156 156 787
Adenomas 2 8 22 29 55 116
(%)
(1.1) (5.3) (15.5) ( 18.6) (35.3) (14.7)
GASTROENTEROLOGY Vol. 104, No. 1
BAYERDORFFER ET AL.
140
Table 3. Observed
High-Risk
and Low-Risk
Adenomas
in Different
All patients
BMI
”
>50.5
In
and ~58.6
Men, and Women)
Men High-nsk
Low-nsk
adenomas
adenomas
In WI
[n (Dl
BMI
Adenomas (%)
Age Groups (All Patients,
WI
Women
Adenomas w) n
In
(WI
High-risk
Low-nsk
adenomas
adenomas
(WI
[n
-
Adenomas @)
In (WI
BMI
n
[n
High-nsk
Low-nsk
adenomas
adenomas
cwl
[n WI
In (WI
years of age
<21.35
59
I5 (25.4)
7 (11.9)
8 (13.5)
<21.97
36
9 (25.0)
3 (8.3)
6 (16.7)b
~20.77
24
6 (25.0)
4(16.7)
2 (8.3)
<23.31
69
17 (24.6)
8 (I 1.6)
9 (13.0)
~23.73
29
9 (31.0)
5 (17.2)
4 (13.8)
<22.80
33
7 (21.2)
3 (9.1)
4(12.1)
<24.91
52
I3 (25.0)
13 (25.0)
0 (0.0)
c25.09
32
8 (25.0)
7 (21.9)
l(3.1)
<24.75
26
7 (26.9)
7 (26.9)
0 (0.0)
<27.00
72
I9 (26.4)
IO (13.9)
9 (12.5)
<27.20
37
12 (32.4)
6(16.2)
6 (16.2)
<26.86
35
6f17.1)
3 (8.5)
3 (8.5)
>27.00
43
7 (16.3)
5 (I 1.6)
2 (4.7)
>27.20
19
6(31.6)
5 (26.3)
I (5.3)0
>26.86
24
1 (4.2)
0 (0.0)
1 (4.2)
2 (5.7)
>58.6
and ~68.1
years of age
<21.35
58
I2 (20.7)
6 (10.3)
6 (10.3)
<21.97
28
9 (32.1)
3 (10.7)C
6 (21.4)’
~20.77
35
5 (14.3)
3 (8.6)
~23.31
66
I6 (24.2)
7 (10.6)
9 (13.6)
~23.73
29
IO (34.5)
6 (20.7)
4 (13.8)
<22.80
30
3 (10.0)
2 (6.7)
1 (3.3)
<24.91
65
I I (16.9)
7 (10.8)
4 (6.1)
<25.09
31
4(12.9)
2 (6.4)
2 (6.4)
~24.75
36
8 (22.2)
4(11.1)
4(ll.l)
~27.00
58
I7 (29.3)
11 (19.0)
6 (10.3)
<27.20
27
11(40.7)
7 (25.9)
4 (14.8)
<26.86
29
5 (17.2)
4 (13.8)
1 (3.4)
>27.00
51
13 (25.5)
I2 (23.5)
I (2.0)
>27.20
27
8 (29.6)
7 (25.9)
I (3.7)d
>26.86
26
6 (23.1)
5 (19.2)
1 (3.8)
0 (0.0)
>68.
I years of age
<21.35
60
I6 (26.7)
I4 (23.4)
2 (3.3)
<21.97
24
7 (29.2)
5 (20.8)
2 (8.4)
<20.77
32
9 (28.1)
9 (28.1)
<23.31
69
23 (33.3)
20 (29.0)
3 (4.3)
<23.73
31
I3 (41.9)
IO (32.3)
3 (9.6)
<22.80
44
10 (22.7)
10 (22.7)
0 (0.0)
<24.91
60
34 (56.7)
I9 (31.7)
I5 (25.0)
<25.09
27
13 (48.1)
I I (40.7)
2 (7.4)
~24.75
33
20 (60.6)
8 (24.2)
I2 (36.4)
<27.00
56
I8 (32.1)
15 (26.8)
3 (5.3)
~27.20
27
9 (33.3)
8 (29.6)
1 (3.7)
<26.86
27
IO (37.0)
7 (25.9)
3(11.1)
>27.00
47
I3 (27.6)
I2 (25.5)
l(2.1)
>27.20
27
9 (33.3)
7 (25.9)
2 (7.4)
>26.86
20
4 (20.0)
4 (20.0)
0 (0.0)
<21.97
64
I8 (28.1)
6 (9.4)?
I2 (18.7)’
~23.73
58
I9 (32.7)
I I (19.0)
8 (13.7)
<25.09
63
12 (19.1)
9(14.3)
3 (4.8)
<27.20
64
23 (35.9)
13 (20.3)
10(15.6)
>27.20
46
I4 (30.4)
I2 (26. I)
>50.5
and ~68.1
years of age
NOTE. In the bglstlc model the P values for each prognostic respectively. n = number of patients. Odds ratio for the lOWeSt and highest e3.21
(Cl. 1.15-8.98;
associated (Table
P < 0.05);
with
quIntlIe
‘0.19
was ‘3.17
(Cl, 0.017-0.92;
the presence
5). The data were
factor for tugh-risk adenomas
(Cl, 0.73-
13.70:
P = 0.19);
(for men I” the age quintlIes
b0.27
(Cl, 0.034-2.26:
of adenomas
split for males
in women
model.
Cholesterol
Table 4. Ratio of Low-Risk
the Combination
BMI
n
Adenomas (n)
Ratio low-risk/ high-risk adenomas
<21.97 <23.73 c25.09 <27.20 >27.20
36 29 32 37 19
9 9 8 12 6
2.0 0.8 0.14 1.0 0.2
NOTE. n = number of patients.
nor
low-risk
Adenomas and High-Risk of III and IV
>50.5 and ~58.6 years of age
<2.92
with adenomas,
(C,, 0.69-12.4,;
for BMI and
P = 0.14):
4O.,4
p =0.185for cholesterol,
(C,, O.Olg_o.g9;
p < 0.05);
III to form
with
28 29 31 27 27
weight
small
numbers
of patients
the BMI quintiles group
(men
II and
with
a BMI
The
association
logistic
between
analysis high-risk
showed
a
adenomas
and overweight (P < 0.041) for men in the age quintiles ~50.5 and <68.1 years (Table 6). The odds ratios
as a Function of Age for Men in the Age Quintiles
~58.6 and ~68.1 years of age
n
a normal
a BMI >25.09).
significant
ade-
Adenomas
we combined
from 21.97 to 25.09) according to Bray,32 and the quintiles IV and V to form an overweight group (men
and BMI were significantly high-risk
P = 0.23);
were p = 0.106
In view of the relatively
and females,
risk and low-risk adenomas, only age was significantly associated with the presence of adenoma applying a logistic
3 and 4 combmed)
P < 0.025).
because they are known to have different rates of colorectal adenoma (Table 2). 23 In the subgroups of high-
associated with neither nomas (Table 5).
2 (4.3)’
Adenomas (n)
Ratio low-risk/ high-risk adenomas
9 10 4
2.0 0.67 1.0
11 8
0.57 0.14
268.1
years of age
n
Adenomas (n)
Ratio low-risk/ high-risk adenomas
24 31 27 27 27
7 13 13 9 9
0.4 0.3 0.18 0.13 0.28
>50.5
Ill, IV, V, and
and ~68.1 years of age
n
Adenomas (n)
Ratio low-risk/ high-risk adenomas
64 58 63 64 46
18 19 12 23 14
2.0 0.72 0.33 0.76 0.17
OVERWEIGHT AND COLORECTAL ADENOMAS
January 1993
in All Patients, in Men, and in Women
Table 5. Observed High-Risk and Low-Risk Adenomas All patients
BMI <21.35 <23.31 <24.9 1 <27.00 >27.00 Total Age Cholesterol BMI
All adenoma [n (%)I 53 63 71 61 40 288
(17.9) (21.4) (24.1) (20.6) (13.7) (19.5)
Men
Hugh-risk adenoma [n (%)I
Low-nsk adenoma
27 30 36 33
26 (8.8) 33 (1 1.2) 35 (11.9)
(9.1) (10.2) (12.2)
(11.1) 24 (8.2) 150 (10.1)
BMI
In (%I
28 (9.5) 16 (5.5) 138 (9.4)
141
c21.97 ~23.73 c25.09 ~27.20 >27.20
Women
All adenoma [n @)I
High-risk adenoma
Low-nsk adenoma
In (%)I
In (%)I
BMI
30 (22.3) 40 (28.6) 34 (24.4) 38 (27.3) 30 (2 1.5) 172 (24.8)
13 (9.7) 17 (12.1) 24(17.2) 22 (15.8) 21 (15.0) 97 (14.0)
17 (12.6) 23 (16.5) 10 (7.2) 16 (1 1.5) 9 (6.5) 75 (10.8)
~20.77 <22.80 ~24.75 <26.86 >26.86
All adenoma
Low-risk adenoma
[nWI
Hugh-nsk adenoma [n (%)I
24 (15.3) 22 (13.9) 37 (22.8) 21 (13.4) 12 (7.9) 116 (14.7) 10.00 1 0.32 co.025
13 (8.3) 12 (7.6) 17 (10.5) 9 (5.7) 5 (3.3) 56 (7.1)
11 10 20 12 7
In WI (7.0) (6.3) (12.3) (7.7) (4.6)
60 (7.6) 10.001 0.68 0.3 I
NOTE.The frequency of adenomas is grven in percent. The lower part of the table shows the P values of each prognostic factor for each category of patients as calculated in the logisbc model (stepwise elimination procedure).
for normal weight men vs. overweight
men were 0.95
(95% CI 0.61-1.48; Table 6). In the age group 58.668.1, overweight men had the highest relative risk of developing
high-risk
Table 6). The increased
men of developing
adenomas was still significant
Table
Studies in the past have reported either a positive association lorectal
6). We also analyzed the question various conditions
of high-risk
ently associated with overweight to tubulovillous
as to whether
the
adenomas were differin men. With respect
or villous adenomas,
we identified
a
trend toward a positive association with the BMI which, however, was not significant (Table 7). In the subgroup of adenomas with severe dysplasia and for the subgroup with large adenomas (22 cm) or numer-
Table 6. Observed High-Risk and Low-Risk Adenomas
High-risk adenoma Normal weighta Ovenveightb Odds ratio (confidence limits) P Low-risk adenoma Normal weight” Overweight* Odds ratio (confidence limits) P
between BMI and the incidence
carcinoma”-”
reviewing
or no association.20
of co-
Waaler”
1,700,OOO subjects found a twofold
higher
frequency of colorectal cancer in the highest BMI group in men aged 50-64 years but no association in women. Nomura et al.,l’ who reported on 8000 men, found a 2.5-fold increased frequency of colorectal cancer 255
in the highest BMI quintile
years but no association
in the age group
in the age group 45-54
years. On the basis of data obtained from 750,000
sub-
in Men With Normal Weight and Overweight Men 50.5-58.6 years of age
All men
we found no associa-
BMI and Mortality From Colorectal Cancer
when both age groups CI 1.02-3.90;
adenomas),
Discussion
risk
high-risk colorectal
were pooled (odds ratio 2.00,95%
(26
tion with BMI.
adenomas, with an odds ratio of
3.53 (95% CI 1.28-9.74; for overweight
ous adenomas
Men 50.5-68.1 years of age
Men 58.6-68.1 years of age
No adenoma
Adenoma
No adenoma
Adenoma
No adenoma
Adenoma
No adenoma
Adenoma
232 208
50 47
52 38
12 11
53 35
6 14
105 73
18 25
0.95 (0.61- 1.48) 0.83 232 208
23 20
1.03 (0.55- 1.93) 0.92
1.25
5 7
52 38 1.91
3.53 (1.28-9.74) <0.025
(0.50-3.15) 0.63
(0.57-6.44) 0.29
a 5
53 35 1.06
2.00 (1.02-3.90) <0.05
(0.32-3.50) 0.93
105 73 1.33
13 12 (0.57-3.07) 0.51
NOTE. In the logistic model, the P values for high risk adenomas in overweight men for each prognostic factor (for men in the age quintiles 3 and 4) were P = 0.302 for cholesterol and P = 0.041 for BMI. ‘As defined by Braya and corresponding to BMI quintiles 21.97-25.09 in men. bOverweight means all men above these limits.
142
BAYERDtiRFFER ET AL.
GASTROENTEROLOGY Vol. 104, No. 1
Table 7. Subgroup
Analysis of High-Risk Normal Weight and Overweight
Adenomas:
Observed
Normal weightb Overweight” Odds ratio (confidence limits) P
Adenoma”
232 208
1.06
Men 50.5-68.1
years of age
years of age
jects, Lew and Garfinkel” in the frequency weight
not significant,
4 5
are considered. to BMI quintiles limits.
reported
2 1.97-25.09
a 1.7-fold
cancer
increase
in the highest
in men of all ages and a 1.2-fold, increase
in women.
but
Our study investi-
gated colorectal adenomas as the benign precursor lesion of colorectal cancer and found a very similar relationship
between
adenomas
BMI
and
the
in men. The number
observed
high-risk
of high-risk
adenomas
with an enhanced potential for malignant development in all age groups was increased in the highest BMI quintile
by a factor
same magnitude
of 1.9-fold,
as that reported
kel” for colorectal
cancer.
years was analyzed,
the relative
colorectal
adenomas
risk adenomas nomas
increased,
the potential spective
When
increased
decreased for malignant
which
2.27
creased
an influence
of BMI on
transformation
in the re-
age group.
Adenomaa
105 73
(0.6 l-8.33) 0.22
risk for high-risk
A more marked ber of observed
1.96
8 11
(0.77-5.10) 0.16
colorectal
adenomas
relation
between
high-risk
other
diseases.
overweight
It has been
argued
risks
BMI and the num-
colorectal
has a major influence
in older
by competing
adenomas
4) or carcinoma**,‘” in the age group 50-68 also been reported for the relation between that
(Table years has BMI and
longstanding
on the general
health
risk.‘4*20
Different Potential for Developing Malignancy
50-68
ade-
4 6
overweight women may be masked in this age group (Table 4).
risk factor for high-risk to 3.2. The rate of low-
No adenoma
in men.
by Lew and Garfinthe age group
Adenoma”
53 35
is of the
as the rate of high-risk
suggesting
No adenoma
0.44
of colorectal
category
Adenomaa
1.72 (0.43-6.66)
aOnly tubulovillous and villous adenomas bAs defined by 8ray33 and corresponding ‘Overweight means all men above these
With
Men 58.6-68.1
52 38
(0.54-2.04) 0.86
in Patients
years of age No adenoma
20 19
or Villous Adenomas
Men 50.5-58.6 All men No adenoma
Tubulovillous
Interestingly, only
between
BMI
have an increased increased
this and
found
colorectal
risk of malignant
risk of developing
ber of factors,
study
cancer
an association
adenomas,
which
degeneration. is linked
such as the histological
An
to a num-
type tubulo
vil-
lous, or villous adenoma, adenoma with severe dysplasia, adenoma larger than 2 cm, or more than six
In the case of women, the situation was different. When the number of observed high-risk adenomas in the highest BMI quintile was compared with that in the lowest BMI quintile, we found a small, nonsignifi-
concomitant adenomas. 34 Though not significant, our data suggest that the development of adenomas and their potential for malignant transformation are in-
cant negative nonsignificant,
fluenced by a number cal factors.
association positive
for all ages but a small, also association in women aged
58-68 years. This finding is more or less in agreement with former studies of colorectal cancer, which found either no association’” or a weakly positive, but not significant, association ” between BMI and the risk of colorectal carcinoma for women. These data may be illustrated by the age dependence of colorectal adenomas, which shows that the risk of colorectal adenomas in women is 3.5-6 times lower in those <50 years of age and increases at age >68 years to a risk similar to that in men (Table 3). This suggests that menopausal women have a risk of developing colorectal cancer similar to that of men. However, an in-
of epidemiological
and etiologi-
BMI and Mortality From Other Diseases The influence of overnutrition on health in general has been investigated extensively, especially for coronary heart disease.35-39 Reports by Simopoulos et a1.,14 Lew,37 and Harris et al.,” suggested that overweight has to be persistent to have an effect on the frequency of colorectal cancer. A number of limitations of the use of BMI as a measure of overnutrition have been discussed.24 However, BMI has been accepted as the best means of estimating overweight. 39 Self-reported data on height and
January
1993
weight
OVERWEIGHT
may not be very reliable.24
Not all the authors
of previous epidemiologic studies that investigated the relation of BMI and colorectal cancer” were able to measure
weight
ported
data.
and height
The
data
sively on clinically other
major
studies
measured
source
itself. 24 BMI
tional
factors,
monal
study
balance.41
in BMI colorectal
of advanced
is influenced
Calories
exclu-
and heights.
An-
cancer
cancer
on body
not only
by nutri-
activity4’
and hor-
but also by physical
may influence
are based
weights
of error
may be the impact
weight
but had to rely on self-re-
in this
derived
from
alcohol
also
BMI.42
The
frequency
were controlled
data
for
for the major
and cholesterol
colorectal risks,
level in the logistic
risk of colorectal
especially increased dures, have
such as age, sex,
cancer.
However, factors that with an in-
For instance
alcohol,
in the form of beer, was associated with an risk of colorectal cancer.43 Surgical proce-
such as gastric also been
resection
reported
with
an in-
risk of colorectal cancer and colorectal adenomas. 44,45Because we were unable to obtain the family histories of all the patients, some important risk might
Risks
The imponderable the results of this study risks. Colorectal seventh decade,
encountered on considering is the influence of competing
by numerous
with an earlier peak mortality cade. Early insurance studies association
diseases
in the sixth or fifth deraised the question of a
of overweight
that from
competing
in the cancer
cancer.47
and general Coronary
mortalheart
dis-
ease (CHD) and cancer were found to be the most common causes of mortality. Other epidemiologic studies showed similarities in the incidence rates of CHD and cancer studies primarily
smoking
subclinical
in industrialized countries.‘,* Later designed to investigate risk factors
for CHD, partly confirmed the positive association between overweight and cancer mortality,” but also produced conflicting data. 23 However, the main source of bias may be attributed to the much earlier mortality peak in CHD. Death rates of coronary heart disease in the fifth through the seventh decade are much higher than those of colorectal cancer. Manson et a1.24 also noted
disease
secondary diseases hypertension. We did not rectal cancer.
the associahad at least
biases.
They
the
controlled
for
such as diabetes
and
of overweight
any of these
cancer.
diseases
than the peak
The death
are rather
of colorectal
cancer
mortality the relation
is less
inof
of colorectal
by 5-20
mortality
early the
lesion
that
years. 48 The
very
precedes impact
cancer
major
mortality
has been underestimated
cause of a lack of monitoring
low,
of BMI
risks. The design
biases,
or lower
important
in our study is the investigation
precursor
noma
rates of the
to the association
and BMI had at least one of these three
provement
but as-
were 50 years old on
at ages <50
which are associated with BMI quintiles. 24 Therefore
by
lesion of colo-
15 years younger
that our approach
investigating
factors,
precursor
our patients
is about
by competing
studies
for
and inappropriately
frequency
fluenced
to control loss caused
Thus,
competing
failed
and to assess weight
of colorectal
suggesting and
investigating and mortality
as the benign
which
143
weight
consider
sessed adenomas
im-
of the be-
colorectal
carci-
of overweight
on
in most studies
be-
for competing
risks.
Conclusions pothesis
cancer has its peak mortality so that the risk of colorectal
may be influenced
ity, including
other
have been missed.46
Competing
possible
one of three cigarette
nign
and cholecystectomy,
to be associated
creased
factors
body
ADENOMAS
risk of dying from coronary
all the studies
between
major
model.
we could not control for several minor have also been reported to be associated creased
tion
disease,
mortality adenomas
COLORECTAL
that besides the competing heart
average,
Risk Factors for Colorectal Cancer
AND
This prospective investigation supports the hythat the BMI influences the risk of developing
colorectal adenomas at least in men, who have a high risk for malignant degeneration. This study investigated the benign precursors of colorectal carcinomas, and the results confirm epidemiologic data found to apply to colorectal men with a high additional
risk
carcinoma. The data suggest that BMI, especially if they have such
factors
as a high
level, ought to receive prophylactic prevent colorectal carcinoma.
serum
cholesterol
colonoscopy
to
References 1. Doll R. The geographical distribution of cancer. Br J Cancer 1969;23: l-8. 2. Burkitt DP. Epidemiology of cancer of the colon and rectum. Cancer 197 1;28:313. 3. Weisburger JH, Reddy BS, Wynder EL. Colon cancer: its epidemiology and experimental production. Cancer 1977;40:24 142420. 4. Wynder EL, Reddy BS. Metabolic epidemiology of colorectal cancer. Cancer 1974;34:80 l-806. 5. Howell MA. Diet as an etiological factor in the development of cancers of the colon and rectum. J Chron Dis 1975;28:67-80. 6. Graham S, Haenszel W, Bock FG, Lyon JL. Need to pursue new leads in the epidemiology of colorectal cancer. JNCI 1979; 63:879-88 1.
144
BAYERDGRFFER
7. Wynder EL, Kajitani T, lshikawa S, Dodo H, Takano A. Envrronmental factors of cancer of the colon and rectum. II. Japanese epidemrological data. Cancer 1969;23: 12 IO- 1220. 8. Haenszel W, Berg JW. Segi M, Kurihara M, Locke FB. Large-bowel cancer in Hawaiian Japanese. JNCI 1973;5 1: 1765- 1779. factors and cancer incidence 9. Armstrong D, Doll R. Environmental and mortality in different countries with special reference to dietary practices. Int J Cancer 1975; 15:617-63 1. 10. Wynder EL. The epidemrology of large bowel cancer. Cancer Res
11. 12. 13. 14. 15. 16. 17. 18. 19. 20.
21. 22.
23.
24. 25. 26.
27.
28. 29.
30.
GASTROENTEROLOGY
ET AL.
1975;35:3388-3394. Tannenbaum A. The genesis and growth of tumors. II. Effects of caloric restriction per se. Cancer Res 1942;2:460-467. Mann GV. The influence of obesrty on health. New Engl J Med 1974;29 1:178-232. Lowenfels AB, Anderson ME. Diet and cancer. Cancer 1977;39:1809-1814. Simopoulos AP, Van ltallie T. Body weight, health, and longevrty. Ann Int Med 1984; 100:285-295. Bray GA. Complications of obesity. Ann Intern Med 1985; 103: 1052-1062. Garfinkel L. Overweight and cancer. Ann Int Med 1985; 103: 1034- 1036. Lew EA. Garfinkel L. Variations in mortality by weight among 750000 men and women. J Chron Dis 1979;32:563-576. Nomura A, Heilbrun LK, Stemmermann GN. Body mass index as predictor of cancer in men. JNCI 1985;74:3 19-323. Waaler HT. Height, werght and mortality: The Norwegran experience. Acta Med Stand Suppl 1984;679: l-56. Harris T, Cook E, Garrison R, Higgins M, Kannel W, Goldman L. Body mass index and mortality among nonsmoking older persons. JAMA 1988;259: 1520- 1524. Feinleib M. Epidemiology of obesity in relation to health hazards. Ann Intern Med 1985; 103: 10 19- 1024. Garrison RJ. Castelli WP. Weight and thirty-year mortality of men In the Framingham Study. Ann Intern Med 1985; 103: 10061009. Mannes GA, Maier A, Thieme C, Wiebecke B, Paumgartner G. Relation between the frequency of colorectal adenoma and the serum cholesterol level. New Engl J Med 1986;3 15: 1634- 1638. Manson JA, Stampfer MJ, Hennekens CH, Willett WC. Body weight and longevity. JAMA 1987;257:353-358. Muto T, Bussey HJR, Morson BC. The evolution of cancer of the colon and rectum. Cancer 1975;36:225 l-2270. Day DW, Morson BC. The adenoma-carcinoma sequence. In: Morson BC, ed. The pathogenesis of colorectal cancer. Phrladelphia: Saunders, 1978: 126- 152. Tornberg SA, Holm LE, Carstensen JM, Eklund GA, Odont D. Risks of cancer of the colon and rectum in relation to serum cholesterol and beta-lipoprotein. New Engl J Med 1986;315: 1629-1633. Wynder EL. Nutrition and cancer. Federation Proc 1976;35: 1309-1315. National Institute of Health Consensus Development Conference Statement. Health implrcations of obesity. Ann Int Med 1985; 103: 1073- 1077. Morson BC, Sobin LH. Histological typing of intestinal tumors. lnternattonal histological classification of tumours. Geneva: World Health Organization, 1976.
31.
32. 33.
34.
35.
36. 37. 38. 39. 40.
41.
42. 43.
44.
45.
46.
47. 48.
Vol. 104,
No. 1
Siedel J, Hagele EO, Ziegenhorn J, Wahlefeld AW. Reagent for the enzymatic determination of serum total cholesterol with Improved lrpolytic efficiency. Clin Chem 1983;29: 1075- 1080. Schlesselmann JJ. Case control studies: design, conduct, analysis New York: Oxford University, 1982. Bray GA. Obesity in America. Bethesda, Maryland: National Institutes of Health, 1979. National Institutes of Health Publication No. 79-359. Hermanek P, Fruhmorgen P. Guggenmoos-Holzmann I, Altendorf A, Matek W. The malignant potential of colorectal polyps-A new statrstical approach. Endoscopy 1983; 15: 16-20. Glueck CJ, Taylor HL, Jacobs D. Morrison JA, Beaglehole R, Williams OD. Plasma high-density lipoprotein cholesterol: Association with measurements of body mass. Circulation 1980; 62(Suppl 4):62-69. Jarrett RJ, Shipley MJ, Rose G. Weight and mortality in the Whrtehall Study. Br Med J 1982;285:535-537. Lew EA. Mortality and Weight: Insured lives and the American Cancer Society Studies. Ann Intern Med 1985; 103: 1024- 1029. Sidney S, Friedman GD, Siegelaub AB. Thinness and Mortalrty. Am J Public Health 1987;77:3 17-322. Van ltallie TB. Health implications of overweight and obesity in the United States. Ann Int Med 1985; 103:983-988. Johnson ML, Burke BS, MayerJ. Relative importance of Inactivity and overeating in the energy balance of obese high school girls. Am J Clin Nutr 1956;4:37-44. Pariza MW. Boutwell RK. Hrstorical perspective: calories and energy expenditure in carcinogenesis. Am J Clin Nutr 1987;45: 151-156. Ziegler RG. Alcohol-nutrient interactions in cancer etiology. Cancer 1986;58: 1942- 1948. Pollack ES, Nomura AMY, Heilbrun LK, Stemmermann GN, Green SB. Prospective study of alcohol consumption and cancer. New Engl J Med 1984;310:617-621. Caygill CPJ, Hill MJ, Hall CN. Kirkham JS, Northfield TC. Increased risk of cancer at multiple sites after gastric surgery for peptic ulcer. Gut 1987;28:924-928. Mannes GA, Weinzrerl M, Stellaard F, Thieme Ch, Wiebecke B, Paumgartner G. Adenomas of the large intestine after cholecysctectomy. Gut 1984;25:863-866. Cannon-Albright. Skolnick MH, Bishop DT, Lee RG, Burt RW. Common inheritance of susceptibility to colonic adenomatous polyps and associated colorectal cancers. New Engl J Med 1988;3 19:533-537. Tannenbaum A. Relationship of body weight to cancer incidence. Arch Path 1940;30:509-5 17. Stryker SJ. Wolff BG, Culp CE, Libbe SD, llstrup DM, McCarty RL. Natural history of untreated colonic polyps. Gastroenterology 1987;93: 1009- 10 13.
Received January 4, 1991. Accepted August 18, 1992. Address requests for reprints to: Ekkehard Bayerdorffer, M.D., Medical Department II, Klinikum GroRhadern, University of Munich, MarchioninistraDe 15, D-8000 Munchen 70, Germany. The authors thank Dr. P. Dirschedl and P. Kampe for valuable help in performing statistical calculations and Prof. M. Knedel (Institute of Clinical Chemistry, University of Munich) for his cooperation.
1993:104:137-144
Increased Risk of ‘High-Risk’ Colorectal Adenomas Overweight Men EKKEHARD WOLFGANG
BAYERDoRFFER,* GERD ALEXANDER MANNES,* THOMAS OCHSENKtiHN,* KoPCKE,* BALDUR WIEBECKE,’ and GUSTAV PAUMGARTNER*
*Medical Department Germany
II, Klmikum Grol3hadern. *lnstltute
of Biometry and Epidemiology,
Background: Epidemiologic studies have suggested that the incidence of colorectal carcinoma may be related to overnutrition, but retrospective analysis of its relation to the body mass index (BMI: kg/m2) has produced conflicting data. Methods: To avoid as many sources of statistical bias as possible, the relation between BMI and the presence of colorectal adenomas was investigated in a cross-sectional study. Results: Two thousand twelve consecutive colonoscoped patients were investigated (532 patients with malignancies or other conditions associated with weight loss were excluded). The relation between BMI and observed colorectal adenomas was evaluated by a logistic model controlling for other prognostic factors such as age, sex, and serum cholesterol level. The subgroup of “high-risk” adenomas with an increased risk of malignant transformation was positively associated with the BMI in men of the age group 50.5-68.1 years (quintiles Ill and IV: odds ratio for the top quintile vs. the lowest quintile, 3.21; 95% confidence interval, 1.15-8.98). Conclusions: It was concluded that the risk of developing high-risk adenomas tends to be increased in men who are overweight and that this association is independent of the positive association with the serum cholesterol level recently described.
T
he frequency
of colorectal
cancer
in industrialized
countries is several times higher than in developing countries.lV3 Nutritional factors have been sug-
gested to be a major reason for this difference.4*5 Epidemiologic studies have indicated a strong correlation between lesterol,
in
the per capita consumption of animal fat, choand total caloric intake on the one hand and
the incidence of colorectal cancer on the other.4*7-‘0 In support of the theory that nutritional factors are important in the genesis of colorectal cancer, other investigations have found an increased body mass index (BMI) to be associated with an increased mortality from cancer including colorectal cancer.“-” Conflicting data have been obtained on the association between BMI and colorectal cancer. Some studies de-
scribed others
and “Institute
a positive found
of Pathology, University of Munich,
association
no association
with
BMI”-“,
between
while
BMT and colorec-
tal cancer.20-24 Many
studies
elevated
studies
vascular
colon
Because
higher
colorectal
studies
studies
rates
of an
or case-
risks of cardio-
cardiovascular
death
risks
diseases
in younger
have
patients
cancer
and because
they are also posi-
with
an increased
BM1,‘2,21,22 these
associated
may be a major cancer
as cohort
to assess the mortality
disease.
significantly tively
on the health
BMI were designed
control
than
reporting
is being
source studied
quately designed to examine and colon cancer. Another
of statistical and
influence
not
ade-
the relation between BMI problem with interpreta-
tion may derive from the possibility cancer undetected at the baseline have a metabolic
are thus
bias when
that an advanced investigation may
on the BMI itself.23*24
To control for potential confounding risk factors, we applied a different approach to the problem. There is a lot of evidence from adenomatous
that most colorectal cancers arise po1yps2’ (this is referred to as the
adenoma-dysplasia-carcinoma tempted
to correlate
BMI with
sequence’“), the different
and we atstages in
this sequence. Using this approach, Mannes et al. have recently shown that the serum cholesterol level is positively
correlated with the risk for colorectal adewith a report by nomas,23 which is in good agreement for Tijrnberg et al., who found the same association colorectal carcinomas in a case control study.27 While nutritional factors are considered important in the genesis of colorectal cancer,28 the role of total caloric intake, unlike food composition, has not been investigated so far. Because BMI is a good quantitative measure of total caloric intake, we investigated the risk of colorectal
adenomas
in relation
to BMI.
Abbreviations used in this paper: 95% Cl, 95% confidence interval; BMI, body mass index; CHD, coronary heart disease. cc‘ 1993 by the American Gastroenterological Association 0016-5085/93/$3.00
138
SAVERDORFFER
GASTROENTEROLOGV
ET AL.
Patients and Methods
serum
All patients
(n = 2012) who underwent
January
addition
to personal
weight
1987 and March
data and medical
were recorded,
blood
(patients
cepted),
stool
diarrhea, origin,
also performed polypectomy tients
histories,
and in all patients for colonoscopy
in the
with
constipation,
occult
acute
as follow-up
examination for colorectal
inflammatory
and
bowel
or overt
hemorrhage
abdominal neoplasia.
or resection
In
height
pain
of
moved
CF-IB, In
polyps
logically.
CF-ITI,
addition
Because
after
or biopsy
material were
tubulovillous,
with
quency
teria of the World
Health
sia of adenomas
were also examined
corehisto-
histologically
in accordance
0rganization3’
was graded
unit
with the cri-
Epithelial
mild, moderate,
as
body weight
Fasting
analyses, weight
initiated.
About
day after
admission
and
Total serum cholesterol matically
at the Department
commercial heim,
again
before
concentration
kit (CHOD-PAP
8 AM and was
preparation.
was determined
of Clinical method,
quintile.
Chemistry
using a
Boehringer,
Mann-
Statistics To control
for potential
with
diseases
with
weight,
such
as previous
chronic
inflammatory
confounding
an obvious or current
bowel disease,
ciated with weight
loss, were excluded,
partial
Patients
colectomy.
excluded, influence
variables,
all
influence
on body
malignant
disease,
or other diseases
asso-
as were patients
with
with familial
polyposis
were also
because genetic factors may have a predominant on the development of adenomas.
The reasons
for exclusion
were recorded
in the protocol.
The remaining patients were divided into quintiles body mass range. To control for potential confounding ables, all factors
known
to be associated
of the vari-
with an increased
frequency of colorectal adenomasage, sex, serum cholesterol level, and indication for colonoscopy - were included in the analyses. The presence of adenomas, “high-risk” adenomas, and “low-risk” adenomas in patients in the individual BMI quintiles
were compared
with
respect
numbers
subjects
were
between
was also per-
the BMI
quintiles
small,
we com-
colorectal subjects
as defined
and the quintiles
Ninety-five
were computed values
ined
percent
in accordance
adenoma
with
the fre-
by Bray et a1.32 For
2 and
3 were
taken
4 and 5 as overweight.
as All
confidence
intervals
with Woolf
(95% CI)
s method.33
All P
were two tailed.
to age, sex,
were excluded
ulcerative colitis, partial colectomy, ing
enzy-
Germany).3’
patients
adenomas
Results
were taken on the bowel
of the frequency
of the association
dyspla-
bowel preparation
10% of the blood samples
age dependency colorectal
pro-
rel. 1988).
data were analyzed for men and women separately and for both together. Odds ratios for crude and adjusted percentages were calculated for BMI as categorical data in the highest and lowest
or severe.30
were taken between before
package,
in normal-weight
Five hundred
blood samples
for these
the BMDP-LR
of the observed
in overweight
Laboratory Tests 9 AM on the day of admission
was adjusted
using
analysis
and when
the association
or familial
continuous
in the age quintiles.
pared
and of pa-
fiber-optic
adenomas,
Additionally,
cancer
evaluation,
classified
or villous
for colonoscopy.
of age and serum choles-
software
of the marked
of colorectal
disease
macroscopic
Statistical
formed
of patients
and CF-101
to
Adenomas
tubular,
(BMDP
unwas
Endoscopy and Histology with Olympus
by logistic
BMI and the observed
Colonoscopy
in our endoscopy
for the influence analysis
gram
No. 1
data, respectively.
parameters
these
lonoscopes.
indication
of adenomas
normal
was performed
and
terol level the presence
ex-
polyposis.
Colonoscopy
and frequency
To standardize
medicine
the BMI was calcu-
were
and suspected
with chronic
colonoscopy
of internal
1988 were evaluated.
lated.‘” Indications
known
variables
the cecum in our department
between
level,
The t test and the x2 test were used to examine
Patients and Study Design including
cholesterol
Vol. 104.
108 with
chronic
diseases
of 2012 patients examthe study: 59 patients with
thirty-two
from
71 with Crohn’s disease, 48 with 344 with malignant disease, includ-
colorectal associated
The characteristics
cancer,
and
with
weight
of the remaining
10 with
other
loss. 1480 patients,
such as mean age, mean serum cholesterol mean BMI are shown in Table 1. Patients
level, and with ade-
nomas were statistically significantly older and had a higher serum cholesterol level; however there was no difference in BMI (Table 1). With respect to the distributions
of the patients’
age and serum cholesterol
level
in the BMI quintiles, it was found that patients in the highest BMI quintile were younger and had lower serum cholesterol levels than those in the lower BMI quintiles. Because the frequency of adenomas is associated with age and serum cholesterol leve1,23 these parameters were considered in the statistical analysis. As Table 2 shows, the number of observed colorectal adenomas is clearly related to older age. The observed numbers of patients with colorectal adenomas were significantly higher in men than in women except in the oldest age quintile where the percentage of adenomas was similar for both sexes (Table 2). The relation
between
BMI and the presence
of colo-
January
Table
OVERWEIGHT
1993
of 1480
1. Characteristics
Patients
With (n = 288)
and Without
AND
(n = 1 192) Colorectal
COLORECTAL
With adenoma (n = 172) 60.5 237 24.57
Age W Cholesterol (mg/dL) BMI (kg/m2) NOTE. Values
are means
rectal adenomas
r 11.5 I!Z 60 -+ 3.23
in the age quintiles
no overall
the presence
of colorectal
With adenoma (n = 116)
f 15.3 + 67 f 3.33
64.5 242 23.22
ble 3). In the oldest
in Table
between
adenoma
in the younger
years and >39/<50.5
is shown
association
and >58.6/<68.1
association
51.3 218 24.70
BMI and
association showed
in the age quintiles
(Table
i 10.7 f 75 + 3.26
We found
no
with the high-risk
age quintiles,
i.e., under
39
0.14,
years neither
for men
nor for
(Table
for develop-
found
no association
analyzed
nomas,
for a slight
association
dysplasia,
adenomas
of 22 cm, and multiple
been
considered
adenomas
to harbor
malignant.
teria
were
and for women
evaluated
Next we analyzed
separately (Table
(26
an increased
All adenomas
men,
with
a cm)
risk of
that met these crifor all patients,
for
and low-risk
To
BMI quintile
of 3.17 for high-risk
13.70,
3). In the age quintile
adenoma
obtain
58.6 and 68.1 years of age, the odds ratio lated to be 2.92 (CI, 0.69-12.41,
Table
of high-risk positive
we
and BMI and or low-risk
ade-
but not significant
between
58.6 and
68.1
with
high-risk
information
between
low-risk
we calculated
for
adenomas.
adenoma
influence
adenomas
The
ratio
of low-risk
in the BMI quintiles
of age as indicated influence
and those
reveals
and an ob-
by the highest
of the BMI,
val-
which
is
indicated by the rapidity by which the ratio declines. As calculated by the log linear model, the influence seems
between
and
was calcu-
3). When
further
the ratio
ues and also a clear
(CI, 0.73-
of men
mentioned
3).
men
vious
adenomas.
For the age quintile 50.5-58.6 years in men we calculated an odds ratio between the lowest and the highest Table
years (Table
in men
in age quintiles,
adenoma
in the age group
high-risk
3).
high-risk
except
in the above
between
nomas
severe
was no such
adenomas
3).
In the case of women adenomas
there
with odds ratios of 0.27,
respectively,
subgroup
becoming
k 16.5 +- 57 + 3.82
to the BMI when compared
adenomas,
and 0.19,
age groups
relation
also in neither
have
49.8 223 23.97
age quintile
ing malignancy varies, depending on a number of different factors. Villous or tubulovillous adenomas, adediameter
Without adenoma (n = 671)
3). The low-risk
an inverse
years.
women (data not shown). The potential of colorectal
with
(n = 787)
f SD.
3. We found >50.5/<58.6
Women
Without adenoma (n = 521)
139
Adenomas
Men (n = 693)
Characteristic
ADENOMAS
we
to be strongest 68.1 years
next older
in the age group
and weaker
BMI quintiles
in the next
(Table
between
58.6
younger
and
4).
pooled the two age groups, i.e., men from 50.5 to 68.1 years, the odds ratio was significant at 3.21 and a CI of
The total of observed colorectal adenomas of all age groups and the subgroups of high-risk and low-risk
1.15-8.98
adenomas in the BMI quintiles of men and women are shown in Table 5. Statistical analysis applying a logis-
(Table
age quintiles revealed
3). Statistical
combined
no significant
analysis
employing association
of these
the logistic between
two
model
tic model
the overall
group of adenomas and BMI (P = 0.106), or between adenomas and cholesterol (P = 0.185) for men in age quintiles >50.5 and <58.6 and >58.6 and <68.1 (Ta-
Table 2. Numbers
Age <39 <50 <58.6 <68.1 >68.1
of Patients
All patients 300 294 296 298 292 1480
and Those With Adenomas Adenomas 10 34 61 77 106 288
(3.3) (11.6) (20.6) (25.8) (36.3) (19.5)
(%)
showed
a significant
association
between
the
frequency of adenomas sexes. The association
in all patients and age for both between cholesterol and ade-
nomas
in men.
was significant
BMI was significantly
in the Age Quintiles Men
Adenomas
119 142 154 142 136 693
8 26 39 48 51 172
(%)
(6.7) (18.3) (25.3) (33.8) (37.5) (24.8)
Women 181 152 142 156 156 787
Adenomas 2 8 22 29 55 116
(%)
(1.1) (5.3) (15.5) ( 18.6) (35.3) (14.7)
GASTROENTEROLOGY Vol. 104, No. 1
BAYERDORFFER ET AL.
140
Table 3. Observed
High-Risk
and Low-Risk
Adenomas
in Different
All patients
BMI
”
>50.5
In
and ~58.6
Men, and Women)
Men High-nsk
Low-nsk
adenomas
adenomas
In WI
[n (Dl
BMI
Adenomas (%)
Age Groups (All Patients,
WI
Women
Adenomas w) n
In
(WI
High-risk
Low-nsk
adenomas
adenomas
(WI
[n
-
Adenomas @)
In (WI
BMI
n
[n
High-nsk
Low-nsk
adenomas
adenomas
cwl
[n WI
In (WI
years of age
<21.35
59
I5 (25.4)
7 (11.9)
8 (13.5)
<21.97
36
9 (25.0)
3 (8.3)
6 (16.7)b
~20.77
24
6 (25.0)
4(16.7)
2 (8.3)
<23.31
69
17 (24.6)
8 (I 1.6)
9 (13.0)
~23.73
29
9 (31.0)
5 (17.2)
4 (13.8)
<22.80
33
7 (21.2)
3 (9.1)
4(12.1)
<24.91
52
I3 (25.0)
13 (25.0)
0 (0.0)
c25.09
32
8 (25.0)
7 (21.9)
l(3.1)
<24.75
26
7 (26.9)
7 (26.9)
0 (0.0)
<27.00
72
I9 (26.4)
IO (13.9)
9 (12.5)
<27.20
37
12 (32.4)
6(16.2)
6 (16.2)
<26.86
35
6f17.1)
3 (8.5)
3 (8.5)
>27.00
43
7 (16.3)
5 (I 1.6)
2 (4.7)
>27.20
19
6(31.6)
5 (26.3)
I (5.3)0
>26.86
24
1 (4.2)
0 (0.0)
1 (4.2)
2 (5.7)
>58.6
and ~68.1
years of age
<21.35
58
I2 (20.7)
6 (10.3)
6 (10.3)
<21.97
28
9 (32.1)
3 (10.7)C
6 (21.4)’
~20.77
35
5 (14.3)
3 (8.6)
~23.31
66
I6 (24.2)
7 (10.6)
9 (13.6)
~23.73
29
IO (34.5)
6 (20.7)
4 (13.8)
<22.80
30
3 (10.0)
2 (6.7)
1 (3.3)
<24.91
65
I I (16.9)
7 (10.8)
4 (6.1)
<25.09
31
4(12.9)
2 (6.4)
2 (6.4)
~24.75
36
8 (22.2)
4(11.1)
4(ll.l)
~27.00
58
I7 (29.3)
11 (19.0)
6 (10.3)
<27.20
27
11(40.7)
7 (25.9)
4 (14.8)
<26.86
29
5 (17.2)
4 (13.8)
1 (3.4)
>27.00
51
13 (25.5)
I2 (23.5)
I (2.0)
>27.20
27
8 (29.6)
7 (25.9)
I (3.7)d
>26.86
26
6 (23.1)
5 (19.2)
1 (3.8)
0 (0.0)
>68.
I years of age
<21.35
60
I6 (26.7)
I4 (23.4)
2 (3.3)
<21.97
24
7 (29.2)
5 (20.8)
2 (8.4)
<20.77
32
9 (28.1)
9 (28.1)
<23.31
69
23 (33.3)
20 (29.0)
3 (4.3)
<23.73
31
I3 (41.9)
IO (32.3)
3 (9.6)
<22.80
44
10 (22.7)
10 (22.7)
0 (0.0)
<24.91
60
34 (56.7)
I9 (31.7)
I5 (25.0)
<25.09
27
13 (48.1)
I I (40.7)
2 (7.4)
~24.75
33
20 (60.6)
8 (24.2)
I2 (36.4)
<27.00
56
I8 (32.1)
15 (26.8)
3 (5.3)
~27.20
27
9 (33.3)
8 (29.6)
1 (3.7)
<26.86
27
IO (37.0)
7 (25.9)
3(11.1)
>27.00
47
I3 (27.6)
I2 (25.5)
l(2.1)
>27.20
27
9 (33.3)
7 (25.9)
2 (7.4)
>26.86
20
4 (20.0)
4 (20.0)
0 (0.0)
<21.97
64
I8 (28.1)
6 (9.4)?
I2 (18.7)’
~23.73
58
I9 (32.7)
I I (19.0)
8 (13.7)
<25.09
63
12 (19.1)
9(14.3)
3 (4.8)
<27.20
64
23 (35.9)
13 (20.3)
10(15.6)
>27.20
46
I4 (30.4)
I2 (26. I)
>50.5
and ~68.1
years of age
NOTE. In the bglstlc model the P values for each prognostic respectively. n = number of patients. Odds ratio for the lOWeSt and highest e3.21
(Cl. 1.15-8.98;
associated (Table
P < 0.05);
with
quIntlIe
‘0.19
was ‘3.17
(Cl, 0.017-0.92;
the presence
5). The data were
factor for tugh-risk adenomas
(Cl, 0.73-
13.70:
P = 0.19);
(for men I” the age quintlIes
b0.27
(Cl, 0.034-2.26:
of adenomas
split for males
in women
model.
Cholesterol
Table 4. Ratio of Low-Risk
the Combination
BMI
n
Adenomas (n)
Ratio low-risk/ high-risk adenomas
<21.97 <23.73 c25.09 <27.20 >27.20
36 29 32 37 19
9 9 8 12 6
2.0 0.8 0.14 1.0 0.2
NOTE. n = number of patients.
nor
low-risk
Adenomas and High-Risk of III and IV
>50.5 and ~58.6 years of age
<2.92
with adenomas,
(C,, 0.69-12.4,;
for BMI and
P = 0.14):
4O.,4
p =0.185for cholesterol,
(C,, O.Olg_o.g9;
p < 0.05);
III to form
with
28 29 31 27 27
weight
small
numbers
of patients
the BMI quintiles group
(men
II and
with
a BMI
The
association
logistic
between
analysis high-risk
showed
a
adenomas
and overweight (P < 0.041) for men in the age quintiles ~50.5 and <68.1 years (Table 6). The odds ratios
as a Function of Age for Men in the Age Quintiles
~58.6 and ~68.1 years of age
n
a normal
a BMI >25.09).
significant
ade-
Adenomas
we combined
from 21.97 to 25.09) according to Bray,32 and the quintiles IV and V to form an overweight group (men
and BMI were significantly high-risk
P = 0.23);
were p = 0.106
In view of the relatively
and females,
risk and low-risk adenomas, only age was significantly associated with the presence of adenoma applying a logistic
3 and 4 combmed)
P < 0.025).
because they are known to have different rates of colorectal adenoma (Table 2). 23 In the subgroups of high-
associated with neither nomas (Table 5).
2 (4.3)’
Adenomas (n)
Ratio low-risk/ high-risk adenomas
9 10 4
2.0 0.67 1.0
11 8
0.57 0.14
268.1
years of age
n
Adenomas (n)
Ratio low-risk/ high-risk adenomas
24 31 27 27 27
7 13 13 9 9
0.4 0.3 0.18 0.13 0.28
>50.5
Ill, IV, V, and
and ~68.1 years of age
n
Adenomas (n)
Ratio low-risk/ high-risk adenomas
64 58 63 64 46
18 19 12 23 14
2.0 0.72 0.33 0.76 0.17
OVERWEIGHT AND COLORECTAL ADENOMAS
January 1993
in All Patients, in Men, and in Women
Table 5. Observed High-Risk and Low-Risk Adenomas All patients
BMI <21.35 <23.31 <24.9 1 <27.00 >27.00 Total Age Cholesterol BMI
All adenoma [n (%)I 53 63 71 61 40 288
(17.9) (21.4) (24.1) (20.6) (13.7) (19.5)
Men
Hugh-risk adenoma [n (%)I
Low-nsk adenoma
27 30 36 33
26 (8.8) 33 (1 1.2) 35 (11.9)
(9.1) (10.2) (12.2)
(11.1) 24 (8.2) 150 (10.1)
BMI
In (%I
28 (9.5) 16 (5.5) 138 (9.4)
141
c21.97 ~23.73 c25.09 ~27.20 >27.20
Women
All adenoma [n @)I
High-risk adenoma
Low-nsk adenoma
In (%)I
In (%)I
BMI
30 (22.3) 40 (28.6) 34 (24.4) 38 (27.3) 30 (2 1.5) 172 (24.8)
13 (9.7) 17 (12.1) 24(17.2) 22 (15.8) 21 (15.0) 97 (14.0)
17 (12.6) 23 (16.5) 10 (7.2) 16 (1 1.5) 9 (6.5) 75 (10.8)
~20.77 <22.80 ~24.75 <26.86 >26.86
All adenoma
Low-risk adenoma
[nWI
Hugh-nsk adenoma [n (%)I
24 (15.3) 22 (13.9) 37 (22.8) 21 (13.4) 12 (7.9) 116 (14.7) 10.00 1 0.32 co.025
13 (8.3) 12 (7.6) 17 (10.5) 9 (5.7) 5 (3.3) 56 (7.1)
11 10 20 12 7
In WI (7.0) (6.3) (12.3) (7.7) (4.6)
60 (7.6) 10.001 0.68 0.3 I
NOTE.The frequency of adenomas is grven in percent. The lower part of the table shows the P values of each prognostic factor for each category of patients as calculated in the logisbc model (stepwise elimination procedure).
for normal weight men vs. overweight
men were 0.95
(95% CI 0.61-1.48; Table 6). In the age group 58.668.1, overweight men had the highest relative risk of developing
high-risk
Table 6). The increased
men of developing
adenomas was still significant
Table
Studies in the past have reported either a positive association lorectal
6). We also analyzed the question various conditions
of high-risk
ently associated with overweight to tubulovillous
as to whether
the
adenomas were differin men. With respect
or villous adenomas,
we identified
a
trend toward a positive association with the BMI which, however, was not significant (Table 7). In the subgroup of adenomas with severe dysplasia and for the subgroup with large adenomas (22 cm) or numer-
Table 6. Observed High-Risk and Low-Risk Adenomas
High-risk adenoma Normal weighta Ovenveightb Odds ratio (confidence limits) P Low-risk adenoma Normal weight” Overweight* Odds ratio (confidence limits) P
between BMI and the incidence
carcinoma”-”
reviewing
or no association.20
of co-
Waaler”
1,700,OOO subjects found a twofold
higher
frequency of colorectal cancer in the highest BMI group in men aged 50-64 years but no association in women. Nomura et al.,l’ who reported on 8000 men, found a 2.5-fold increased frequency of colorectal cancer 255
in the highest BMI quintile
years but no association
in the age group
in the age group 45-54
years. On the basis of data obtained from 750,000
sub-
in Men With Normal Weight and Overweight Men 50.5-58.6 years of age
All men
we found no associa-
BMI and Mortality From Colorectal Cancer
when both age groups CI 1.02-3.90;
adenomas),
Discussion
risk
high-risk colorectal
were pooled (odds ratio 2.00,95%
(26
tion with BMI.
adenomas, with an odds ratio of
3.53 (95% CI 1.28-9.74; for overweight
ous adenomas
Men 50.5-68.1 years of age
Men 58.6-68.1 years of age
No adenoma
Adenoma
No adenoma
Adenoma
No adenoma
Adenoma
No adenoma
Adenoma
232 208
50 47
52 38
12 11
53 35
6 14
105 73
18 25
0.95 (0.61- 1.48) 0.83 232 208
23 20
1.03 (0.55- 1.93) 0.92
1.25
5 7
52 38 1.91
3.53 (1.28-9.74) <0.025
(0.50-3.15) 0.63
(0.57-6.44) 0.29
a 5
53 35 1.06
2.00 (1.02-3.90) <0.05
(0.32-3.50) 0.93
105 73 1.33
13 12 (0.57-3.07) 0.51
NOTE. In the logistic model, the P values for high risk adenomas in overweight men for each prognostic factor (for men in the age quintiles 3 and 4) were P = 0.302 for cholesterol and P = 0.041 for BMI. ‘As defined by Braya and corresponding to BMI quintiles 21.97-25.09 in men. bOverweight means all men above these limits.
142
BAYERDtiRFFER ET AL.
GASTROENTEROLOGY Vol. 104, No. 1
Table 7. Subgroup
Analysis of High-Risk Normal Weight and Overweight
Adenomas:
Observed
Normal weightb Overweight” Odds ratio (confidence limits) P
Adenoma”
232 208
1.06
Men 50.5-68.1
years of age
years of age
jects, Lew and Garfinkel” in the frequency weight
not significant,
4 5
are considered. to BMI quintiles limits.
reported
2 1.97-25.09
a 1.7-fold
cancer
increase
in the highest
in men of all ages and a 1.2-fold, increase
in women.
but
Our study investi-
gated colorectal adenomas as the benign precursor lesion of colorectal cancer and found a very similar relationship
between
adenomas
BMI
and
the
in men. The number
observed
high-risk
of high-risk
adenomas
with an enhanced potential for malignant development in all age groups was increased in the highest BMI quintile
by a factor
same magnitude
of 1.9-fold,
as that reported
kel” for colorectal
cancer.
years was analyzed,
the relative
colorectal
adenomas
risk adenomas nomas
increased,
the potential spective
When
increased
decreased for malignant
which
2.27
creased
an influence
of BMI on
transformation
in the re-
age group.
Adenomaa
105 73
(0.6 l-8.33) 0.22
risk for high-risk
A more marked ber of observed
1.96
8 11
(0.77-5.10) 0.16
colorectal
adenomas
relation
between
high-risk
other
diseases.
overweight
It has been
argued
risks
BMI and the num-
colorectal
has a major influence
in older
by competing
adenomas
4) or carcinoma**,‘” in the age group 50-68 also been reported for the relation between that
(Table years has BMI and
longstanding
on the general
health
risk.‘4*20
Different Potential for Developing Malignancy
50-68
ade-
4 6
overweight women may be masked in this age group (Table 4).
risk factor for high-risk to 3.2. The rate of low-
No adenoma
in men.
by Lew and Garfinthe age group
Adenoma”
53 35
is of the
as the rate of high-risk
suggesting
No adenoma
0.44
of colorectal
category
Adenomaa
1.72 (0.43-6.66)
aOnly tubulovillous and villous adenomas bAs defined by 8ray33 and corresponding ‘Overweight means all men above these
With
Men 58.6-68.1
52 38
(0.54-2.04) 0.86
in Patients
years of age No adenoma
20 19
or Villous Adenomas
Men 50.5-58.6 All men No adenoma
Tubulovillous
Interestingly, only
between
BMI
have an increased increased
this and
found
colorectal
risk of malignant
risk of developing
ber of factors,
study
cancer
an association
adenomas,
which
degeneration. is linked
such as the histological
An
to a num-
type tubulo
vil-
lous, or villous adenoma, adenoma with severe dysplasia, adenoma larger than 2 cm, or more than six
In the case of women, the situation was different. When the number of observed high-risk adenomas in the highest BMI quintile was compared with that in the lowest BMI quintile, we found a small, nonsignifi-
concomitant adenomas. 34 Though not significant, our data suggest that the development of adenomas and their potential for malignant transformation are in-
cant negative nonsignificant,
fluenced by a number cal factors.
association positive
for all ages but a small, also association in women aged
58-68 years. This finding is more or less in agreement with former studies of colorectal cancer, which found either no association’” or a weakly positive, but not significant, association ” between BMI and the risk of colorectal carcinoma for women. These data may be illustrated by the age dependence of colorectal adenomas, which shows that the risk of colorectal adenomas in women is 3.5-6 times lower in those <50 years of age and increases at age >68 years to a risk similar to that in men (Table 3). This suggests that menopausal women have a risk of developing colorectal cancer similar to that of men. However, an in-
of epidemiological
and etiologi-
BMI and Mortality From Other Diseases The influence of overnutrition on health in general has been investigated extensively, especially for coronary heart disease.35-39 Reports by Simopoulos et a1.,14 Lew,37 and Harris et al.,” suggested that overweight has to be persistent to have an effect on the frequency of colorectal cancer. A number of limitations of the use of BMI as a measure of overnutrition have been discussed.24 However, BMI has been accepted as the best means of estimating overweight. 39 Self-reported data on height and
January
1993
weight
OVERWEIGHT
may not be very reliable.24
Not all the authors
of previous epidemiologic studies that investigated the relation of BMI and colorectal cancer” were able to measure
weight
ported
data.
and height
The
data
sively on clinically other
major
studies
measured
source
itself. 24 BMI
tional
factors,
monal
study
balance.41
in BMI colorectal
of advanced
is influenced
Calories
exclu-
and heights.
An-
cancer
cancer
on body
not only
by nutri-
activity4’
and hor-
but also by physical
may influence
are based
weights
of error
may be the impact
weight
but had to rely on self-re-
in this
derived
from
alcohol
also
BMI.42
The
frequency
were controlled
data
for
for the major
and cholesterol
colorectal risks,
level in the logistic
risk of colorectal
especially increased dures, have
such as age, sex,
cancer.
However, factors that with an in-
For instance
alcohol,
in the form of beer, was associated with an risk of colorectal cancer.43 Surgical proce-
such as gastric also been
resection
reported
with
an in-
risk of colorectal cancer and colorectal adenomas. 44,45Because we were unable to obtain the family histories of all the patients, some important risk might
Risks
The imponderable the results of this study risks. Colorectal seventh decade,
encountered on considering is the influence of competing
by numerous
with an earlier peak mortality cade. Early insurance studies association
diseases
in the sixth or fifth deraised the question of a
of overweight
that from
competing
in the cancer
cancer.47
and general Coronary
mortalheart
dis-
ease (CHD) and cancer were found to be the most common causes of mortality. Other epidemiologic studies showed similarities in the incidence rates of CHD and cancer studies primarily
smoking
subclinical
in industrialized countries.‘,* Later designed to investigate risk factors
for CHD, partly confirmed the positive association between overweight and cancer mortality,” but also produced conflicting data. 23 However, the main source of bias may be attributed to the much earlier mortality peak in CHD. Death rates of coronary heart disease in the fifth through the seventh decade are much higher than those of colorectal cancer. Manson et a1.24 also noted
disease
secondary diseases hypertension. We did not rectal cancer.
the associahad at least
biases.
They
the
controlled
for
such as diabetes
and
of overweight
any of these
cancer.
diseases
than the peak
The death
are rather
of colorectal
cancer
mortality the relation
is less
inof
of colorectal
by 5-20
mortality
early the
lesion
that
years. 48 The
very
precedes impact
cancer
major
mortality
has been underestimated
cause of a lack of monitoring
low,
of BMI
risks. The design
biases,
or lower
important
in our study is the investigation
precursor
noma
rates of the
to the association
and BMI had at least one of these three
provement
but as-
were 50 years old on
at ages <50
which are associated with BMI quintiles. 24 Therefore
by
lesion of colo-
15 years younger
that our approach
investigating
factors,
precursor
our patients
is about
by competing
studies
for
and inappropriately
frequency
fluenced
to control loss caused
Thus,
competing
failed
and to assess weight
of colorectal
suggesting and
investigating and mortality
as the benign
which
143
weight
consider
sessed adenomas
im-
of the be-
colorectal
carci-
of overweight
on
in most studies
be-
for competing
risks.
Conclusions pothesis
cancer has its peak mortality so that the risk of colorectal
may be influenced
ity, including
other
have been missed.46
Competing
possible
one of three cigarette
nign
and cholecystectomy,
to be associated
creased
factors
body
ADENOMAS
risk of dying from coronary
all the studies
between
major
model.
we could not control for several minor have also been reported to be associated creased
tion
disease,
mortality adenomas
COLORECTAL
that besides the competing heart
average,
Risk Factors for Colorectal Cancer
AND
This prospective investigation supports the hythat the BMI influences the risk of developing
colorectal adenomas at least in men, who have a high risk for malignant degeneration. This study investigated the benign precursors of colorectal carcinomas, and the results confirm epidemiologic data found to apply to colorectal men with a high additional
risk
carcinoma. The data suggest that BMI, especially if they have such
factors
as a high
level, ought to receive prophylactic prevent colorectal carcinoma.
serum
cholesterol
colonoscopy
to
References 1. Doll R. The geographical distribution of cancer. Br J Cancer 1969;23: l-8. 2. Burkitt DP. Epidemiology of cancer of the colon and rectum. Cancer 197 1;28:313. 3. Weisburger JH, Reddy BS, Wynder EL. Colon cancer: its epidemiology and experimental production. Cancer 1977;40:24 142420. 4. Wynder EL, Reddy BS. Metabolic epidemiology of colorectal cancer. Cancer 1974;34:80 l-806. 5. Howell MA. Diet as an etiological factor in the development of cancers of the colon and rectum. J Chron Dis 1975;28:67-80. 6. Graham S, Haenszel W, Bock FG, Lyon JL. Need to pursue new leads in the epidemiology of colorectal cancer. JNCI 1979; 63:879-88 1.
144
BAYERDGRFFER
7. Wynder EL, Kajitani T, lshikawa S, Dodo H, Takano A. Envrronmental factors of cancer of the colon and rectum. II. Japanese epidemrological data. Cancer 1969;23: 12 IO- 1220. 8. Haenszel W, Berg JW. Segi M, Kurihara M, Locke FB. Large-bowel cancer in Hawaiian Japanese. JNCI 1973;5 1: 1765- 1779. factors and cancer incidence 9. Armstrong D, Doll R. Environmental and mortality in different countries with special reference to dietary practices. Int J Cancer 1975; 15:617-63 1. 10. Wynder EL. The epidemrology of large bowel cancer. Cancer Res
11. 12. 13. 14. 15. 16. 17. 18. 19. 20.
21. 22.
23.
24. 25. 26.
27.
28. 29.
30.
GASTROENTEROLOGY
ET AL.
1975;35:3388-3394. Tannenbaum A. The genesis and growth of tumors. II. Effects of caloric restriction per se. Cancer Res 1942;2:460-467. Mann GV. The influence of obesrty on health. New Engl J Med 1974;29 1:178-232. Lowenfels AB, Anderson ME. Diet and cancer. Cancer 1977;39:1809-1814. Simopoulos AP, Van ltallie T. Body weight, health, and longevrty. Ann Int Med 1984; 100:285-295. Bray GA. Complications of obesity. Ann Intern Med 1985; 103: 1052-1062. Garfinkel L. Overweight and cancer. Ann Int Med 1985; 103: 1034- 1036. Lew EA. Garfinkel L. Variations in mortality by weight among 750000 men and women. J Chron Dis 1979;32:563-576. Nomura A, Heilbrun LK, Stemmermann GN. Body mass index as predictor of cancer in men. JNCI 1985;74:3 19-323. Waaler HT. Height, werght and mortality: The Norwegran experience. Acta Med Stand Suppl 1984;679: l-56. Harris T, Cook E, Garrison R, Higgins M, Kannel W, Goldman L. Body mass index and mortality among nonsmoking older persons. JAMA 1988;259: 1520- 1524. Feinleib M. Epidemiology of obesity in relation to health hazards. Ann Intern Med 1985; 103: 10 19- 1024. Garrison RJ. Castelli WP. Weight and thirty-year mortality of men In the Framingham Study. Ann Intern Med 1985; 103: 10061009. Mannes GA, Maier A, Thieme C, Wiebecke B, Paumgartner G. Relation between the frequency of colorectal adenoma and the serum cholesterol level. New Engl J Med 1986;3 15: 1634- 1638. Manson JA, Stampfer MJ, Hennekens CH, Willett WC. Body weight and longevity. JAMA 1987;257:353-358. Muto T, Bussey HJR, Morson BC. The evolution of cancer of the colon and rectum. Cancer 1975;36:225 l-2270. Day DW, Morson BC. The adenoma-carcinoma sequence. In: Morson BC, ed. The pathogenesis of colorectal cancer. Phrladelphia: Saunders, 1978: 126- 152. Tornberg SA, Holm LE, Carstensen JM, Eklund GA, Odont D. Risks of cancer of the colon and rectum in relation to serum cholesterol and beta-lipoprotein. New Engl J Med 1986;315: 1629-1633. Wynder EL. Nutrition and cancer. Federation Proc 1976;35: 1309-1315. National Institute of Health Consensus Development Conference Statement. Health implrcations of obesity. Ann Int Med 1985; 103: 1073- 1077. Morson BC, Sobin LH. Histological typing of intestinal tumors. lnternattonal histological classification of tumours. Geneva: World Health Organization, 1976.
31.
32. 33.
34.
35.
36. 37. 38. 39. 40.
41.
42. 43.
44.
45.
46.
47. 48.
Vol. 104,
No. 1
Siedel J, Hagele EO, Ziegenhorn J, Wahlefeld AW. Reagent for the enzymatic determination of serum total cholesterol with Improved lrpolytic efficiency. Clin Chem 1983;29: 1075- 1080. Schlesselmann JJ. Case control studies: design, conduct, analysis New York: Oxford University, 1982. Bray GA. Obesity in America. Bethesda, Maryland: National Institutes of Health, 1979. National Institutes of Health Publication No. 79-359. Hermanek P, Fruhmorgen P. Guggenmoos-Holzmann I, Altendorf A, Matek W. The malignant potential of colorectal polyps-A new statrstical approach. Endoscopy 1983; 15: 16-20. Glueck CJ, Taylor HL, Jacobs D. Morrison JA, Beaglehole R, Williams OD. Plasma high-density lipoprotein cholesterol: Association with measurements of body mass. Circulation 1980; 62(Suppl 4):62-69. Jarrett RJ, Shipley MJ, Rose G. Weight and mortality in the Whrtehall Study. Br Med J 1982;285:535-537. Lew EA. Mortality and Weight: Insured lives and the American Cancer Society Studies. Ann Intern Med 1985; 103: 1024- 1029. Sidney S, Friedman GD, Siegelaub AB. Thinness and Mortalrty. Am J Public Health 1987;77:3 17-322. Van ltallie TB. Health implications of overweight and obesity in the United States. Ann Int Med 1985; 103:983-988. Johnson ML, Burke BS, MayerJ. Relative importance of Inactivity and overeating in the energy balance of obese high school girls. Am J Clin Nutr 1956;4:37-44. Pariza MW. Boutwell RK. Hrstorical perspective: calories and energy expenditure in carcinogenesis. Am J Clin Nutr 1987;45: 151-156. Ziegler RG. Alcohol-nutrient interactions in cancer etiology. Cancer 1986;58: 1942- 1948. Pollack ES, Nomura AMY, Heilbrun LK, Stemmermann GN, Green SB. Prospective study of alcohol consumption and cancer. New Engl J Med 1984;310:617-621. Caygill CPJ, Hill MJ, Hall CN. Kirkham JS, Northfield TC. Increased risk of cancer at multiple sites after gastric surgery for peptic ulcer. Gut 1987;28:924-928. Mannes GA, Weinzrerl M, Stellaard F, Thieme Ch, Wiebecke B, Paumgartner G. Adenomas of the large intestine after cholecysctectomy. Gut 1984;25:863-866. Cannon-Albright. Skolnick MH, Bishop DT, Lee RG, Burt RW. Common inheritance of susceptibility to colonic adenomatous polyps and associated colorectal cancers. New Engl J Med 1988;3 19:533-537. Tannenbaum A. Relationship of body weight to cancer incidence. Arch Path 1940;30:509-5 17. Stryker SJ. Wolff BG, Culp CE, Libbe SD, llstrup DM, McCarty RL. Natural history of untreated colonic polyps. Gastroenterology 1987;93: 1009- 10 13.
Received January 4, 1991. Accepted August 18, 1992. Address requests for reprints to: Ekkehard Bayerdorffer, M.D., Medical Department II, Klinikum GroRhadern, University of Munich, MarchioninistraDe 15, D-8000 Munchen 70, Germany. The authors thank Dr. P. Dirschedl and P. Kampe for valuable help in performing statistical calculations and Prof. M. Knedel (Institute of Clinical Chemistry, University of Munich) for his cooperation.