Rationale: The incidence of cancers affected almost 13 million people and caused more than 7 million deaths worldwide in 2008. Incidence is expected to increase to 15 million in 2015 and death to more than 9 million due to demographic effects alone. However increased longevity is not the only explanation, e.g. in France the incidence of cancers in males increased by 35 % and in females by 43 % after controlling for the demographic effect(Reference Belot, Grosclaude and Bossard1) Thus it is generally acknowledged that changes in exposure to carcinogenic environment and in nutrition are factors of this evolution.
If changes in food patterns are more often associated with an increased incidence of cancers, as illustrated in migrant studies(Reference Nelson2), it happens that nutritional recommendations are followed by a decreased incidence(Reference Puska3). This underlines the search for beneficial nutrients. Several epidemiological studies have shown a risk reduction of some cancers associated with long chain omega3 fatty acids (ω3 LC-PUFA) or fish intake(Reference Gerber4), but the limited evidence or the absence of consistency required further investigations.
A systematic review of the epidemiological studies published since 2007 is undertaken here focusing on ω3 LC-PUFA either from dietary intake (but not considering fish) or from plasma or cellular markers. As in the FAO/WHO joint expert consultation(Reference Gerber4), the most common cancers, colorectal, prostate and breast cancers are covered, and a paragraph on other cancers has been added. Use of ω3 LC-PUFA as adjuvant therapy of cancers will also be considered.
Objectives
This update review focused on studies published not taken into account in the previous FAO/OMS expert consultation(Reference Gerber4) to determine whether there exists any progress in the evaluation of the causal relationship between dietary ω3 PUFA and cancers and whether it is possible to propose preventive and/or adjuvant therapeutic recommendations.
Methods
Types of studies and eligibility criteria
In the complex field of cancer and nutrition, taking into consideration all studies available (mosaic or portfolio approach(Reference Gerber5)) is necessary. All prospective and case-control observational studies published since the ones reported in the FAO/WHO joint expert consultation(Reference Gerber4) were considered. Intervention studies and randomised controlled trials (RCT) when they exist were also considered. In vivo or in vitro experimental studies are mentioned to provide for biological plausibility. All these studies must meet validity criteria, such as population and sample size, ascertainment of disease diagnosis, quality of exposure measurement, (questionnaire characteristics – interview, self-administered, number of items, food groups-, or relevancy of biological markers), quality of statistics (adjustment for confounding factors). Population and sample size, quality of exposure measurement, quality of statistics (adjustment for confounding factors) and specific remarks are shown in corresponding tables. Excluded invalid studies were referenced and their exclusion is justified.
When evaluating the level of evidence, a hierarchy among studies has been proposed to help to establish a causal inference(Reference Smit, Mozaffarian and Willett6).Top of the hierarchy is data from prospective studies, which then might be supported by intervention studies, when they exist. Case-control studies are judged by these authors to be in third position, followed by experimental studies. However, each study of the portfolio have to be weighted, and in this perspective, it is acknowledged that prospective studies have the highest weighting.
Finally, the World Cancer Research Fund/American Institute for Cancer Research in 2007 proposed criteria for grading evidence(7):
● Convincing (unlikely to be modified by further studies): evidence from more than one type of study and from at least two prospective cohort studies; no substantial unexplained heterogeneity within or between studies types or in different populations; valid studies (as defined above); dose response effect, not necessarily linear as long as the explanation is biologically plausible; strong experimental evidence (human or animal) that exposure to the factor can lead to the disease.
● Probable: the same as the points above except for the first one: evidence from at least two prospective cohort studies or at least five case-control studies;
● Limited-suggestive: not enough studies, or studies with methodological flaws, but show generally consistent direction of effect, in spite of some unexplained heterogeneity.
● Limited-no conclusion evidence so limited that no firm conclusion can be made.
● Substantial effect on risk unlikely: the same as convincing but with studies showing absence of effect.
When the level of evidence is judged convincing or probable, preventive recommendations should be made in the perspective of public health.
Information sources
MEDLINE, via PubMed©, was searched between 1–15 April, 2011 back to 2007 in order to update the FAO/OMS expertise(Reference Gerber4) with the following strategy for each considered cancer : omega 3 fatty acids, fish intake, fish oil. For the database LILACS, the same word with the all words strategy was used. The studies reporting on fish were only included if the relationship with cancer incidence was specifically expressed as omega3 fatty acids. Whenever possible, distinction is made between α-linolenic acid (18 : 3 n-3, ALA) and LC ω3 PUFA, and among them eicosapentaenoic acid (20 : 5 n-3, EPA) and docosahexaenoic acid (22 : 6 n-3, DHA)
Colorectal cancer
Food and nutrition play an important role in colorectal cancer development among the factors related to high income, industrialization and urbanization, hence prevention may be implemented. The studies covered in the expert consultation of FAO on fatty acids(Reference Gerber4) suggest a probable causal relationship between fish intake and CRC. However, evidence was too limited to draw any firm conclusion on the effect of LC ω3 PUFA.
Four case-control studies have been published: Kato et al(Reference Kato, Majumdar and L8) was excluded for insufficient characterisation of the omega 3 intake, the others(Reference Kimura, Kono and Toyomura9–Reference Kim, Sandler and Galanko11) are presented in Table 1, Table 2 describes the results of the recent prospective studies(Reference Weijenberg, Luchtenborg and de Goeij12–Reference Butler, Wang and Koh17
Table 1 Omega 3 fatty acids and colorectal cancer risk (incidence): case-control studies
OR (CI): estimated relative risk (confidence interval); ALA, alpha-linolenic acid; omega 3 LC-PUFA, omega3 long chain-polyunsaturated fatty acids; CC, colon cancer; RC, rectum cancer; M, males; F, females; FFQ, food frequency questionnaire; CRC, colorectal cancer; H, highest quantile; L, lowest quantile; EPA, eicosapentaenoic acid; DHA, docosahaxaenoic acid; T, trend; NIH, National Institute of Health.
Table 2 Omega 3 fatty acids and colorectal cancer risk (incidence): prospective studies
* RR (CI): relative risk (confidence interval); ALA, alpha-linolenic acid; omega 3 LC-PUFA, omega 3 long chain-polyunsaturated fatty acids; CC, colon cancer RC, rectum cancer; M, males; F, females; FFQ, food frequency questionnaire; CRC, colorectal cancer; H, highest quantile; L, lowest quantile; EPA, eicosapentaenoic acid; DHA, docosahaxaenoic acid; TFA: total fatty acids; T, trend; NS, non significant.
ALA
Two case-control studies(Reference Theodoratou, McNeill and Cetnarskyj10, Reference Kim, Sandler and Galanko11) out of 2, (Table 1) showed no effect, as did one European(Reference Weijenberg, Luchtenborg and de Goeij12), and one Japanese(Reference Sasazuki, Inoue and Iwasaki15) prospective cohorts, whereas another American one(Reference Daniel, McCullough and Patel16) showed a significant increase in risk (Table 2). However, when further adjustment was made for meat, the RR decreases, and the trend was no longer significant, alluding to the confounding effect of meat rich in ALA, because of the soya feed given to livestock. Thus, there is no firm conclusion, but a limited evidence for an absence of relationship.
LC ω3 PUFA
The 3 case-control studies(Reference Kimura, Kono and Toyomura9–Reference Kim, Sandler and Galanko11) showed a decreased risk associated with the highest quantile of LC-ω 3PUFA intake, in spite of strong difference in intake between the 3 studies (Table 1). Results of the 6 prospective studies are conflicting with 1 study(Reference Daniel, McCullough and Patel16) study reporting no effect, one(Reference Butler, Wang and Koh17 ) an increased risk, and 3(Reference Hall, Campos and Li13–Reference Sasazuki, Inoue and Iwasaki15)reporting a reduced risk. One of these 3 studies(Reference Hall, Campos and Li13), a subgroup of the Physician Health Study(Reference Hall, Chavarro and Lee14) based on biological markers, reported a reduced risk, only in subjects not taking aspirin. The Japanese study(Reference Sasazuki, Inoue and Iwasaki15) reporting a reduced risk, analysed a large number of sub-groups, giving rise to different results between men and women, stage and sites of the disease, which might lead spurious findings, in spite of the quality of the study. A subsequent study(Reference Stern, Butler and Corral18) of the Chinese population of Singapore(Reference Butler, Wang and Koh17), showing an increased risk associated with a high intake of LC ω3 PUFA, reported that the positive association between high intake of marine n-3 PUFA and rectal cancer risk was observed among carriers of at least one PARP codon 762 Ala allele (OR: 1·7, CI: 1·1–2·7) without indicating whether this SNP is frequent in Singapore Chinese. The role of eventual chemical contaminants might be also evoked to explain this increased risk(Reference Hoyer, Grandjean and Jørgensen19)
Compared to the complex colorectal cancer picture, with various subsites and stages, and avoiding the measurement error of questionnaire, an endoscopy-based case-control study on colorectal adenomas with serum level measurement of fatty acids represents a simpler situation to apprehend(Reference Pot, Geelen and van Heijningen20). In such a study, total serum n-3PUFA levels were inversely associated with colorectal adenoma risk, the OR comparing the third tertile with the first tertile was 0·67 (0·46–0·96), trend: 0·03.
Given that results are rather consistent when fish intake is considered(Reference Gerber4), one could expect comparable observations, serum LC-ω 3PUFA being highly correlated with fish intake(Reference Hall, Chavarro and Lee14, Reference Gerber, Scali and Michaud21). Several experimental studies support the hypothesis of CRC risk reduction by LC ω3 PUFA(Reference Moreira, Sabarense and Dias22–Reference Rogers, Kikawa and Mouradian25) through different mechanisms In addition, the hypothesis of the anti-inflammatory effect of LC-ω3PUFA and especially EPA(Reference Galli and Calder26) is supported by the subgroup study nested in the PHS, based on biomarkers(Reference Hall, Campos and Li13), showing a significant interaction with the absence of aspirin treatment (Table 2). However, epidemiological studies are as yet inconsistent, revealing the difficulty of studies focusing on one specific type of nutrient within the context of a multifactorial disease and subjected to confounding and imprecision of exposure measurement.
Thus, it can be said that limited evidence, supported mainly by case-control studies, experimental studies and biological plausibility, suggests a possible relationship between LC-ω 3PUFA and CRC.
Prostate cancer
Prostate cancer (PC) is the 2nd most common cancer in men, accounting for around 12 % of all new cases of cancers in the world. Screening for prostate specific antigen (PSA) is in part responsible for its high incidence in high-income countries. As for CRC, the expert consultation of FAO on fatty acids (4) recognised heterogeneity in the considered studies and could not conclude on the existence of a relationship.
Prostate cancer is generally recognized as a hormone-dependent cancer, and to vary with ethnic characteristics, a higher incidence being observed in Afro-American men. However, comparative studies of African Americans in Washington, D.C. and Nigerians in Ibadan demonstrated similar incidence of latent prostate cancer, although the African Americans recorded a 10-fold higher incidence of clinical prostate cancer. Differences observed between healthy indigenous Africans and African Americans in their levels of estrogen and androgen metabolites and urinary steroids were reported to depend on their respective diets and could explain their disparate PC rates as suggested by a comparative study conducted in these populations (48 cases, 96 population based controls Afro-Americans; 66 cases, 266 population based controls Nigerians)(Reference Ukoli, Fowke and Akumabor27). This study was designed as a case-control study based on fatty acids biomarkers showed that total omega 3 fatty acids were significantly higher in Nigerian than in Afro-Americans (p < 0·01). However estimation of the relative risk suffered from the low power of the study because of the small sample (12 to 15 cases by quartile) and could not be retained in the evaluation of the association of PC with omega 3 fatty acids. Another case-control study(Reference Williams, Whitley and Hoyo28) with a small sample (79 cases, 12 to 19 cases by tertiles of the subgroups) cannot be interpreted, with findings bearing only on the ratio ω6/ω3. Retained studies are displayed on Tables 3 and 4.
Table 3 Omega 3 fatty acids and prostate cancer risk (incidence): case-control studies
OR (CI): estimated relative risk (confidence interval); ALA, alpha-linolenic acid; omega 3 LC-PUFA, omega 3 long chain-polyunsaturated fatty acids; FFQ, food frequency questionnaire; H, highest quantile; L, lowest quantile; EPA, eicosapentaenoic acid; DPA, docosapentaenoic acid; T, trend; DHA, docosahaxaenoic acid; PSA, prostate specific antigen; TFA, total fatty acids; NS, non significant.
Table 4 Omega 3 fatty acids and prostate cancer risk (incidence): cohort studies
* RR (CI): relative risk (confidence interval); ALA, alpha-linolenic acid; omega 3 LC-PUFA, omega3 long chain-polyunsaturated fatty acids; FFQ, food frequency questionnaire; CRC, colorectal cancer; H, highest quantile L, lowest quantile; EPA, eicosapentaenoic acid; DHA, docosahaxaenoic acid; PHS, Physician Health Study; TFA, total fatty acids; T: trend; NS, non significant.
ALA
Two case-control studies(Reference Fradet, Cheng, Casey and Witte29, Reference Shannon, O'Malley and Mori30) in Table 3 and 4 prospective ones in Table 4(Reference Giovannucci, Liu and Platz31–Reference Chavarro, Stampfer and Li34reported conflicting results: the 2 case-control studies(Reference Fradet, Cheng, Casey and Witte29, Reference Shannon, O'Malley and Mori30) and one(Reference Wallstrom, Bjartell and Gullberg33) among the 4 prospective cohorts showed no association of PC with ALA. One prospective study(Reference Park, Murphy and Wilkens32) showed a reduced risk and 2(Reference Fradet, Cheng, Casey and Witte29, Reference Chavarro, Stampfer and Li34) an increased PC risk related to ALA intake. A pooled analysis(Reference Carayol, Grosclaude and Delpierre35) concluded that the absence of relationship between ALA and PC was very likely, with a limited probability of a weak positive effect. There are neither biological plausibility nor experimental models supporting a deleterious effect of ALA. Therefore, it is likely that the risk associated to high intake in some studies results from residual confounding. Cattle being fed with soy, rich in ALA, could be responsible for the residual confounding, this fatty acid could be confounded by the limited but suggestive risk associated with processed meat, milk and dairy products.
LC ω3 PUFA
One(Reference Shannon, O'Malley and Mori30) of the 2 case-control studies(Reference Fradet, Cheng, Casey and Witte29, Reference Shannon, O'Malley and Mori30) and one(Reference Park, Murphy and Wilkens32) of the 3 prospective studies(Reference Park, Murphy and Wilkens32–Reference Chavarro, Stampfer and Li34) showed no association with the sum of LC-ω 3PUFA, the other case-control study(Reference Fradet, Cheng, Casey and Witte29) and one prospective study(Reference Chavarro, Stampfer and Li34), showed a risk reduction of PC for the highest quantile of LC-ω3PUFA and of each of LC-ω 3PUFA intake. The results of the case-control study(Reference Fradet, Cheng, Casey and Witte29) showed an interaction between the LC-ω3PUFA and each LC-ω 3PUFA, and a variant of the COX-2, modifying the effect of the intake. The results of the prospective study are of interest because the exposure measurement is assessed by biomarkers, avoiding measurement error, and results are consistent in spite of the many analyses conducted. One prospective(Reference Wallstrom, Bjartell and Gullberg33) study showed an increased risk for a very high intake (median of the quartile 1·3 g/day) of LC ω3 PUFA. This result might be considered in the light of possible fish environmental contaminants(Reference Hoyer, Grandjean and Jørgensen19).
The biological plausibility from experimental studies(Reference Rogers, Kikawa and Mouradian25, Reference Karmali, Reichel and Cohen36), especially involving the anti-inflammatory effects of PUFAs, possibly through mediation of cyclooxygenase (COX), a key enzyme in fatty acid metabolism and inflammation(Reference Galli and Calder26), is supported by the case-control study(Reference Fradet, Cheng, Casey and Witte29) showing that the subjects carrying a mutation of the gene COX 2 needed a higher intake of EPA+DHA to be protected than the subjects carrying the most common gene.
Thus, epidemiological studies provide inconsistent results suggesting an inverse association of LC ω3 PUFA. However, the heterogeneity could be explained by measurement errors, insufficient food composition tables, high intake of fish possibly contaminated by endocrine disruptors, statistical difficulty at disentangling specific fatty acids, which are correlated in the intake and expressed as % when measured in blood.
Breast cancer
Breast cancer (BC) is the most common cancer in women worldwide. Hormone metabolism is the preponderant influential factor for BC: high estradiol, either from endogenous (high synthesis and/or altered regulation of binding proteins) or exogenous sources is the major risk factor. Most of the lifestyle factors modify BC risk through their effect on hormone metabolism as well as others factors related to industrialization and urbanization. These factors encompass societal changes such as women's sexual liberation and entrance in the working world, resulting in late age at births, low parity and low frequency of lactation; they encompass also, possible environmental factors such as endocrine disruptors. Altogether, these lifestyle, societal, and possible environmental factors participate in the high incidence of BC in high-income countries, and in the rapidly increasing incidence in emergent and low-income countries. Nutrition is only one amongst these numerous factors, but also one that is the easiest to modify. So far, except for energy imbalance and development of obesity, there is not strong evidence for nutritional recommendation with regard to BC. Four case-control studies are reported in Table 5 and 6 prspective studies in Table 6.
Table 5 Omega 3 fatty acids and breast cancer risk (incidence): case-control studies
RR (CI): relative risk (confidence interval); ALA, alpha-linolenic acid; omega 3 LC-PUFA, omega3 long chain-polyunsaturated fatty acids; EPA, eicosapentaenoic acid; DHA, docosahaxaenoic acid; FFQ, food frequency questionnaire; H, highest quantile; L, lowest quantile; TFA: total fatty acids; Mnp, menopausal; PFC. Proliferative cystic disase.
Table 6 Omega 3 fatty acids and breast cancer risk (incidence): cohort studies
RR (CI): relative risk (confidence interval); ALA, alpha-linolenic acid; omega 3 LC-PUFA, omega3 long chain-polyunsaturated fatty acids;; FFQ, food frequency questionnaire; H, highest quantile; L, lowest quantile; EPA, eicosapentaenoic acid; DHA, docosahaxaenoic acid; TEI, total energy intake; BC, breast cancer; TFA, total fatty acids; Mnp, menopausal.
ALA
There is no association between ALA and BC in Asian countries(Reference Kuriki, Hirose and Wakai37, Reference Shannon, King and Moshofsky41, Reference Murff, Shu and Li44) It is found in the French study(Reference Thiébaut, Chajès and Gerber40) that BC risk is reduced when the source of ALA is vegetable oil, but increased when the source of ALA is processed food (Table 6). This suggests, as in other cancers, that the relationship of ALA with cancers is likely to be confounded by some components in foods. These foods could be part of a Western dietary pattern and not of Asian dietary pattern. Hence, ALA per se is probably not associated with breast cancer.
LC ω3 PUFA
Results are rather consistent both from case controls and from prospective cohorts studies from Asian countries showing a risk reduction associated with intake of LC ω3 PUFA(Reference Kuriki, Hirose and Wakai37–Reference Shannon, King and Lampe39, Reference Shannon, King and Moshofsky41) with one exception(Reference Murff, Shu and Li44). There is also a risk reduction in a study conducted in USA for the fish oil current users(Reference Brasky, Lampe and Potter43). There is no association observed in European countries(Reference Thiébaut, Chajès and Gerber40, Reference Witt, Christensen and Schmidt42) (Tables 5 and 6). The heterogeneity could be explained by a difference in exposure. The median intake reported in the negative Chinese study(Reference Murff, Shu and Li44) is 200 mg/day whereas it is from 300 mg up to more than 500 mg in other Asian studies. The North-American(Reference Brasky, Lampe and Potter43) study showing an association is based on the effect of supplementation with fish oil. Considered together, these studies indicate the possible necessity of a high intake of LC-ω3PUFA to show an association.
These observations might be reinforced by 2 studies on BC and fibrocystic disease(Reference Shannon, King and Lampe39, Reference Dijkstra, Lampe and Ray45) and one on BC prognosis(Reference Patterson, Flatt and Newman46) The first one was a case-control study that determined erythrocyte fatty acid concentrations in 155 women with non proliferative fibrocystic disease (NPFC), 185 women with proliferative fibrocystic disease (PFC), 241 women with BC, and 1,030 control subjects. The results related to BC are reported in Table 5(Reference Shannon, King and Lampe39) and showed that EPA reduced the risk of progressing from PFC to BC. The study showed also that EPA reduced the risk of NPFC (0·33 CI: 0·18–0·61, T: 0·0001). In a more recent study(Reference Dijkstra, Lampe and Ray45), the same authors observed an inverse association between fibroadenoma and higher percentages of the RBC EPA and DHA (0·38 CI 0·19–0·77, T: 0·007 and 0·32 CI 0·15–0·70 T: 0·024, respectively) in a case (248)-control (1035) study on fibroadenoma risk. Association between dietary intake of EPA and DHA from food and supplements, and disease-free survival and overall survival was examined(Reference Patterson, Flatt and Newman46) Women with higher intakes of EPA and DHA from food had an approximate 25 % reduced risk of additional breast cancer events: (tertile 3: HR: 0·72, 0·57–0·90) compared with the lowest tertile of intake. Women with higher intakes of EPA and DHA from food had a dose-dependent reduced risk of all-cause mortality: tertile 3: HR = 0·59 (95 % CI = 0·43–0·82). EPA and DHA intake from fish oil supplements was not associated with breast cancer outcomes. Thus, EPA and DHA but also another nutrient in fish, appeared to be associated with reduced risk of additional breast cancer events and all-cause mortality. Experimental studies(Reference Giros, Grzybowski and Sohn24, Reference Altenburg and Siddiqui47) support the biological plausibility of this association. Some of the experimental studies tend to allot the most important role to DHA, through regulation of gene transcription(Reference Rogers, Kikawa and Mouradian25, Reference Sun, Berquin and Owens48).
Thus, there is increasing evidence suggesting an association between BC and LC ω3 PUFA, with a possible explanation of the heterogeneity by the amount of the intake and the dietary pattern context. These more recent results seem to confirm an earlier meta-analysis(Reference Saadatian-Elahi, Norat and Goudable49) showing an association between LC-ω3PUFA and BC (RR: 0·61, CI 0·40–0·93) for all women in the considered cohorts, and more especially for post memopausal women (RR: 0·58, CI 0·52–0·64).
Other cancers
There are far fewer studies on other cancers and omega 3 fat acids. One study(Reference Zuijdgeest-van Leeuwen and van der Heijden50) compared the level of ALA and LC-ω 3PUFA in plasma phospholipids and cholesteryl esters in 71 newly diagnosed, untreated cancer patients of three tumour types: oesophageal or cardia cancer (n 35), non-small cell lung cancer (n 22) and pancreatic cancer (n 15) and in 45 healthy subjects. Only patients with pancreatic cancer presented significantly lower levels of EPA and DPA compared to healthy subjects and to other cancers. In a systematic review(Reference MacLean, Newberry and Mojica51), it was concluded that there were no significant associations between omega-3 fatty acid consumption and cancer incidence for upper respiratory-digestive cancers, bladder cancer, lymphoma, ovarian cancer, pancreatic cancer, or stomach cancer. Not enough studies have been undertaken to modify this conclusion.
A study(Reference MacLean, Newberry and Mojica51) (532 cases and 1701 population-based controls) was conducted in the USA and showed an increased risk for pancreatic cancer (OR: 1·5, CI: 1·1–2·0, T: 0·02) for an ALA intake ≥ 1·4 g/day compared to 0·850 g/day. The consumption of LC-ω3PUFA being very low in this population, the authors computed tertiles in the highest quartile. In this group of high consumers (representing 90th and 95th percentile, they showed that a LC-ω3PUFA consumption ≥ 0·850 g/day was associated with decreased risk (OR: 0·47 CI: 0·25–0·90), compared to an intake < 0·120 g/day. Experimental studies provide support and mechanistic hypotheses(Reference Park, Lim and Kim53, Reference Strouch, Ding and Salabat54). However data are as yet insufficient to draw firm conclusions.
For gastric cancer, a Japanese case-control study(Reference Kuriki, Wakai and Matsuo55) (179 cases and 532 hospital-based controls) showed also a reduced risk (OR: 0·39, CI: 0·23-0·68, T: < 0·005) associated with LC-ω 3PUFA consumption (>7·98 % vs < 5·61 % of fatty acids in erythrocytes membranes.
Relationship to the incidence of lung cancer was not recently investigated. A rather old prospective cohort study(Reference Veierød, Laake and Thelle56) in Norway (153 cases/25,956 men and 25,496 women aged 16–56 years) showed an inverse association with intake of cod liver oil supplement (RR: 0·5, CI: 0·3–1·0).
Indeed, because of the anti-inflammatory, apoptotic and oxidative effects of LC-ω3PUFA shown in cell cultures and in animal models(Reference Galli and Calder26, Reference Sun, Berquin and Owens48, Reference Lu, Nie and Witt57–Reference Lindskog, Gleissman and Ponthan59) more studies were conducted exploring the possible use of LC-ω3PUFA as adjuvant therapy.
LC-ω3PUFA as adjuvant in anti-cancer therapy
Several experimental in vitro and in vivo studies, demonstrated that omega 3 fatty acids sensitize tumour cells to effects of anticancer drugs in culture or in tumor-bearing animals(Reference Germain, Chajes and Cognault60–Reference Cha, Lin and Meckling62). Because of these first observations in animal models, a phase II trial(Reference Bougnoux, Hajjaji and Ferrasson63) was undertaken on 25 patients with metastatic breast cancer, treated 3 times/day with 600 mg of DHA from algae origin, 7 to 10 days as a loading period before chemotherapy and during the 5 months of chemotherapy. As for a phase II protocol, there was no control group, but it was observed that patients DHA plasma levels presented a Gaussian distribution, reflecting a different ability to incorporate DHA. When stratifying the patients on the median of plasma levels (2·5 %), it could be observed that the overall survival was significantly greater in the patients group showing plasma level >2·5 % with a median survival time of 34 months vs 18 months in the patients showing plasma level < 2·5 % group (p = 0·007). This is an indication of the beneficial effect of DHA on chemotherapy treatment, however it has to be confirmed in a randomised controlled trial.
With regard to EPA, the Cochrane review published in 2007, covering the randomised controlled trials up to February 2005 concluded that there were ‘insufficient data to establish whether oral EPA was better than placebo. Comparisons of EPA combined with a protein energy supplementation versus a protein energy supplementation (without EPA) in the presence of an appetite stimulant provided no evidence that EPA improves symptoms associated with the cachexia syndrome often seen in patients with advanced cancer’(Reference Dewey, Baughan and Dean64).
A multicenter double-blind, randomised placebo controlled trial was conducted on 518 weight-losing patients with advanced gastrointestinal or lung cancer(Reference Fearon, Barber and Moses65). Patients received a novel preparation of pure EPA at a dose of 2 g or 4 g daily or placebo (2 g EPA, n 175; 4 g EPA, n 172; placebo, n 171). Patients were assessed at 4 weeks and 8 weeks. The best results were obtained in the 2 g EPA group, with a borderline significance (p = 0·066) for weight gain at 8 weeks (mean weight gain 1·2 kg, CI, 0·0 kg to 2·3 kg) compared with placebo. Physical function improved by approximately 7 % compared with placebo in those receiving 2 g EPA (p: 0·04) and fell by around 5 % in those receiving 4 g EPA. Thus, there was no evidence of a dose response beyond 2 g per day, and if anything a suggestion of either a plateau or at worst a degree of deterioration with 4 g per day. However, in this study, EPA was given alone and not in combination with oral nutritional supplements.
The effects of an oral nutritional supplement containing omega 3 fatty acids on nutritional status and inflammatory markers were investigated in 40 patients with non-small cell lung cancer (NSCLC) stage III undergoing multimodality treatment in a double-blind, randomised, placebo controlled trial(Reference van der Meij, Langius and Smit66). The intervention patients showed a higher energy intake and a better weight maintenance than the control group (intention to treat basis, p: 0·02 at 4 weeks, and p: 0·04 at 8 weeks).
Other recent studies by a Canadian group aimed at analysing the effect of EPA on patients with non small cell lung cancer (NSCLC), especially with regard to the muscle mass. They first reported on the use of Computed tomography (CT) images to measure muscle mass and illustrate the relationship of muscle mass amount with serum fatty acids(Reference Murphy, Mourtzakis and Chu67). Initially they followed about 600 solid lung tumours with longitudinal CT. In a first subset of 41 patients with lung cancer receiving chemotherapy, 25 were sarcopenic. Omega3 fatty acids were the only fatty acids to be different between the sarcopenic and the non sarcopenic, and patients with the maximal muscle loss presented the lowest concentration of omega3 fatty acids (p = 0·005). An open-label study with a contemporaneous control group was reported later(Reference Murphy, Mourtzakis and Chu68): 40 patients who were receiving first-line chemotherapy (platinum-based doublet chemotherapy with either curative or palliative intent) consented to participate in a nutritional intervention study: 14 patients received 2·2 g EPA (I) and 16 the standard of care regimen (C); a reference group (n 104) was established to ensure the representativity of the I and C groups. The primary endpoint was change in muscle mass between baseline and the end of chemotherapy. Adipose tissue, body weight, and plasma EPA at baseline and at the end of chemotherapy were secondary endpoints. Patients in the C group experienced an average weight loss of 2·3 ± 0·9 kg whereas I patients maintained their weight (0·5 ± 1·0 kg) (p: 0·05). Patients with the greatest increase in plasma EPA concentration after fish supplementation were found to have the greatest gains in muscle mass (r 2: 0·55, p: 0·01). Approximately 69 % of I patients gained or maintained muscle mass vs 29 % of C patients who, overall, lost 1 kg of muscle. Another subset of patients with a clinical diagnosis of stage IIIB or IV NSCLC, who were receiving first-line chemotherapy (platinum-based doublet chemotherapy with palliative intent) was enrolled in a study(Reference Murphy, Mourtzakis and Chu69) designed as the one described above(Reference Murphy, Mourtzakis and Chu68) The primary endpoint was chemotherapy response rates. Clinical benefit, chemotherapy toxicity, and survival were secondary endpoints. Sixty % of the I group had a increased response rate to chemotherapy vs 26 % of the C group (p = 0·008) and a greater clinical benefit (80 % vs 42 %, p = 0·02). Toxicity did no differ, and one-year survival tend to be more frequent (60·0 % vs 38·7 %; p = 0·15).
Thus, limited evidence suggests that supplementaion of fish oil of patients with NSCLC is beneficial.
General conclusion
The recent studies reported here did not increase the consistency of the results and do not permit to draw firm conclusions, except for ALA, which, probably, is neither a risk factor nor a beneficial factor with regards to cancers.
Thus, these new studies do not permit to go much further than the conclusion of the FAO/OMS(Reference Gerber4): observational studies only provided limited evidence on the possible role of LC-ω3PUFA for colon cancer prevention. The same level of heterogeneity is observed for prostate cancer. The evidence is somewhat stronger for breast cancer when the exposure is as high as in Asian countries.
An interesting point is that 2 of these epidemiological studies(Reference Hall, Campos and Li13, Reference Fradet, Cheng, Casey and Witte29) brought about data in agreement with the mechanistic hypotheses developed by experimental data(Reference Galli and Calder26, Reference Altenburg and Siddiqui47, Reference Park, Lim and Kim53), thereby increasing the biologic plausibility of the beneficial anti-inflammatory effect of LC-ω3PUFA on cancers.
Another point is the evocation of explanations for heterogeneity: In addition to the inherent difficulties associated with epidemiology (measurement error, relevance of biomarkers, genetic polymorphism, cancer stages), the review of these recent the studies calls the attention on confounding: confounding with nutriments of other foods (essentially meat products for colorectal(Reference Daniel, McCullough and Patel16), or processed food for breast cancer(Reference Thiébaut, Chajès and Gerber40)). The same can be hypothesised in the case of meat and dairy products for prostate cancers. Beyond a nutriment or a food, a favourable dietary pattern might confound the relationship between LC-ω3PUFA and cancers as suggested by the frequent homogeneity in favourable results of Asian studies(Reference Kimura, Kono and Toyomura9, Reference Kim, Sandler and Galanko11, Reference Sasazuki, Inoue and Iwasaki15, Reference Kuriki, Hirose and Wakai37–Reference Shannon, King and Lampe39, Reference Shannon, King and Moshofsky41).
Thus, nutritional recommendation will focus on fish consumption, or even better on a healthy dietary pattern, traditional Asian or Mediterranean.
With regard to cancer cachexia, inflammation and accompanying cytokines appear to be at the heart of the situation. Knowing the anti-inflammatory activity of LC-ω3PUFA, their role as adjuvant, in view of the latest RCT on lung cancers, might be considered as of possible use, even if other RCT on other cancers still need to be undertaken, especially in breast cancers with regard to the possible beneficial effect on the chemotherapy outcome(Reference Bougnoux, Hajjaji and Ferrasson63).
Acknowledgements
I thank my colleagues from the FAO/WHO expert consultation, Dariush Mozaffarian, Tom Sanders and Murray Skeaff for fruitful discussion on the topic. There are neither conflicts of interest nor funding.
