303
Clinical Science ( 1998) 94,303-3I I (Printed in Great Britain)
Dietary fish oil suppresses human colon tumour growth in
athymic mice
P.C. CALDER J. DAVIS*, P. YAQOOB*, H. PALA*, F. THIES* and E.A NEWSHOLME*
Division of Human Nutrition, School of Biological Sciences, University of Southampton, Bossett Crescent
East, Southampton SO16 7PX. U.K., and *Department of Biochemistry. University of Oxford, South Parks
Road, Oxford OX1 3QU, U.K.
(Received 7August/l2 November 1997; accepted 19 November 1997)
1. Human colon tumour growth, initiated by subcutaneous inoculation of HT29 cells, was measured
in athymic mice fed ad libitum on high-fat (210 g/kg)
diets rich in coconut oil (CO), olive oil (OO), safflower oil (SO) or fish oil (FO); a low fat (LF; 25
g/kg) diet was used as the control. In one experiment the mice were fed the experimental diets for 3
weeks before HT29 cell inoculation and were killed
2 weeks post-inoculation. In a second experiment
the mice were maintained on the LF diet until 4
days post-HT29 cell inoculation; they were then fed
the experimental diets for 17 days.
2. Compared with mice fed the LF diet, tumour size
was increased in mice fed the CO, 00 or SO diets
for 3 weeks before HT29 cell inoculation; FO feeding did not significantly increase tumour size.
3. Feeding mice the CO or 00 diets from 4 days
post-inoculation increased tumour growth rate and
tumour size compared with feeding the LF, SO or
FO diets; tumour growth rate and size did not differ
among mice fed the latter diets.
4. The fatty acid composition of the tumours was
markedly influenced by the fatty acid composition of
the diet.
5. We conclude that human colon tumour growth is
influenced by the type of fat consumed in the diet.
Human colon tumour growth in this model is promoted by feeding high fat diets rich in medium
chain saturated fatty acids (CO) or monounsaturated fatty acids (00).A high fat diet, rich in long
chain n - 3 polyunsaturated fatty acids (FO), does
not promote colon tumour growth. The effect of a
high fat diet rich in n - 6 polyunsaturated fatty
acids (SO) depends upon the time at which it is fed
if fed before tumour cell inoculation such a diet
promotes tumour growth, whereas if fed once
tumour growth is initiated it does not. This suggests
that n -6 polyunsaturated fatty acids promote the
initiation of colon tumour growth, but do not exert
growth-promoting effects on colon tumours once
they are established.
INTRODUCTION
Colon cancer is one of the most prevalent cancers
in the Western world. Epidemiological studies have
established a link between total fat consumption and
the incidence of colon cancer [l-51. In addition to
the amount, the type of fat consumed appears to be
important in influencing colon cancer development.
Several studies have identified a link between
animal fat and/or saturated fat consumption and the
incidence of colon cancer [3-51. Furthermore, the
increase in colon cancer in Japan, where the incidence is lower than in the Western world [2, 61, has
been attributed to dietary change, especially
increased total fat and saturated fat consumption
[6]. A prospective study found a positive association
between intake of the monounsaturated fatty acid
(MUFA) oleic acid (ClS:l,n-9), a component of
animal fats and the main fatty acid found in olive
oil, with incidence of colon cancer [4]. There
appears to be no correlation between the incidence
of colon cancer and total intake of vegetable fat,
polyunsaturated fatty acids (PUFAs) or n-6
PUFAs ([4]; see also [7]). However, colon cancer is
negatively correlated with intake of n-3 PUFAs
(see [7]). The principal n-3 PUFA found in the
Western diet is a-linolenic acid (C18:3,n-3), a component of some vegetable oils and of green leaves.
Colon cancer rates are lower in countries where
consumption of fish oils is high, such as Japan [6],
and Kromann and Green [8] noted a lowered incidence of cancers in general among Greenland Eskimos who consume large amounts of seal and fish
oils. Such oils are rich in the n -3 PUFAs eicosapentaenoic
(Cm :5 , - 3) and docosahexaenoic
(C22:6.n-3) acid. A prospective study found a nega-
Key words: Cdon tumour. athpk mouse, fish oil, fatty aids, dietary lipids
A b b r e v h b MUFA, monounsaturatedfatty acid; PUFA, polyunsaturatedfatty acid; LF, low fat; CO. coconut oil; 00, dive oil; SO,safflower oil; FO, fish oil; MEM, minimal
ertemal medium; MTBE, methyl-t-butyl e t h q ANOVA, analyw of variance.
Comspondencc:Dr P. C. Calder.
304
P. C.Calder e-t al.
tive association between fish consumption and the
incidence of colon cancer [4]. Thus, epidemiological
studies suggest that total fat and saturated fat
increase the incidence of colon cancer, whereas
n-3 PUFAs decrease the incidence, with n-6
PUFAs, at levels consumed in the human diet, being
neutral [7].
Studies in rats indicate that the type as well as the
amount of fat in the diet influences the incidence of
chemically induced colon cancer (see [9] for a
review). The incidence of colon tumours is higher in
animals fed high-fat diets, where the dietary source
is saturated fatty acid- or n -6 PUFA-rich, than in
animals fed low-fat diets containing these fats, or
fed high-fat diets rich in n-9 MUFAs (see [9] for
references). Dietary fish oil (FO) has been shown to
lower the incidence of chemically induced colon
cancer in rats compared with saturated fats or oils
rich in n - 6 PUFAs [lo-151. FO was also found to
reduce the growth of implanted murine colon
tumours in mice [16]. Similarly, Tisdale and Dhesi
[ 171 reported lower growth of subcutaneously inoculated murine colon tumours in NIMR mice fed FO
than in those fed other diets; in addition FO
reduced the cachexia associated with growth of the
MAC16 tumour used.
While these animal studies have proven very useful in gaining understanding of the relationship
between dietary fat consumption and colon cancer,
they have studied either tumours whose growth is
initiated by significant levels of a carcinogen, or
murine tumours initiated by inoculation into susceptible mouse strains. The availability of immunodeficient strains of mice, such as the athymic (nulnu or
‘nude’) mouse, allows the study of human tumour
growth in animals. Athymic mice readily accept
human tumour cells or transplants of human
tumours because they are T lymphocyte deficient
[18], and they have been used to study the effect of
dietary fat on human mammary tumour growth
[19-24].
Given the prevalence of colon cancer in man, and
the evidence of differing influences of various
dietary fats upon its development in man and in
some animal models, we thought it important to use
the athymic mouse to study the effects of dietary
fats upon human colon tumour growth. To our
knowledge, human colon tumour growth has not
been studied previously in athymic mice. Therefore,
we compared the growth of colon tumours, initiated
by the subcutaneous inoculation of HT29 cells, in
athymic mice fed a low-fat diet or high-fat diets rich
in medium-chain saturated fatty acids, MUFAs,
n - 6 PUFAs or n - 3 PUFAs.
MATERIALS AND METHODS
Sources of animals, diets and chemicals
Female athymic (nulnu) mice weighing 15-20 g
were purchased from Bantin and Kingman (Hull,
Yorkshire, U.K.). They were housed in the Department of Zoology, University of Oxford in purposebuilt facilities which provided a continuous germfree laminar airflow over the cages. A 12 h lightll2 h
dark cycle was used and the temperature was maintained at 24°C. The animals were allowed free
access to sterile water which was replaced every 3
days (water bottles were autoclaved before use); in
addition the cages and bedding were autoclaved and
replaced every 3 days. Only J.D. had access to the
animals prior to their being killed. Animals were fed
on either a low fat (LF) diet (25 g of corn oilkg) or
on one of four high fat (210 g/kg) diets (purchased
from ICN Biomedicals, High Wycombe, Bucks.,
U.K.). The high fat diets contained 200 g/kg of
hydrogenated coconut oil (CO), olive oil (00),safflower oil (SO) or fish (menhaden) oil (FO) plus 10
g/kg of corn oil to prevent essential fatty acid deficiency. All diets contained identical amounts of protein (200 g/kg), starch (200 g/kg), sucrose (295.8
g k g ) and vitamin E (1.2 g/kg). The fatty acid composition of these diets is shown in Table 1. Because
the oils used to make the diets might contain different levels of constituents such as cholesterol, the
diets may differ in the composition of components
other than the fatty acids under study. The metabolizable energy content of the high-fat diets was
18986 kJ/kg (of which 41% was derived from fat),
while that of the LF diet was 12133 kJ/kg (of which
7.7% was derived from fat). The FO diet was stored
at -20°C until use; the other diets were stored at
room temperature. All procedures involving animals
were approved under the Animals (Scientific Procedures) Act 1986 by the Home Office.
Minimal essential culture medium (MEM), glutamine, fetal calf serum and antibiotics were purchased from Sigma Chemical Co., Poole, Dorset,
Table I. Fatty acid compositionof the diets used. nd, not detected.
Fatty acid (g/I00 g of total fatty acids)
Diet
100
I2:O
14:O
160
LF
CO
00
nd
0.1
47.3
3.0
2.2
0.9
18.2
SO
5.0
nd
nd
FO
nd
nd
19.3
10.5
11.1
11.8
24.7
1.5
0.4
10.9
161 n-7
1.7
2.2
nd
nd
16.9
180
181n-9
182n-6
163n-6
183n-3
205n-3
22611-3
3.6
11.0
3.9
4.4
4.4
33.4
2.4
69.9
20.8
13.8
39.9
3.5
15.3
60.4
5.5
nd
nd
nd
nd
nd
nd
nd
nd
nd
nd
nd
nd
nd
nd
nd
nd
11.7
nd
5.2
1.3
Fish oil and tumouir growth
U.K. Solvents were purchased from Fisher Scientific,
Loughborough, Leics., U.K.
305
of C02. Tumours were carefully excized and
weighed. Parametrial adipose tissue was dissected
out and weighed.
HT29 cell maintenance
Human colon carcinoma cells (HT29 cells) were
cultured at 37°C in an air/COz (19: 1) atmosphere in
MEM supplemented with 2 mmolfl glutamine, 10%
(vk) fetal calf serum and antibiotics. They were subcultured when a confluent cell layer formed
(approximately every 3 days); PBS containing 1
mmol/l EDTA was used to remove the cells before
sub-culturing.
Effects of dietary lipids on HT29 tumour growth in
athymic mice
Two different experiments were performed to
investigate the effect of dietary lipid manipulation
upon human colon tumour growth in athymic mice.
Experiment I. Athymic mice were fed the diets
described above for 3 weeks. They were then
injected subcutaneously in the flank region with
HT29 cells (1 x lo6) suspended in 200 pl of sterile
PBS; control mice received PBS alone. The mice
were maintained on the same diets for a further 2
weeks.
Experiment 2. Athymic mice were fed on the LF
diet for 2 weeks; they were then injected with HT29
cells (2 x lo6 cells/200 pl) as described above. They
were maintained on the LF diet for a further 4 days
(until tumour growth was visible) and were then
transferred to one of the diets described above for a
further 17 days.
Mice were weighed regularly and their food
intake was monitored. Tumour growth was monitored by measuring the size of the tumour in three
dimensions, using microcalipers, and calculating its
volume. That this is a satisfactory measure of
tumour size is shown by the strong correlation
(r = 0.925; P<O.001) between tumour volume 1 day
pre-sacrifice and tumour weight at sacrifice (Figure
1). Mice were killed in the fed state by an overdose
O
Fatty acid composition analysis of tumoun
Tumours were homogenized in chloroform/methano1 (2: 1, vk). Neutral lipids and phospholipids were
separated using Sep-Pak columns (Waters C o p ,
Milford, MA, U.S.A.). The columns were flushed
with 12 ml of hexane and the sample, dissolved in
1 ml methyl-t-butyl ether (MTBE), was applied.
Neutral lipids were eluted using 10 ml MTBE/acetic
acid (100:0.2, vk) and phospholipids were eluted
using 10 ml MTBE/methano1/0.001 M ammonium
acetate (5 :8 :2, vkk.) Fatty acids were prepared by
saponification for 15 min at 80°C in methanolic 0.5
molfl NaOH. Then, fatty acid methyl esters were
prepared by incubation with 14% (wk) boron trifluoride at 80°C for 15 min. Fatty acid methyl esters
were isolated by solvent extraction, dried and separated by GC in a Hewlett-Packard 6890 gas chromatograph (Hewlett Packard, Avondale, PA, U.S.A.)
fitted with a 30 m x 0.32 mm BPX70 capillary column, film thickness 0.25 pm. Helium at 2.0 ml/min
was used as the carrier gas and the split/splitless
injector was used with a split/splitless ratio of 10: 1.
Injector and detector temperatures were 170°C and
250°C respectively. The column oven temperature
was maintained at 170°C for 12 min after sample
injection, and was programmed to then increase
from 170 to 200°C at 5"C/min before being maintained at 200°C for 15 min. The separation was
recorded with HP GC Chem Station software (Hewlett Packard, Avondale, PA, U.S.A.). Fatty acid
methyl esters were identified by comparison with
standards run previously.
Statistical analysis
Data are meanskSEM of the indicated number of
observations. Effects of diet were determined by
one-way analysis of variance (ANOVA) and apost
hoc least significant difference test; in all cases
P <0.05 was considered to indicate a statistically significant difference.
I
RESULTS
i3t
J
Effects of dietary lipids on HT29 tumour growth in vivo
Experiment I. The food intake of mice fed the LF
diet was greater than that of mice fed the high fat
diets (Table 2); despite this, the energy intake in
high fat-fed mice (approx. 220-260 Hhveek) was
approximately 35% higher than in the LF-fed mice
(175-190 Hiweek). Food intake was unaffected by
tumour bearing (Table 2). Before inoculation of
tumour cells, mice fed the high-fat diets gained
more weight than those fed the LF diet (Table 3).
P. C.Calder et al.
306
Table 2. Food intake p m and port-HRPtumour cell inoculation in athymii mice fed dlierent lipids. Athymic mice were fed
one of f i i experimental diets. After 3 weeks they were injjed subcutaneously with either sterile PBS w I x 106 HT19 tumour cells.
The mice were maintained on the same diet for a further 2 weeks. Food intake was monitored throughout. Data for weeks I, 2 and 3
are means fSEM from I 2 animals fed on each diet; data for weeks 4 and 5 are means kSEM for six animals in each group. Statistical
significance between dietary group was determined by one-way ANOVA; values in the same cdumn not sharing a common
superscript letter are significantly different.
Food consumption @week)
Week 4
Week 5
Diet
Week I
Week 2
week 3
Saline
Tumour
Saline
Tumour
LF
16.0kO.B
I 1.4f0.3b
I I .6 +0.6b
12.9k0Sb
12.6 f0.7
14.8 f 0.4=
12.3+0.3b
I I .9 f0.2b
12.3f l.4*
12.7 f0.6b
14.9 k0.3l
12.9+0.9b
12.5 k0.5b
13.0k0.6b
12.0 k0.9
14.5 k0.5’
12.7kO.p
12.9 k0.4’
12.9+0.4b
12.4 fOSb
13.6 k 0 . 4
12.9k0.4
12.5 f0.8
13.3f0.4
12.6 f0.5
15.2kO.B
12.1 k0.3b
12.8 k0.4b
13.3+0.5&
12.4 f 0.6b
13.6 f0.6
12.8 f0.5
12.6 k0.7
13.4k0.4
12.3k0.3
co
00
so
FO
This is most likely due to the greater energy intake
among these animals. There were no differences in
weight-gain among mice fed the different high-fat
diets pre-inoculation (Table 3). Saline-injected mice
gained more weight than those inoculated with
tumour cells, irrespective of the diet fed, although
this difference was statistically significant only in
mice fed the CO diet (Table 3). Mice fed the highfat diets gained more weight than those fed the LF
diet post-saline injection or tumour cell inoculation
(Table 3).
Tumour growth became apparent within 4 days,
after which tumours increased in size (results not
shown); the mice were killed 2 weeks post-inoculation. Tumour weights at sacrifice were greater in
mice fed the CO, 00 or SO diets compared with
those fed the LF diet (Table 3). Among mice fed
the high-fat diets, tumour weights were lowest in
those fed the FO diet (Table 3), and there was no
significant difference in tumour weight between
animals fed the LF or FO diets (Table 3). When
expressed as a proportion of body weight, tumour
growth was not increased in FO-fed mice compared
with those fed the LF diet, although it was increased
in mice fed the other high-fat diets (Table 3).
Tumour weight did not exceed 0.75% of body
weight in any animals in this experiment.
Parametrial adipose depot weight was greater in
high-fat-fed mice than in those fed the LF diet,
whether they had received the saline injection or
tumour cell inoculation (Table 3). Thus, the greater
weight of high-fat-fed mice might be accounted for
by increased adipose deposition. Among the salineinjected, high-fat-fed mice, parametrial adipose
depot weight was lowest in the FO-fed mice (Table
3); however, there were no significant differences in
parametrial adipose depot weight among mice fed
the different high-fat diets (Table 3). Tumour-bearing resulted in decreased parametrial adipose mass
among animals fed each of the diets, although this
was not statistically significant (Table 3); the
decrease in adipose mass appeared to be smallest in
mice fed the LF or FO diets (Table 3). Thus, the
lower weight gain of tumour-bearing mice compared
with those injected with saline might be due to
reduced adipose tissue deposition and/or increased
adipose tissue mobilization. FO-fed mice appear to
maintain adipose tissue mass better than those fed
each of the other diets. In absolute terms, the parametrial adipose tissue mass was approx. 13% lower
Table 3. Body, tumour and adipose tissue weights of athymii mice fed different lipids pre- and after-Hn9 tumwr cell inoculation. Athymic mice were fed
one of fve experimental diets. After 3 weeks they were injected subcutaneously with either sterile PBS or I x I06 HT29 tumour cells. The mice were maintained on the
same diet for a further 2 weeks. Food intake was monitored throughout. Data for weight gain pre-inoculation are means kSEM from I 2 animals fed on each diet; other
data are means SEM for six animals in each group. Statistical significance between dietary groups was determined by o w w a y ANOVA; values in the same cdumn not
sharing a common superscript letter are significantlydifferent.
Parametrial adipose depot weight
Tumour weight at sacrifice
Diet
LF
CO
00
SO
FO
Weight gain
pre-inoculation
(9)
Weight gain
post-saline
Weight gain posttumour cell inoculation
0
0
5.4+0.5’
6.9t_0.Sb
7.1 +0.6b
7.2kO.P
7.3+0.6b
2.4k0.5’
5.0k0.9
3.2*0.5*
3.8kO.P
3.8kO.k
1.3kO.6’
2.4+0.S*
2.1 *0.4*
3.0+0.5b
2.4*0.6*
(md
82+ I I’
143kIp
135k17b
143+2Ib
Il 4 k l P
(Percentage of
body weight)
0.36k0.05’
0.S2&0.0Sb
0.58+0.0P
0.59k0.07b
0.42+0.07*
Saline injected
(mg)
(Percentage of
bodyweight)
15OkIP
O.Mk0.08’
368kSab
380+46b
400k82b
320f54b
I.27k0.19b
1.34k0.13b
1.33+0.22b
1.10f0.15b
Tumour cell inoculated
(mg)
(Percentage of
body weight)
130k16’
253k28b
280+53b
274k40b
293f35b
0.53k0.05’
0.92kO.W
1.04i.0.18b
0.95k0.12b
I.OS+O.llb
Fish oil and turnour growth
in tumour bearing LF-fed mice than in LF-fed mice
that received a saline injection. In mice fed the CO,
00 or SO diets tumour bearing decreased parametrial adipose mass by between 26 and 31%. In contrast, the reduction was only 8% in tumour bearing
FO-fed mice. When the parametrial adipose mass is
expressed as a percentage of body weight the reduction in size that accompanies tumour growth is 17%
(LF), 22-28% (CO, 00, SO) and 4.5% (FO).
Feeding experiment 2. In this experiment all mice
were fed the LF diet pre-tumour cell inoculation
and for 4 days post-inoculation; no saline-injected
control group was included. Before inoculation of
tumour cells, there was no difference in food intake
or weight gain among mice that were later transferred to the different diets (Tables 4 and 5).
Tumour growth became apparent within 4 days, at
which time the mice were transferred to the different diets on which they were maintained for 17
days. Mice fed the high-fat diets consumed less food
(but slightly more energy) (Table 4) and gained
more weight than those fed the LF diet posttumour-cell inoculation (Table 5); there were no differences in weight gain among mice fed the different
high fat diets (Table 5 ) .
Tabk 4. Food intake p m and post.HT29 tumour dl inoculation in
& h p i mice fed di&rant lipids. +ic
mice were fed the LF diet for
2 weeks and were then injected subcutaneously with 2 x 106 HR9 turnour
cells. The mice were maintained on the LF diet for a further 4 dayx and
were then transferred to one of fnre experimentaldiets on whiih they were
maintained for a further 17 days. Food intake was monitored throughout.
Data are meansfSEM from rix animals fed on each diet. Statistical
s g i i m e between dietary groups was determined by one-way ANOVA;
values in the same cdumn not sharing a common superscript letter are
significantly different.
Food consumption(g/week)
Diet
Week I
Week 2
Week 4
Week 5
LF
13.6 k0.6
13.6 k0.7
14.0f0.8
13.6 f0.4
14.0 k0.7
14.8k0.4
13.6 f0.8
14.4 f0.8
13.4 k0.7
13.4kO.8
18.0 f 0.3'
12.0f 0.4b
I 2.8 f0.8b
18.4f0.8
I 2.4 f0.p
12.4 f 0.7
12.0f 0.4b
12.8f0.8b
co
00
so
FO
13.2+0.lb
14.4f0.9b
307
Tumours increased in size over time (results for
LF-fed mice are shown in Figure 2); there were considerable differences in the rate of tumour growth
among animals fed the same diet (e.g. Figure 2).
Tumour volume at day 4 did not differ among the
different groups (at this stage all animals had consumed only the LF diet) (Table 6). By day 7, tumour
size in mice fed the CO or 00 diets was already
increasing at a greater rate than in mice fed the
other diets (Table 6). Tumour volume in mice fed
the LF, SO or FO diets did not differ at any time
point (Table 6). In contrast, tumour volume in mice
fed the CO or 00 diets was greater than in mice
fed each of the other diets at all times points, significantly so at days 19 and 21 (Table 6).
Tumour weights at sacrifice were greater in this
experiment (Table 5 ) than in experiment 1 (Table
3). This is most likley due to the greater number of
tumour cells inoculated in this experiment (2 x lo6
compared with 1 x106 in experiment 1) and the
longer period of tumour growth (21 days compared
with 14 days). In this experiment, tumour weight
accounted for as much as 10% of body weight in
some mice (Table 5). Tumours were heavier in mice
fed the CO or 00 diets (Table 5); these weights
were significantly greater than tumour weights in
mice fed the LF, SO or FO diets. Tumour weight,
expressed as a proportion of body weight, was significantly greater in CO- or 0 0 - f e d mice than in mice
fed the other diets, among which there were no differences (Table 5 ) . Parametrial adipose depot
weight was greater in high-fat-fed mice than in those
fed the LF diet, although the differences were not
statistically significant (Table 5). Saline-injected control mice were not included in this experiment and
so it is not clear what the adipose tissue mass would
have been in non-tumour-bearing mice maintained
under identical conditions. Based upon data from
experiment 1, it might be expected that the parametrial adipose tissue depot might account for 0.5% of
body weight in LF-fed mice and for 1% of body
weight in those fed high-fat diets, if they had been
fed for about 21 days without a tumour load. Thus,
tumour bearing might have reduced parametrial adipose mass, by as much as 50%, in this experiment.
T f i 5. Bcdy, tumour and d i p tissue weights of athymii mice fed diflcrent lipids post.HT29 tumour cell inoculrtion. A t h p K mice were fed the LF diet
for 2 weeks and were then injected subcutaneously with 2 x I06 HR9 turnour cells. The mice were maintained on the LF diet for a further 4 days and were then
transferred to one of fnre experimental dim on whiih they were maintained for I7 days. Animal we@ was monitored throclghout. At sacrifke, tumour and psrametnd
adipose tissue weight were determined. Data are means f SEM from six animals fed on each diet. Statistical signifme between dietary groups was determined by oneway ANOVA; values in the same cdumn not sharing a cmmon superscript letter are signifmtly different.
Diet
LF
co
00
so
FO
Tumwr weight
Parameaidadipose depot we@
Weight gain
pre-inoculation
g
Weight gain
post-inoculation
Weight gain minus
turnour w&t
g
g
mg
%orbodywelght
mg
%ofbodywelght
3.1 k0.3
3.6 k0.6
3.5 f 0.6
2.9 f0.2
3.0f0.6
3.2ko.r
4.9 k I .p
7.6 f0.3b
5.8 f 0.p
6.1 f I.@
2.7 f 0.6'
2.6 k I .ZX
6.2 f 0.4b
4.4 f 0.4k
5.4 f 0 . p
776 k95'
2155 +394b
1651 f289b
3.2k0.3'
9.2f 1.76
6.6f l.2b
3.1 f0.5'
2.4k0.2'
82k I7
106+41
164f39
I14f 18
132f36
0.39k0.06
0.41 k0.14
0.63k0.14
0.46f0.07
0.48fO.l I
7~
lor
509f 105'
308
P. C.Calder e-t al.
Fatty acid composition of tumour neutral lipids and
phospholipids
both the neutral lipid and phospholipid fractions,
while the 00 diet resulted in a marked increase in
the proportion of oleic acid in both lipid fractions.
The SO diet caused an elevation in the proportion
of linoleic acid (C18:2,n-6) in both fractions. Feeding
the FO diet resulted in a number of changes in fatty
acid composition. The proportion of myristic acid
was increased in both neutral lipids and phospholipids, whereas the proportions of palmitic (c16:O)
and palmitoleic ( c 1 6 : 1,n-7) acids were elevated in
the neutral lipid fraction. FO feeding resulted in
markedly reduced proportions of linoleic and arachidonic (c20:4, n -6) acids in both lipid fractions. There
was a significant increase in the proportions of
eicosapentaenoic and docosahexaenoic acids in the
neutral lipid and phospholipid fractions after FO
feeding (Tables 7 and 8).
The fatty acid compositions of the neutral lipid
and phospholipid fractions of the tumours obtained
in experiment 2 were significantly affected by diet
(Tables 7 and 8). The CO diet resulted in an
increase in the proportion of myristic acid (c14:O) in
/
DISCUSSION
0
14
7
The athymic mouse is T-lymphocyte deficient [ 181
and, as such, is unable to mount an efficient cellmediated immune response. Thus, these mice
present the opportunity to study the growth of
21
Days wl-tumoor inoculation
4 . 2 . Growth cum of subcutuKour tumours growing in athymii
mice. Data are for six different mice maintainedon the LF diet
Table 6. Time course of the-i
in tumour size in a w i c mice fed dfiuent lipids. Athymic mice were fed the LF diet for
2 w& and were then i n w subcutaneously with 2 x lob HT29 tumour cells. The mice were maintained on the LF diet for a
further 4 days and were then transferred to one of five experimental diets on which they were maintained for a further 17 days.
Tumour size was measured using calipers and tumour volume was calculated. Data are means+SEM from six animals fed on each
diet. Statistical signifiie between dietary groups was determined by oneway ANOVA; values in the same column not sharing a
common supenaipt letter are significantlydifferent.
Tumwr volume (mm))
Diet
Day 4
Day 7
Day I I
Day 14
Day 19
Day 21
LF
42f8
co
55k I5
31 f 4
so
28k11
FO
3 9 5 14
151 f 2 0
298 k89
313f 104
103k31
121 f 3 9
197f26
407f 133
522 f I70
157f45
183k54
36 I f 54a
MI k 145b
00
66f 16
167f52
167f48
61 k 2 2
82&25
517+W
1021 f 17Eb
I317 +334b
463 f 104'
357 f 88'
977 k 245b
31 I k75'
297 f 74'
Table 7. Fatty acid composition of the neubal lipids of tumoun taken from mice fed dfierent lipids. Athymic mice were fed the LF diet for 2 weeks and were
then in@ subcutaneouslywith 2 x 106 HT29 tumwr cells. The mice were maintained on the LF diet diet for a further 4 days and were then tansferred to one of five
experimental diets on which t h y were maintaned for 17 days. At d i c e t u m ~ were
~ r ~removed. Lipid was ermcted,neutral lipid and phospholipid classes separated
and the fatty acid composition determined (see Materials and medrods section). Data are meansfSEM from 3 to 5 animals fed on each diet Statistical signifKance
between dietary p u p s was determined by meway ANOVA; values in the same column not sharing a common superscript letter are significantly different. n.d.. not
detected.
Fatty acid (g1100 g of total fatty acids)
Diet
140
I60
161.11-7
160
18:l.n-9
18:2,n-6
18:3,n-3
18:3,n-6
204,n-6
205,n-3
226,n-3
20.3kO.P
19.4kO.P
16.7k0.6b
19.0kO.P
25.9k0.4
10.2k0.5'
8.8k0.6'
6.4f0.5b
5.4f0.3b
10.3k0.8
8.2kO.P
8.9k1.1
6.9k1.1'
10.7k0.6b
10.9k1.2b
33.6kl.P
29.2k2.r
47.7k2.p
19.0k0.5'
23.2il.P
16.2kO.P
14.8kI.P
13.0+0.4b
31.6k1.4'
6.9k0.1d
0.2kO.l
0.2k0.1
0.5k0.2
0.3k0.1
0.4+0.1*
0.3H.l'
0.7f0.1b
0.4*0.1*
0.3kO.la
5.3kO.P
5.9kO.P
4.0kO.P
5.9k0.5'
2.0f0.1b
0.3kO.I'
0.8+0.1'
0.9k0.I'
0.7k0.2'
0.6kO.I'
5.3k0.2b
LF
I.9kO.P
CO
5.5kO.P
00
SO
FO
I.5kO.l'
2.0+0.2*
4.3k0.2b
0.4k0.1
n.d.
n.d.
n.d.
6.9+0.6b
Fish oil and tumour growth
309
* *
Tabla 8. Fatty add comporidon dtha
ofturnours taken hwn mica fed dillcrrnt lipid+ Athymic mice were fed the LF dm for 2 weeks and were
then injected rubcutaneoudywith Z x 10 HR9 turnour cells. The rnicewere maintained on the LF diet for afunher 4 day and were then transferred to me of frve
experimental die0 on which they were maintained for 17 day.At g a i f ~ eturnours were removed. Lipid was extracted, neutral lipid and phosphdipid dasses repaated
and the fatty acid composition determined (see Materiak and methods section). Data are means f SEM from 3 to 5 animals fed on each diet. Starisaal signifme
between dietacy groups was determined by one-way ANOVA; values in the sane cdurnn not sharing a wmmon supencript letter are signKkandy differem
Fatty acid @/I00g total fatty acids)
Diet
14:O
I60
16l,n-7
180
18:l. n-9
182, n-6
183,n-3
183,n-6
LF
l.9k0.1x
3.8k0.4b
1.7k0.1'
1.9k0.2x
2.5k0.3'
24.4k0.7
24.8k0.2
22.2k5.0
23.6k0.9
24.9k1.4
7.9k0.5'
7.4kO.F
5.5k0.5k
4.3k0.2b
6.1f0.4b
14.0k0.6'
15.450.k
13.750.3'
15.8k0.4'
17.3+0.3b
21.3k0.5'
19.8kO.P
28.!~+_1.5~
13.2k0.3'
14.5k0.3'
14.6k0.4'
14.5k0.3'
12.9k0.5'
25.4k0.7"
5.6k0.2'
0.350.1
0.3k0.1
0.3k0.1
0.3kO.l
0.3kO.l
0.3k0.1
0.3k0.1
0.4kO.l
0.3k0.2
CO
00
SO
FO
human tumours [18], which are destroyed in
immuno-competent rodents. The effect of dietary
fats of different compositions on human mammary
tumour growth in athymic mice has been investigated [19-241. However, to our knowledge, the current study is the first to examine the influence of
dietary fat on human colon tumour growth in athymic mice. Two approaches to studying the effect of
dietary fat on human colon tumour growth in vivo
were used. The first approach involved feeding the
mice for 3 weeks on different diets before inoculation with human colon cancer cells, and maintainance on those diets post-inoculation. The second
approach involved feeding the mice on an LF diet
before and for 4 days after tumour-cell inoculation,
when tumour growth became apparent; at this stage
the mice were transferred to the different diets.
Both experiments indicated that some high fat
diets promote human colon tumour growth compared with an LF diet (Tables 3 and 5). One contributor to enhanced tumour growth during high-fat,
compared with LF feeding, is likely to be the higher
energy intake among the high-fat-fed animals. However, the type of fat present in the high-fat diet was
found to be important in determining the precise
effect: both experiments indicated significant colon
tumour growth-promoting effects of high fat diets
rich in medium chain saturated fatty acids (CO) or
in n - 9 MUFAs (00),and both experiments indicated that a high-fat diet, rich in n - 3 PUFAs (FO),
does not enhance tumour growth compared with an
LF diet. These observations agree with the findings
of epidemiological studies in man (i.e. increased
incidence of colon cancer with increased total fat
[l-51, saturated fat [3-51 and n-9 MUFA [4] consumption, and decreased incidence of colon cancer
with increased fish and n - 3 PUFA consumption [4,
6, 71). Furthermore, they agree with the increased
incidence of carcinogen-induced colon tumours in
rats fed high-fat diets where the fat source is saturated (see [9] for references) and the lowered incidence of such tumours if rats are fed FO [lo-151.
These experiments also indicate that the effect of
a high-fat diet, rich in n -6 PUFAs (SO), on human
colon tumour growth is dependent upon the stage at
0.5+0.1
204, n-6
205,n-3
9.4k0.5'
0.6+0.3*
1l.lkl.P
0.1+0.1'
0.2+_O.Za
8650.4'
8.850.6'
3.7k0.4b
1.5kO.P
14.3kI.l'
226, n-3
l.2kO.l'
1.5k0.2'
1.3+_0.1'
0.6k0.Zb
5.6kO.I'
which the host is exposed to the diet. If the SO diet
was fed to the mice before tumour cell inoculation
there was a tumour-growth promoting effect equivalent to that of CO or 00 (Table 3). This observation agrees with the tumour-growth promoting
effects of n - 6 PUFA-rich high-fat diets (e.g. corn
oil, SO) seen in rats exposed to carcinogens (see [9]
for references). However, if the mice were transferred to the SO diet once tumour growth became
apparent, the tumours did not grow any faster or
larger than those in mice fed the LF or FO diets
(Table 5 and 6).
The different effects of the high-fat diets upon
tumour growth cannot be due to differences in
energy intake between the animals, since the highfat diets were isoenergetic and the mice fed the different high-fat diets consumed the same amount of
food (and thus energy) both before and after
tumour-cell inoculation (Tables 2 and 4). Similarly,
the adipose tissue mobilization observed in mice
bearing tumours is clearly not due to reduced food
or energy intake (Tables 2 and 4), and is most likely
due to the production of one or more catabolic
mediators. Interestingly, in addition to the reduction
in tumour growth compared with other high-fat
diets, FO appeared to decrease the adipose tissue
mobilization that accompanied tumour bearing.
Whether this is simply due to the lower tumour burden in FO-fed mice, or due to reduced production
of, or insensitivity to, the mediators responsible for
adipose tissue mobilization, is unclear. Such mediators might include cytokines, such as tumour necrosis factor-a or interleukind, or novel protein
mediators which are at present not fully characterized (see [25] for a discussion of such factors). It is
possible that n - 3 PUFAs will influence the production of these mediators by tumours.
This study did not investigate the mechanisms
whereby different dietary fats might affect colon
tumour growth. Mechanisms which can be considered include, effects on the production of eicosanoids within the growing tumour, effects upon lipid
peroxidation within the tumour and effects on the
immune system, which provides host surveillance
against tumours.
310
P. C.Calder et al.
Effects of diet upon T-lymphocyte-mediated
immunity are unlikely in the current model, since
athymic mice are T-cell-deficient [MI. Athymic mice
do retain some natural killer cell activity. However,
several studies have shown that rodent natural killer
cell activity is diminished by FO feeding [26-301,
suggesting that enhanced natural killer cell activity is
unlikely to explain the reduction in tumour growth
in FO-fed mice. Interestingly, 00 has also been
shown to reduce rodent natural killer cell activity
[29-311; this may partly explain the enhanced
tumour growth in athymic mice fed this diet. Athymic mice also retain macrophage-mediated cytotoxic
activities, although these are likely to be reduced
because of the absence of help from T lymphocytes
(e.g. interferon-y production). FO feeding has been
shown to markedly reduce the ability of inflammatory macrophages to produce tumour necrosis factor
[32-341 and nitric oxide [35, 361 and to kill tumour
cells [33, 37-39]. Therefore it seems unlikely that
FO is exerting its effects on colon tumour growth
through macrophage-mediated activities. Thus the
reduced colon tumour growth observed in mice fed
FO is probably not a result of effects of FO on the
immune response in these animals. It is more likely
to be due to effects exerted within the tumour, such
as altered eicosanoid production or lipid peroxidation, both of which are known to influence tumour
growth [40-471 and both of which are affected by
increasing the proportion of n - 3 PUFAs in the
diet.
Tumour growth requires a supply of fatty acids for
synthesis of new cell membranes. These fatty acids
will be supplied from the diet and from adipose
tissue mobilization; in turn, the fatty acids stored in
adipose tissue will be largely of dietary origin,
especially in animals fed high-fat diets. Thus, the
fatty acid composition of the circulating pool of fatty
acids, in the form of triacylglycerols, phospholipids
or non-esterified fatty acids, will come to resemble
the fatty acid composition of the diet. Therefore, in
experiment 1, where the mice were fed different
diets for 3 weeks before tumour-cell innoculation,
the fatty acids available from the very earliest stages
of tumour initiation were different among animals
fed the different diets. Thus, this experiment does
not differentiate between effects that diet might
have upon tumour initiation or upon subsequent
tumour growth. Therefore, in this experiment the
SO diet could promote tumour initiation or tumour
growth or both. In contrast, in experiment 2 the
mice were transferred to the different diets once
tumour growth was apparent. Therefore, tumour initiation occurred in the presence of the same pool of
circulating fatty acids, influenced by the LF diet, in
all animals irrespective of the diet which they were
subsequently fed. Thus, this experiment investigated
the effects of dietary fat on the growth of an established tumour. That the SO diet resulted in
increased tumour weight in experiment 1, but not in
experiment 2, suggests that dietary n - 6 PUFAs, or
at least linoleic acid, play a role in tumour initiation,
but that, once a tumour is established, they do not
exert growth-promoting effects. The mechanism by
which the SO diet exerts different effects depending
upon the timing at which it is fed in relation to the
timing of tumour cell innoculation was not investigated in the current study. However, it may relate to
the availability of linoleic acid as a precursor to
arachidonic acid. This is because arachidonic acidderived eicosanoids such as prostaglandin Ez have
been implicated in the development of some
tumours [40, 411 including colon tumours [42-441. A
tumour-promoting effect of prostaglandin E2 and/or
related eicosanoids might account for the effect of
SO when fed before tumour innoculation, since
animals fed this diet would have been consuming
large amounts of the arachidonic acid precursor
linoleic acid. If the key effects of arachidonic acidderived eicosanoids are at the early stages of tumour
promotion, rather than on growth of an established
tumour (prostaglandin levels are highest in the early
stages of tumour growth [45]), then an SO-rich diet
will have less of an effect on tumour growth if fed to
animals with an already established tumour; this is
what was observed in the current study.
If dietary fatty acids do exert their effects upon
colon tumour growth through changing eicosanoid
production and/or lipid peroxidation, then their
incorporation into tumour lipids must occur before
they can exert their effects. Thus, the rate at which
changes in tumour lipid fatty acid composition occur
would be important in determining the precise effect
of the different diets. In the current study, the fatty
acid composition of tumour lipids was measured
after 17 days of feeding the different diets. Significant changes in fatty acid composition occurred. The
time course of the changes in tumour fatty acid
composition was not investigated. However, since
the tumours are rapidly growing, it would be
expected that their fatty acid composition would
change readily in response to changes in dietary
fatty acid composition.
In summary, this study has shown that human
colon tumour growth in athymic mice is enhanced
by feeding high-fat diets rich in medium-chain saturated or n - 9 MUFAs, and that this high-fatinduced tumour promotion is absent if FO is used as
the fat source. The effect of a high-n - 6 PUFA diet
depends upon the timing of feeding: if fed before
tumour inoculation, this diet promotes tumour
growth, but if fed once a tumour is established there
is no growth promoting effect. The mechanisms by
which these diets influence tumour growth warrant
further investigation.
ACKNOWLEDGEMENTS
This work was supported by a grant to P.C.C. and
E.A.N. from the Ministry of Agriculture, Fisheries
and Food (Grant number AN0215). P.Y. and F.T.
Fish oil and tumour growth
hold Post-doctoral Research Fellowships from the
Ministry of Agriculture, Fisheries and Food. J.D.
held a BBSRC Graduate Studentship when this
work was completed. We are indebted to Professor
Chris Graham of the Department of Zoology, University of Oxford, for allowing access to specialized
animal housing facilities.
References
I. Carroll KK, Khor HT. Dietary fat in relation to tumorigenesis. Progr Biochem
Pharmacd 1975; 1 0 308-53.
2. Wynder EL, Kajiii T, lshikawa S, Dodo H, Takano A Environmentalfaaors of
cancer of the cdon and rectum. II.Japeneseepidemido& data. Cancer 1969
2 3 I 2 lo-20.
3. Morgan JW,Fracer GE, Phillips RL. A
ndm MH. Dietary factors and cdon
cancer incidence among Sevemhday Mventists. Am J Epidemid 1988; 128: 918.
4. Willett WC, Smpfer MJ, Cddin GA,Rosner BA,S p e r FE. Relation of meas
fat, and fiber intake to the risk of colon cancer in a prospeaive study among
women. New End J Med 1990,323: 1664-72
5. McKeown-Eyssen GE, Bright-See E. Dietary factors in cdon cancer:
international relationships: an update. Nutr Cancer 1985; 7: 25 1-3.
6. Lands WEM, Hamazaki T, Yam& K, Okuyama H, Sakai K, Got0 Y, Hubbard
6.
C w n g dietary patterns. Am J Clin Nutr 1990; 5 I:991-3.
7. Hendenon MM. Corrdationr between fatty acid intake and cancer incidence. In:
Nelson GJ, ed. Hedth effectsof dietary fatty acids. Champagn: AmericM Oil
Chemists' Society, I99 I: I 36-49.
8. KKwnann N. Green A E p i d e m i i studies in the Upernavik District,
Greenland. Acta Med stand 1980.208: 401-6.
9. Reddy BS. Dietary fat and colon cancec animal studies. Lipids 1992; 27:
807-813.
10. Reddy BS. Margama H. Effect of dietary fish oil on azoxymethane-inducedcolon
carcinogenesis in male F-344 rats. Cancer Res 1986; 46:3367-70.
I I. Reddy BS, %e S. Effect of different levels of omega-3 and omega41fatty acids
on azoxpethaminducedcdon carcinogeneris in F344 rats. Cancer Res 1988.
48:6642-7.
12. Dexhner EE, LydeIS,Wong G, Rupert0JF, Newmark HL. The effect of dietary
omega-3 fatty acids (fish oil) on azoxpethad-induced focal areas of +plasia
and cdon tumor incidence. Cancer 1990; 66:USO-6.
13. Minoura T, Takata T. -hi
M, Takada H. Yamamura M, Hioki K,
Yamamoto M. Effect of dietary eicosapentaenoic acid on azoxymethane-induced
cdon carcinogenesis in rats. Cancer Res 1988; 48:4790-4.
14. Nelson RL, Tanure JC, Andrianopouloa G, Souza G, Lands WEM. A comparison
of dietary fish oil and corn oil in experimental c o l o r e d carcinogeneris. Nutr
Cancer 1988; II:215-20.
IS. Lindner MA A fish oil diet inhibits colon cancer in mice. Nutr Cancer 199 I ; 15:
1-1 I.
16. Cannino F. Broitman SA Postpromotionaleffects of dietary marine or safflower
oils on large bowel or pulmonary implants of CT-26 in mice. Cancer Res 1989:
49: 4289-94.
17. Ti& MJ, DhesiJK. Inhibition of we&t loss by omega-3 fatty acids in an
experimentalcachexia model. Cancer Res 1990; 5 0 5022-6.
18. GMvandla BC, Fogh J.The nude mouse in cancer rerearch. Adv Cancer Res
1985; 44:69-120.
19. Gonzalez MJ, schemmel PA, Gray JI,Dugan L. Sheffield LG, Welxh CW.Effect
of dietary fat on growth of MCF-7 and MDA-MB23I human breast carcinomas in
athymic nude mice: dationship between carcinoma growth and lipid
peroxidadon productlevels. Carcinogenesis 1991: 12: 1231-5.
20. Borgeson CE, Pardini L, Pardini RS, R e i C. Effect of d i q f i oil on human
mammary carcinoma and on lipld-metabding enzymes. Lipids 1989: 2 4 290-5.
21. Gabor H, Blank EW, Ceriani RL. Effect of dietary fat and monoclonal antibody
therapyon thegrowth of human mammary adenocaKinoma MX-I gdted in
athymic mice. Cancer Lett 1990; 52: 173-8.
22. Rose DP, Condly JM. Effects of d q omega-3 fatty acids on human breast
cancer growth and memtases in nude mice. J Nad Cancer Inn 1993; 8 5
1743-7.
23. W e l d CW,odd^ CS,Chang C-C, Wdxh MA Sup~ressionof growth by
31 I
dietary.fish oil of human breast carcinomas maintained in three different strains
of immunodeficient mice. Nutr Cancer 199% 2 0 I 19-27.
24. Rose DP, Condly JM. RayburnJ. Coleman M. Influence of dm containing
eicwapentaenoicor docoghexaenol'c acid on growth and metactaris of breast
cancer cells in nude mice. J Nad Cancer lnst 1995; 87: 587-92.
25. Tidale MJ. lsdationof a novel cancer cacheaic f w . P m Nutr Soc 1997; 5 6
777-83.
26. Meydani SN, Yogeemaran G. Liu S, Barkar S, Meydani M. F i i oil and
tocopherol-inducedchanges in natural killer cell-mediatedcytotoxicity and PGEl
synthesis in young and dd mice. J Nutr 1988; 118: 1245-52.
27. Fritsche KL, JohnstmeW.Effect of dietary omega-3 fatty acids on cell-mediated
cytotoxic aaivity in BALM mice. Nutr Res 1990; 10 577-88.
28. Berger A G e m JB, Chiang BL, Ansari Ak Keen CL, Fletcher MP, Gershwin
MR Influence of feeding unsaturated fats on growth and immune status of mice.
J Nutr 1993; 123: 225-33.
29. Yaqoob P, Newshdme EA, cakler PC. Inhibition of natural killer cell aaivity by
dietary lipids. lmmud Lett 1994; 41: 241-7.
30. Sandwson P, Yaqoob P, Calder PC. Effects of dietary lipid manipulation upon rat
spleen lymphocyte functions and the expresion of lymphocvtesurface
mdeculer. J Nutr Environ Med 1995; 5 I 19-32.
31. Jeffery NM, Yaqoob P. Newrhdme E4, cdder PC. The effectsof dw
i oil upon
rat serum lipd kvels and lymphocyte functions appear to be due to deic acid.
AnnNutrMetab 19964&71-80.
32. Billiar T, Bankey P, Svingen 6, Curran RD, West MA, Hdman RT, Simmons RL,
Cena FB. Fatty acid uptake and Kupffer cell funaion: fKh oil alters eicoranoid
and d n e production to endotoxin stirnutation. Surgecy 1988; 104: 343-9.
33. Renii G. Skamene E, de Sanctis J, M i o c h D. Dietary n-3 polyunsaturated
fatty acids prwent the development of atherosclerotic lesions in mice:
Modulation of macrophage secretwy saivities. Artwio Thromb 1993; 13:
1515-24.
34. Yaqoob P, Calder PC. Effects of dietary lipid manipulation upon infbmmatory
m e d i r productionby murine macrophages. Cdl l m m d 1995; 163 120-8.
35. Joe6, Lokesh BR Rde of capsaicin. curcumin and d q n -3 fatty acids in
loweringthegeneation of reactive oxygen species in rat peritoneal
macrophages. Biochim Biophys Acta 1994; IW:255-63.
36. BoutardV, Fouqwry 6, Philippe C.Perez J, Baud L Fish oil supplementation and
essential fatty acid deficiency reduce nitric oxide synthesis by rat macrophap.
Kidney Int 1994;4& 1280-6.
37. Somen SD, Chapkin RS, E r i c h KL. Alteation of in vim murine pentMleal
ma crop^ funaion by dietary enrichmentwith eicosapemaenoicand
docosahexswKwc
' acids in menhaden f i oil. Cdl lmmud 1989; 123: 201-1 I.
38. Black JM, Kindla JE. Dietary.n-3 fatty adb alter mouse peritoneal mavophage
cytotoxicity.Ann Nutr Metab 1993; 3 7 I 10-20.
39. Hubbard NE, Chapkm RS, Erickron KL. Effect of dietary linseed oil on
tumoricidal Mivity and eicoanoid production in murine macrophages. Lip&
19% 29: 651-5.
40. F u h AM. The role of eicoranoidr in tumor megstaw's. Prost Leuk Ers Fatty
M i 1988.34 229-37.
4 I. Abou-El-ElaSH, Prase KW, Carroll R, Wade AE,Dharwadkar S. Bunce OR
Eicoanoid synthesis in 7, I2dimethylbenzene(a)anthraceminduced mammary
carcinomas in Sprqsw-Dawley rats fed primrose oil, menhaden oil or corn oil
diet. Lipld~1988; U:948-54.
42. Nariawa T, Sat0 M, Tani M. Kudo T, Takahashi T, Goto A Inhibition of
devdopment of methyinitrozurea-induced rat colon tumors by indomethacin
k.
Cancer 6 1981; 41: 1954-7.
43. Moser AR, Ptot HC, Dove WF. A dominant mutation that p r e d i i to
multiple intestinal neoplasia in the moose. Science 1990; 247: 322-4.
44. M b d SK, DannenbergAJ,Chadburn A Mam~ciC, Guo XJ, Ramonetti JT,
Abreu-Gotis M, Newmark HL, Lipkin ML, DeCosseJJ,Bertagndli MM.
C y c l o o ~ a s e - 2OYecexpceSSiOn and tumor formation are blocked by sulindac
in a murine model of familial admommu polyposis. Cancer Res 1996; 5 6
2556-60.
45. Smith BJ. Wills MR Savory J. Prostaglandins in mer.Ann Lab Sci 1983; 13
359-65.
46. Horrobin DF. Essentialfatty acids, lipid peroxidation, and cancer. In: Horrobin
DF, ed. Omega4 fatty acids pathophVsidogV and mler in dinicd medicine. New
York: Ahn R Lis, 1990: 351-77.
47. Gonzalez MJ,schemmel PA, Dugan L, Gray JI,W e l d CW. Dietary fish oil
inhibits human breast carcinoma growth:a funaion of i
d lipid
~ x i d a t b n Lipids
.
1993; u):827-32.