1-s2.0-S0163725899000261-main

8/2/2019 1-s2.0-S0163725899000261-main

1/28

Pharmacology & Therapeutics 83 (1999) 217244

0163-7258/99/$ see front matter 1999 Elsevier Science Inc. All rights reserved.

PII: S0163-7258(99)00026-1

Associate editor: S.T. Mayne

Omega-3 fatty acids as cancer chemopreventive agentsDavid P. Rose*, Jeanne M. Connolly

Division of Nutrition and Endocrinology, American Health Foundation, One Dana Road, Valhalla, NY 10595, USA

Abstract

There is both epidemiologic and experimental evidence that the long-chain omega-3 fatty acids (FAs), which occur at high levels in

some fish oils, exert protective effects against some common cancers, notably those of breast, colon, and, perhaps, prostate. Multiple

mechanisms are involved in this chemopreventive activity, including suppression of neoplastic transformation, cell growth inhibition and

enhanced apoptosis, and antiangiogenicity; however, a common feature of most of these biological effects is the inhibition of eicosanoid

production from omega-6 FA precursors. Several of the known risk factors for breast, and colon, cancer may be favorably modified by di-

etary omega-3 FA supplementation, and the implementation of clinical chemoprevention trials is now feasible. 1999 Elsevier ScienceInc. All rights reserved.

Keywords: Breast cancer; Colon cancer; Chemoprevention; Omega-3 fatty acids; Eiscosanoids

Abbreviations: AA, arachidonic acid; COX, cyclooxygenase; DHA, docosahexaenoic acid; DMBA, dimethylbenz[

a

]anthracene; DPA, docosapentaenoic

acid; EGF, epidermal growth factor; EPA, eicosapentaenoic acid; FA, fatty acid; HETE, hydroxyeicosatetraenoic acid; HPETE, hydroperoxyeicosatetraenoic

acid; 13-HODE, 13(

S

)-hydroxyoctadecadienoic acid; LA, linoleic acid; LNA,

-linolenic acid; LOX, lipoxygenase; LT, leukotriene; ODC, ornithine decar-

boxylase; PG, prostaglandin; PGI

2

, prostacyclin; PKC, protein kinase C; TX, thromboxane.

Contents

1. Introduction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 218

2. Nomenclature, chemistry, and metabolic conversion of unsaturated fatty acids . . . . . . . . . 218

3. Dietary sources of omega-3 fatty acids, intestinal absorption, and tissue and subcellular

distribution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 220

3.1. Dietary sources. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 220

3.2. Intestinal absorption. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 221

3.3. Tissue and subcellular distribution. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 221

4. The eicosanoids. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 222

4.1. Arachidonic acid-derived eicosanoids . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 222

4.2. Eicosanoids derived from eicosapentaenoic acid. . . . . . . . . . . . . . . . . . . . . . . . . . . . 223

4.3. Eicosanoid biosynthesis and n-3:n-6 fatty acid ratio . . . . . . . . . . . . . . . . . . . . . . . . . 224

4.4. 13(S)-Hydroxyoctadecadienoic acid . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 225

5. Biomarkers ofn-3 fatty acid nutritional status . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 225

6. Omega-3 fatty acids and cancer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 226

6.1. Breast cancer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 226

6.1.1. Epidemiology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 226

6.1.2. Experimental studies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2276.1.3. The relative potency of eicosapentaenoic acid and docosahexaenoic

acid in breast cancer animal models. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 228

6.2. Colon cancer. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 228

6.2.1. Epidemiology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 228

6.2.2. Experimental studies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 229

6.3. Mechanisms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 230

6.3.1. Breast cancer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 230

6.3.2. Colon cancer. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 231

* Corresponding author. Tel.: 914-789-7145; fax: 914-592-6317.

E-mail address:

[email protected] (D.P. Rose)

8/2/2019 1-s2.0-S0163725899000261-main

2/28

218

D.P. Rose, J.M. Connolly / Pharmacology & Therapeutics 83 (1999) 217244

7. Omega-3 fatty acids and cancer chemoprevention. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 234

7.1. Breast cancer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 234

7.1.1. Mammographic densities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 234

7.1.2. Atypical hyperplasia and carcinoma in situ . . . . . . . . . . . . . . . . . . . . . . . . . 234

7.1.3. Chemoprevention . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 234

7.1.4. Omega-3 fatty acid supplementation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 235

7.1.5. Dietary modification targeting other fatty acids:

The n-3:n-6 fatty acid ratio. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2357.1.6. Intermediate response biomarkers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 236

8. Commentary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 237

Acknowledgments. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 237

References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 238

1. Introduction

It is becoming increasingly recognized that, at least in thelong term, future advances in clinical cancer research willcome from an emphasis on prevention rather than the treat-ment of metastatic disease. This research effort encompasses

epidemiological studies, such as those responsible for the rec-ognition that high consumption of vegetables and fruits is as-

sociated with a reduced risk of some cancers, laboratory ex-periments to evaluate potential natural and synthetic productsas chemopreventive agents, and the execution of clinical pre-ventive trials (Greenwald et al., 1993). As always, these differ-ent approaches are complementary: for example, work withanimal models showed that nonsteroidal anti-inflammatorydrugs such as aspirin possess chemopreventive activity againstexperimental colonic carcinogenesis, epidemiological obser-vations supported such a potential role in human populations,

and several of this class of compounds are now undergoing

evaluation in clinical trials for colon cancer prevention.The progression from the initial steps in the carcinogenicprocess to the establishment of clinically manifest, invasivedisease occurs in a series of steps, each of which merges intothe next (Fig. 1). Intervention strategies follow much thesame path: primary chemoprevention merges with chemo-suppression, which itself is indistinguishable from the che-motherapy of early disease. However, the distinction be-tween chemoprevention and chemosuppression is useful

because it recognizes that it is not necessary to prevent thedevelopment of premalignant, or even frankly cancerous,foci in order to suppress progression to symptomatic, life-threatening cancer. Thus, while the antiestrogen tamoxifenmay have a place as a truly chemopreventive agent, its actionin preventing the development of a new primary cancer inthe contralateral breast of a postmenopausal woman previ-ously treated surgically for primary breast cancer is mostlikely one of chemosuppression (Jordan, 1991). Experimen-

tal studies performed in our laboratory suggest that theomega-3 fatty acids (FAs), which are the subject of this re-view, can have a similar chemosuppressive effect on the pro-gression of microscopic metastatic foci (Rose et al., 1996).

Some drugs that were introduced originally for the treat-ment of metastatic cancer are now having their role ex-tended to include chemoprevention. This changing applica-

tion carries with it the need for increased vigilance in orderto avoid the introduction of undesirable side effects; for ex-

ample, the long-term risk of causing a tumor at another or-gan site, while acceptable when treating advanced cancer,will prohibit the use of the same drug as a chemopreventiveagent in healthy individuals.

As cancer chemoprevention research has progressed, in-terest has turned to the investigation of natural products, anumber of which have been found to be active in animalmodels when administered at levels that appear to lack sys-temic toxicity. Prominent among these compounds are nu-

trients, including vitamins and the omega-3 FA; food addi-tives such as curcumin; and food-associated natural products,such as indole 3-carbinol and lycopene.

This review deals exclusively with the omega-3 FAs.These remarkable nutrients have attracted interest becauseof their importance in normal brain development (Neuringeret al., 1988), as dietary supplements for the prevention and

treatment of chronic cardiovascular disease (Simopoulos,1991), and the treatment of arthritic disorders (Kremer, 1991)

and diabetes mellitus (Malasanos & Stacpoole, 1991). Theyhave been, and are currently, the subject of various clinicaltrials in which they have shown a lack of complications orserious side effects. Epidemiological studies, investigationsutilizing a range of animal models, and mechanistic experi-ments in vitro all support their potential as cancer chemo-preventive and chemosuppressive agents, and as auxiliaryagents for cancer therapy. The tumor targets include carci-nomas of the breast and colon, which are two of the most

common cancers in North America and Europe.

Our objective is to provide a background general discussionof the omega-3 FAs, review the current status of research intotheir chemopreventive activities and the biological mecha-nisms involved, and to consider in some depth the develop-ment of a clinical trial in women at high breast cancer risk.

2. Nomenclature, chemistry, and metabolic conversion

of unsaturated fatty acids

The unsaturated FAs comprise monounsaturates andpolyunsaturates. The conventional chemical nomenclatureis to begin the systematic numbering of carbon atoms from

8/2/2019 1-s2.0-S0163725899000261-main

3/28


219

the carboxyl terminal group. The carbon atoms number 2and 3 from the carboxyl group are referred to as the

and

carbons, respectively, the last carbon is the

- or n

-carbon,and the position of a double bond is indicated by the symbol

, followed by a number: for example,

9

refers to a doublebond between carbon atoms 9 and 10 from the carboxylgroup. However, an accepted practice in describing thechemical structure of FA molecules is to start numbering

the carbons at the methyl group (

- or n

-).Oleic acid has its single double bond located between the9th and 10th carbon atoms from the methyl end, and so it is

designated an

9 (or n

-9) monounsaturated FA (Fig. 2). Itcan be synthesized by all mammals, including humans. Theomega-3 (

n

-3) and omega-6 (

n

-6) polyunsaturated FAs can-not be synthesized by mammals, and because they must beobtained from the diet, they are referred to as essential fattyacids. In Fig. 2, the n

-3 FAs are represented by

-linolenicacid (LNA) and the n

-6 FAs by linoleic acid (LA). Both LNAand LA are metabolized to longer-chain FAs, largely in theliver; LNA is converted to eicosapentaenoic acid (EPA), and

thence to docosahexaenoic acid (DHA), while LA is the met-

abolic precursor of arachidonic acid (AA). These events,which are summarized in Fig. 3, involve increases in chainlength and degree of unsaturation that are achieved by addingextra double bonds between the existing double bond and thecarboxyl group (de Gomez Dumm & Brenner, 1975). Com-petition exists between the n

-3 and the n

-6 FAs for the

4

and

6

desaturases, with the n

-3 FAs having greater affinities forthe enzymes. In consequence, increasing the dietary intake of

n

-3 FAs reduces the desaturation of LA and so, the produc-tion of AA (de Gomez Dumm & Brenner, 1975; Hague &Christoffersen, 1984), an inhibition that is achieved by EPAand DHA, as well as LNA (Christiansen et al., 1991).

Humans can produce EPA from LNA, but the extent of thecapacity to perform this conversion has not been well defined

(Hunter, 1990), and there is evidence that human

6

desatu-rase activity decreases with age (de Gomez Dumm & Bren-ner, 1975). In one study (Mantzioris et al., 1994), feeding alow LA-containing diet to human volunteers, thus reducingthe enzymic competition referred to above, did allow effec-tive conversion of LNA to EPA, with corresponding eleva-

tions in the tissue DHA levels. However, Grimsgaard et al.(1997) observed a 15% decrease in the serum phospholipidDHA level in a group of adult males whose diet was supple-

mented with 3.8 g EPA/day. In addition to any forward meta-bolic sequence, retroconversion of DHA to EPA does occurin humans (Von Schacky & Weber, 1985; Conquer & Holub,1997; Nelson et al., 1997; Grimsgaard et al., 1997).

Dietary LA generally is considered to be the major sourceof tissue AA, although lean meats and meat fat are directsources in the human diet (Li et al., 1998). LA is the principalpolyunsaturated FA in the American diet, amounting to 1015 g/day (Jonnalagadda et al., 1995), and most of the AA in

serum and platelets is derived from dietary LA (Grnn et al.,

1991). However, in another human study, James et al. (1993)found no changes in the neutrophil phospholipid AA levels inresponse to LA intakes ranging from 2.5 to 17.5% of energy.Similarly, Whelan et al. (1993) showed that feeding AA tomale Syrian hamsters increased the phospholipid AA levelsin a variety of normal tissues, whereas when dietary LA wasincreased by as much as 50%, there was no such effect. Theyconcluded that while dietary LA may influence eicosanoid

formation by increasing the tissue AA pool, this contributiondiminishes as dietary AA intake increases.

In rats, incremental elevations in LA intake did producecorresponding increases in liver AA (Lands et al., 1990b;

Fig. 1. The progression of premalignant benign breast disease to metastatic carcinoma.

8/2/2019 1-s2.0-S0163725899000261-main

4/28

220


Marangoni et al., 1992), but while increasing dietary LA in-take results in greater hepatic conversion of LA to AA, in ex-trahepatic tissues, the enhanced availability of LA can resultin displacement of AA. For example, when LA in the human

diet was increased to about 8% of total energy, the levels ofAA in the platelet phospholipids, and especially the AA/LAratio, were decreased (Tremoli et al., 1986). One conse-quence of the AA displacement is that feeding sufficientlyhigh levels of LA can actually result in a tissue-specific re-duction in the biosynthesis of AA-derived eicosanoids (Galliet al., 1981; Croft et al., 1984; Tremoli et al., 1986).

3. Dietary sources of omega-3 fatty acids, intestinal

absorption, and tissue and subcellular distribution

3.1. Dietary sources

To a considerable extent, the FA content of fish oils isdetermined by the food that is available to the fish, and thisis influenced strongly by the geographic area in which thefish live, the season of the year, and the fluctuations that oc-

cur from year to year. In consequence, published data on theclasses and levels of FAs in various species of fish varywidely, and they still may not represent the true situation

Fig. 2. Structures of oleic acid (an n-9 monounsaturated FA), LA (an n-6 polyunsaturated FA), and LNA (an n-3 polyunsaturated FA). The n-numbers are

counted from the methyl terminal.

Fig. 3. The metabolism of LNA to EPA and DHA and of LA to -linolenic acid (n-6; C20:3) and AA. Each metabolic sequence competes for the same enzyme

systems, with the affinity being greater for n-3 FA than n-6 FA. -Linolenic acid, AA, and EPA are the substrates for the biosynthesis of 1-series PG and

2-series LT, 2 series-PG and 4-series LT, and 3-series PG and 5-series LT, respectively.

8/2/2019 1-s2.0-S0163725899000261-main

5/28


221

(Stansby, 1986). Puustinen et al. (1985) analyzed the total

lipid content and the FA composition of 12 commonly eatennorthern European fish. They found that the range of totallipid amounts in different fish species was greater than thevariation of the percentage amount of EPA and DHA, whichled them to conclude that any beneficial dietary effectswould be best achieved by the inclusion of the fattier fish.The United States Department of Agriculture published fatand FA values for 30 fish of different species or location(Simopoulos, 1991); 14 of these are listed in Table 1. The

total amount of fat in these fish ranged from a low of 0.7 g/100 g in cod and haddock to highs of 13.8 and 13.9 g/100 gin Greenland halibut and Pacific herring, respectively. Therelative amounts of EPA and DHA contained in fish oilsvary considerably between species (Childs et al., 1990).Moreover, some freshwater fish oils contain relatively lowlevels of EPA, but substantial amounts of AA, comparedwith those present in marine fish oils and so, their consump-tion will have very different effects on the cell membrane

FA profiles and prostanoid production (Innis et al., 1995).In addition to fish and fish oils, soybean and canola (low

erucic acid rapeseed) oils may provide a significant sourceof dietary n

-3 FA in the form of LNA (Hunter, 1990), themajor FA in chloroplast lipids. In North American diets, theprincipal food sources of LNA are salad and cooking oilsand salad dressing products. The per capita intake in theUnited States has been estimated to be

1620 g/day formen and 12 g/day for women (Kim et al., 1984).

The changes in the FA content of the typical human diet thathave occurred over time since the Paleolithic period were dis-cussed by Simopoulos (1991). She pointed out that the early

diet contained small, but approximately equal, amounts ofn

-6and n

-3 FAs, whereas the modern Western diet contains a rela-

tive excess ofn

-6 FA. It is this imbalance ofn

-6 FA to n

-3 FA

that some investigators have associated with increased risks forboth cardiovascular disease and some cancers, including carci-noma of the breast. While this aspect of nutritional cancer epi-demiology will be discussed in detail in Section 7.1.5, one in-structive example may be mentioned at this point: the increasedbreast cancer risk in Japanese women, which has taken placeover the past four decades and which correlates with a decreasein the dietary n

-3:

n

-6 FA ratio (Rose, 1997a).

3.2. Intestinal absorption

The form in which dietary n

-3 FAs are ingested affectstheir absorption from the intestinal tract. In human subjects,EPA ethyl ester was not absorbed as efficiently as the free

FA (El Boustani et al., 1987). Lawson & Hughes (1988)found the relative absorption capacities to be: docosapen-taenoic acid (DPA; C

22:5

)

EPA and DHA. Also, the rela-tive absorption of the different forms of EPA, DHA, and

C

22:5

, compared with that of an 18:3 n

-3 control, was freeFA

triglyceride

ethyl ester. Fish and fish oil productsare mostly ingested as the triglycerides; much of the EPAand DHA in the Western diet comprises long-chain n-3 FAsthat are absorbed as 2-monoglyceride (Nelson & Ackman,

1988).Hudson & Tisdale (1994) compared the effect of free EPA

and EPA ethyl ester on cachectic weight loss and MAC16 ad-enocarcinoma growth in mice, and related this to their rates ofintestinal absorption. Oral administration of the free EPA re-sulted in prevention of body weight loss and suppression oftumor growth, effects that were associated with elevations inplasma and tumor EPA concentrations. In contrast, the ethyl

ester of EPA failed to exert a favorable effect on body weight,and the tumor volumes were not significantly different fromthose of control animals. This lack of therapeutic efficacywas reflected in the lower tumor EPA concentrations andplasma EPA responses to EPA ethyl ester.

Nordy et al. (1991) assessed the intestinal absorption ofboth EPA and DHA as the triglycerides or ethyl esters inhealthy adults, and in contrast to the reports by El Boustaniet al. (1987) and Lawson & Hughes (1988), found them tobe equally well absorbed.

3.3. Tissue and subcellular distribution

In humans and other mammals, DHA occurs at relativelyhigh levels in the cerebral cortex (OBrien & Sampson,

1965), the retina (Anderson, 1970), and testis (Poulos et al.,1975). Dietary deficiency of LNA in developing animals re-sults in reduced CNS DHA levels, which have been associ-ated with learning disabilities and impaired visual function(Innis, 1991). In response to a DHA-supplemented diet, the

n

-3 FA also concentrates in the heart muscle, as well as theliver and lung (Innis et al., 1995).

Both n

-3 and n

-6 FAs are important constituents of the cellmembranes, and cell membrane FA composition of both nor-

mal and neoplastic tissues undergoes modification in animals

Table 1

The total lipid and omega-3 fatty content

a

of 14 edible fish that are

widely available in the United States

Fish species

Total lipid

(g/100 g)

EPA

(g/100 g)

DHA

(g/100 g)

LNA

(g/100 g)

Herring, Pacific 13.9 1.0 0.7 0.1

Halibut, Greenland 13.8 0.5 0.4 Tr

Mackerel, king 13.0 1.0 1.2 ()

Salmon, chinook 10.4 0.8 0.6 0.1

Bluefish 6.5 0.4 0.8 ()Tuna, albacore 4.9 0.3 1.0 0.2

Herring, round 4.4 0.4 0.8 0.1

Salmon, pink 3.4 0.4 0.6 Tr

Trout, rainbow 3.4 0.1 0.4 0.1

Halibut, Pacific 2.3 0.1 0.3 0.1

Swordfish 2.1 0.1 0.1 ()

Plaice, European 1.5 0.1 0.1 Tr

Cod, Atlantic 0.7 0.1 0.2 Tr

Haddock 0.7 0.1 0.1 Tr

a

Per 100-g raw, edible, portion.

(), lack of reliable data; Tr, trace (

0.05 g/100 g).

Reproduced from Simopoulos (1991), with permission of the author

and the copyright holder, Williams & Wilkins, Baltimore.

8/2/2019 1-s2.0-S0163725899000261-main

6/28

222


fed different high-fat diets, reflecting the FA content of the

consumed lipid. The n

-3 FAs accumulate in phosphatidylcho-line, phosphatidylethanolamine, and triglycerides (de Bravoet al., 1991). The phospholipid pool is the major site of FA in-corporation in both cultured normal and transformed cells. Ina human breast cancer cell line, LA was found to be preferen-tially esterified to phosphatidylcholine, whereas EPA was in-corporated largely into phosphatidylethanolamine. EPA wasalso effective in channeling LA from phospholipid to the neu-tral lipid pool (Hatala et al., 1994).

4. The eicosanoids

4.1. Arachidonic acid-derived eicosanoids

AA (5,8,11,14-eicosatetraenoic acid) is an omega-6 polyun-saturated FA metabolically derived from LA. It is the substrate

for two classes of enzymes, the prostaglandin (PG) synthases(EC 1.14.99.1), which produce the prostanoidsPGs, prosta-

cyclin (PGI

2

), and thromboxanes (TXs), and the lipoxygenases(LOXs) (EC 1.13.11.12), which catalyze the biosynthesis ofthe hydroxyeicosatetraenoic acids (HETEs) and leukotrienes(LTs). These eicosanoids, so called because of their common

origin from a 20-carbon (eicosa-) polyunsaturated FA, behave

as local hormones (autacoids), and are chemical transmittersfor a variety of intercellular and intracellular signals. The bio-synthesis of the n

-6 FA-derived eicosanoids is initiated by therelease of AA from membrane stores, this cleavage from phos-pholipids being mediated by membrane-associated phospholi-pase A

2

activity (Verheij et al., 1981).PG synthase catalyzes two sequential reactions: first, the

cyclooxygenase (COX) activity of the enzyme converts AAto PGG

2

, and then the peroxidase activity reduces PGG

2

to

PGH

2

(Oates et al., 1988) (the subscripts indicate the pres-ence of double bonds). However, in spite of these two dis-tinct reactions, it has become commonplace for the com-plete entity to be referred to as COX. The PGH

2

formedfrom AA is then converted to the various prostanoids (Fig. 4).

There are two forms of COX, of which COX-1 is consti-tutively expressed in most tissues and is considered to gen-erate PG for normal physiological function (Herschman,1994; Kargman et al., 1995a). COX-2 undergoes rapid in-

duction in response to a variety of stimuli, including mito-gens, cytokines, and hormones (Herschman, 1994). WhileCOX-2 is undetectable in normal intestine (Kargman et al.,1996) and breast tissue (Hwang et al., 1998), it is expressed

Fig. 4. The metabolism of AA to the prostanoid subfamily of eicosanoids (2-series PG and TXA 2).

8/2/2019 1-s2.0-S0163725899000261-main

7/28


223

in the majority of colonic carcinomas (Kargman et al.,1995b) and in some primary breast cancers (Parrett et al.,1997; Hwang et al., 1998).

The PGs possess a broad spectrum of biological activities,but once PGH

2

has been produced, specific cell types are usu-

ally highly selective in their formation of bioactive pros-tanoids. For example, PGI

2

is synthesized primarily by vascu-lar endothelial cells and inhibits platelet activation andaggregation, while TXA

2

, a product of TX synthase-mediatedmetabolism of PGH

2

(Hamberg et al., 1975), is the principalCOX product in platelets, has opposing actions to PGI

2

, andpromotes platelet aggregation. Both PGI

2

and TXA

2

have im-portant roles in cancer biology because they are involved intumor cell-vascular endothelial cell and tumor cell-platelet in-teractions, but whereas PGI

2

inhibits metastasis, TXA

2

exerts

an enhancing effect (Honn, 1983; Chen et al., 1992).The 5-, 12-, and 15-LOXs insert molecular oxygen into

polyunsaturated FA (Fig. 5). The incorporation of 1 mol of

oxygen forms the corresponding 5-, 12-, or 15-hydroperox-yeicosatetraenoic acid (HPETE), which is reduced to thehydroxy analogue (5-, 12-, and 15-HETE), but in addition,5-HPETE is converted to the bioactive LT (Malle et al.,1987; Spector et al., 1988) (Fig. 6).

The HETEs are involved in the regulation of a wide range

of biological activities, which are only now being understood,but include ion transport, hormone secretion, and the immuneresponse. They have been associated with rheumatic and col-lagen diseases, atherosclerosis, and ischemic injury. Withinthe lung, LOX activity and 5-, 12-, and 15-HETE productionoccur in alveolar macrophages (Peters-Golden & Thebert,

1987), and LTs are believed to have a role in the pathogenesisof asthma, cystic fibrosis, and pulmonary hypertension. Theremay be a place for n

-3 FAs in the management of these respi-ratory diseases: in one study, the levels of 12-HETE in lunghomogenates were reduced by approximately 60% in mice

fed fish oil compared with those in mice fed LA-rich saf-flower oil (Zhang & German, 1996), and AA, 5-HETE, and12-HETE concentrations were all reduced in the lungs of ratsfed a range ofn

-3 FA levels with a constant amount of n

-6FA (Hwang et al., 1988).

Both LTs and HETEs influence various aspects of tumorcell biology, and there is considerable interest in the poten-tial for pharmacological inhibitors of their biosynthesis as anew approach to cancer chemoprevention and therapy(Rose & Hatala, 1994; Rioux & Castonguay, 1998).

4.2. Eicosanoids derived from eicosapentaenoic acid

EPA is a precursor of the prostanoids of the 3-series,with 3 double bonds (Needleman et al., 1979; Fischer &Weber, 1984), and LTs of the 5-series, with 5 double bonds(Prescott, 1984; Lee et al., 1984; Strasser et al., 1985). The

n

-3 FA is converted by COX to the 3-series endoperoxidePGH

3

, by way of PGG

3

, which is further metabolized to

PGE

3

, PGI

3

, and TXA

3

, and thence to the inactive metabo-lites

17

-6-keto-PGF

1

and TXB

3 (Fig. 7). These pathwayscorrespond to those by which AA is converted to the 2-seriesprostanoids, and EPA competes effectively with the n-6 FAfor available COX activity (Needleman et al., 1979). In hu-mans, feeding an EPA-supplemented diet caused an in-

Fig. 5. The metabolism of LA to 13-HODE and of AA to 5-, 12-, and 15-HETE. Not shown are the intermediate compounds 5-, 12-, and 15-HPETE, which

are converted to the corresponding HETE by peroxidase activity.

8/2/2019 1-s2.0-S0163725899000261-main

8/28

224 D.P. Rose, J.M. Connolly / Pharmacology & Therapeutics 83 (1999) 217244

crease in PGI3 production, as reflected in the urinary excre-tion of 17-2,3-dinor-6-keto-PGF1 (Fischer & Weber,1984). Because the PGI3 derived from EPA is as antiaggre-gatory as PGI2 derived from AA, whereas EPA-derivedTXA3 has only weak aggregatory activity, the net result is

that the COX products of EPA metabolism are associatedwith diminished platelet aggregation (Fischer & Weber,1984; Kramer et al., 1996).

EPA can also be metabolized by way of the 5-LOX path-way, which parallels the metabolism of AA to the 4-seriesLT (Fig. 8). The immediate product of this 5-LOX activity,LTA5, is metabolized to LTB5, which possesses only 510%of the activity of LTB4 as a chemotactic and aggregatingfactor for polymorphonuclear leukocytes (Lee et al., 1984).Some tumor cell types have been shown to produce LTs of

the 5-series from EPA (Jakschik et al., 1980; Murphy et al.,1981), and these biosynthetic pathways are subject to di-etary manipulation (Murphy et al., 1981; Strasser et al.,

1985). The synthesis of LTB5 and LTC5 by mastocytomasin fish oil-fed mice was accompanied by pronounced reduc-tions in LTB4 and LTC4 (Murphy et al., 1981). However,we are not aware of any studies that have related these LTsto the cancer chemopreventive properties of EPA.

Although the earlier studies largely attributed the inhibitory

effects of dietary fish oil on eicosanoid biosynthesis to EPA,DHA was found to be a potent inhibitor of PG production invitro (Corey et al., 1983) and of LTB4, LTC4, and 12-HETE,products of LOX-catalyzed AA metabolism (Lokesh et al.,1988). When DHA is incorporated into platelet and other cellmembranes, it is not metabolized to a TX (Aveldao & Spre-

cher, 1983) or to LT (Corey et al., 1983). Even so, it can com-

petitively inhibit eicosanoid biosynthesis from AA.

4.3. Eicosanoid biosynthesis andn-3:n-6 fatty acid ratio

A number of investigators have stressed the importance

of the dietary n-3:n-6 FA ratio, rather than the absolute lev-els of the two classes of unsaturated FAs, in modulating

eicosanoid biosynthesis from AA. This arises because thecompetitive inhibition that exists between n-3 and n-6 FAsfor the desaturases and elongases results in the capacity ofdietary n-3 FA to suppress the conversion of LA to AA, be-ing dependent not only on the amount ofn-3 FA, but also onthe amount ofn-6 FA in the diet. In one study (Boudreau etal., 1991), it was found that there was no dose-dependentchange in tissue AA or in the levels of TXB2, 6-keto-PGF1,and 12-HETE when groups of rats were fed 10% total fat-

containing diets providing different amounts of fish oil, butwith equivalent increases in safflower oil so as to maintain aconstant n-3:n-6 FA ratio of 0.3.

This competition between the two classes of polyunsatu-rated FAs was also demonstrated in human volunteers byfeeding them a fixed level of dietary n-3 FA, but in combi-nation with high or low intakes of n-6 FA. Here, the EPAcontent in serum, platelets, and neutrophil phospholipidswas significantly higher in those fed the low LA diet (Grnn

et al., 1991; Cleland et al., 1992).The importance of the dietary n-3:n-6 FA ratio has been

re-emphasized by two cancer-related studies. In one, con-cerned with dietary n-3 FAs as potential modifiers of coloncancer risk, a relatively high n-3:n-6 FA ratio was required

Fig. 6. The metabolism of AA to the 4-series LTs, with 4 double bonds, via 5-HPETE. The LTA synthase is a dehydrase.

8/2/2019 1-s2.0-S0163725899000261-main

9/28

D.P. Rose, J.M. Connolly / Pharmacology & Therapeutics 83 (1999) 217244 225

to obtain suppression of PGE2 production in human rectalmucosa (Bartram et al., 1995a). In the other, a multi-national epidemiological study, the ratio ofn-3 FAs to n-6FAs in adipose tissue biopsies, but not the absolute levels ofthe two classes of FAs per se, was inversely associated withbreast cancer risk (Simonsen et al., 1998). These two studiesare discussed in more detail in Section 6.

4.4. 13(S)-Hydroxyoctadecadienoic acid

LA can form an oxidation product, 13(S)-hydroxyocta-decadienoic acid (13-HODE), both under the catalytic influ-

ence of 15-LOX (Fig. 5) and by nonenzyme autoxidation(Welsch, 1995). This metabolite is attracting the attention oftumor biologists because it may be involved in both car-cinogenesis and cancer progression.

In some cell types, 15-LOX metabolism of LA is stimu-lated by epidermal growth factor (EGF), and 13-HODEmodulates the EGF mitogenic signal (Glasgow et al., 1992).Welsch (1995) has described the stimulation of mousemammary epithelial cell proliferation by 13-HODE, and the

LA-stimulated growth of these cells in vitro requires the

presence of EGF (Bandyopadhyay et al., 1987). On the

other hand, in contrast to these stimulatory effects, 13-HODE opposes the 15-HETE and 12-HETE stimulation oftumor cell adhesion, which is an integral part of the meta-static process (Bastida et al., 1990; Liu et al., 1991).

5. Biomarkers ofn-3 fatty acid nutritional status

The FA composition of the plasma lipids and the mem-

branes of platelets and erythrocytes provides an assessmentof the dietary intake of the long-chain n-3 FAs over the pre-ceding 2 days to 18 weeks (Bjerve et al., 1993; Andersen etal., 1996; Grnn et al., 1991; Innis et al., 1988; Brown et al.,1991). However, for epidemiological studies aimed at deter-mining the association of dietary FAs with disease risk oroutcome, it is necessary to obtain an index of consumptionover a relatively long period of time. For this purpose, mostinvestigators have utilized adipose tissue obtained by percu-

taneous biopsy, for which the turnover time for FAs hasbeen estimated to be 13 years (Field & Clandinin, 1984).

Marckmann et al. (1995) found adipose tissue DHA andDPA levels to be much higher than that of EPA, althoughthe food record estimates of dietary EPA and DHA intakeswere both considerably higher than that of DPA. The appar-ent discrepancy may have resulted from the elongation ofEPA to the less metabolically active DPA. In this study, theDHA content of adipose tissue provided a better marker of

fish and marine n-3 FA intake than did EPA, a result thatwas also obtained by Tjnneland et al. (1993).

Two other studies provided a comparison of data from

food-frequency questionnaires, diet records, and subcutane-ous adipose tissue FA analyses in American postmeno-pausal women and men, respectively (London et al., 1991;Hunter et al., 1992). Both showed significant correlationsbetween the percentage ofn-3 FAs in adipose tissue, and thepercentage ofn-3 FAs in the diet, as determined by a food-

frequency questionnaire. In the study of American males,dietary intakes were also assessed by 7-day diet records,which are generally regarded as more accurate. However,this method of evaluation gave a similar level of correlationto that obtained with the semi-quantitative food-frequencyquestionnaire, suggesting that the less complex method pro-vides an equally reliable assessment. In neither study wereseparate data collected for EPA and DHA and so, there wereno results upon which to judge the relative merits of the twon-3 FAs as markers ofn-3 FA intake. One point worth mak-ing is that the instruments ofn-3 FA intake used in this andother epidemiological studies do not distinguish betweenthe types of fish consumed, which, given the wide varia-tions in n-3 content, may reduce their sensitivity.

Although the analysis of adipose tissues is necessary toevaluate dietary FA intake over extended periods of timeand is unaffected by temporary variations in diet, their ac-quisition is not amenable to the large-scale sampling re-

quired for many epidemiological studies, and the presence

Fig. 7. The metabolism of EPA to 3-series prostanoids. The only structural

difference between PGH3 and PGH2 is that PGH3 has an additional, 17

,unsaturation.

8/2/2019 1-s2.0-S0163725899000261-main

10/28


of large amounts of nonessential FAs can obscure the rela-tively low levels ofn-3 FAs. With this in mind, it is encour-

aging that a carefully executed comparative study by God-

ley et al. (1996) found that the EPA and DHA contents ofadipose tissue aspirates and erythrocyte membranes hadsimilar correlations with fish consumption, as estimated bya food-frequency questionnaire.

Because the n-3 FAs are very susceptible to peroxida-tion, appropriate storage is a prime concern when accruingsamples for later analysis. Two studies that have addressedthis issue concluded that the DHA and EPA contents of adi-

pose tissue were even stable at room temperature (20)over a 7-month period (Deslypere et al., 1993), and that, atleast when stored at70, the n-3 FA in erythrocytes exhib-ited little loss over 12 months (Stanford et al., 1991). Theauthors suggested that the remarkable stability of DHA andEPA in adipose tissue may be due to the presence of highlevels of vitamin E, which prevents deterioration due tolipid peroxidation because of its antioxidant properties.

6. Omega-3 fatty acids and cancer

6.1. Breast cancer

6.1.1. Epidemiology

Breast cancer is the most common cancer among womanin the United States, and is second to lung cancer as themost frequent cause of cancer-related deaths. It was esti-

mated that 178,700 new cases would be diagnosed in 1998,and that 43,500 American women would die of the disease

(American Cancer Society, 1998). The influence of dietary

fat intake on breast cancer risk continues to be a topic ofconsiderable interest, and controversy (Hunter et al., 1996;Wynder et al., 1997). While this important issue is unlikelyto be resolved in the immediate future, ecologic studies doshow a wide spread in breast cancer incidence and mortalityrates between countries and correspondingly large varia-tions in dietary practices. Moreover, migration studies con-sistently have shown that when women move from coun-

tries such as Japan, in which breast cancer risk and dietaryfat intake are relatively low, to countries such as the UnitedStates, for which the reverse is true, their breast cancer inci-dence rates increase within one generation (Ziegler et al.,1993). Even within Japan, a correlation was found betweendietary total fat intake and breast cancer mortality rates fordifferent districts, with both being higher in urban locations(Hirayama, 1978).

One possible explanation for the conflicting evidence

concerning dietary fat and breast cancer risk is that the em-phasis has been largely on the total amounts consumedrather than the nature of FA contained in those fats (Rose,1997b). Support for this interpretation came when Kaizer etal. (1989) pointed out the inverse association between breastcancer incidence and mortality rates and dietary fish intakefor the same 32 countries that previously had been shown toexhibit a direct relationship for total fat consumption (Arm-

Fig. 8. The metabolism of EPA and AA to their LT products of the 5- and 4-series, via 5-hydroperoxyeicosapentaenoic (5-HPEPE) and 5-HPETE, respec-

tively.

8/2/2019 1-s2.0-S0163725899000261-main

11/28


strong & Doll, 1975). However, this analysis was heavily

dependent on the inclusion of data from Japan, a countrywith an exceptionally high fish consumption and low breastcancer mortality rate. Sasaki et al. (1993) analyzed the datafrom 30 countries, including Japan, by multiple regressionanalyses using dietary animal fat minus fish fat intake.When this was done, there was a highly significant positiverelationship with breast cancer mortality rates for womenover 50 years of age; a separate analysis for fish intakeshowed the expected negative correlation.

Despite these widely quoted results from ecologic stud-ies, most case-control studies have not supported a protec-tive effect of fish consumption (reviewed by Willett, 1997).A similar absence of benefit was found in prospective stud-ies performed in the United States (Stampfer et al., 1987)and in Norway (Vatten et al., 1990), a country with a rela-tively high consumption of n-3 FA-rich-cold water fish.However, as noted in Sections 3.1 and 4.3, it may be the ra-tio ofn-3 FA to n-6 FA in the diet, rather than the absolute

levels of their intake, which is critical in influencing cancerpromotion and progression. This distinction is likely to beparticularly important in Western countries, where the lev-els of dietary n-6 FAs are generally high, whereas those ofn-3 FAs are low. This interpretation is supported by the re-sults of a multi-national case-control study of adipose tissueFA profiles that was performed in 5 European countrieswith an approximately 2-fold range in breast cancer inci-dence (Simonsen et al., 1998). Although the level ofn-3 orn-6 FA showed no consistent association, the ratio of long-chain n-3 FA to total n-6 FA was inversely correlated withbreast cancer in four of the five centers. Interestingly, the

relationship between breast cancer and the ratio of LNA ton-6 FA was inconsistent across centers.Although Japanese women continue to have a relatively

low breast cancer risk, this has risen over the past 40 years,and an increasing breast cancer mortality rate since 1960has been accompanied by dietary changes, prominent

among which is an increase in the average fat consumptionfrom 9% of total energy in 1955 to 25% in 1987 (Wynder etal., 1991). However, in addition to an increase in the con-sumption of red meat, there has been an increase in the useof LA-rich vegetable oils, a decrease in fish consumption,and hence, a reduction in the dietary n-3:n-6 ratio (Lands etal., 1990a).

While ecologic studies have generally associated highbreast cancer risk with high dietary fat intakes, the native

Eskimos of Greenland and Alaska provide an instructive ex-ception. Their diet is high in fat, but this comes largely frommarine mammals (Bang et al., 1976) and fish (Nobmann etal., 1992). In consequence, they have exceptionally high in-takes of n-3 FA, which are reflected in the plasma anderythrocyte membrane EPA and DHA levels, and the n-3:n-6ratios (Innis et al., 1988; Parkinson et al., 1994). A surveyconducted for the years 19691973 showed that althoughthe total cancer incidence rates among Alaska Eskimos and

Aleuts did not differ significantly from that of the United

States white population, there were significantly lower rates

for cancers of the breast, endometrium, and prostate (Lanieret al., 1976). Although these observations were consistentwith earlier reports, a further survey of cancer incidence forthe years 19891993 found that there were significant in-creases in the incidence rates for cancers of the breast andendometrium in women, and for cancers of the prostate andcolon in men (Lanier et al., 1996). A contributing factor tothese changes may be urbanization of the Alaska nativepopulation, and altered dietary practices. Nielsen & Hansen

(1980) had drawn a similar conclusion from an analysis ofbreast cancer incidence among indigenous Greenlandicwomen for the years 19501979. A recent study of Cana-dian Eskimos found that plasma DHA and, to a lesser ex-tent, EPA levels were positively correlated with age,whereas age was negatively correlated with plasma n-6 FA(Rode et al., 1995). This age-related difference in plasmaFA and the n-3:n-6 ratio was due to a high consumption ofmarine mammals and fish by the older Eskimos and re-

placement of this fat source by vegetable oils in the youngermembers of the community.

Although these epidemiological studies of special popu-lations support a protective role for dietary n-3 FA, Miller &Gaudette (1996), while remarking on a low breast cancerrisk among Inuit populations in the circumpolar region,rightly point out that other known protective factors, earlyage at first pregnancy and prolonged lactation, are alsoprevalent among the Inuit. These also may be changing over

time in response to Western influences. Indeed, it has to berecognized that all epidemiological evidence for an associa-tion between dietary n-3 FA intake and cancer risk is infer-

ential and correlative, and liable to confounding by otherfactors.

6.1.2. Experimental studies

Although the emphasis of research has been on the ef-fects ofn-3 FA on the later stages of carcinogenesis in vivo,

there is some direct evidence that they also affect cell trans-formation. Takahashi et al. (1992) used ionizing radiation ofmouse embryo fibroblasts, and, in separate experiments, thetransfection of these and NIH 3T3 cells with the H-ras on-cogene, as models to demonstrate suppression of transfor-mation by EPA and DHA, and they associated this with theinhibition of eicosanoid production from AA. An analogousstudy was performed by Telang et al. (1988) using explantcultures of mammary glands from mice harboring the mu-

rine mammary tumor virus. In their experiment, it wasshown that both LA and AA enhanced the formation of pre-neoplastic hyperplastic alveolar lesions in vitro, whereasEPA exerted inhibitory activity. In association with theseopposing effects, the n-6 FAs increased the levels ofras p21protein, whereas the n-3 FAs decreased this indicator ofrasproto-oncogene expression.

It is well established that chemically induced rat mam-mary carcinogenesis is promoted by dietary n-6 FA, but is

inhibited by feeding n-3 FA-rich menhaden oil (Jurkowski &

8/2/2019 1-s2.0-S0163725899000261-main

12/28


Cave, 1985; Braden & Carroll, 1986; Abou-El-Ela et al., 1988,

1989; Cohen et al., 1993). Abou-El-Ela et al. (1988) showedthat high tumor production rates of PGE2, 6-keto-PGF1, andLTB4 were associated with enhanced mammary carcinomadevelopment in dimethylbenz[a]anthracene (DMBA) exposedrats fed a 20% corn oil, LA-rich diet, whereas the levels ofthese eicosanoids were reduced and tumorigenesis was par-tially suppressed, in animals fed a 20% menhaden oil diet.In their second study, Abou-El-Ela et al. (1989) obtained in-hibition of mammary tumorigenesis and of PGE2 and LTB4production by tumor cells when they fed a diet containingFAs with an n-3:n-6 FA ratio of 1.2 (15% menhaden oil and5% corn oil) compared with a 20% corn oil diet.

The optimal level of menhaden oil to be fed in order tosuppress experimental mammary tumor development re-mains uncertain, but so also does the relevance of thesefeeding studies to the chemoprevention of human breastcancer by n-3 FA. There is evidence that when menhadenoil is fed as 18% of the diet, it may actually enhance mam-

mary tumor development (Cohen et al., 1993), but it is ques-tionable whether such an extreme situation has any implica-tions for the design of human dietary intervention trials,where there is no possibility of attempting such high in-takes.

In addition to their effects on mammary carcinogenesisin animal models, FAs affect both the growth of trans-planted tumors and the expression of the metastatic pheno-type. Gabor et al. (1985) showed an enhancing effect of

dietary LA on the growth of transplanted mammary carci-nomas in BALB/c mice, and subsequently reported thatfeeding a diet containing 10% menhaden oil suppressed

growth, or more specifically, enhanced tumor cell loss (Ga-bor & Abraham, 1986). Earlier, Karmali et al. (1984) hadalso reported inhibition of R3230AC mammary tumorgrowth in rats fed MaxEPA, a fish oil concentrate that con-tains approximately 18% EPA and 12% DHA.

Human breast cancer cells growing as xenografts in athy-

mic nude mice have proved invaluable for assessing the in-fluence of dietary FAs on metastasis, despite the limitationthat estrogen-dependent cell lines do not metastasize, or doso rarely, so that the model cannot be used to study estro-gen-FA interactions on the metastatic process (Rose & Con-nolly, 1997). A number of studies have shown that diets richin the long-chain n-3 FA inhibit the growth (Borgeson et al.,1989; Gonzalez et al., 1991; Rose & Connolly, 1993; Roseet al., 1995) and metastasis (Rose & Connolly, 1993; Rose

et al., 1995) of human breast cancer cell lines growing assolid tumors in nude mice. While fish oil was used in theearlier studies, when pure EPA and DHA ethyl esters be-came available, these were shown to be effective in sup-pressing tumor progression when fed at levels of 8% and4% (wt/wt) in a diet containing a total of 20% fat (Rose etal., 1995). As generally used, the nude mouse model is oflimited clinical relevance because the tumor growing at theinjection site is left in situ throughout the experiment. A

more realistic approach was introduced when the ethyl es-

ters of EPA or DHA were used as nutritional adjuvant ther-

apy after the surgical removal of the primary breast cancer(Rose et al., 1996). In this way, the focus was on the chemo-suppression of residual disease and of microscopic me-tastases.

6.1.3. The relative potency of eicosapentaenoic acid

and docosahexaenoic acid in breastcancer animal models

The recent availability of EPA and DHA in bulk quanti-ties and of unique DHA-rich oils derived from microalgaehas offered a new approach to n-3 FA chemoprevention andthe suppression of cancer progression. It is also now possi-ble to compare the efficacy of the two long-chain n-3 FAs.Noguchi et al. (1997) used the DMBA-induced mammarytumor model to examine the effects of low doses of EPAand DHA ethyl esters on tumor incidence and growth in rats

fed a high-fat, 20% corn oil, diet. In this experiment, 0.5 mLof the 98% pure n-3 FA was given by gastric gavage twice a

week to provide a calculated dietary n-3:n-6 FA ratio of1:1.8. The DHA-treated rats showed a 69% reduction andthe EPA-treated rats a 47% reduction in tumor incidencecompared with the incidence in the high-fat-fed controls.These relative degrees of efficacy of the two n-3 FAs wereparalleled by a partial, but statistically insignificant, sup-pression of growth once the tumors had developed.

A transplantable mouse mammary carcinoma was usedby Kinoshita et al. (1994) to compare the effects of EPA andDHA, again given at low dose by intragastric intubation, onprimary tumor growth and metastasis to the lungs. Somepartial suppression was observed, with no significant differ-ences between the two n-3 FAs.

When we compared the two n-3 FAs in experiments us-ing nude mice bearing human breast cancer xenografts, wefound no difference between EPA and DHA ethyl esters in

achieving partial suppression of tumor growth. There was,however, some indication that EPA was more effective ininhibiting metastasis (Rose et al., 1995). A second study inwhich the n-3 FAs were employed in an experimental proto-col designed to mimic postsurgery, adjuvant chemotherapyproduced opposite results. The extent of lung metastasiswas reduced to a greater degree by feeding 2% or 4% DHA,compared with the same levels of EPA (Rose et al., 1996).More recently, we have had the opportunity to evaluate

DHA triglyceride prepared from microalgae in the MDA-MB-435 breast cancer cell/nude mouse model (Connolly &Rose, 1998a) The control of primary tumor growth and me-tastasis appeared to be superior to that obtained with EPA inour earlier experiment.

6.2. Colon cancer

6.2.1. Epidemiology

Colon cancer is the fourth most common cancer in theUnited States and ranks second among cancer-relatedcauses of death. It was estimated that 24,600 American

women and 23,100 American men would die of the disease

8/2/2019 1-s2.0-S0163725899000261-main

13/28


in 1998 (American Cancer Society, 1998). Some of the in-

ternational comparisons of breast cancer mortality rates ap-ply also to colon cancer (Rose et al., 1986), and a positivecorrelation was found between the geographic distributionof the breast and colorectal cancer mortality rates within theUnited States (Jansson et al., 1975). In Finland, an excessrisk of colon cancer was observed in women who had a pre-vious history of breast cancer (Teppo et al., 1985), and otherstudies have shown an increased incidence of precancerouscolonic polyps in breast cancer patients (Bremond et al.,

1984; Jouin et al., 1989). Rozen et al. (1990) reported thatrectal mucosal cell proliferation, a presumptive marker ofcolon cancer risk, was elevated in treated breast cancer pa-tients. These relationships suggest the presence of either acommon risk factor(s) or the same relative lack of factor(s)that are protective against the two tumor types. Amongthese, it is tempting to focus on dietary FA, but recent epi-demiological studies indicate that we cannot ignore sharedendocrine mechanisms (Fernandez et al., 1996; Hbert-

Croteau, 1998).In an analysis involving 24 European countries, an in-

verse correlation was found in males between colorectalcancer mortality and current fish intake. A similar trend forfemales did not achieve statistical significance, and therewas no association for breast cancer (Caygill & Hill, 1995).However, the importance of the dietary n-3:n-6 FA ratiowas stressed in Section 4.3, and so it is of particular interestthat when those data were reanalyzed for breast and col-

orectal cancer, there was evidence of a protective effect of ahigh fish intake relative to that of dietary sources ofn-6 FAsfor both tumor types (Caygill et al., 1996).

Familial adenomatous polyposis is an example of a ge-netically determined precancerous condition in which nu-merous polyps develop in the colon and rectum. These ap-pear in young adulthood, and virtually all of theseindividuals develop colon adenocarcinoma by age 40 years(DeCosse et al., 1977). The standard therapy is total colec-

tomy. In a controlled clinical trial, the nonsteroidal anti-inflammatory drug and unselective COX inhibitor sulindacreduced the number and size of polyps, but its effect was in-complete (Giardiello et al., 1993). This precancerous condi-tion is an obvious target for chemopreventive intervention,and while the present focus is on selective COX-2 inhibi-tors, the place for dietary n-3 FA supplementation meritscareful evaluation.

6.2.2. Experimental studies

A number of studies have shown that experimental ratcolon carcinogenesis is inhibited by feeding a diet contain-ing a high level of fish oil (Reddy & Maruyama, 1986; Nel-son et al., 1988; Reddy & Sugie, 1988; Deschner et al.,1990; Reddy et al., 1991) or supplemented with n-3 FAs(Minoura et al., 1988). In the experiment performed by Mi-noura et al. (1988), a diet containing 5% total fat and sup-plemented with 4.7% EPA (91% pure; without DHA) to-

gether with 0.3% LA to prevent essential FA deficiency

inhibited azoxymethane-induced colon carcinogenesis in

rats compared with tumorigenesis in rats fed a 5% LA con-trol diet. It should be noted that the LA content of the con-trol diet generally would be considered as providing a lowintake of the n-6 FA. Therefore, in human terms, at least forWestern countries, the EPA was exerting its protective ef-fect in the presence of an already favorable low-fat, low-LA diet.

The effects of different fish oil-containing diets, formu-lated to provide several n-3:n-6 FA ratios, on the early

pathological stages of experimental colon carcinogenesiswere studied by Deschner et al. (1990). They used anazoxymethane-induced rat model fed diets that all containeda total of 20.4% fat; the fish oil concentrate MaxEPA wasthe source ofn-3 FA. Epithelial cell proliferation in the co-lonic crypts showed the expected elevation as a result of ex-posure to azoxymethane, and this increase was partiallysuppressed by feeding the highest level of MaxEPA (16%wt/wt) with an n-3:n-6 FA ratio of approximately 1.5. The

16% MaxEPA-containing diet also inhibited the increase inDNA-synthesizing, S-phase, cells in the middle and upperthirds of the crypts, which is a feature of a carcinogenic re-sponse. Similarly, the occurrence of early dysplastic fociwas reduced significantly in carcinogen-exposed animalsfed either 16% or 10.2%, but not 4.4%, MaxEPA.

In a second part of this study, Deschner et al. (1990) ex-amined the effect of the same levels of MaxEPA on the fulldevelopment of azoxymethane-induced colonic carcinomas.

The highest tumor incidence occurred in rats fed 20.4%corn oil (5055% LA) or 16% corn oil with the lowest(4.4%) level of MaxEPA. Diets containing 4.4% corn oil

without any n-3 FA addition, or 10.2% or 16% MaxEPA, allproduced similar reductions in colonic tumor incidence.We have described the study by Deschner et al. in some

detail because it is one of the few in which different levelsof n-3 FAs, with differing n-3:n-6 FA ratios, have beencompared for their chemopreventive efficacy. One addi-

tional point that came out in analyzing the results of this ex-periment is that both tumor incidence and number of tumorsper animal were actually higher in rats fed the 16% cornoil:4.4% MaxEPA diet than in those fed 20.4% corn oil.This could be interpreted as a direct adverse effect of dietscontaining supplemental n-3 FA, at relatively low levels, oncolon carcinogenesis. However, as suggested by the au-thors, it could also be a reflection of the dietary LA leveland the reduced production of AA-derived eicosanoids that

can occur with very high LA intakes (Section 2), such asthat provided by a diet containing 20.4% corn oil.

The effect ofn-3 FA on the early stages of colon carcino-genesis was also examined by Takahashi et al. (1993). How-ever, the experimental design was somewhat unusual in thatDHA ethyl ester (0.7 mL) was given by intragastric intuba-tion in single doses 5 times a week. This regimen reducedthe formation of 1,2-dimethylhydrazine-induced aberrantcrypt foci, and when these colonic preneoplastic lesions did

occur, there was retardation of their subsequent growth.

8/2/2019 1-s2.0-S0163725899000261-main

14/28


Various germ line mutations of the murineApc gene re-

sult in multiple intestinal polyposis, and provide a model forhuman familial adenomatous polyposis (Moser et al., 1990;Fodde et al., 1994). The inhibitory effects of DHA and anEPA/DHA-enriched fish oil concentrate on the develop-ment and progression of intestinal polyps have been demon-strated in two of these models. Oshima et al. (1995) fed a3% DHA ethyl ester-containing diet, of unspecified totaland n-6 FA content, to female and male Apc716 heterozy-gotes. The females showed a 69% reduction in polyp forma-

tion and a significant suppression of polyp growth. In themales, there was no effect on polyp number and a lesser ef-fect on polyp size. The mechanism for this sex difference isobscure and was not reproduced in a similar study per-formed by Paulsen et al. (1997). These investigators usedthe Min (multiple intestinal neoplasia) mouse, which isheterozygous for a nonsense mutation of the Apc gene atcodon 850, and fed a 12% total fat diet and a fish oil con-centrate containing 54.4% EPA, 30.3% DHA, and 8.3%

other n-3 FAs. Both tumor numbers and size were reducedin male and female mice fed the n-3 FA-rich diet.

As in the case of breast cancer, the administration of pureEPA or DHA has also been shown to inhibit colon cancercell metastasis to the lungs in experimental animal models(Iigo et al., 1997; Suzuki et al., 1997). In both cases, theseinhibitory effects on tumor progression were associatedwith a reduction in matrix metalloproteinase (EC 3.4.24.7)activity, which is a critical part of the invasive phenotype

(Liu & Rose, 1995; Suzuki et al., 1997).

6.3. Mechanisms

6.3.1. Breast cancer6.3.1.1. Mitogenesis and apoptosis Both EPA and DHAinhibit the growth of human breast cancer cell lines in vitro.In experiments performed by Rose & Connolly (1990),DHA was considerably more effective than EPA in sup-pressing the growth of estrogen-independent MDA-MB-231 breast cancer cells. While there was a 50% reduction incell number when 2 g/mL of DHA was added to the cul-

ture medium, trypan blue exclusion indicated that the cellspresent were viable, suggesting that the cell loss was due toinhibition of mitosis rather than cytotoxic cell death. Thetwo long-chain n-3 FAs, and to a lesser degree LNA, alsoinhibited the growth of the estrogen-dependent MCF-7breast cancer cell line (Grammatikos et al., 1994). Again,this inhibition was not due to cytotoxicity, which can occurdue to the formation of peroxidation products from polyun-saturated FAs (Welsch, 1995).

Although nonspecific cytotoxicity does not appear to beresponsible for the growth inhibition induced by n-3 FA atlow concentration, at high levels, both n-3 FA and n-6 FAcaused lysis of human breast cancer cells in vitro (Begin etal., 1986, 1988; Begin & Ells, 1987). This cytotoxic activitywas shown to correlate with the levels of intracellularthiobarbituric acid-reactive substances, the determination ofwhich provides a crude index of lipid peroxidation. How-

ever, DHA produced the least thiobarbituric acid-reactive

substances, and the highest levels were generated by -lino-lenic acid and AA (Begin et al., 1988), a result that is notconsistent with the overall experience concerning the rela-tive effects ofn-3 FAs and n-6 FAs on breast cancer cellgrowth (Rose & Connolly, 1990; Buckman et al., 1991),and brings into question the physiological and clinical rele-vance of these cell culture experiments performed in vitrowith very high FA concentrations.

The acquisition of solid tumor mass and increasing cell

number is the consequence of active mitogenesis, reducedprogrammed cell death (apoptosis), or a combination ofthese two cellular events (McDonnell, 1993; Kerr et al.,1994). In contrast to cytotoxin-induced necrosis, apoptosisis a physiological process important to tissue developmentand organ modeling, and is carefully regulated by the ex-pression of specific genes. Although there is direct evidencethat the long-chain n-3 FAs can induce apoptosis, this ap-pears to be cell-type specific. Thus, while culture with EPA

resulted in apoptosis in a human pancreatic cancer cell line(Lai et al., 1996), this n-3 FA produced an inhibition ofMorris hepatocarcinoma 3924A growth in vivo that was pri-marily due to reduced proliferation. In contrast, DHA hadno effect on mitosis, but it induced apoptosis (Calviello etal., 1998).

We are not aware of any published studies that have in-vestigated the effects ofn-3 FAs on breast cancer cell apop-tosis, but this may be inferred from a publication by Gabor

& Abraham (1986), which predated the current intense in-terest in apoptosis and cancer. They reported that the inhibi-tion of growth of a transplantable mouse mammary adeno-

carcinoma by dietary n-3 FA was a result of cell loss andsuggested the involvement of PG in this process.As discussed in Section 6.1.2, the inhibition of DMBA-

induced rat mammary carcinogenesis by feeding n-3 FA-rich menhaden oil is associated with a reduction in tumorPG and in LOX-mediated AA product levels. The relative

importance of these eicosanoid families in mammary tum-origenesis and breast cancer progression is uncertain, al-though LOX rather than COX inhibitors were effective ininhibiting human breast cancer cell growth in vitro (Rose &Connolly, 1990; Buckman et al., 1991) and DMBA-inducedtumor growth in vivo (Kitagawa & Noguchi, 1994). Simi-larly, Abou-El-Ela et al. (1989) concluded that a reductionin tumor LTB4 production due to 5-LOX inhibition wascausally associated with n-3 FA inhibition of DMBA-

induced rat mammary tumorigenesis.Both COX-2 (Tsujii & DuBois, 1995) and 12-LOX

(Tang et al., 1996) expression have been shown to down-regulate the apoptotic pathway, and overexpression of 12-LOX in MCF-7 breast cancer cells resulted in increasedgrowth in nude mice and suppressed apoptosis (Connolly &Rose, 1998b). Treatment of W256 cells, a monocytoid cellline that arose spontaneously in the mammary gland of apregnant rat, with LOX, but not COX, inhibitors induced

apoptosis that was partially reversed by exogenous 12-

8/2/2019 1-s2.0-S0163725899000261-main

15/28


HETE or 15-HETE (Tang et al., 1996). With this back-

ground, it seems reasonable to postulate that dietary n-3FAs, which inhibit 12-HETE, 15-HETE, and PGE2 produc-tion by human breast cancer cells growing as solid tumorsin nude mice, may exert their suppressive effect on tumormass acquisition, at least in part, by activation of the apop-totic pathway (Rose & Connolly, 1997).

6.3.1.2. Angiogenesis Neovascularization, or angiogenesis,

is essential for solid tumor growth (Folkman, 1990), andalso provides the tumor cells with access to the vascular cir-culatory system, thus establishing the potential for meta-static disease progression. Angiogenic activity in primarybreast cancers has been associated with metastasis and apoor prognosis (Heimann et al., 1996; Fox et al., 1997) inbenign breast lesions with a propensity for progression tomalignant disease (Guinebretire et al., 1994; Fregene et al.,1994) and in breast ductal carcinoma in situ with later inva-

siveness (Heffelfinger et al., 1996).

The process of neovascularization is promoted by a varietyof angiogenic factors, which include both PGs (Form & Auer-bach, 1983; Spisni et al., 1992) and 12-HETE (Tang et al.,1995). We have shown that transfection of the MCF-7 humanbreast cancer cell line so that it stably overexpresses 12-LOXand secretes high levels of 12-HETE, causes rapid growth withnot only reduced apoptosis, but also high angiogenic activity(Connolly & Rose, 1998b). Again, dietary n-3 FAs may prove

to be antiangiogenic because of their suppressive effects oneicosanoid biosynthesis (Connolly et al., 1997b).

6.3.1.3. Estrogen metabolism The natural estrogen 17-estradiol, in combination with other steroid and polypeptide

hormones and growth factors, stimulates not only normalmammary development, but also neoplastic transformation.A series of studies has shown that a shift in estradiol metab-olism in favor of 16-hydroxylation and away from the for-mation of catechol derivatives produces aberrant hyperpro-liferation in mammary explant experiments, and clinicalstudies suggest that an elevated C16-hydroxylation of estra-diol may provide a biomarker for breast cancer risk (Telanget al., 1997).

In a preliminary study, Osborne et al. (1988) found that

feeding an n-3 FA-rich fish oil supplement to women witheither a family history of breast cancer, breast atypical hy-

perplasia, or carcinoma in situ (Section 7.1.2) reduced theextent of 16-hydroxylation. This potentially important ob-servation should be pursued in any future trials ofn-3 FAsas breast cancer chemopreventive agents.

6.3.1.4. Mevalonate synthesis 3-Hydroxy-3-methylglutarylco-enzyme A reductase (EC 1.1.1.34) is the rate-limitingenzyme in the cholesterol biosynthetic pathway and cata-lyzes the synthesis of mevalonate. Mevalonate exerts anumber of effects that signal increased cell proliferation,and compounds such as dietary isoprenoids, which inhibitmevalonate synthesis, are inhibitors of tumor cell growth

(Elson, 1995).

El-Sohemy & Archer (1997) extended these observations

to n-3 FAs, which were known to reduce hepatic 3-hydroxy-3-methylglutaryl co-enzyme A reductase activity, and foundthat both immunoreactive enzyme expression and meva-lonate production were suppressed in the mammary glandsof menhaden oil-fed female rats. Thus, the down-regulationof mevalonate synthesis in preneoplastic mammary epithe-lial cells provides another plausible mechanism that maycontribute to the chemopreventive activity ofn-3 FAs.

6.3.1.5. Lipid peroxidation The role of lipid peroxidation inthe suppressive effects ofn-3 FA on mammary tumorigene-sis was reviewed by Welsch (1995), and in this section, wehave already noted that the lytic effects of polyunsaturatedFAs on cultured tumor cells correlated with the degree oflipid peroxidation product formation. Gonzalez et al. (1991)found that when diets containing 19% menhaden oil were

fed to mice, there was both an inhibition of growth of hu-man breast cancer cell xenografts and an accumulation of

secondary products of lipid peroxidation in these tumors.Moreover, the inhibitory effects of the fish oil was reversedwhen antioxidants were added to the diet in sufficientamounts to reduce lipid peroxidation.

The mechanism(s) by which these degradative productsofn-3 FAs inhibit breast cancer cell growth is uncertain, asis the relevance of this effect of very high fish oil intakes to

the clinical application ofn-3 FA for breast cancer chemo-prevention. We found that feeding DHA as the triglycerideat a level of 2% of diet retarded the growth of MDA-MB-435 human breast cancer cell solid tumors in nude mice, anddid so without increasing the tumor levels of secondarylipid peroxidation products to those that were necessary to

inhibit growth in experiments reported by Gonzalez et al.(1993) (J. M. Connolly, R. Mongru, & D. P. Rose, unpub-lished data).

6.3.2. Colon cancer

6.3.2.1. Cyclooxygenase-2 inhibition Sakaguchi et al. (1984)

showed that a high-fat, LA-rich diet enhanced colon car-cinogenesis in rats treated with the chemical colonic carcin-ogen azoxymethane, and that the tumors contained elevatedlevels of AA. The colonic mucosal levels of PGE2 and 6-keto-PGF1 are elevated during the initiation and postinitia-tion stages of carcinogenesis in this model (Kulkarni et al.,1992), and in human surgically excised colon cancers, thePGE2 concentrations are increased compared with the corre-sponding normal colonic mucosa (Rigas et al., 1993). In

Section 6.2.2, we referred to a study by Minoura et al.(1988), and here, partial suppression of rat colon carcino-genesis by EPA was associated with a reduction in the PGE2content of the tumors that did appear compared with thatpresent in normal mucosa.

Increased production of PG in colon cancers is associ-ated with the up-regulation of COX-2 (Eberhart et al., 1994;Kargman et al., 1995b), and selective COX-2 inhibitors areeffective in experimental colon cancer chemoprevention

(Reddy et al., 1996a; Yoshimi et al., 1997). Singh et al.

8/2/2019 1-s2.0-S0163725899000261-main

16/28


(1997) showed that there was a time-related increase in

COX-2 expression in rat colonic mucosa after azoxymethaneadministration, and particularly high levels were present inthe tumors that subsequently developed. This enzyme in-duction was enhanced further by feeding a high-fat, n-6 FA-rich diet, but was suppressed by a high-fat, 20% menhadenoil diet. These elevations in COX-2 expression correlatedwith the incidence and multiplicity of colon tumors, whilethe expression of COX-1 was unchanged.

Thus, it appears that the suppressive effects ofn-3 FAs

on colon carcinogenesis, including the reduced formationand progression of aberrant crypts described by Takahashiet al. (1993), are intimately related to the down-regulationof COX-2. In this context, it should be noted that a selectiveinhibitor of COX-2 suppressed intestinal polyposis in the

Apc716 knockout mouse model for familial adenomatouspolyposis (Oshima et al., 1996). However, while the currentemphasis is on COX-2 inhibitors as potential colon cancerchemopreventive agents, the involvement of LOX products,

the synthesis of which is also blocked by n-3 FA, deservesattention. Hussey & Tisdale (1994) showed that the growthin two murine colon adenocarcinoma cell lines was stimu-lated by both LA and AA in vitro, and that this effect couldbe blocked by a selective LOX inhibitor. In later experi-ments, these investigators obtained similar results in vivoand demonstrated the particular importance of 12-LOX in-hibition in achieving growth suppression (Hussey & Tis-dale, 1996).

6.3.2.2. Apoptosis Chang et al. (1997) assessed both cellproliferation and apoptosis in the colonic crypts of rats ex-

posed to azoxymethane and fed either a high-corn oil or ahigh-fish oil diet, and found that the tumor-inhibitory effectof dietary n-3 FA was associated with increased apoptoticactivity, but not with reduced cell proliferation.

The apoptotic pathway in colon cancer cells is regulatedby COX-2 expression. In one study, treatment of a humancolon cancer cell line that possessed a high constitutivelevel of COX-2 expression and PGE2 production with a se-lective COX-2 inhibitor increased apoptotic activity, whichwas reversed by exogenous PGE2 (Sheng et al., 1998).

6.3.2.3. Angiogenesis As in the case of breast cancer andsome other solid tumors, a high degree of angiogenic activ-

ity in colonic carcinomas has been associated with a poorprognosis (Takebayashi et al., 1996). Consistent with thisrelationship is the finding that microvessel density is higherin aggressively invasive colon cancers (Saeki et al., 1997).

The calcium-independent and -inducible nitric oxidesynthase-2 was found to be expressed at high level in ap-proximately 60% of human colon adenomas and 2025% of

colon carcinomas (Ambs et al., 1998), and both the induc-ible and endothelial constitutive nitric oxide synthases wereelevated in rat colon tumors induced by azoxymethane(Takahashi et al., 1997). While the role for nitric oxide intumor angiogenesis is currently under investigation (Garca-

Cardea & Folkman, 1998), it is noteworthy in the present

context that nitric oxide is known to activate COX-2 andregulate PG biosynthesis (Salvemini et al., 1993; Sautebinet al., 1995).

Given the relationships between both COX and LOX andangiogenesis referred to in Section 6.3.1.2, it seems verylikely that the inhibition of PG and HETE production in co-lon cancer cells by n-3 FAs provides an antiangiogenicmechanism for their chemopreventive activity.

6.3.2.4. Protein kinase C (EC 2.7.10) Under physiologicalcircumstances, protein kinase C (PKC) is activated by thecombined effects of phospholipids and diacylglycerol (Han-nun et al., 1986). The activation of PKC has been implicatedin the stimulation of colonic epithelial cell growth by unsat-

urated FAs. Both LA and AA, as well as the monounsat-urated oleic acid, stimulated [3H]thymidine incorporation,with concomitant PKC translocation from the soluble to theparticulate fraction of colonic cells, the hallmark of PKC ac-

tivation. When these FAs were instilled into the colon invivo; in both cases, AA was the most effective of the threeFAs (Craven & DeRubertis, 1988).

Total PKC was found to be reduced in human colon ade-nocarcinomas compared with the uninvolved mucosa

(Guillem et al., 1987) and in the histologically normal mu-cosa from colon cancer patients compared with colonic mu-cosa from patients without cancer (Sakanoue et al., 1991).Moreover, the proportion of the total PKC activity associ-ated with the particulate, membrane, fraction was greater inmucosa from the colon cancer patients (Sakanoue et al.,1991). Baum et al. (1990) found that colonic mucosal PKC

activity exhibited translocation to the membranes in frac-tions prepared from the colons of rats treated with the pro-

carcinogen 1,2-dimethylhydrazine and examined prior tothe appearance of microscopic neoplastic change. Overtime, the extent of this translocation increased and there wasa concomitant reduction in total PKC activity. In a secondstudy, this same group of investigators showed that thesechanges in PKC were also present in the tumors that subse-quently developed when they were accompanied by in-creases in 1,2-diacylglycerol (Wali et al., 1991).

Reddy et al. (1996b) found that the exposure of rats to

the direct-acting chemical colonic carcinogen azoxy-methane increased diacylglycerol kinase activity in the co-

lonic mucosa. This effect was enhanced by feeding a high-fat, n-6 FA-rich diet, but not by a high-fat, n-3 FA-rich diet.In contrast to the experience of others, total PKC activitywas higher in the colon tumors that developed comparedwith the corresponding mucosa. The reason for this contra-dictory result is unclear. Certainly, the reported down-regu-lation of total PKC activity in human and rat colon tumors

contrasts with the elevated total activity and changes inPKC isoform expression seen in breast cancer (Morse-Gaudio et al., 1998).

Chapkin et al. (1993) also examined the effect of dietaryfish oil on PKC activity and intracellular distribution, but

8/2/2019 1-s2.0-S0163725899000261-main

17/28


they found that the ratio of membrane to cytosolic PKC ac-

tivity in normal colonic mucosa increased in rats fed a dietcontaining 11.5% menhaden oil compared with the activityratio in animals fed a 15% corn oil, n-6 FA-rich diet. Also,there was a positive correlation between the activity ratio,the total membrane PKC activity. and colon cell prolifera-tion. Thus, in spite of the report by Craven & DeRubertis(1988) that LA and AA increased the membrane PKC activ-ity in colon cells in vivo and in vitro, n-3 FA appeared tohave the potential for reversing the PKC changes associated

with colon carcinogenesis.Aside from the total PKC activity, the colonic mucosa

exhibits a distinct pattern of isoform expression (Doi et al.,1994; Davidson et al., 1994), and altered PKC isoform sig-naling appears to be involved in colon cancer development.In the present context, the report by Jiang et al. (1997) is ofparticular interest. These investigators found that dietaryfish oil blocked the decrease in the steady-state levels of theatypical PKC isoforms, which is otherwise induced in the

rat colonic mucosa by azoxymethane. There is clearly aneed for further research on PKC activity, expression of thevarious isoforms in colon cancer, and the biological andmolecular significance of the changes resulting from dietaryFA modification.

6.3.2.5. Ornithine decarboxylase (EC 4.1.1.17) The poly-amines that are produced from the metabolism of ornithinehave a role in cell proliferation, and the intracellular content

of these compounds is highly regulated. The first and gen-erally the rate-limiting step in polyamine biosynthesis iscatalyzed by ornithine decarboxylase (ODC), an inducible

enzyme that is elevated during the promotion phase of multi-stage carcinogenesis, such as that which occurs in the colon.Increased ODC activity is present in premalignant adenomas,including adenomatous polyps, and in colorectal carcinomatissues, compared with the corresponding uninvolved colonicmucosa (Narisawa et al., 1989; Hixson et al., 1993).

Craven & DeRubertis (1988) found that in addition toPKC activation, there was an increase in ODC activity whencolonic epithelial cells were exposed to n-6 FA. Reddy &Sugie (1988) assayed ODC activity in their study of the in-fluence of dietary FA on azoxymethane-induced rat coloncarcinogenesis and observed an inverse relationship be-tween the enzyme and the level of menhaden oil in the diet.In a second study, Rao & Reddy (1993) demonstrated astimulatory effect of an n-6 FA-rich diet on ODC activity

and a relative suppression of the enzyme by dietary n-3 FAin colonic mucosa exposed to azoxymethane for 2 weeks.

6.3.2.6. Fecal bile acid and neutral sterol excretion Bothepidemiologic and animal studies have provided evidencethat colon cancer risk and the adverse influence of dietaryfat on risk are associated with increased levels of secondarybile acids within the colonic lumen (Reddy & Wynder,1973; Reddy et al., 1977).

Bartram et al. (1995b) performed a short-term experi-

ment to examine the effects of dietary fish oil on fecal bile

acid excretion in which the daily diet of healthy volunteerswas supplemented with either 11 g of fish oil (4.4 g n-3 FA)or 11 g of corn oil. After 4 weeks, there was no change inthe total bile acid excretion. The excretion of the secondarybile acid lithocholic acid was lower after n-3 FA administra-tion compared with that after dietary supplementation withthe n-6 FA-rich corn oil, although the difference was notstatistically significant. There was, however, a significantreduction in the excretion of the putative colon carcinogen

4-cholesten-3-one (Bartram et al., 1998).While these preliminary data need to be confirmed, they

are supported by animal experimentation. Reddy et al.(1996b) determined fecal bile acid excretion in their studiesof dietary fats and azoxymethane-induced rat colon carcino-genesis. They found that although the total bile acid excre-tion was unaffected by the type of FA in the diet, the fecalsecondary bile acid concentrations, including lithocholicacid, were reduced in rats fed a high-fat, n-3 FA-rich diet

compared with those in rats fed a high-fat, n-6 FA-rich diet.Moreover, the activity of cecal bacterial 7-dehydroxylase,which is involved in the transformation of primary to sec-ondary bile acids, was higher in the rats fed a high, 23.5%,corn oil-containing diet than that in animals fed a 20.5%menhaden oil and 3.0 % corn oil diet.

Various studies have related these effects of bile acidsand the associated fecal neutral sterols to colonic cell PKCand ODC activities, and to LOX products of AA metabo-

lism. Bile salts stimulate colonic epithelial cell growth, andthis potentially tumor-promoting effect is preceded by boththe translocation of PKC from the cytosolic to the mem-

brane fraction and increased ODC activity (Craven et al.,1987).Fitzer et al. (1987) showed that the unconjugated bile ac-

ids chenodeoxycholate and deoxycholate stimulate conven-tional (calcium and phospholipid-dependent) PKC activityin vitro when assayed in the presence of a high calcium

concentration. Without this ionic excess, inhibition occursbecause the bile acids interact with the Ca2 and the phos-phatidylserine. The investigators postulated that the tumor-promoting activity of bile acids in colon carcinogenesis maybe modified by the relative absorption of Ca2 and bile ac-ids by the colonic mucosa, with calcium exerting a chemo-preventive function and bile acids a promoting one.

The expression of several biological characteristics of tu-mor cells has been shown to involve the regulation of PKC

activity by the LOX-mediated product of AA metabolism12-HETE. These include tumor cell adhesion to endothelialcells (Liu et al., 1991), cytoskeletal rearrangement (Timar etal., 1993), and invasive capacity (Liu et al., 1994). In thepresent context, it is noteworthy that tumor cell 12-HETEproduction is inhibited by dietary n-3 FA (Rose et al.,1995), and that EPA and DHA (Connolly & Rose, 1993)and a pharmacological inhibitor of PKC activity (Connollyet al., 1994) inhibited the invasive capacity of a human

breast cancer cell line in an in vitro assay.

8/2/2019 1-s2.0-S0163725899000261-main

18/28


This association of dietary FA, ODC activity, 12-HETE,

bile salts, and colonic epithelial cell proliferation wasbrought together by DeRubertis et al. (1984). They showedthat when deoxycholate stimulated the release of AA fromrat colon in vivo, there was an accumulation of 12-HETE,which was followed by an elevation in colonic mucosalODC activity and increased [3H]thymidine incorporationinto DNA. Interestingly, treatment with indomethacin, aCOX inhibitor, increased 12-HETE production, presumablyby shifting AA metabolism into the LOX-mediated path-

way, and this resulted in a further increase in ODC activityand thymidine incorporation.

7. Omega-3 fatty acids and cancer chemoprevention

7.1. Breast cancer

A reduction in the breast cancer mortality rate is now evi-dent in the United States, which is most likely attributable to

the beneficial effect of mammographic screening on the clin-ical stage at diagnosis, together with the success of adjuvantchemohormonal therapy in the management of the postsurgi-cally resected breast cancer patient (Chu et al., 1996).

One consequence of routine breast cancer screening by

mammography is that abnormalities that are known to con-stitute risk factors for invasive carcinoma of the breast arebeing recognized with increasing frequency. Some of theseare direct radiological abnormalities, while others constitutehistopathological features that are observed in breast tissueobtained by biopsy, itself performed in response to a mam-mographic abnormality.

7.1.1. Mammographic densitiesWolfe (1976) originally defined four categories of mam-

mographic parenchymal patterns and breast cancer risk.Risk is lowest in the N1 category, the parenchyma beingcomposed largely of fat, and highest in the DY category,which is characterized by extensive nodular densities, so-

called mammographic dysplasia. Later, attention focusedon mammographic density per se, which, while approximat-ing to the Wolfe classification in significance, is a distinctentity. Here, the proportion of the breast volume

1-s2.0-S0163725899000261-main

Documents