Skip to main content
NCBI home page
Advances in Nutrition logoLink to Advances in Nutrition
. 2020 Aug 13;12(1):128–140. doi: 10.1093/advances/nmaa086

Comparing the Effects of Docosahexaenoic and Eicosapentaenoic Acids on Inflammation Markers Using Pairwise and Network Meta-Analyses of Randomized Controlled Trials

Cécile Vors 1, Janie Allaire 2, Sonia Blanco Mejia 3,4, Tauseef A Khan 5,6, John L Sievenpiper 7,8,9,10,11, Benoît Lamarche 12,✉
PMCID: PMC7850108  PMID: 32790827

ABSTRACT

Recent data from randomized clinical trials (RCTs) suggest that DHA may have stronger anti-inflammatory effects than EPA. This body of evidence has not yet been quantitatively reviewed. The aim of this study was to compare the effect of DHA and EPA on several markers of systemic inflammation by pairwise and network meta-analyses of RCTs. MEDLINE, EMBASE, and The Cochrane Library were searched through to September 2019. We included RCTs of ≥7 d on adults regardless of health status that directly compared the effects of DHA with EPA and RCTs of indirect comparisons, in which the effects of DHA or EPA were compared individually to a control fatty acid. Differences in circulating concentrations of C-reactive protein (CRP), interleukin-6 (IL-6), tumor necrosis factor-alpha (TNF-α) and adiponectin were the primary outcome measures. Data were pooled by pairwise and network meta-analysis and expressed as mean differences (MDs) with 95% CIs. Heterogeneity was assessed (Cochran Q statistic) and quantified (I2 statistic) in the pairwise meta-analysis. Inconsistency and transitivity were evaluated in the network meta-analysis. The certainty of evidence was assessed using the GRADE (Grading of Recommendations Assessment, Development, and Evaluation) approach. Eligibility criteria were met by 5 RCTs (N = 411) for the pairwise meta-analysis and 20 RCTs (N = 1231) for the network meta-analysis. In the pairwise meta-analysis, DHA and EPA had similar effects on plasma CRP [MDDHA versus EPA = 0.14 mg/L (95% CI: –0.57, 0.85); I2 = 61%], IL-6 [MDDHA versus EPA = 0.10 pg/mL (–0.15, 0.34); I2 = 40%], and TNF-α [MDDHA versus EPA = –0.10 pg/mL (–0.37, 0.18); I2 = 40%]. In the network meta-analysis, the effects of DHA and EPA on plasma CRP [MDDHA versus EPA = –0.33 mg/L (–0.75, 0.10)], IL-6 [MDDHA versus EPA = 0.09 pg/mL (–0.12, 0.30)], and TNF-α [MDDHA versus EPA = –0.02 pg/mL (–0.25, 0.20)] were also similar. DHA and EPA had similar effects on plasma adiponectin in the network meta-analysis. Results from pairwise and network meta-analyses suggest that supplementation with either DHA or EPA does not differentially modify systemic markers of subclinical inflammation.

Keywords: DHA, EPA, omega-3, inflammation, C-reactive protein, interleukin, cardiovascular disease, systematic review, meta-analysis

Introduction

Subclinical or chronic inflammation is now indisputably recognized as a key factor in the development of atherosclerosis and subsequent cardiovascular disease (CVD) (1–3). Chronic inflammation is characterized by elevated blood concentrations of acute-phase proteins and cytokines, including C-reactive protein (CRP), interleukin-6 (IL-6), and tumor necrosis factor-alpha (TNF-α), and by low concentrations of plasma anti-inflammatory markers such as adiponectin (4–6). Systemic inflammation can be attenuated through the intake of specific nutrients and foods (7). Among those, long-chain ω-3 PUFAs (LCn–3PUFAs), mainly EPA and DHA, have raised tremendous interest for their purported anti-inflammatory effects. DHA and EPA are naturally present in most seafood, especially fatty fish, but their cardiometabolic effects have been mainly demonstrated when consumed in high quantity as supplements (7, 8). Fish oil but also krill and algal oils are commonly used in many ω-3 fatty acid supplements. Most randomized controlled trials (RCTs) so far have used a mix of DHA and EPA in various forms and proportions, as DHA and EPA occur concomitantly and naturally in food and dietary supplements. A meta-analysis of these RCTs demonstrated the anti-inflammatory effect of LCn–3PUFA supplementation as evidenced by significant reductions in plasma CRP, IL-6, and TNF-α concentrations compared with the control conditions (9). However, several RCTs have tested the hypothesis that individual LCn–3PUFAs are not equally effective in modulating markers of inflammation (10–14). Accordingly, there is increasing evidence from individual trials that DHA and EPA have distinct effects on systemic inflammation, as recently reviewed (15). Specifically, the review indicated that despite greater beneficial effects of DHA compared with EPA on triglycerides, blood pressure, heart rate, and vascular function, the differential effect of EPA and DHA on inflammation markers remained inconclusive. Such data have not yet formed the basis of a systematic review and meta-analysis, which are considered the gold standard of evidence, to inform dietary guidelines. We thus conducted a systematic review and meta-analysis of available RCTs to assess and compare the individual effects of DHA and EPA on surrogate markers of systemic inflammation, namely plasma CRP, IL-6, TNF-α, and adiponectin concentrations. As data from studies having directly compared DHA and EPA are limited, the “network meta-analysis” approach (16, 17) was also used to include both direct and indirect comparisons in the meta-analysis. Thus, the pairwise meta-analysis included RCTs that compared directly the effects of DHA and EPA (direct comparisons) on inflammation, whereas the network meta-analysis included all direct comparison RCTs as well as RCTs that assessed the effects of DHA or EPA individually compared with a control oil or fatty acid (indirect comparisons). To the best of our knowledge, no study has yet compared the independent effect of DHA and EPA on systemic inflammation, using both pairwise and network meta-analysis methodologies. This analysis aimed to answer the following question: do DHA and EPA have similar effects on systemic markers of low-grade inflammation?

Methods

Protocol and registration

The present systematic review and meta-analyses have been conducted according to the Cochrane Handbook for Systematic Reviews of Interventions (18). Results are reported in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) and The PRISMA Extension Statement for conducting Network Meta-analyses (19). The protocol was registered at clinicaltrials.gov as NCT03520556 but not with the International Prospective Register of Systematic Reviews (PROSPERO) because the analysis was proposed and undertaken before 1 July, 2018.

Search strategy and data sources

MEDLINE, EMBASE, and The Cochrane Library were searched until 27 September, 2019 using the strategy presented in Supplementary Table 1. Manual searches based on references of selected studies and reviews were performed to supplement the electronic search.

Eligibility criteria and study selection

Studies were included in pairwise and/or network meta-analyses if they met all the following criteria: 1) RCTs that directly compared the effects of DHA to those of EPA, or assessed the effects of DHA or EPA individually compared with a suitable control (i.e., fatty acids other than EPA and DHA as control); 2) DHA or EPA supplementation with a minimum proportion of 80% for the LCn–3PUFAs; 3) minimum intervention period of 7 d; 4) RCTs in adults (aged 19 y and older); 5) outcomes including plasma concentration of CRP, IL-6, TNF-α, and adiponectin. The following studies were excluded: 1) RCTs including neonates, children, or adolescents; 2) RCTs of acute postprandial effects only; 3) RCTs based on enteral/parenteral nutrition; 4) co-intervention (e.g., drug, dietary supplement, diet, or exercise) not applied in all intervention arms. Unpublished trials and literature published in languages other than English or French were not considered.

Two authors (CV, JA) independently screened titles and abstracts of studies retrieved using the search strategy to identify the ones that met the inclusion criteria outlined above. The same authors (CV, JA) independently assessed the full text of preselected studies for eligibility. Disagreements were resolved by consensus and by involvement of another author (BL) if required.

Data extraction

One author (CV) extracted relevant data from included studies. The extracted information was verified by a second author (JA or BL). Relevant data included the following characteristics: name of first author, title, year of publication, journal, RCT design (parallel or crossover; single or double-blind), sample size (randomized, completers; male/female), study duration, washout duration (for crossover only), health status (i.e., healthy, normal-weight/obese, metabolic or other peripheral disease), mean age, mean baseline BMI, mean waist circumference, specification of DHA and EPA supplementation (dose, proportions in capsules, form, source), specification of control or placebo (type of fatty acids, dose), weight loss (yes versus no), primary outcome of the study, outcomes extracted for the present meta-analyses, data analysis (intent-to-treat or per-protocol), funding source, and conflict of interest. Outcome data were also extracted from each study and included mean differences (± SEM) between postintervention values (for pairwise meta-analysis) and mean (± SD) postintervention values (for network meta-analysis). Study authors were contacted to obtain additional information and missing outcome data. Six authors provided additional data (11, 13, 20–23). In the absence of numerical values for outcome measurements or the inability to contact study authors, values were extracted from figures using Plot Digitizer, version 2.6.8 (Free Software Foundation).

Risk of bias assessment

Two authors (CV, BL) independently assessed risk of bias of each study included in this work using the Cochrane Risk of Bias Tool (18). The following categories of bias were assessed: random sequence generation, allocation concealment, blinding, incomplete outcome data, selective outcome reporting, and other bias. Trials were considered at: 1) high risk of bias if ≥3 out of a maximum of 6 items were rated as “high risk,” 2) low risk of bias if ≥4 out of a maximum of 6 items were rated as “low risk” and a maximum of 1 item rated with a high risk of bias, and 3) moderate/unclear risk of bias for all other studies (24). Disagreements were resolved by consensus. Risk of bias for network meta-analysis also included assessment of transitivity. To evaluate the assumption of transitivity, we compared the distribution of the potential effect modifiers (BMI, age, study duration, percent male, DHA dose, EPA dose) across the available direct comparisons.

Outcomes

There were 4 prespecified outcomes in the present study: 1) plasma CRP (mg/L), 2) plasma IL-6 (pg/mL), 3) plasma TNF-α (pg/mL), and 4) plasma adiponectin (mg/L). The pairwise meta-analysis was performed for the first 3 outcomes because only 1 pairwise comparison was available for adiponectin. The network meta-analysis was conducted for the 4 outcomes.

Data synthesis and analysis

We used standard Cochrane methods for pairwise meta-analysis and augmented this evidence using network meta-analysis methods. The latter compare multiple treatments simultaneously in a single analysis by combining direct and indirect evidence within a network of studies. Direct evidence refers to the existing comparison of 2 interventions to each other within a study (e.g., A compared with B). Indirect evidence refers to the evidence obtained through 1 or more common comparators (e.g., in the absence of studies that directly compare A and B, direct evidence of A compared with C, and B compared with C can be used to provide indirect evidence about A compared with B). The combination of direct and indirect evidence is called mixed evidence.

Pairwise meta-analysis

Data were managed and analyzed with the use of Review Manager (RevMan) version 5.3.2 (The Nordic Cochrane Centre, The Cochrane Collaboration) for the pairwise meta-analysis. The generic inverse variance method with random-effects models was used to synthesize the overall effect estimate of DHA and EPA on plasma CRP, IL-6, and TNF-α. DerSimonian and Laird random-effects models were used even in the absence of statistically significant between-study heterogeneity, as they yield more conservative summary effect estimates in the presence of residual heterogeneity (25, 26). For each outcome, mean differences (MDs) between DHA and EPA were extracted for each pairwise comparison when provided in the publications. If not provided, postintervention values after each treatment (DHA and EPA) in each study were used to calculate MDs between the 2 treatments. MDs were calculated by subtracting means of postintervention values and SEs were calculated from the available data and statistics using published formulas (25). When only medians and IQRs were available, means were estimated from median values using the new method of Luo et al. (27) and SDs were estimated from IQR using the method described by Wan et al. (28). The SEs were then calculated from SD values using Cochrane Formulas (25). Paired analyses were applied to all crossover trials with the use of a conservative correlation coefficient of 0.5 in determining the missing variance data of crossover trials (29). For studies with multiple interventions (e.g., different doses of DHA or EPA) or controls, we did not combine groups but split the “shared” group into 2 or more groups with sample size divided between groups, and included 2 or more comparisons as proposed in the Cochrane Handbook to overcome unit-of-analysis error (18). The pooled estimates are expressed and presented as MDs with 95% CIs. Interstudy heterogeneity was assessed with the Cochran Q statistic and quantified by the I2 statistic, where I2 ≥50% at P <0.10 was considered as substantial heterogeneity. Sources of heterogeneity were explored through sensitivity analyses. To determine the particular influence of any single study on the results, each trial comparison was individually removed from the pairwise meta-analysis and the overall effect size and heterogeneity were recalculated. We did not perform a priori subgroup analyses nor investigated publication bias, as there were fewer than 10 direct comparisons available for the analyses of each outcome in the pairwise meta-analysis (25).

Network meta-analysis

Data for the network meta-analysis were managed and analyzed with the use of STATA/SE version 13 (StataCorp). We performed a frequentist network meta-analysis using a multivariate meta-analysis model and the “network” suite of commands available in STATA (30). For each inflammation outcome, postintervention values after each treatment in each study (DHA versus control and EPA versus control) were extracted and used to calculate MDs between DHA and EPA directly in STATA. Random-effects network meta-analysis was used for plasma CRP, IL-6, and TNF-α. We used the fixed option for adiponectin data, as there was no source of heterogeneity. Network diagrams were performed for each inflammation outcome to show the interactions among the studies included in the network meta-analysis and to illustrate the available direct comparisons between treatments (30). We checked the consistency in the data using both local and global approaches in STATA. We applied the loop-specific and the side-splitting approaches as local methods to evaluate the presence of statistical inconsistency. The loop-specific approach looks at the inconsistency in each closed loop in the network (31) whereas the side-splitting approach detects comparisons for which direct estimates disagree with indirect evidence from the entire network (32). The global approach tests for overall inconsistency from all possible sources in the whole network simultaneously using a design-by-treatment interaction model (33). If inconsistency was suggested, sensitivity analyses were performed to explore the sources of heterogeneity. Publication bias was assessed in the network meta-analysis through generation of comparison-adjusted funnel plots that were visually inspected for asymmetry, but only when 10 trial comparisons or more were available for a specific outcome. Box plots representing the distributions of the potential effect modifiers (transitivity analyses) were obtained using GraphPad Prism version 7.0c for Mac OS X (GraphPad Software).

Grading of the evidence

The certainty of evidence was assessed for each outcome independently by 2 authors (CV, BL) as “very low,” “low,” “moderate,” or “high” using the GRADE (Grading of Recommendations Assessment, Development, and Evaluation) approach. Evidence from RCTs received by default a grade of “high” certainty and was then downgraded on the basis of prespecified criteria related to risk of bias (weight of trials shows evidence of serious risk of bias by the Cochrane Risk of Bias Tool), inconsistency (unexplained substantial heterogeneity, I2 ≥50%, P <0.10), indirectness (presence of factors that limit the generalizability of the results), imprecision [95% CIs for pooled effect estimates cross a prespecified minimally important difference (MID)], and publication bias (evidence of small-study effects). The GRADE approach was also used to evaluate the certainty of evidence from the network meta-analysis according to the new procedure described by Brignardello-Petersen et al. (34). Briefly, this approach considers rating the certainty of all contributing evidence (direct, indirect, and network) for each pairwise comparison in the network meta-analysis. Intransitivity and incoherence were considered in addition to conventional criteria (risks of bias, inconsistency, imprecision, indirectness, and publication bias).

Results

Search results

The search strategy identified 3354 articles, of which 387 were reviewed in full and 20 were included in the final analysis (Figure 1) (10–14, 20–23, 35–45). Only 5 trials compared DHA and EPA directly with data on markers of inflammation as clinical outcomes (10–14). These were included in the pairwise meta-analysis. All 20 trials were included in the network meta-analysis combining the direct and indirect comparisons of DHA and EPA.

FIGURE 1.

FIGURE 1

Open in a new tab

Summary of the search and selection process of the studies comparing the effect of DHA and EPA on inflammation markers in adults.

Trial characteristics

Characteristics of RCTs included are summarized in Table 1. From the 20 RCTs, 17 were parallel-arm trials (10–13, 20–23, 36–42, 44, 45) and 3 had a crossover design (14, 35, 43). Of the 20 RCTs, 10 trials were conducted in North America (USA, Canada) (12–14, 20, 22, 23, 35, 39, 40, 45), 6 in Europe (Spain, UK) (21, 37, 38, 41, 43, 44), 2 in Oceania (Australia) (10, 36), and 2 in Asia (Iran) (11, 42). The duration of the RCTs ranged between 4 and 48 wk. Generally, participants were middle-aged and overweight. The age of study participants included in these studies ranged from 19.3 to 61.5 y. Baseline BMI was between 21.3 and 35 kg/m2. Eleven RCTs (55%) were conducted in healthy individuals (12, 13, 20, 21, 23, 37, 39–41, 43, 44). There was a relatively equal distribution of men and women across trials. Fourteen RCTs (70%) assessed the effect of DHA on inflammation outcomes (10–14, 20, 22, 23, 35, 37, 38, 42, 43, 45), 11 RCTs (55%) assessed the effect of EPA on inflammation outcomes (10–14, 21, 36, 39–41, 44), and 5 RCTs (25%) reported the effects of both DHA and EPA (10–14). One RCT did not indicate the given dose of DHA (37); for all other RCTs, the median dose was 2000 mg/d for DHA (range: 465–4000 mg/d) and 1869 mg/d for EPA (range 627–3840 mg/d). Of the 14 RCTs on DHA, 5 used DHA from algae (13, 20, 22, 23, 43), 4 from fish oil (10, 14, 42, 45), 1 from enzyme synthesis (38), and the remaining trials did not report the source of DHA. Of the 11 RCTs on EPA, 7 used EPA from fish oil (10, 14, 21, 36, 39, 40, 44), 1 from yeast (13), and the remaining trials did not report the information. Of the 9 RCTs that reported the form of DHA and EPA, 6 used a triglyceride formulation (13, 14, 21, 36, 38, 43), 2 provided the supplements in the form of ethyl esters (10, 41), and 1 used DHA as monoglycerides (45). Olive oil or oleic acid was the most frequently used control in the studies (>50% RCTs). CRP was investigated in 14 RCTs (70%) (10–14, 20, 22, 23, 35, 36, 38, 42–44), IL-6 was investigated in 17 RCTs (85%) (10, 13, 14, 20–23, 35–41, 43, 44), TNF-α was investigated in 11 RCTs (47%) (10, 13, 14, 22, 23, 36–41), and adiponectin was investigated in 2 RCTs (10%) (14, 35). Subclinical inflammation was specified as the primary outcome in 6 trials (30%) (11, 14, 22, 39, 40, 42). The majority of trials (85%) performed a per-protocol analysis of data; an intent-to-treat analysis of the results was used in only 3 trials (13, 20, 39). Nine RCTs received funding from public agencies (14, 22, 36–41, 44), 3 from private partners (20, 21, 43), 4 from both public and private sources (12, 13, 23, 35), 2 received no funding (11, 45), and the information was not reported for 2 RCTs (39, 42).

TABLE 1.

Characteristics of included randomized controlled trials comparing the effect of DHA and EPA on inflammation markers in adults

First author, year, arms Design1 Duration (weeks) Health conditions Participants: n randomized (% males) Drop out (%) Mean age (y)2 Mean BMI (kg/m2)2 Dose (mg/d) Source and form Inflammation as primary outcome Outcomes extracted Data analysis Funding source3 Country
Allaire, 2016 (14) Crossover 10 (WO: 9) Abdominal obesity, low-grade inflammation Total: 154 (31) 20 M: 57 F: 50 M: 30 F: 29 Yes CRP, IL-6, TNF-α, adiponectin Per-protocol Public Canada
 EPA 2700 FO, TG
 DHA 2700 FO, TG
 Control: corn oil 3000
Asztalos, 2016 (13) Parallel 6 Healthy Total: 121 9 No CRP, IL-6, TNF-α ITT Public-private USA
 EPA low dose 30 (67)4 10 52.8 27.4 627 Yeast, TG
 EPA high dose 31 (66)4 6 52.2 27.5 1869 Yeast, TG
 DHA 30 (68)4 7 52.3 27.0 758 Micro-algae, TG
 Control: olive oil 30 (69)4 13 52.2 27.7 2000
Azizi-Soleiman, 2013 (11) Parallel 12 T2D Total: 60 25 Yes CRP Per-protocol None Iran
 EPA 20 (50)4 30 51.4 26.9 980 NR
 DHA 20 (50)4 30 56.1 27.0 964 NR
 Control: canola oil 20 (47)4 15 56.9 28.4
Baril-Gravel, 2015 (35) Crossover 4 (WO: 2 to 4) Abdominal obesity Total: 151 (57)4 24 47.5 29.9 No CRP, IL-6, adiponectin Per-protocol Public-private Canada
 DHA 3500 NR
 Control I: canola
 Control II: oleic acid
Bradbury, 2017 (36) Parallel 12 Chronic work stress Total: 90 (29) 17 No CRP, IL-6, TNF-α Per-protocol Public Australia
 EPA 45 (NR) 16 43.24 25.3 2200 FO, TG
 Control: olive oil 45 (NR) 18 45.38 26.4 950
Capo, 2014 (37) Parallel 8 Healthy Total: 22 32 NR IL-6, TNF-α Per-protocol Public Spain
 DHA 11 (100) 18 20.4 23.5 Unspecified NR
 Control: olive oil 11 (100) 45 19.3 24.0 Unspecified
DiLorenzo, 2014 (20) Parallel 4 Healthy Total: 50 18 21.8 NR CRP, IL-6 ITT Private USA
 DHA 25 (100) 16 NR 2000 Algae, NR
 Control: corn oil 25 (100) 20 NR 2000
Domingo, 2018 (38) Parallel 48 HIV, high fasting TG Total: 39 (90)4 NA5 No CRP, IL-6, TNF-α Per-protocol Public Spain
 DHA 18 (83) 44 26.2 4000 Enzyme synthesis, TG
 Control: olive oil 21 (95) 45 25.1 7000
Gray, 2012 (21) Parallel 6 Healthy Total: 16 0 NR IL-6 Per-protocol Private UK
 EPA 8 (100) 0 24 24.2 1300 FO, TG
 Control: olive oil 8 (100) 0 24 23.5 3000
Kelley, 2009 (22) Parallel 13 Hyperlipidemic Total: 40 15 Yes CRP, IL-6, TNF-α Per-protocol Public USA
 DHA 20 (100) 15 55 27.8 3000 Microalgae, NR
 Control: olive oil 20 (100) 15 53.1 30.6 7500
Kiecolt-Glaser, 2011 (39) Parallel 12 Healthy Total: 68 1 Yes IL-6, TNF-α ITT NR USA
 EPA 34 (59) NR 23.9 NR 2085 FO, NR
 Control: oil mixture6 34 (53) NR 23.4 NR NR
Kiecolt-Glaser, 2012 (40) Parallel 16 Healthy, OW Total: 138 4 Yes IL-6, TNF-α Per-protocol Public USA
 EPA low dose 46 (39) 2 51.1 NR 1042.5 FO, NR
 EPA high dose 46 (37) 4 51.0 NR 2085 FO, NR
 Control: oil mixture6 46 (22) 4 51.1 NR 3000
Mori, 2003 (10) Parallel 6 T2D Total: 59 14 NR CRP, IL-6, TNF-α Per-protocol Public Australia
 EPA (82)3 NR 61.2 27.9 3840 FO, EE
 DHA (72)3 NR 60.9 30.6 3680 FO, EE
 Control: olive oil (75)3 NR 61.5 29.9 4000
Morin, 2018 (45) Parallel 4.3 Cystic fibrosis Total: 15 (NR) 27 NR IL-6 Per-protocol None Canada
 DHA NR NR 32.7 20.1 3200 FO, MAG
 Control: sunflower oil NR NR 24.4 20.0 5000
Neff, 2011 (23) Parallel 18 Healthy, OW Total: 49 27 No CRP, IL-6, TNF-α Per- protocol Public-private USA
 DHA (32)4 NR 43 35 2000 Algal oil, NR
 Control: corn-soybean oils (53)4 NR 44 34 5 mL
Shahbakhti, 2004 (41) Parallel 12 Healthy Total: 28 (43) 50 447 NR NR IL-6, TNF-α Per-protocol Public UK
 EPA 14 (NR) 50 NR NR 3800 NR, EE
 Control: oleic acid 14 (NR) 50 NR NR 4000
Shidfar, 2016 (42) Parallel 12 Iron deficiency anemia Total: 76 0 Yes CRP Per-protocol NR Iran
 DHA 38 (0) 0 33.03 21.30 465 FO, NR
 Control: corn oil 38 (0) 0 36.61 22.08 500
Theobald, 2007 (43) Crossover 12 (WO: 16) Healthy Total: 40 (50) 2.5 M: 51.1 F: 46.2 M: 23.6 F: 24.4 No CRP, IL-6 Per-protocol Private UK
 DHA NR NR 700 Algae, TG
 Control: olive oil NR NR 1500
Tsunoda, 2015 (12) Parallel 6 Healthy Total: 90 9 No CRP Per-protocol Public-private USA
 EPA 30 (87)4 10 53.6 27.5 1800 NR
 DHA 30 (72)4 13 51.6 27.8 1800 NR
 Control: olive oil 30 (69)4 3 49.7 28.1 6000
Yusof, 2008 (44) Parallel 8 Healthy Total: 21 5 NR CRP, IL-6 Per-protocol Public UK
 EPA 10 (100) 10 43.7 25.7 1800 FO, NR
 Control: coconut oil 11 (100) 0 44.7 26.5 2600
Open in a new tab
1

All trials were double-blind fashioned except for Capo et al. (37) and DiLorenzo et al. (20) where information was not reported.

2

Mean age and BMI of subjects who completed the study.

3

Public funding includes government, university sources, or not-for-profit health agencies; other sources are included in private funding.

4

In these studies, details on percent male were not provided according to the number of randomized subjects but to the number of subjects who completed the study or were included in the final analysis.

5

In this trial, 87 participants were randomized and 39 were included in the present inflammatory substudy (here dropout rate is not applicable).

6

Mix of palm, olive, soy, canola, and coco butter oils.

7

The median age was reported in this trial.

CRP, C-reactive protein; EE, ethyl ester; F, female; FO, fish oil; ITT, intent-to-treat; MAG, monoglyceride; M, male; NA, not applicable; NR, not reported; OW, overweight; T2D, type 2 diabetes; TG, triglyceride; WO, washout period.

Network diagrams

Figure 2 shows the network diagrams for plasma CRP, IL-6, TNF-α, and adiponectin of all available direct comparisons between pairs of interventions (DHA compared with EPA, DHA compared with control, and EPA compared with control) used in the network meta-analysis. The largest number of RCTs assessed the impact of DHA on plasma CRP and IL-6 compared with olive oil as a control (n = 8), whereas the same number of RCTs (n = 5) compared the effects of EPA and DHA to olive oil on plasma TNF-α. The contribution plots showing the percentage of statistical contribution coming from direct and indirect evidence for each direct comparison in the network for CRP, IL-6, and TNF-α are presented as Supplementary Figures 1, 2, and 3, respectively. The NMA estimate for the comparison of interest DHA compared with EPA is informed by the direct comparison DHA versus EPA with a contribution of around 40% for CRP and IL-6 compared with 70% for TNF-α.

FIGURE 2.

FIGURE 2

Open in a new tab

Network diagrams from network meta-analysis comparing the effect of DHA and EPA in adults on plasma CRP (Panel A), IL-6 (Panel B), TNF-α (Panel C), and adiponectin (Panel D). The size of the nodes is proportional to the number of participants and the thickness of the lines is proportional to the number of studies available for a particular comparison. CRP: C-reactive protein; oil mixture: mix of palm, olive, soy, canola, and coco butter oils.

Risk of bias

Supplementary Figure 4 shows the individual Cochrane risk of bias assessments for all 20 RCTs included in this work. Of the 20 RCTs included in the network meta-analysis, 11 (55%) were judged to be at low risk of bias (12–14, 35, 36, 38–40, 42–44), 9 had a moderate/unclear risk of bias (10, 11, 20–23, 37, 41, 45), and none were classified as being at high risk of bias. Of the 5 RCTs included in the pairwise meta-analysis, 3 trials were judged to be at low risk of bias (12–14) and the others were classified as having a moderate/unclear risk of bias (10, 11).

Supplementary Figures 5 and 6 show the risk of bias proportions for RCTs included specifically in the pairwise meta-analysis and the network meta-analysis, respectively. Among RCTs included in the pairwise meta-analysis, 100% indicated a low risk of bias for blinding selective reporting and other bias, 0% for allocation concealment, 75% for incomplete outcome data, and 50% for random sequence generation (Supplementary Figure 5). With regard to the network meta-analysis, 35% of all RCTs indicated a low risk of bias for random-sequence generation, 30% for allocation concealment, 90% for blinding, 55% for incomplete outcome data, 65% for selective reporting, and 100% for “other bias” (Supplementary Figure 6). High risks of bias were found for incomplete outcome data and for selective outcome reporting for 2 distinct RCTs (41, 45) (Supplementary Figure 6).

Transitivity analyses showed minor differences among studies for subjects’ characteristics (i.e., BMI, age, percent male) and RCTs’ characteristics (i.e., study duration, DHA dose, EPA dose) (Supplementary Figure 7).

Inconsistency

For comparisons in the network meta-analysis, the side-splitting approach suggested no significant inconsistency for plasma CRP, IL-6, and TNF-α (Supplementary Tables 2, 3, and 4). The loop-specific approach identified 3 loops for CRP, IL-6, and TNF-α without statistical inconsistency (Supplementary Table 5). Due to the lack of available data on adiponectin specifically, we were not able to conduct the loop-specific nor the side-splitting approaches for this outcome. The design-by-treatment model showed no significant inconsistency for plasma CRP (P = 0.79), IL-6 (P = 0.93), TNF-α (P = 0.85), and adiponectin (P = 0.98).

Inflammation outcomes

Figure 3 summarizes the pooled estimates of plasma CRP, IL-6, TNF-α, and adiponectin for the comparison of DHA compared with EPA from both pairwise and network meta-analyses. The main results of the pairwise meta-analysis are presented as forest plots for CRP, IL-6, and TNF-α in Supplementary Figures 8, 9, and 10. The main results of the network meta-analysis are presented as interval plots for CRP, IL-6, TNF-α, and adiponectin in Supplementary Figures 11–14. Sensitivity analyses were not performed in the network meta-analysis as no inconsistency was observed for all comparisons.

FIGURE 3.

FIGURE 3

Open in a new tab

Summary plot of pooled effect estimates from the pairwise and network meta-analyses comparing the effect of DHA and EPA on inflammation outcomes in adults. Pooled effect estimates from the pairwise meta-analysis are expressed as standardized mean differences, represented by diamonds and 95% CIs by the line through the diamond, and were estimated with the use of a generic inverse variance random-effect model. Interstudy heterogeneity was detected with the use of the Cochran's Q statistic and quantified with the use of the I2 statistic. Mean differences (MDs, 95% CI) for the inflammation outcomes as estimated from the network meta-analysis for the comparison of DHA compared with EPA are also presented. CRP: C-reactive protein; MD, mean difference; SMD, standardized mean difference.

Plasma CRP

Pairwise meta-analysis of direct comparisons of the effects of DHA and EPA revealed no significant difference on plasma CRP (MDDHA versus EPA = 0.14 mg/L; 95% CI: –0.57, 0.85 mg/L; P = 0.70), with evidence of significant heterogeneity (I2 = 61%, PHeterogeneity = 0.02). Removal of Mori et al. 2003 (10) explained all of the heterogeneity [I2 = 0%, PHeterogeneity = 0.82; new MDDHA versus EPA = –0.26 mg/L (95% CI: –0.54, 0.02 mg/L), P = 0.07; Supplementary Table 6]. Network meta-analysis of all direct and indirect comparisons of DHA and EPA also suggested comparable effects of the 2 modalities on plasma CRP [MDDHA versus EPA = –0.33 mg/L (–0.75, 0.10 mg/L); –12.8%].

Plasma IL-6

Pairwise meta-analysis of direct comparisons of DHA and EPA showed no significant difference in their effects on plasma IL-6 (MDDHA versus EPA = 0.10 pg/mL; 95% CI: –0.15, 0.34 pg/mL; P = 0.44) without evidence of heterogeneity (I2 = 40%, PHeterogeneity = 0.17). Network meta-analysis of all direct and indirect comparisons of DHA and EPA also suggested similar effects on plasma IL-6 [MDDHA versus EPA = 0.09 pg/mL (–0.12, 0.30 pg/mL); +4.6%].

Plasma TNF-α

Pairwise meta-analysis of direct comparisons of the effects of DHA and EPA revealed no significant difference on plasma TNF-α (MDDHA versus EPA = –0.10 pg/mL; 95% CI: –0.37, 0.18 pg/mL; P = 0.49) without evidence of heterogeneity (I2 = 40%, PHeterogeneity = 0.17). Similar results were obtained in the network meta-analysis [MDDHA versus EPA for TNF-α = –0.02 pg/mL (–0.25, 0.20 pg/mL); –0.8%].

Plasma adiponectin

The only trial that directly compared the effects of DHA and EPA on plasma adiponectin showed a greater increase with DHA than with EPA (MDDHA versus EPA = 0.30 mg/L; 95% CI: 0.10, 0.50 mg/L; P = 0.003). However, DHA and EPA had similar effects on plasma adiponectin [MDDHA versus EPA = 0.29 pg/mL (–1.08, 1.66 mg/L); +4.1%] according to the network meta-analysis.

Small-study effects

Supplementary Figure 15 shows the comparison-adjusted funnel plots including all studies of the network meta-analysis for plasma CRP, IL-6, and TNF-α. There was no visual evidence of funnel-plot asymmetry for CRP and TNF-α (Supplementary Figure 15A–C). The funnel plot for IL-6 appears to be slightly asymmetric (Supplementary Figure 15B), due mainly to 1 comparison (DHA versus olive oil). Publication bias could not be assessed for adiponectin in the network meta-analysis due to the limited number of RCTs available (<10). Likewise, publication bias was not assessed in the pairwise meta-analysis, as all outcomes had fewer than 10 trial comparisons available for analyses.

GRADE assessment

The certainty of evidence assessment for the pairwise meta-analysis directly comparing the effect of DHA and EPA on markers of systemic inflammation is shown in Supplementary Table 7. The certainty of evidence was rated high for IL-6 and TNF-α and moderate for CRP owing to downgrades for serious imprecision. With regard to network meta-analysis, the certainty of network evidence for CRP (Supplementary Table 8), IL-6 (Supplementary Table 9), and TNF-α (Supplementary Table 10) was rated as moderate for most of the pairwise comparisons. Serious imprecision for most of the comparisons drove judgments of moderate quality of evidence. There was no incoherence between direct and indirect evidence for the comparison of interest DHA versus EPA regarding plasma CRP and IL-6. Regarding TNF-α, incoherence was observed for the comparison DHA versus EPA but the direct evidence was of moderate quality and contributed more than the indirect evidence to the network evidence. It was not possible to generate the direct and indirect estimates with the side-splitting approach due to lack of available data. Also, publication bias could not be assessed due to lack of power for assessing funnel plot asymmetry and small-study effects (<10 trial comparisons). Putting aside those concerns, the certainty of evidence was not downgraded with respect to study limitations, indirectness, or inconsistency. The certainty of estimates comparing the effects of DHA and EPA on plasma adiponectin was considered low only due to very serious imprecision.

Discussion

The present work is the first to quantitatively evaluate the differential effect of EPA and DHA on 4 inflammation markers using both pairwise and network meta-analysis methodologies. This work is important because these inflammation-related cardiometabolic risk factors have all been associated etiologically to the risk of CVD. Results suggest that DHA and EPA have similar effects on CRP, IL-6, TNF-α, and adiponectin. Assessment of overall data heterogeneity and quality reveals no major limitations in the interpretation of such results.

The few meta-analyses that have assessed the impact of LCn–3PUFAs in general on markers of inflammation have yielded inconsistent results. Although some meta-analyses of RCTs have shown no effect of LCn–3PUFAs on plasma CRP concentrations (46, 47), others have shown an anti-inflammatory effect of LCn–3PUFA on plasma CRP (9, 48) and IL-6 and TNF-α concentrations (9). As emphasized earlier, most of the RCTs included in these meta-analyses have either used a mix of EPA and DHA in various forms and proportions or have investigated only 1 of the 2 LCn–3PUFAs. The assumption that DHA and EPA trigger different anti-inflammatory responses has been proposed based on data from several in vitro studies (49, 50), but limited clinical evidence supports this claim.

Innes and Calder recently published a systematic review on the differential effects of DHA and EPA on cardiometabolic risk factors (15). Their qualitative analysis of the available data from RCTs revealed inconsistent results regarding the potentially greater anti-inflammatory effects of DHA compared with EPA. Data from these various RCTs were not quantitatively analyzed in this authoritative review. A recent meta-analysis of 20 RCTs was conducted to assess the individual effects of EPA and DHA on blood pressure and inflammatory factors (51). This was, to the best of our knowledge, the first meta-analysis on the topic but issues pertaining to the methodological approach used by the authors need to be addressed. The authors concluded, based on their analysis of the available data, that EPA and DHA have similar effects on CRP concentration, whereas limited evidence on circulating IL-6 and TNF-α did not allow them to draw conclusions. However, they assessed the effects of DHA and of EPA independent of each other and compared summary estimates of the effect of DHA to that of the effect of EPA to assess whether both LCn–3PUFAs have similar or different effects on subclinical inflammation markers. Comparing and drawing inferences about comparative effectiveness from different summary estimates in studies that have not directly compared against each other is of questionable validity and has been strongly discouraged (18). To assess the question of whether EPA and DHA have similar or differential effects on inflammatory factors, it is best to either use studies that have compared these interventions directly, or to use a network meta-analysis approach, through which 2 interventions can be compared indirectly via a common comparator. Here, we were able to fill this gap by comparing the effect of EPA and DHA on markers of subclinical inflammation using all data available and a rigorous methodological approach. The calculation of summary estimates in our study also took into account data from the control treatment in each RCT, which make the results more rigorous. We stress that the present study does not address whether DHA or EPA have a significant anti-inflammatory effect on their own.

The combined approach of pairwise and network methodologies is a strength of the present study, as are the inclusion of a larger number of studies using the network approach, the inclusion of 4 outcomes (CRP, IL-6, TNF-α, and adiponectin), the risk of bias assessment including transitivity analyses for the network meta-analysis, heterogeneity and inconsistency testing, sensitivity analyses, and judgment of the overall certainty of evidence. Although the evidence from pairwise meta-analysis is limited, the network approach included data from 20 RCTs, thus yielding a much larger sample size and stronger statistical power for the pooled effects being compared. A majority of the trials (11 out of 20) were rated as having a low risk of bias. Relying on RCTs lowered the risk of selection bias due to randomization, and all studies had a double-blind design, thereby decreasing the risk of performance bias. No inconsistency was observed in the network meta-analysis and the only substantial heterogeneity observed for CRP in the pairwise analysis was explained by the removal of 1 study, indicating the overall stability and homogeneity of the results. Accordingly, transitivity analyses did not reveal important effect modifiers, indicating relative stability across RCTs. The certainty of evidence was rated mainly high or moderate in both meta-analyses for plasma CRP, IL-6, and TNF-α. Only a small number of RCTs reported the effect of DHA and EPA on adiponectin concentrations, and results on that outcome need to be analyzed with caution because of the high degree of imprecision. Although our search strategy may have missed studies due to exclusion of those reported in languages other than English or French, this risk is low as we have used a highly rigorous search process as proposed in meta-analysis guidelines. Most of the RCTs included in both meta-analyses did not report information about allocation concealment, which seems to be often confounded with blinding among studies. Despite that, the majority of studies were graded as being at “low risk” of bias. Different doses (≤ ≈4g/d) and durations (4 to 48 wk) of EPA and DHA supplementation were used in the RCTs included in the present review. Half of the included RCTs were conducted among healthy individuals whereas the other half included participants with metabolic disorders or health problems. The possibility that dose, study duration, and health conditions of participants may have modified the response of inflammation markers to DHA and EPA cannot be excluded but the limited number of RCTs and participants included in this review did not allow us to conduct rigorous sensitivity analyses based on these factors.

The present work suggests that DHA and EPA have comparable effects on markers of systemic inflammation. From a prevention perspective, EPA and DHA have always been considered to have comparable health benefits and dosing recommendations usually describe a combination of the 2, without differentiation. There is an increasing range of supplements available to help individuals with a typically low fish intake achieve the recommended dietary intake in LCn–3PUFAs. A large proportion of the existing research has thus been conducted using mixed LCn–3PUFAs due to the availability of combined EPA and DHA products. However, the development of separation and purification technologies in recent years has allowed research on the biological effects of purified EPA and DHA. To date, 2 major large intervention trials have been conducted to study the cardioprotective effects of EPA supplements provided as highly purified ethyl esters (52, 53): the Japan EPA lipid intervention study (JELIS) and the Reduction of Cardiovascular Events with Icosapent Ethyl—Intervention Trial (REDUCE-IT). JELIS included 18,645 patients with a plasma cholesterol ≥6.5 mmol/L randomly assigned to receive either 1.8 g/d of EPA daily with statin or statin alone for 5 y. REDUCE-IT included 8179 patients on statin therapy randomly assigned to receive daily either 4 g of EPA or placebo (mineral oil) for 5 y. Both studies showed significant 20% to 30% reductions in the risk of major coronary events (52) and in cardiovascular death, heart attacks, and stroke (53). Interestingly, the observed cardiovascular risk reduction with EPA supplementation was not explained by changes in plasma lipid concentrations in both studies, suggesting benefits on other key processes involved in atherogenesis, including inflammation. Future research should examine how supplementing high-risk populations with DHA alone modifies the risk of CVD.

Supplementary Material

nmaa086_Supplemental_File
Click here for additional data file. (6.9MB, docx)

ACKNOWLEDGEMENTS

The authors’ responsibilities were as follows—CV and BL designed the search strategy; CV performed the search strategy, study selection, data extraction, quality assessment, statistical analyses, and drafted the manuscript; JA and SBM were involved in the search strategy and study selection; TAK participated in the interpretation and analysis of the data obtained from the network meta-analysis; SBM, TAK, and JLS reviewed the final content of the manuscript prior to submission and provided feedback; BL had primary responsibility for the final content of the manuscript; and all authors: read and approved the final manuscript.

Notes

CV is a postdoctoral fellow supported by the European Marie Skłodowska-Curie Actions. JA is a recipient of PhD scholarships from the Canadian Institutes of Health Research (CIHR) and the Fonds de recherche du Québec—Santé (FRQ-S). This work was supported by the Canadian Institutes of Health Research (funding reference number, 129920) through the Canada-wide Human Nutrition Trialists’ Network (NTN). The Diet, Digestive tract, and Disease (3D) Centre, funded through the Canada Foundation for Innovation (CFI) and the Ministry of Research and Innovation's Ontario Research Fund (ORF), provided the infrastructure for the conduct of this project. JLS was funded by a PSI Graham Farquharson Knowledge Translation Fellowship, Diabetes Canada Clinician Scientist Award, CIHR INMD/CNS New Investigator Partnership Prize, and Banting and Best Diabetes Centre Sun Life Financial New Investigator Award. None of the sponsors had a role in any aspect of the present study, including design and conduct of the study; collection, management, analysis, and interpretation of the data; and preparation, review, approval of the manuscript, or decision to publish.

Author disclosures: TAK has received research support from the CIHR and an unrestricted travel donation from Bee Maid Honey Ltd. He was an invited speaker at a Calorie Control Council annual general meeting for which he received an honorarium. JLS has received research support from the Canadian Foundation for Innovation, Ontario Research Fund, Province of Ontario Ministry of Research and Innovation and Science, CIHR, Diabetes Canada, PSI Foundation, Banting and Best Diabetes Centre (BBDC), ASN, International Nut and Dried Fruit Council Foundation (INC), National Dried Fruit Trade Association, The Tate and Lyle Nutritional Research Fund at the University of Toronto, The Glycemic Control and Cardiovascular Disease in Type 2 Diabetes Fund at the University of Toronto (a fund established by the Alberta Pulse Growers), and the Nutrition Trialists Fund at the University of Toronto (a fund established by the Calorie Control Council). He has received in-kind food donations to support a randomized controlled trial from the Almond Board of California, California Walnut Commission, American Peanut Council, Barilla, Unilever, Unico/Primo, Loblaw Companies, Quaker (Pepsico), Kellogg Canada, and WhiteWave Foods. He has received travel support, speaker fees, and/or honoraria from Diabetes Canada, Mott's LLP, Dairy Farmers of Canada, FoodMinds LLC, PepsiCo, The Ginger Network LLC, International Sweeteners Association, Nestlé, Pulse Canada, Canadian Society for Endocrinology and Metabolism (CSEM), GI Foundation, Abbott, Biofortis, ASN, Health Sciences North, Inc. Nutrition Research & Education Foundation, and Physicians Committee for Responsible Medicine. He has or has had ad hoc consulting arrangements with Perkins Coie LLP, Tate & Lyle, and Wirtschaftliche Vereinigung Zucker e.V. He is a member of the European Fruit Juice Association Scientific Expert Panel. He is on the Clinical Practice Guidelines Expert Committees of Diabetes Canada, European Association for the study of Diabetes (EASD), Canadian Cardiovascular Society (CCS), and Obesity Canada. He serves as an unpaid scientific advisor for the Food, Nutrition, and Safety Program (FNSP) and the Technical Committee on Carbohydrates of the International Life Science Institute (ILSI) North America. He is a member of the International Carbohydrate Quality Consortium (ICQC), Executive Board Member of the Diabetes and Nutrition Study Group (DNSG) of the EASD, and Director of the Toronto 3D Knowledge Synthesis and Clinical Trials foundation. His wife is a former employee of Unilever Canada. BL has received funding in the last 2 y from the CIHR and Atrium Innovations for his work on LCn–3PUFA. All other authors report no conflicts of interest.

Supplemental Tables 1–10 and Supplemental Figures 1–15 are available from the “Supplementary data” link in the online posting of the article and from the same link in the online table of contents at https://academic.oup.com/advances.

Abbreviations used: CRP, C-reactive protein; CVD, cardiovascular disease; GRADE, Grading of Recommendations Assessment, Development, and Evaluation; JELIS, Japan EPA lipid intervention study; LCn–3PUFA, long-chain ω-3 PUFA; MD, mean difference; RCT, randomized controlled trial; REDUCE-IT, Reduction of Cardiovascular Events with Icosapent Ethyl – Intervention Trial.

Contributor Information

Cécile Vors, Centre Nutrition, Santé et Société (NUTRISS), Institut sur la Nutrition et les Aliments Fonctionnels (INAF), Laval University, Quebec City, Canada.

Janie Allaire, Centre Nutrition, Santé et Société (NUTRISS), Institut sur la Nutrition et les Aliments Fonctionnels (INAF), Laval University, Quebec City, Canada.

Sonia Blanco Mejia, Department of Nutritional Sciences, Faculty of Medicine, University of Toronto, Toronto, Canada; Toronto 3D Knowledge Synthesis and Clinical Trials Unit, Risk Factor Modification Center, St. Michael's Hospital, Toronto, Canada.

Tauseef A Khan, Department of Nutritional Sciences, Faculty of Medicine, University of Toronto, Toronto, Canada; Toronto 3D Knowledge Synthesis and Clinical Trials Unit, Risk Factor Modification Center, St. Michael's Hospital, Toronto, Canada.

John L Sievenpiper, Department of Nutritional Sciences, Faculty of Medicine, University of Toronto, Toronto, Canada; Toronto 3D Knowledge Synthesis and Clinical Trials Unit, Risk Factor Modification Center, St. Michael's Hospital, Toronto, Canada; Li Ka Shing Knowledge Institute, St. Michael's Hospital, Toronto, Canada; Division of Endocrinology and Metabolism, St. Michael's Hospital, Toronto, Canada; Department of Medicine, Faculty of Medicine, University of Toronto, Toronto, Canada.

Benoît Lamarche, Centre Nutrition, Santé et Société (NUTRISS), Institut sur la Nutrition et les Aliments Fonctionnels (INAF), Laval University, Quebec City, Canada.

References

  • 1. Rocha VZ, Libby P. Obesity, inflammation, and atherosclerosis. Nat Rev Cardiol. 2009;6(6):399–409. [DOI] [PubMed] [Google Scholar]
  • 2. Libby P Inflammation in atherosclerosis. Nature. 2002;420(6917):868–74. [DOI] [PubMed] [Google Scholar]
  • 3. Ridker PM, Everett BM, Thuren T, MacFadyen JG, Chang WH, Ballantyne C, Fonseca F, Nicolau J, Koenig W, Anker SDet al. Antiinflammatory therapy with canakinumab for atherosclerotic disease. N Engl J Med. 2017;377(12):1119–31. [DOI] [PubMed] [Google Scholar]
  • 4. Trayhurn P, Wood IS. Adipokines: inflammation and the pleiotropic role of white adipose tissue. Br J Nutr. 2004;92(3):347–55. [DOI] [PubMed] [Google Scholar]
  • 5. Ridker PM From C-reactive protein to interleukin-6 to interleukin-1: moving upstream to identify novel targets for atheroprotection. Circ Res. 2016;118(1):145–56. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 6. Calder PC, Ahluwalia N, Albers R, Bosco N, Bourdet-Sicard R, Haller D, Holgate ST, Jonsson LS, Latulippe ME, Marcos Aet al. A consideration of biomarkers to be used for evaluation of inflammation in human nutritional studies. Br J Nutr. 2013;109(Suppl 1):S1–34. [DOI] [PubMed] [Google Scholar]
  • 7. Calder PC n-3 polyunsaturated fatty acids, inflammation, and inflammatory diseases. Am J Clin Nutr. 2006;83(6 Suppl):1505S–19S. [DOI] [PubMed] [Google Scholar]
  • 8. Calder PC Very long-chain n-3 fatty acids and human health: fact, fiction and the future. Proc Nutr Soc. 2018;77(1):52–72. [DOI] [PubMed] [Google Scholar]
  • 9. Li K, Huang T, Zheng J, Wu K, Li D. Effect of marine-derived n-3 polyunsaturated fatty acids on C-reactive protein, interleukin 6 and tumor necrosis factor alpha: a meta-analysis. PLoS One. 2014;9(2):e88103. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 10. Mori TA, Woodman RJ, Burke V, Puddey IB, Croft KD, Beilin LJ. Effect of eicosapentaenoic acid and docosahexaenoic acid on oxidative stress and inflammatory markers in treated-hypertensive type 2 diabetic subjects. Free Radic Biol Med. 2003;35(7):772–81. [DOI] [PubMed] [Google Scholar]
  • 11. Azizi-Soleiman F, Jazayeri S, Eghtesadi S, Rajab A, Heidari I, Vafa MR, Gohari MR. Effects of pure eicosapentaenoic and docosahexaenoic acids on oxidative stress, inflammation and body fat mass in patients with type 2 diabetes. Int J Prev Med. 2013;4(8):922–8. [PMC free article] [PubMed] [Google Scholar]
  • 12. Tsunoda F, Lamon-Fava S, Asztalos BF, Iyer LK, Richardson K, Schaefer EJ. Effects of oral eicosapentaenoic acid versus docosahexaenoic acid on human peripheral blood mononuclear cell gene expression. Atherosclerosis. 2015;241(2):400–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 13. Asztalos IB, Gleason JA, Sever S, Gedik R, Asztalos BF, Horvath KV, Dansinger ML, Lamon-Fava S, Schaefer EJ. Effects of eicosapentaenoic acid and docosahexaenoic acid on cardiovascular disease risk factors: a randomized clinical trial. Metabolism. 2016;65(11):1636–45. [DOI] [PubMed] [Google Scholar]
  • 14. Allaire J, Couture P, Leclerc M, Charest A, Marin J, Lepine MC, Talbot D, Tchernof A, Lamarche B. A randomized, crossover, head-to-head comparison of eicosapentaenoic acid and docosahexaenoic acid supplementation to reduce inflammation markers in men and women: the Comparing EPA to DHA (ComparED) Study. Am J Clin Nutr. 2016;104(2):280–7. [DOI] [PubMed] [Google Scholar]
  • 15. Innes JK, Calder PC. The differential effects of eicosapentaenoic acid and docosahexaenoic acid on cardiometabolic risk factors: a systematic review. Int J Mol Sci. 2018;19(2):532. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 16. Mills EJ, Thorlund K, Ioannidis JP. Demystifying trial networks and network meta-analysis. BMJ. 2013;346:f2914. [DOI] [PubMed] [Google Scholar]
  • 17. Salanti G, Higgins JP, Ades AE, Ioannidis JP. Evaluation of networks of randomized trials. Stat Methods Med Res. 2008;17(3):279–301. [DOI] [PubMed] [Google Scholar]
  • 18. Higgins JPT, Green S, Cochrane Handbook for Systematic Reviews of Interventions. The Cochrane Collaboration; 2011. [Google Scholar]
  • 19. Hutton B, Salanti G, Caldwell DM, Chaimani A, Schmid CH, Cameron C, Ioannidis JP, Straus S, Thorlund K, Jansen JPet al. The PRISMA extension statement for reporting of systematic reviews incorporating network meta-analyses of health care interventions: checklist and explanations. Ann Intern Med. 2015;162(11):777–84. [DOI] [PubMed] [Google Scholar]
  • 20. DiLorenzo FM, Drager CJ, Rankin JW. Docosahexaenoic acid affects markers of inflammation and muscle damage after eccentric exercise. J Strength Cond Res. 2014;28(10):2768–74. [DOI] [PubMed] [Google Scholar]
  • 21. Gray P, Gabriel B, Thies F, Gray SR. Fish oil supplementation augments post-exercise immune function in young males. Brain Behav Immun. 2012;26(8):1265–72. [DOI] [PubMed] [Google Scholar]
  • 22. Kelley DS, Siegel D, Fedor DM, Adkins Y, Mackey BE. DHA supplementation decreases serum C-reactive protein and other markers of inflammation in hypertriglyceridemic men. J Nutr. 2009;139(3):495–501. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 23. Neff LM, Culiner J, Cunningham-Rundles S, Seidman C, Meehan D, Maturi J, Wittkowski KM, Levine B, Breslow JL. Algal docosahexaenoic acid affects plasma lipoprotein particle size distribution in overweight and obese adults. J Nutr. 2011;141(2):207–13. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 24. Schwingshackl L, Chaimani A, Schwedhelm C, Toledo E, Punsch M, Hoffmann G, Boeing H. Comparative effects of different dietary approaches on blood pressure in hypertensive and pre-hypertensive patients: a systematic review and network meta-analysis. Crit Rev Food Sci Nutr. 2019;59(16):2674–87. [DOI] [PubMed] [Google Scholar]
  • 25. Moher D, Liberati A, Tetzlaff J, Altman DG, PRISMA Group Preferred reporting items for systematic reviews and meta-analyses: the PRISMA statement. J Clin Epidemiol. 2009;62(10):1006–12. [DOI] [PubMed] [Google Scholar]
  • 26. DerSimonian R, Laird N. Meta-analysis in clinical trials. Control Clin Trials. 1986;7(3):177–88. [DOI] [PubMed] [Google Scholar]
  • 27. Luo D, Wan X, Liu J, Tong T. Optimally estimating the sample mean from the sample size, median, mid-range, and/or mid-quartile range. Stat Methods Med Res. 2018;27(6):1785–805. [DOI] [PubMed] [Google Scholar]
  • 28. Wan X, Wang W, Liu J, Tong T. Estimating the sample mean and standard deviation from the sample size, median, range and/or interquartile range. BMC Med Res Methodol. 2014;14:135. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 29. Elbourne DR, Altman DG, Higgins JP, Curtin F, Worthington HV, Vail A. Meta-analyses involving cross-over trials: methodological issues. Int J Epidemiol. 2002;31(1):140–9. [DOI] [PubMed] [Google Scholar]
  • 30. Chaimani A, Higgins JP, Mavridis D, Spyridonos P, Salanti G. Graphical tools for network meta-analysis in STATA. PLoS One. 2013;8(10):e76654. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 31. Bucher HC, Guyatt GH, Griffith LE, Walter SD. The results of direct and indirect treatment comparisons in meta-analysis of randomized controlled trials. J Clin Epidemiol. 1997;50(6):683–91. [DOI] [PubMed] [Google Scholar]
  • 32. Dias S, Welton NJ, Caldwell DM, Ades AE. Checking consistency in mixed treatment comparison meta-analysis. Statist Med. 2010;29(7–8):932–44. [DOI] [PubMed] [Google Scholar]
  • 33. Jackson D, Barrett JK, Rice S, White IR, Higgins JP. A design-by-treatment interaction model for network meta-analysis with random inconsistency effects. Statist Med. 2014;33(21):3639–54. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 34. Brignardello-Petersen R, Bonner A, Alexander PE, Siemieniuk RA, Furukawa TA, Rochwerg B, Hazlewood GS, Alhazzani W, Mustafa RA, Murad MHet al. Advances in the GRADE approach to rate the certainty in estimates from a network meta-analysis. J Clin Epidemiol. 2018;93:36–44. [DOI] [PubMed] [Google Scholar]
  • 35. Baril-Gravel L, Labonte ME, Couture P, Vohl MC, Charest A, Guay V, Jenkins DA, Connelly PW, West S, Kris-Etherton PMet al. Docosahexaenoic acid-enriched canola oil increases adiponectin concentrations: a randomized crossover controlled intervention trial. Nutr Metab Cardiovasc Dis. 2015;25(1):52–9. [DOI] [PubMed] [Google Scholar]
  • 36. Bradbury J, Myers SP, Meyer B, Brooks L, Peake J, Sinclair AJ, Stough C. Chronic psychological stress was not ameliorated by omega-3 eicosapentaenoic acid (EPA). Front Pharmacol. 2017;8:551. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 37. Capo X, Martorell M, Llompart I, Sureda A, Tur JA, Pons A. Docosahexanoic acid diet supplementation attenuates the peripheral mononuclear cell inflammatory response to exercise following LPS activation. Cytokine. 2014;69(2):155–64. [DOI] [PubMed] [Google Scholar]
  • 38. Domingo P, Gallego-Escuredo JM, Fernandez I, Villarroya J, Torres F, Del Mar Gutierrez M, Mateo MG, Villarroya F, Vidal F, Giralt Met al. Effects of docosahexanoic acid supplementation on inflammatory and subcutaneous adipose tissue gene expression in HIV-infected patients on combination antiretroviral therapy (cART). A sub-study of a randomized, double-blind, placebo-controlled study. Cytokine. 2018;105:73–9. [DOI] [PubMed] [Google Scholar]
  • 39. Kiecolt-Glaser JK, Belury MA, Andridge R, Malarkey WB, Glaser R. Omega-3 supplementation lowers inflammation and anxiety in medical students: a randomized controlled trial. Brain Behav Immun. 2011;25(8):1725–34. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 40. Kiecolt-Glaser JK, Belury MA, Andridge R, Malarkey WB, Hwang BS, Glaser R. Omega-3 supplementation lowers inflammation in healthy middle-aged and older adults: a randomized controlled trial. Brain Behav Immun. 2012;26(6):988–95. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 41. Shahbakhti H, Watson RE, Azurdia RM, Ferreira CZ, Garmyn M, Rhodes LE. Influence of eicosapentaenoic acid, an omega-3 fatty acid, on ultraviolet-B generation of prostaglandin-E2 and proinflammatory cytokines interleukin-1 beta, tumor necrosis factor-alpha, interleukin-6 and interleukin-8 in human skin in vivo. Photochem Photobiol. 2004;80(2):231–5. [DOI] [PubMed] [Google Scholar]
  • 42. Shidfar F, Amani S, Vafa M, Shekarriz R, Hosseini S, Shidfar S, Eshraghian M, Mousavi SN. Effects of iron supplementation with and without docosahexaenoic acid on the cardiovascular disease risk based on paraoxonase-1, hs-CRP, and ApoB/ApoA-I ratio in women with iron deficiency anemia. Biol Trace Elem Res. 2016;169(1):34–40. [DOI] [PubMed] [Google Scholar]
  • 43. Theobald HE, Goodall AH, Sattar N, Talbot DC, Chowienczyk PJ, Sanders TA. Low-dose docosahexaenoic acid lowers diastolic blood pressure in middle-aged men and women. J Nutr. 2007;137(4):973–8. [DOI] [PubMed] [Google Scholar]
  • 44. Yusof HM, Miles EA, Calder P. Influence of very long-chain n-3 fatty acids on plasma markers of inflammation in middle-aged men. Prostaglandins Leukotrienes Essent Fatty Acids. 2008;78(3):219–28. [DOI] [PubMed] [Google Scholar]
  • 45. Morin C, Cantin AM, Vezina FA, Fortin S. The efficacy of MAG-DHA for correcting AA/DHA imbalance of cystic fibrosis patients. Mar Drugs. 2018;16(6):184. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 46. Balk EM, Lichtenstein AH, Chung M, Kupelnick B, Chew P, Lau J. Effects of omega-3 fatty acids on serum markers of cardiovascular disease risk: a systematic review. Atherosclerosis. 2006;189(1):19–30. [DOI] [PubMed] [Google Scholar]
  • 47. Xin W, Wei W, Li X. Effects of fish oil supplementation on inflammatory markers in chronic heart failure: a meta-analysis of randomized controlled trials. BMC Cardiovasc Disord. 2012;12:77. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 48. AbuMweis S, Jew S, Tayyem R, Agraib L. Eicosapentaenoic acid and docosahexaenoic acid containing supplements modulate risk factors for cardiovascular disease: a meta-analysis of randomised placebo-control human clinical trials. J Hum Nutr Diet. 2018;31(1):67–84. [DOI] [PubMed] [Google Scholar]
  • 49. Weldon SM, Mullen AC, Loscher CE, Hurley LA, Roche HM. Docosahexaenoic acid induces an anti-inflammatory profile in lipopolysaccharide-stimulated human THP-1 macrophages more effectively than eicosapentaenoic acid. J Nutr Biochem. 2007;18(4):250–8. [DOI] [PubMed] [Google Scholar]
  • 50. De Caterina R, Liao JK, Libby P. Fatty acid modulation of endothelial activation. Am J Clin Nutr. 2000;71(1 Suppl):213S–23S. [DOI] [PubMed] [Google Scholar]
  • 51. Guo XF, Li KL, Li JM, Li D. Effects of EPA and DHA on blood pressure and inflammatory factors: a meta-analysis of randomized controlled trials. Crit Rev Food Sci Nutr. 2019;59:3380. [DOI] [PubMed] [Google Scholar]
  • 52. Yokoyama M, Origasa H, Matsuzaki M, Matsuzawa Y, Saito Y, Ishikawa Y, Oikawa S, Sasaki J, Hishida H, Itakura Het al. Effects of eicosapentaenoic acid on major coronary events in hypercholesterolaemic patients (JELIS): a randomised open-label, blinded endpoint analysis. Lancet. 2007;369(9567):1090–8. [DOI] [PubMed] [Google Scholar]
  • 53. Bhatt DL, Steg PG, Miller M, Brinton EA, Jacobson TA, Ketchum SB, Doyle RT Jr, Juliano RA, Jiao L, Granowitz Cet al. Cardiovascular risk reduction with icosapent ethyl for hypertriglyceridemia. N Engl J Med. 2019;380(1):11–22. [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

nmaa086_Supplemental_File
Click here for additional data file. (6.9MB, docx)

Articles from Advances in Nutrition are provided here courtesy of American Society for Nutrition

RESOURCES