Professional Documents
Culture Documents
A Meta-Analysis of N-3 Polyunsaturated Fatty Acids Effects On
A Meta-Analysis of N-3 Polyunsaturated Fatty Acids Effects On
Clinical Nutrition
journal homepage: http://www.elsevier.com/locate/clnu
Meta-analyses
a r t i c l e i n f o s u m m a r y
Article history: Background & aims: Chronic inflammation is related with cancer and leads to worsening prognosis in
Received 15 November 2016 cancer patients. n-3 polyunsaturated fatty acids (PUFAs) supplementation has been proposed as adjuvant
Accepted 8 May 2017 treatment in cancer due anti-inflammatory properties. In the present meta-analysis, we pooled ran-
domized clinical trials (RCTs) assessing the effects of n-3 PUFAs (from fish oil isolated or added in an
Keywords: immunonutrition formula) on inflammatory markers in gastric cancer.
Inflammation
Methods: A comprehensive literature search was performed in Medline, Scopus, Cochrane library, Sci-
Eicosapentaenoic acid
ence Direct and Web of Science, besides GOOGLE Scholar and a hand searching of reference lists, through
Docosahexaenoic acid
Stomach neoplasm
July 2016. We pooled the effect size from individual studies using a random-effect model and carried out
heterogeneity and sensitivity analyses.
Results: Nine trials (698 patients) fulfilled the entry criteria and were included in the synthesis of the
systematic review. Eight were carried out in surgical patients and one in patients that received
chemotherapy. Four used only fish oil as intervention and five used an immunonutrition formula. Global
meta-analysis demonstrated higher albumin (7 studies, SMD 0.28; 95% CI 0.07, 0.48) and prealbumin
(4 studies, SMD 0.56; 95% CI 0.12, 1.00) concentrations, and lower IL-6 (2 studies, SMD 0.71; 95%
CI 1.15, 0.27) and TNF-a (2 studies, SMD 0.92; 95% CI 1.58, 0.26) concentrations in patients of the
intervention group as compared to control group. However, total protein, transferrin and CRP concen-
trations were not improved by n-3 PUFAs supplementation.
Conclusion: This study provides evidence that n-3 PUFAs supplementation from fish oil or added an
immunonutrition formula has favorable effects on inflammatory markers in gastric cancer patients
undergoing surgical procedures.
© 2017 Elsevier Ltd and European Society for Clinical Nutrition and Metabolism. All rights reserved.
1. Introduction Cytokines such as IL-6, TNF-a and IL-1b can serve as growth and
survival factors that act on malignant cells [3e5]; induce an acute-
The association between cancer and inflammation is recognized phase protein response by hepatocytes and changes in metabolism
more than a century ago and well discussed in scientific literature. [6]; stimulate angiogenesis, tumour progression and metastasis
In gastric cancer the chronic inflammation has a key role in carci- [4,5]; cause chemoresistance and increase the toxicity to anti-
nogenesis, and most tumours, if not all, secrete molecules that cancer drug or radiation therapy [5]; contribute with cachexia-
activate inflammatory cells promoting an inflammatory status [1,2]. related to cancer [6]; and also, maintain tumour-promoting
inflammation in a kind of vicious-cycle [4].
The n-3 polyunsaturated fatty acids (PUFAs), especially eicosa-
Abbreviations: CI, Confidence Interval; CRP, C-Reactive Protein; DHA, Docosa-
pentaenoic (EPA) and docosahexaenoic (DHA), have been linked to
hexaenoic acid; EPA, Eicosapentaenoic acid; GPR, G-Protein Receptor; IFN, Inter-
feron; IL, Interleukin; NF-kB, Nuclear factor kappa B; PPAR, Peroxisome Proliferator-
numerous beneficial effects on cancer such as anti-inflammatory,
Activated Receptor; PUFA, polyunsaturated fatty acids; RBP, Retinol Binding Protein; pro-apoptotic, antineoplastic and anti-catabolic properties [7e10].
RCT, Randomized Clinical Trial; SD, Standard Deviation; SMD, Standardized Mean Regarding their influences on inflammation related to cancer, there
Difference; TNF, Tumor Necrosis Factor. is a previous meta-analysis showing a reducing action of n-3 PUFAs
* Corresponding author. Graduate Program of Nutrition, Nutrition department,
on several inflammatory markers in colorectal cancer [11]. How-
Health Science Center, University Campus e Trindade, Federal University of Santa
Catarina, Florianopolis, Santa Catarina, 88040-900, Brazil. Fax: þ55 48 3721 9542. ever, in gastric cancer, there are many published clinical trials
E-mail address: michel.mocellin@hotmail.com (M.C. Mocellin). assessing the effects of n-3 PUFAs on inflammatory response with
http://dx.doi.org/10.1016/j.clnu.2017.05.008
0261-5614/© 2017 Elsevier Ltd and European Society for Clinical Nutrition and Metabolism. All rights reserved.
Please cite this article in press as: Mocellin MC, et al., A meta-analysis of n-3 polyunsaturated fatty acids effects on circulating acute-phase
protein and cytokines in gastric cancer, Clinical Nutrition (2017), http://dx.doi.org/10.1016/j.clnu.2017.05.008
2 M.C. Mocellin et al. / Clinical Nutrition xxx (2017) 1e11
discrepant results. Therefore, the aim of the present study was to by patients diagnosed with gastric cancer by an histological tech-
systematically review the evidence on the effects of n-3 PUFAs nique; 5) n-3 PUFAs offered in capsules (EPA and/or DHA isolate or
supplementation on inflammatory markers in gastric cancer pa- fish oil) or liquid diet enriched with these fatty acids (cocktails with
tients, through a pooled effect size from all studies and from spe- n-3 PUFA plus antioxidants, glutamine, arginine or nucleic acids
cific subgroups. were not excluded); 6) control group that not received in parallel n-
3 PUFAs supplementation or the amounts of n-3 PUFAs was much
2. Materials and methods less than intervention group (we included studies which control
group received a placebo without n-3 PUFAs, i.e. a supplement of
We planned and performed this systematic review and meta- the paraffin; or, a control/placebo formula with nutritional char-
analysis in accordance with Cochrane Handbook for Systematic acteristics similar to the intervention formula with the exception of
Reviews of Intervention [12] and Preferred Reporting Items for the amount of n-3 PUFAs present); 7) assessment the effect of n-3
Systematic Review and Meta-analysis statement [13] to answer the PUFAs on any biomarkers of our interest (acute phase protein, cy-
following question: does n-3 PUFAs supplementation in gastric tokines and/or eicosanoids).
cancer alter circulating biomarkers related to inflammatory
response? The protocol of this study was registered on PROSPERO 2.3. Data selection and extraction
database (registration number: CRD42016037283).
Records retrieved by search strategy in databases, except
2.1. Literature search GOOGLE Scholar, were exported to a reference manager software
and those repeated were quantified and excluded. One investigator
A systematic search of published literature was conducted in (MM), initially screened the records by title analyzing. If it was very
Cochrane Central Register of Controlled Trials (CENTRAL), Medline, unlikely to be relevant, the abstract was not assessed. Records with
Scopus, Web of Science and Science Direct to identify all relevant titles denoting relevance had their abstract read and, those
trials until July, 14th 2016. We used search terms, keywords or potentially eligible, full texts were assessed to confirm entry
medical subject headings relating to the groups of PICO acronym criteria. This selection process was performed two times by the
(patient, intervention, comparison and outcome): Patient (P) e same investigator. When eligibility was not clear by data presented
patients diagnosed with gastric cancer: (Gastric OR stomach OR in the article, original authors were contacted by e-mail to obtain
gastrointestinal) AND (cancer OR neoplasm* OR malignan* OR more information. If there were doubts about the eligibility and we
tumour OR tumor); Intervention (I) e n-3 PUFAs: (“fish oil” OR “n-3” did not have return from authors clarifying our doubts, the article
OR “unsaturated fatty acids” OR “eicosapentaenoic acid” OR EPA OR was not included in this systematic review and meta-analysis.
“docosahexaenoic acid” OR DHA OR “marine oil” OR “krill oil” OR The following information about the methodology and contin-
omega3 OR “omega-3” OR “omega 3” OR PUFA OR “u-3”); Compar- uous data for specific outcomes were extracted independently from
ison (C): standard diet or placebo/control supplement, or none the eligible articles by two investigators (MM, TC) and were
(omitted in the search strategy); Outcomes (O): circulating acute registered in a predesigned data extraction table: study identifi-
phase proteins (“acute phase protein” OR albumin OR “prealbumin” cation (authors and publication year); country; blinding; random-
OR transferrin OR “retinol binding protein” OR RBP OR CRP OR “C- ization technique; number of patients in each group that were
reactive protein” OR ceruloplasmin OR fibrinogen) and/or others randomized, dropped and malnourished; number of male and fe-
biomarkers of inflammatory status (Cytokin* OR interleukin* OR male; mean age; anti-cancer treatment; n-3 PUFAs supplement
inflammat* OR “inflammatory markers” OR “inflammatory mediators” data (dose, formula, time and route of administration); n-6:n:3
OR eicosanoid* OR prostaglandin* OR leukotriene* OR chemokine* OR fatty acid ratio of control and intervention supplements; antibiotic
TNF OR “tumor necrosis factor” IFN OR interferon). use during the study period; and, results of inflammatory bio-
We combined the search terms using Boolean operators (OR was markers concentrations (mean and standard deviation). Divergence
used to combine search terms from the same PICO's group and, data from two investigators were discussed after new consult to the
AND, to combine search terms from different groups). Also, we used original article and if no reached consensus, a third investigator
truncation symbol (*) to find all multiple word endings and spell- (ET) resolved it.
ings, and, quotation markers (“ ”) to force the database match a To obtain data for meta-analysis from articles that mentioned
term exact enclosed by this symbol (search for exact phrases). No effects on inflammatory markers without reporting numerical data,
filter was used and in the Science Direct database we added the we contacted the authors by e-mail. For continuous data presented
term “clinical trial” to refine the search. in median and interquartile range or minimum and maximum
Additionally, reference lists of eligible studies and topic-related values, we estimated their mean and standard deviation by the
reviews were also manually searched to find others relevant trials. equation proposed by Hozo et al. (2005) [14] (if the original author
GOOGLE Scholar was also searched with a reduced strategy did not provide it by e-mail). Studies were excluded of the meta-
[(“omega-3” OR “n-3 PUFA” OR “fish oil” OR EPA OR DHA) AND analysis (only used in systematic review synthesis) when data
(“gastric cancer” OR “gastric neoplasm” OR “stomach tumor”)] and were not possible to estimate and the data could not be obtained
only the first 10 pages of results were reviewed. The number of from the authors.
articles reviewed in this database were not counted in the total
number of records retrieved, because the search mechanism not 2.4. Risk of bias assessment
permit a combination of all variations of the intervention, clinical
setting and outcomes terms performing a precise search strategy. Risk of individual study bias was assessed using the Cochrane
Collaboration's tool, version 5.1.0 [12]. The tool comprises seven
2.2. Entry criteria domains to evaluate the risk of six known bias and other unknown:
1) adequacy of sequence generation and, 2) allocation concealment
Trials were included if they met the following criteria: 1) (selection bias); 3) blinding of participants and researchers (per-
controlled or randomized clinical trial carried out in humans; 2) full formance bias); 4) blinding of outcome assessment (detection
articles published (published abstracts were not included); 3) bias); 5) incomplete outcome data (attrition bias); 6) selective
language of publication in English; 4) study sample composed only outcome reporting (reporting bias); and, 7) other potential sources
Please cite this article in press as: Mocellin MC, et al., A meta-analysis of n-3 polyunsaturated fatty acids effects on circulating acute-phase
protein and cytokines in gastric cancer, Clinical Nutrition (2017), http://dx.doi.org/10.1016/j.clnu.2017.05.008
M.C. Mocellin et al. / Clinical Nutrition xxx (2017) 1e11 3
3. Results
control group (the reason reported by authors to dropout 23.1% of
3.1. Search results patients from this group was “incomplete data”). None of the
studies applied intention to treat analysis.
From the literature search, 2057 records were retrieved (71 from Two RCTs were double-blind [17,22] and a third [20] reported
CENTRAL, 690 from Scopus, 881 from Science Direct, 210 from Web that only researchers were blinded. All comprised patients of both
of Science and 205 from Pubmed). After exclusion of repeated ar- sexes and a pool of cancer stages. Mean age of participants was
ticles, 1631 records were screened and five met the entry criteria. similar among the RCTs. Only one [22] was performed with patients
Additionally, four studies were found from other sources (search in undergoing chemotherapy treatment, while the eight remaining
GOOGLE scholar and reviewing the references of important origi- RCTs [16e21,23,24] were carried out in surgical settings.
nals articles and reviews about the theme), totalling nine studies Details of n-3 PUFAs supplementation protocol used in each RCT
included in our systematic review and meta-analysis. The complete are described in Table 2. Five trials [16e18,21,23] used an immu-
flowchart is shown in Fig. 1. nomodulatory diet enriched with n-3 PUFAs, arginine and antiox-
idants (besides these immunonutrients, one [16] added glutamine
3.2. Trials characteristics and four [17,18,21,23], ribonucleic acids). n-3 PUFA from fish oil
without others immunonutrients was used in four trials
An overview of study design and demographic parameters of [19,20,22,24]. Equal number of RCTs (n ¼ 3) administered n-3
patients enrolled in the eligible nine studies [16e24] is shown in PUFAs orally [18,22,23], enterally [16,17,21] and parenterally
Table 1. All were randomized clinical trials (RCTs) published from [19,20,24]. In studies where n-3 PUFA was offered orally, amounts
2005 and conducted in Europe [17,19e21] or Asia [16,18,20,22e24]. of EPA and DHA were similar for all patients, while in studies where
In total, 355 patients were allocated to intervention group and 343 n-3 PUFAs were offered parenterally or enterally, dosages of EPA
to control group. The sample size (excluded the losses of follow-up) and DHA varied among patients within the same RCT because it was
ranged from 26 [20] to 231 [18] patients. Four studies [17e19,24] established according to the individual body weights or energy
had losses of participants during follow-up: in three [17e19] the requirements.
proportion of losses was balanced between intervention and con- For those studies with surgical patients, in six [16,17,19e21,24]
trol groups (main causes were: death [17], surgical and feeding n-3 PUFA was offered at postoperatively period (5e7 days), and
complications [17,19], withdrew of consent [17,19], not fulfilled in two [18,23], pre-operatively (5 or 7 days). Use of antibiotics
criteria of permanence in the study [18] and randomization of non- during the study period was reported in five [17,18,21,23,24] of
eligible patients [18]); and, in one [24], losses occurred only in eight RCTs in surgical patients.
Please cite this article in press as: Mocellin MC, et al., A meta-analysis of n-3 polyunsaturated fatty acids effects on circulating acute-phase
protein and cytokines in gastric cancer, Clinical Nutrition (2017), http://dx.doi.org/10.1016/j.clnu.2017.05.008
4 M.C. Mocellin et al. / Clinical Nutrition xxx (2017) 1e11
Table 1
Basic characteristics of the randomized clinical trials included in the meta-analysis.
Trial (Year) Country Blinding Randomization technique n randomized: n n male: n female n malnourished Mean age,
analyzed (% dropouts) (%)a (years)
Abbreviation: NR e No reported; Bold and italic type corresponds to control group. No italic and non bold type corresponds to intervention group.
¥
Significant difference between control and intervention group (p < 0.05).
a
Criteria for diagnostic of malnutrition: Klel et al.: any of these parameters: weight loss (last 3e6 months) 10%, body mass index 18 kg/m2, serum albumin
concentration 3.2 g/dL; serum prealbumin concentration 15 mg/dL; total lymphocytes counts 1200/mm3; Farreras et al.: >10% of lost body weight; Marano et al.: no
reported; Fujitani et al.: subjective global assessment.
b
Values are median.
A placebo supplement without n-6 or n-3 PUFAs was used in results about the nutrients intake/infused during the follow-up
one study [22]. In this trial, Nemati and cols (2015) [22]offered to period. Still, in these RCTs, four [16,19,20,24] did not report
the control group a supplement containing paraffin. In another RCT important information about the surgery procedures that could
[18], control group did not receive any supplement. The remaining influence the outcomes. Baseline characteristics (demographic and
seven studies [16,17,19e21,23,24], the formula offered to inter- clinical) of both groups investigated were not provided in four RCTs
vention group containing high dose of n-3 PUFA was replaced by a [16,19,22,23], which does not guarantee the comparability of the
formula rich in n-6 PUFAs that was offered to control group. We groups into the RCTs.
emphasize that in Chen's RCT [16] the intervention group received a
supplement with more n-6 PUFAs than n-3 (n-6:n-3 ratio ¼ 3.4:1), 3.4. Effect of n-3 PUFA on biomarkers of inflammatory status
while in all other studies, intervention group received a formula
with more n-3 than n-6 PUFAs (n-6:n-3 ratio 0.9:1). Considering the individual studies, we observed a small number
of studies that described statistical differences in circulating con-
3.3. Risk of bias assessment centrations of albumin [22], prealbumin [16,19], transferrin [16,22],
total protein [17] and retinol binding protein [23] between study
The assessment of risk of bias in the included studies using the groups after n-3 PUFAs supplementation (Table 2).
Cochrane tool is shown in Fig. 2. None of RCTs included was free of One study for albumin [23], prealbumin [23], total protein [23]
potential bias. Although they are randomized trials, only three and RBP [23], and, two for transferrin [22,23], did not present in
[17,18,20] reported the technique used to generate the random the article the mean values and/or respectively standard deviation
sequence. Regarding allocation concealment, none provided clear for these outcomes and we were unable to contact authors to
information to judge as adequate. Double-blinding design was re- obtain these data. Thus, meta-analysis was performed with seven
ported in two studies [17,22], but was not mentioned who was RCTs for albumin; four for prealbumin and transferrin; and three
blinded (participants, researches or statistician). In Makay's trial for total protein.
[20] only researchers were blinded. Low risk of reporting bias was Global meta-analysis of data from included RCTs showed a sig-
attributed for all trials, except for Okamoto's RCTs [23], where re- nificant increase in albumin levels (SMD ¼ 0.28; 95% CI ¼ 0.07, 0.48;
sults for some outcomes were not provided (albumin, prealbumin, p ¼ 0.008; Fig. 3 and Supplemental Table 1) and in prealbumin
transferrin and retinol binding protein). In the long-period of (SMD ¼ 0.56; 95% CI ¼ 0.12, 1.00; p ¼ 0.013; Fig. 3 and
supplementation in Nemati's trial [22] (6 weeks) the compliance on Supplemental Table 2) with presence of heterogeneity (Chi2
n-3 PUFAs supplements consumption (10 tablets/day) was impor- test ¼ 0.063; I2 ¼ 59.0%), in favor to intervention group, while for
tant, but was not assessed. In these same study, the nutrients transferrin and total protein, there were no significant results
consumption from foods and beverages was assessed during the (Fig. 3; Supplemental Tables 3 and 4), but with substantial het-
trial and shown a significant lower consumption for control group erogeneity. When the RCTs that assessed albumin were stratified
at the end of week 6 for energy, carbohydrate, fat, protein and the there were significant effects for: RCTs with surgical patients with
majority of vitamins and minerals, what could explain the positive n-3 supplementation at post-operative period (SMD ¼ 0.24; 95%
effects found in intervention group for the outcomes assessed. CI ¼ 0.02, 0.45; p ¼ 0.030; Supplemental Table 1); in double-blind
In RCTs performed with surgical patients, due to infusion/con- RCTs (SMD ¼ 0.59; 95% CI ¼ 0.16, 1.01; p ¼ 0.007; Supplemental
sumption monitoring of the diet enriched with n-3 PUFAs during Table 1); in studies with lower than 50 patients enrolled
the hospital stay by researchers, the compliance probably was (SMD ¼ 0.36; 95% CI ¼ 0.02, 0.69; p ¼ 0.037; Supplemental Table 1);
secured. However, only one [17] of the eight surgical RCT's shown and in RCTs that offered n-3 PUFAs from fish oil alone (SMD ¼ 0.34;
Please cite this article in press as: Mocellin MC, et al., A meta-analysis of n-3 polyunsaturated fatty acids effects on circulating acute-phase
protein and cytokines in gastric cancer, Clinical Nutrition (2017), http://dx.doi.org/10.1016/j.clnu.2017.05.008
M.C. Mocellin et al. / Clinical Nutrition xxx (2017) 1e11 5
Table 2
Characteristics of the n-3 PUFAs supplementation protocol and outcomes assessed in included clinical trials.
Trial (year) Antineoplastic EPA þ DHA, g/d Time and period of Formulation Route of Content of Outcomes
treatment supplementation administration n-6:n-3 in control
formulaj
intervention
formulae
Chen et al. Surgical NRa 7 postoperative days Immunonutrient- Enteral 5:1 j 3.4:1 [ prealbumin,
(2005) [16] enriched and high IL-2, transferrin
protein liquid diet 4 albumin
(arginine, Y IL-6, TNF-a
glutamine and n-3
PUFAs)
Farreras et al. Surgical 4.2b 7 postoperative days Immunonutrient- Enteral 6.8:1 j 0.7:1d [ total protein
(2005) [17] enriched liquid diet 4 prealbumin,
(arginine, n-3 PUFAs albumin, transferrin
and RNA)
Klek et al. Surgical 0.2 g/kgc 7 postoperative days Fish oil Parenteral 6.6:1 j 0.1:1 4 albumin
(2005) [19] [ prealbumin
Okamoto et al. Surgical 3.1 7 preoperative days Immunonutrient- Oral 3:1 j 0.7:1 4 prealbumin,
(2009) [23] enriched liquid diet transferrin,
(arginine, n-3 PUFAs albumin,
and RNA) total protein
[ retinol binding protein
Makay et al. Surgical 0.2 g/kgc 5 postoperative days Fish oil Parenteral 7:1 j 0.1:1 4 albumin
(2011) [20]
Fujitani et al. Surgical 3.4 (2 g EPA þ 5 preoperative days Immunonutrient- Oral 0:0 j 0.8:1 4 CRP
(2012) [18] 1.4 g DHA) enriched liquid diet
(arginine, n-3 PUFAs
and RNA)
Marano et al. Surgical NRa 7 postoperative days Immunonutrient- Enteral 6.6:1 j 0.7:1 4 albumin,
(2013) [21] enriched liquid diet transferrin,
(arginine, n-3 PUFAs total protein
and RNA)
Wei et al. Surgical 0.2 g/kgc 6 postoperative days Fish oil Parenteral 7.4:1 j 0.2:1 4 albumin,
(2014) [24] transferrin, total
protein, prealbumin
and retinol binding
protein, CRP,
VEGF, IGF-1
Y IL-1b, IL-6, TNF-a
Nemati et al. Chemotherapy 3 (1.8 g EPA þ 6 weeks Fish oil Oral 0:0 j <0.1:1 [ albumin, transferrin
(2015) [22] 1.2 g DHA)
Abbreviations: NR e Not reported; RNA e ribonucleic acids; 4 no significant difference between the intervention and control groups after intervention; Y e significantly
lower than control group after intervention; [ e significantly higher than control group after intervention.
a
Total amount of infused diet was determined according to the weight of the participant considering 35 kcal/kg in the study of Marano et al. and 30 non-protein kcal/kg in
the study of Chen et al..
b
The amount corresponds to total omega-3 of the diet considering the volume administered (mean) from the third postoperative day.
c
The amount corresponds to fish oil only, not to n-3 PUFAs.
d
Control group received a diet containing n-3 PUFAs. The amount of these fatty acids provided by the formula from the third postoperative day was 1.7 g/day.
e
Estimated content.
95% CI ¼ 0.03, 0.65; p ¼ 0.032; Supplemental Table 1). No hetero- the heterogeneity can be explained by blinding and random
geneity was observed for all subgroup analyzes. sequence generation (Supplemental Table 4).
Meta-analysis for prealbumin showed an important heteroge- Sensitivity analysis for albumin, transferrin and total protein
neity in pooled effect sizes. This heterogeneity may be attributed meta-analysis showed that removing a single study included in
to: number of patients enrolled in each study; different formulas pooled analysis for each parameter, there are no significant changes
containing n-3 PUFAs offered; route of n-3 PUFAs administration; in combined effect sizes. This does not apply to prealbumin meta-
and report the use or not of antibiotic during the study period. analysis, where after removing the RCTs of Chen et al. [16], Farre-
Interesting, in subgroups without heterogeneity (RCTs with 50 ras et al. [17] or Klel et al. [19], the effect size loses significance
patients; that offered an enriched-diet with n-3 and others (Supplemental Fig. 1).
immunonutrients; enterally; and trials that did not report the use Only two RCTs assessed the effects of n-3 PUFAs on cytokines
of antibiotics) the combined effect size was significant in favor to [16,24] and on CRP [18,24] (Table 2). All studies were conducted in
intervention group to increase prealbumin levels (Supplemental surgical patients, but others characteristics (dose, route of admin-
Table 2). istration, control formula, time and moment of n-3 PUFAs supple-
Regarding the effects of n-3 PUFAs on transferrin and total mentation) were different among them. Meta-analysis was
protein concentrations, both global and subgroups analysis did not performed only for IL-6, TNF-a e CRP (Fig. 4). The findings showed a
show significant effect sizes, but subgroup analysis suggest that significant reduction in favor to intervention group for IL-6
heterogeneity from RCTs that assessed transferrin can be explained (SMD ¼ 0.71; 95% CI ¼ 1.15, 0.27; p ¼ 0.001) and TNF
by the number of participants and by the report the use of antibiotic (SMD ¼ 0.92; 95% CI ¼ 1.58, 0.26; p ¼ 0.006), without sig-
(Supplemental Table 3), while in RCTs that assessed total protein, nificant heterogeneity.
Please cite this article in press as: Mocellin MC, et al., A meta-analysis of n-3 polyunsaturated fatty acids effects on circulating acute-phase
protein and cytokines in gastric cancer, Clinical Nutrition (2017), http://dx.doi.org/10.1016/j.clnu.2017.05.008
6 M.C. Mocellin et al. / Clinical Nutrition xxx (2017) 1e11
Please cite this article in press as: Mocellin MC, et al., A meta-analysis of n-3 polyunsaturated fatty acids effects on circulating acute-phase
protein and cytokines in gastric cancer, Clinical Nutrition (2017), http://dx.doi.org/10.1016/j.clnu.2017.05.008
M.C. Mocellin et al. / Clinical Nutrition xxx (2017) 1e11 7
Fig. 3. Forest plot showing standard mean difference (SMD) and 95% CIs for the impact of n-3 PUFAs supplementation on albumin, prealbumin, transferrin and total protein. The p-
value presented in the figure refer to Chi2 test for heterogeneity.
Please cite this article in press as: Mocellin MC, et al., A meta-analysis of n-3 polyunsaturated fatty acids effects on circulating acute-phase
protein and cytokines in gastric cancer, Clinical Nutrition (2017), http://dx.doi.org/10.1016/j.clnu.2017.05.008
8 M.C. Mocellin et al. / Clinical Nutrition xxx (2017) 1e11
Fig. 4. Forest plot showing standard mean difference (SMD) and 95% CIs for the impact of n-3 PUFAs supplementation on IL-6, TNF-a and C-reactive protein. The p-value presented
in the figure refer to Chi2 test for heterogeneity.
Please cite this article in press as: Mocellin MC, et al., A meta-analysis of n-3 polyunsaturated fatty acids effects on circulating acute-phase
protein and cytokines in gastric cancer, Clinical Nutrition (2017), http://dx.doi.org/10.1016/j.clnu.2017.05.008
M.C. Mocellin et al. / Clinical Nutrition xxx (2017) 1e11 9
and concealment of allocation are other essentials approaches that The basal nutritional status of the patient could influence the
minimize the risk of bias in RCTs. In the current meta analyses/re- concentration of some biomarkers assessed in this meta-analysis
view, only three described an adequate technique to generate the after n-3 PUFAs supplementation. Thus, it would be necessary to
randomization sequence and none described concealing in the separate the malnourished patients from those well-nourished,
allocation. Although not evident in our analysis, some studies and so, to evaluate the impact of supplementation in each group.
demonstrated that the beneficial effect from an intervention was This analysis can not be done because the originals articles not
exaggerated in accordance with methodological characteristics: provided results considering the stratification by nutritional status.
open label studies have a range of 13e17% exaggeration of inter- Of the nine studies included in this work, four presented the
vention effect estimates, on average, than double-blind RCTs; number of malnourished patients. We can observe (Table 1) that
adequate/unclear sequence generation have 5e11% exaggeration the proportion of malnutrition in control and intervention groups
than adequate; and inadequate/unclear allocation concealment was similar in each of these four studies, what suggest that the
have 7e30% exaggeration than adequate [35e37]. For these results found in this work may not be influenced by nutritional
empirical evidences, we alert to researchers for their future studies status of the patients.
be well designed to avoid bias, as well as, provide sufficient The nutrients consumption could be another potential
methodological detail (in the article or in trial register) to ensure confounder of the results found in this work. But, eight of nine
reproducibility and judgment of adequacy on processes adopted. studies included were performed in a hospital setting, where the
In sensitivity analysis for prealbumin, removing any of the dietetic ingestion is standardized.
studies included in this analysis, we observed a significant change Strength points also can be observed in the present study:
on the pooled effect size. When the RCTs of Chen et al. (2005) [16], extensive literature search; we estimated the numerical data (mean
Farreras et al. (2005) [17] or Klek et al. (2005) [19] were removed, the and SD) in one article [18] from a validated equation (when
combined effect lost significance, and when Wei's RCT was removed, possible) to avoid losing it in the meta-analysis; subgroups analysis
the effect went from a medium size to large, and the heterogeneity for methodological characteristics that could explain heterogeneity
disappeared. This change occurred due mainly for influences of the among included studies; sensitivity analysis to assess the effect of
trial of Wei et al. in the combined effect size (we can observe in Fig. 3 individual studies on pooled effect size, that demonstrated that
that Wei's RCT was the only study that presented a lower mean for none of the studies when removed altered the conclusion of the
prealbumin in intervention than in control group, pulling the analysis (exception for prealbumin e discussed earlier).
combined effect for the left side of the graph e “intervention re- In conclusion, the supplementation of n-3 PUFAs in gastric
duces prealbumin”). We identified two causes that could explain cancer has a potential effect on increasing the levels of albumin and
this discrepant effect for prealbumin from this trial: losses occurred prealbumin, and decreasing the pro-inflammatory cytokines IL-6
only in the control group, justified by “incomplete data” (Table 1); and TNF-a. In accordance with the studies published until the
and, of the four studies combined in this meta-analysis, all were moment, it seems prudent to advise the use of n-3 PUFAs from fish
carried out in surgical patients, but the RCT of Wei et al. was the only oil isolated or from immune-enriched diet in replacement a for-
that reported the use of antibiotic at post-operative period. mula containing large amounts of n-6 PUFAS at post-operative
We can cite some reasons that may confuse the conclusion of period for patients submitted to gastric resection surgery for
our study: 1) use combined of n-3 PUFAs with arginine, glutamine improve these parameters. It appears that offering n-3 PUFA orally
and antioxidants in most of included RCTs (these immunonutrients or enterally could result in better concentrations of some
have anti-inflammatory action that could amplify the effects of n-3 biochemical parameters evaluated than parenterally, however,
PUFAs on inflammation [38]); 2) use of control supplement rich in there is a low evidence to support this claim. Nevertheless, other
n-6 PUFAs in seven RCTs (previous studies have demonstrated that well-designed clinical trials need to be conducted, especially in
n-6 PUFAs play an opposing role with n-3 PUFAs in the process of chemotherapy patients, to confirm our findings.
inflammation that could overestimate the effect size arising n-3
PUFAs supplementation [39]); 3) use of antibiotic in a moment of 5. Directions for future research
the study period (when it was used, the trials did not show sig-
nificant results from n-3 PUFAs supplementation e it appears to be Based on limitations of the studies included in this systematic
an effect canceler); and, 4) different anti-cancer treatment applied review and meta-analysis we suggest that researchers take careful
(surgical and chemotherapy). We emphasize that the biggest effect in adopt and describe with more details several procedures to
size for n-3 PUFAs effects on albumin was provided by Nemati's RCT improve the methodological quality and ensure the reproducibility
that not used control formula containing n-6 PUFAs (it was the only of studies regarding the use of n-3 PUFAs as intervention in surgical
RCT with more than one week of n-3 PUFAs supplementation, be- patients:
sides to be the only RCT carried out in patients receiving chemo-
therapy), and the patients enrolled in the control group consumed a) Describe the method used to generate the random sequence;
significant lower amounts of energy, carbohydrate, fat and protein, b) Describe the method used to ensure the allocation
besides the majority of minerals and vitamins during the follow-up, concealment;
important facts that confuse the actual effect of the intervention c) Described the procedures of blinding (if applicable);
tested. However, removing this study of the analysis, the effect size d) Implement a control group that not received any type of fatty
was reduced but remain significant in favor to n-3 PUFAs acids, specially n-6 PUFAs, or standardize the profile of lipids
supplementation. on both intervention and control formulas where the n-3
Other limitations include: low number of studies assessing the PUFAs amounts is the unique difference;
effects on n-3 PUFAs on cytokines and CRP (despite the large effect e) Collect and describe data about the surgical procedures such
size found for IL-6 and TNF-a meta-analysis, we suggest caution in as surgical time and blood loss during operation;
its interpretation); some studies did not provide numerical data f) Describe the use of any medicine used before, during, and
(mean and SD) to include in a meta-analysis (we tried to contact the after the surgery, especially anti-inflammatory and antibi-
authors of these articles, but neither returned); all studies pre- otics drugs;
sented risk of bias; and, the publication bias was not assessed due g) Implement a technique to ensure the compliance in the n-3
the low number of included studies. PUFAs consumption such as to determine the fatty acids
Please cite this article in press as: Mocellin MC, et al., A meta-analysis of n-3 polyunsaturated fatty acids effects on circulating acute-phase
protein and cytokines in gastric cancer, Clinical Nutrition (2017), http://dx.doi.org/10.1016/j.clnu.2017.05.008
10 M.C. Mocellin et al. / Clinical Nutrition xxx (2017) 1e11
profile of the plasma, especially in studies conducted in an [3] Grivennikov SI, Karin M. Inflammatory cytokines in cancer: tumour necrosis
factor and interleukin 6 take the stage. Ann Rheum Dis 2011;70(Suppl. 1):
open setting;
i104e8.
h) Stratify the analysis by grade of tumor stage, nutritional [4] Balkwill F, Mantovani A. Inflammation and cancer: back to Virchow? Lancet
status and localization of the tumor (if performed with 2001;357:539e45.
several types of cancer); [5] Sethi G, Shanmugam MK, Ramachandran L, Kumar AP, Tergaonkar V. Multi-
faceted link between cancer and inflammation. Biosci Rep 2012;32:1e15.
i) In enterally or parenterally studies that use a control formula, [6] Tsoli M, Robertson G. Cancer cachexia: malignant inflammation, tumorkines,
is necessary to standardize the quantities of antioxidants; and metabolic mayhem. Trends Endocrinol Metab 2013;24:174e83.
j) For include the study in meta-analysis, it is essential describe [7] D'Eliseo D, Velotti F. Omega-3 fatty acids and cancer cell cytotoxicity: impli-
cations for multi-targeted cancer therapy. J Clin Med 2016;5.
and express all numerical data in mean and standard devi- [8] Pappalardo G, Almeida A, Ravasco P. Eicosapentaenoic acid in cancer improves
ation (or standard error) from basal and final moments. body composition and modulates metabolism. Nutr (Burbank, Los Angeles
Avoid mean differences, interquartile range and confidence County, Calif) 2015;31:549e55.
[9] Wendel M, Heller AR. Anticancer actions of omega-3 fatty acidsecurrent state
interval; and future perspectives. Anticancer Agents Med Chem 2009;9:457e70.
k) Describe all clinical and demographic variables as possible at [10] Murphy RA, Mourtzakis M, Mazurak VC. n-3 polyunsaturated fatty acids: the
the baseline, and compare them between the study groups potential role for supplementation in cancer. Curr Opin Clin Nutr Metab care
2012;15:246e51.
(apply appropriate statistical tests);
[11] Mocellin MC, Camargo CQ, Nunes EA, Fiates GM, Trindade EB. A systematic
l) Avoid not show data. Present all data of outcomes even in review and meta-analysis of the n-3 polyunsaturated fatty acids effects on
supplementary material; inflammatory markers in colorectal cancer. Clin Nutr (Edinburgh, Scotland)
2016;35:359e69.
m) Take careful with statistic test applied. If study groups are
[12] Higgins JPT, Green S. Cochrane Handbook for systematic reviews of in-
comparable, use appropriate and simple statistic tests. For all terventions. The Cochrane Collaboration; 2011.
outcomes or characteristics variables of participants, when [13] Liberati A, Altman DG, Tetzlaff J, Mulrow C, Gotzsche PC, Ioannidis JP, et al. The
comparing groups, described the p-value of the difference PRISMA statement for reporting systematic reviews and meta-analyses of
studies that evaluate healthcare interventions: explanation and elaboration.
test. Bmj 2009;339:b2700.
[14] Hozo SP, Djulbegovic B, Hozo I. Estimating the mean and variance from the
median, range, and the size of a sample. BMC Med Res Methodol 2005;5:13.
[15] Rosenthal JA. Qualitative descriptors of strength of association and effect size.
Conflict of interest J Soc Serv Res 1996;21:37e59.
[16] Chen DW, Fei ZW, Zhang YC, Ou JM, Xu J. Role of enteral immunonutrition in
patients with gastric carcinoma undergoing major surgery. Asian J Surg
The authors declare that they have no conflict of interest.
2005;28:121e4.
[17] Farreras N, Artigas V, Cardona D, Rius X, Trias M, Gonzalez JA. Effect of early
Author's contributions postoperative enteral immunonutrition on wound healing in patients un-
dergoing surgery for gastric cancer. Clin Nutr (Edinburgh, Scotland) 2005;24:
55e65.
Michel Carlos Mocellin: study concept and design; search in [18] Fujitani K, Tsujinaka T, Fujita J, Miyashiro I, Imamura H, Kimura Y, et al.
databases; study selection; data extraction, analysis and interpre- Prospective randomized trial of preoperative enteral immunonutrition fol-
lowed by elective total gastrectomy for gastric cancer. Br J Surg 2012;99:
tation; quality assessment; statistical analysis; drafting of the 621e9.
manuscript. [19] Kłek S, Kulig J, Szczepanik AM, Jedrys J, Kołodziejczyk P. The clinical value of
Ricardo Fernandes: quality assessment; critical revision of the parenteral immunonutrition in surgical patients. Acta Chir Belg 2005;105:
175e9.
manuscript for important intellectual content.
[20] Makay O, Kaya T, Firat O, Sozbilen M, Caliskan C, Gezer G, et al. Omega-3 fatty
Thayz Rodrigues Chagas: data extraction; critical revision of the acids have no impact on serum lactate levels after major gastric cancer sur-
manuscript for important intellectual content. gery. J Parenter Enter Nutr 2011;35:488e92.
[21] Marano L, Porfidia R, Pezzella M, Grassia M, Petrillo M, Esposito G, et al.
Erasmo Benicio Santos de Moraes Trindade: critical revision of
Clinical and immunological impact of early postoperative enteral immuno-
the manuscript for important intellectual content; study nutrition after total gastrectomy in gastric cancer patients: a prospective
supervision. randomized study. Ann Surg Oncol 2013;20:3912e8.
[22] Nemati A, Nachvak SM, Djafarian K, Faizi-Khankandi I. Effect of omega-3 fatty
acid supplementation on nutritional status in patients with gastric cancer
Acknowledgement during chemotherapy. J Nutr Sci Diet 2015;1:7.
[23] Okamoto Y, Okano K, Izuishi K, Usuki H, Wakabayashi H, Suzuki Y. Attenua-
tion of the systemic inflammatory response and infectious complications after
We are grateful to the Graduate Program of Nutrition at Federal gastrectomy with preoperative oral arginine and u-3 fatty acids supple-
University of Santa Catarina and Coordination for the Improvement mented immunonutrition. World J Surg 2009;33:1815e21.
of Higher Education Personnel (CAPES e Brazil) for the scholarship [24] Wei Z, Wang W, Chen J, Yang D, Yan R, Cai Q. A prospective, randomized,
controlled study of omega-3 fish oil fat emulsion-based parenteral nutri-
granted to first three authors.
tion for patients following surgical resection of gastric tumors. Nutr J
2014;13:25.
Funding sources [25] Li K, Huang T, Zheng J, Wu K, Li D. Effect of marine-derived n-3 poly-
unsaturated fatty acids on C-reactive protein, interleukin 6 and tumor ne-
crosis factor alpha: a meta-analysis. PLoS One 2014;9:e88103.
This research did not receive any specific grant from funding [26] Xin W, Wei W, Li X. Effects of fish oil supplementation on inflammatory
agencies in the public, commercial, or not-for-profit sectors. markers in chronic heart failure: a meta-analysis of randomized controlled
trials. BMC Cardiovasc Disord 2012;12:1e11.
[27] Jiang J, Li K, Wang F, Yang B, Fu Y, Zheng J, et al. Effect of marine-derived n-3
Appendix A. Supplementary data polyunsaturated fatty acids on major eicosanoids: a systematic review and
meta-analysis from 18 randomized controlled trials. PLoS One 2016;11:
e0147351.
Supplementary data related to this article can be found at http:// [28] He L, Li MS, Lin M, Zhao TY, Gao P. Effect of fish oil supplement in maintenance
dx.doi.org/10.1016/j.clnu.2017.05.008. hemodialysis patients: a systematic review and meta-analysis of published
randomized controlled trials. Eur J Clin Pharmacol 2016;72:129e39.
[29] Calder PC. Omega-3 polyunsaturated fatty acids and inflammatory processes:
References nutrition or pharmacology? Br J Clin Pharmacol 2013;75:645e62.
[30] Sigal LH. Basic science for the clinician 39: NF-kappaB-function, activation,
[1] Mantovani A, Allavena P, Sica A, Balkwill F. Cancer-related inflammation. control, and consequences. J Clin Rheum: Pract Rep Rheum Musculoskelet Dis
Nature 2008;454:436e44. 2006;12:207e11.
[2] Fox JG, Wang TC. Inflammation, atrophy, and gastric cancer. J Clin Investig [31] Moshage H. Cytokines and the hepatic acute phase response. J Pathol
2007;117:60e9. 1997;181:257e66.
Please cite this article in press as: Mocellin MC, et al., A meta-analysis of n-3 polyunsaturated fatty acids effects on circulating acute-phase
protein and cytokines in gastric cancer, Clinical Nutrition (2017), http://dx.doi.org/10.1016/j.clnu.2017.05.008
M.C. Mocellin et al. / Clinical Nutrition xxx (2017) 1e11 11
[32] Richard D, Kefi K, Barbe U, Bausero P, Visioli F. Polyunsaturated fatty acids as randomised controlled trials: combined analysis of meta-epidemiological
antioxidants. Pharmacol Res 2008;57:451e5. studies. Health Technol Assess (Winchester, England) 2012;16:1e82.
[33] Lakens D. Calculating and reporting effect sizes to facilitate cumulative sci- [37] Schulz KF, Chalmers I, Hayes RJ, Altman DG. Empirical evidence of bias. Di-
ence: a practical primer for t-tests and ANOVAs. Front Psychol 2013;4:863. mensions of methodological quality associated with estimates of treatment
[34] Karanicolas PJ, Farrokhyar F, Bhandari M. Blinding: who, what, when, why, effects in controlled trials. Jama 1995;273:408e12.
how? Can J Surg 2010;53:345e8. [38] Pierre JF, Heneghan AF, Lawson CM, Wischmeyer PE, Kozar RA, Kudsk KA.
[35] Page MJ, Higgins JP, Clayton G, Sterne JA, Hrobjartsson A, Savovic J. Empirical Pharmaconutrition review: physiological mechanisms. JPEN J Parenter Enter
evidence of study design biases in randomized trials: systematic review of Nutr 2013;37:51Se65S.
meta-epidemiological studies. PLoS One 2016;11:e0159267. [39] Schmitz G, Ecker J. The opposing effects of n-3 and n-6 fatty acids. Prog Lipid
[36] Savovic J, Jones H, Altman D, Harris R, Juni P, Pildal J, et al. Influence of re- Res 2008;47:147e55.
ported study design characteristics on intervention effect estimates from
Please cite this article in press as: Mocellin MC, et al., A meta-analysis of n-3 polyunsaturated fatty acids effects on circulating acute-phase
protein and cytokines in gastric cancer, Clinical Nutrition (2017), http://dx.doi.org/10.1016/j.clnu.2017.05.008