Safety of Supplementation of Omega-3 Polyunsaturated Fatty Acids: A Systematic Review and Meta-Analysis of Randomized Controlled Trials

There is no comprehensive review of the evidence to support omega-3 polyunsaturated fatty acids (PUFAs) as a relatively safe and tolerable intervention. This study aimed to provide a meta-analytic and comprehensive review on the adverse effects of all kinds of ω-3 PUFA supplementation reported in randomized controlled trials (RCTs) in human subjects. A systematic review of RCTs published between 1987 and 2023 was carried out based on searches of 8 electronic databases. All RCTs that compared the adverse effects of ω-3 PUFAs containing eicosapentaenoic acid, docosahexaenoic acid, or both compared with controls (a placebo or a standard treatment) were included. The primary outcome was the adverse effects related to ω-3 PUFA prescription. A total of 90 RCTs showed that the ω-3 PUFA group, when compared with the placebo, had significantly higher odds of occurrence of diarrhea (odds ratio [OR] = 1.257, P = 0.010), dysgeusia (OR = 3.478, P < 0.001), and bleeding tendency (OR = 1.260, P = 0.025) but lower rates of back pain (OR = 0.727, P < 0.001). The subgroup analysis showed that the prescription ω-3 PUFA products (RxOME3FAs) had higher ω-3 PUFA dosages than generic ω-3 PUFAs (OME3FAs) (3056.38 ± 1113.28 mg/d compared with 2315.92 ± 1725.61 mg/d), and studies on RxOME3FAs performed more standard assessments than OME3FAs on adverse effects (63% compared with 36%). There was no report of definite ω-3 PUFA-related serious adverse events. The subjects taking ω-3 PUFAs were at higher odds of experiencing adverse effects; hence, comprehensive assessments of the adverse effects may help to detect minor/subtle adverse effects associated with ω-3 PUFAs. This study was registered at PROSPERO as CRD42023401169.


Introduction
Omega-3 polyunsaturated fatty acids (PUFAs), especially EPA and DHA, are essential for humans and a promising natural remedy for many disorders, as demonstrated in many studies [1].
The number of clinical studies using ω-3 PUFAs for treatment has doubled in the past 2 decades, and more trials used higher than the minimum daily recommended nutritional supplementation dosage (200-500 mg DHA þ EPA), which may also vary by country [10,11].In addition, some of the clinical studies focused on patients comorbid with several medical disorders, e.g., patients with type 2 diabetes mellitus (T2DM) or dyslipidemia and pregnant mothers at risk of depression [12].The safety and tolerability of ω-3 PUFAs have been continuously discussed as its strengths as a potential nutritional therapeutic agent [7,13].However, we should be aware that the advantages of safety might be overestimated because there have been no extensive investigations of safety in the trials testing ω-3 PUFA supplementation, especially ω-3 PUFAs with high-quality content (e.g., prescription ω-3 PUFAs [RxOME3FAs]).
In this study, we performed a systematic review followed by a meta-analysis on the subjective and objective tolerability profiles of ω-3 PUFA supplementation in all the clinical trials in humans.
Moreover, we analyzed the adverse effects by considering the quality of ω-3 PUFAs in the clinical trials.Thus, to the best of our knowledge, this is the first large-scale meta-analysis investigating the potential adverse events (AEs) and laboratory abnormalities associated with ω-3 PUFA supplementation, focusing on RxOME3FAs compared with generic ω-3 PUFAs (OME3FAs) in randomized controlled trials (RCTs).

Methods Protocol
The present study followed the PRISMA guidelines [14] (Figure 1).The current meta-analysis fulfilled the certification requirements of the Institutional Review Board of the Kaohsiung Veterans General Hospital (approval # VGHKS17-EM10-01) and has been registered in PROSPERO (CRD42023401169).

Statistical analysis
In the present study, the meta-analytic procedures consisted of the following 2 parts: 1) dichotomous outcomes: the differences in treatment-related AEs by system organ class and by MedDRA Preferred Terms or commonly used terms, and 2) continuous outcomes: the differences in treatment-related AEs on laboratory measurements.
The meta-analytic procedures were performed in the Comprehensive Meta-Analysis software, version 3 (CMA ver.3.0; Biostat).A 2-tailed P value of <0.05 was considered to indicate statistical significance.Please refer to the Supplementary Material for the details of the methods and statistical analysis.

Studies retrieved and characteristics
The full search strategy is illustrated in Figure 1.A total of 331 articles entered the full-text screening stage, and 103 of them were excluded based on our exclusion criteria (Supplementary Tables 1-5).Furthermore, 90 studies were excluded because we could not extract specific AE data from the articles or obtain the unpublished data from the corresponding authors (Supplementary Table 4).Finally, the remaining 90 articles were included in the present meta-analysis.

Description and characteristics of the included studies
The detailed characteristics of the included studies are listed in Supplementary Table 1.In brief, the included studies originated from Argentina, Australia, Brazil, Canada, China, Denmark, France, Germany, India, Iran, Israel, Italy, Japan, Malaysia, Mexico, Netherlands, New Zealand, Norway, Poland, Romania, Russian Federation, South Africa, Sweden Taiwan, UK, Ukraine, and United States.All trials were published between 1987 and 2023.A total of 90 articles were included with a total of 59,940 participants in the ω- The length of the follow-ups in the trials ranged from 1 wk to 385 wk with a median duration of 18.0 (39 interquartile range) weeks.
Among the 90 included RCTs, the characteristics of study participants included 27 studies in participants with dyslipidemia, 22 in participants with cardiovascular diseases, 8 studies in healthy participants, 7 studies in participants with diabetes, 3 studies in pregnant women, 4 studies in patients on hemodialysis, 2 studies in patients with hypertension, 2 studies in patients with Crohn's disease, 2 studies in patients with nonalcoholic liver diseases, and 1 study for each in participants with the following clinical conditions: age-related cognitive decline, colorectal adenomas, cognitive decline, cyclosporine-treated liver transplant, depression, dry eye, dysglycemia, end-stage renal diseases, Huntington disease, metabolic syndrome, migraine, women with obesity, psoriasis, renal transplant, stroke, and ultrahigh risk of psychotic disorder.Among the 90

Statement of Significance
Both prescription and generic omega-3 polyunsaturated fatty acids (PUFAs) might be associated with higher rates of some types of adverse effects.Moreover, although prescription ω-3 PUFAs appear to have more adverse effects than generic ω-3 PUFAs, this difference may be due to the higher dosage and the systematic evaluation of adverse effects commonly performed in the trials using prescription ω-3 PUFAs.included studies, 46 used RxOME3FAs (US Food and Drug Administration-approved ω-3 PUFAs for clinical application; we have updated 25 new studies since our last report [15]), and 44 used generic ω-3 PUFAs (OME3FAs).

Methodologic quality of included studies
The details of the methodologic quality assessment of included studies are provided in Supplementary Table 5.The median of the Jadad scores of the included 90 studies was 4, with a 25%-75% interquartile range 3-5.

Review of serious AEs
Among the 90 included studies, there were no reports of serious AEs.
We then performed subgroup analysis on the types of ω-3 PUFAs (RxOME3FAs or OME3FAs), the ω-3 PUFA dosage (!3000 mg/d compared with <3000 mg/d), and whether the trial performed routine standard AE evaluations (such as periodic AE assessment with scales).Subgroup analysis showed that the RxOME3FAs, a dosage !3000 mg/d, and routine standard AE evaluations were associated with a lower OR of back pain; OME3FAs, a dosage <3000 mg/d, and routine standard AE evaluations were associated with a higher OR of diarrhea; RxOME3FAs were associated with a higher OR of bleeding tendency; both RxOME3FAs and OME3FAs were associated with a higher OR of dysgeusia; and OME3FAs were associated with a higher OR for dysgeusia than RxOME3FAs (Table 3, Supplementary Material, Supplementary Table 2).

Main results of the meta-analysis of differences of
AEs in laboratory data from participants taking ω-3

PUFAs and controls: continuous variables
The detailed results of our meta-analysis of AEs measured by laboratory data in the participants taking ω-3 PUFAs and those taking placebo are listed in Table 2 and Figure 2B.

Discussion
To our knowledge, this is the first large-scale meta-analysis focusing on the safety and tolerability of all types of ω-3 PUFA supplementation.Trials using RxOME3FA and OME3FA supplementation were included.Among the 90 included studies (with a median duration of follow-up of 18 wk), there was no report of definite ω-3 PUFA-related serious AEs.Our study showed that patients taking ω-3 PUFAs experienced some AEs.Moreover, when we compared studies using RxOME3FAs and OME3FAs, we found that studies using RxOME3FAs used a higher mean dosage of ω- In terms of nonserious adverse effects, our findings are comprised of 3 parts: 1) treatment-related AEs; 2) treatmentrelated abnormal laboratory changes of lipid profiles; 3) treatment-related abnormal nonlipid laboratory changes.

Treatment-related AEs (dichotomous items)
Our study showed that subjects receiving ω-3 PUFAs had a higher rate of diarrhea and dysgeusia than those receiving placebo.Moreover, the occurrence of bleeding tendency was associated with RxOME3FAs but not associated with the dosages of ω-3 PUFAs.Our finding of non-dose-dependent bleeding tendency is consistent with the GISSI-Prevenzione study, which showed that daily intake of dosage higher than 3000 mg of ω-3 PUFAs was not associated with reports of higher rates of bleeding; moreover, >80% of patients were taking aspirin concomitantly in the study [16].In another study conducted by Wachira et al. [17], ω-3 PUFAs did not increase the risk of clinically significant bleeding when used alone or in combination with anticoagulation agents.

Treatment-related laboratory changes of lipid profiles (continuous variables)
Table 2 shows that ω-3 PUFAs had significantly beneficial effects on non-HDL, T-Cho, TG, and VLDL when compared with the placebo group.ω-3 PUFAs increase the "good cholesterol" and lower the "bad cholesterol" by having an impact on reverse cholesterol transport.Reverse cholesterol transport is the process in which excess peripheral cholesterol is transported to the liver for hepatobiliary excretion, thus inhibiting foam cell formation and the development of atherosclerosis [18].Both DHA and EPA have been suggested to exert atheroprotective functions by promoting intracellular catabolism of apolipoprotein-B-100-containing lipoproteins, suppressing hepatic apolipoprotein-B production, stimulating plasma TG clearance via lipoprotein lipase, increasing the VLDL to LDL conversion rate, reducing LDL synthesis, and attenuating postprandial lipemia [19,20].Of note, there was a greater improvement in the serum levels of TG and T-Cho in patients receiving the RxOME3FAs but not in patients receiving OME3FAs (Table 3).This finding further supports the notion that RxOME3FAs have more beneficial effects on the lipid profile than OME3FAs.Moreover, regardless of the ω-3 PUFA dosage and ω-3 types (RxOME3FAs compared with OME3FAs), ω-3 PUFAs lowered serum levels of TG. ω-3 PUFAs were also associated with lower TG and T-Cho levels in the trials with routine standard AE evaluations but a higher T-Cho level in the trials lacking standard AE evaluations.Our results suggested that the lipid-lowering effect of ω-3 PUFAs can be well demonstrated only if ω-3 PUFAs are RxOME3FAs and only when the trials performed routine and standard AE evaluations.

Treatment-related nonlipid laboratory changes (continuous variables)
The results showed that participants in the ω-3 PUFA group, when compared to those in the placebo group, had lower serum levels of ALP and CRP but a higher level of ALT and BUN (Table 2).The CRP-lowering effects in the ω-3 PUFA group are consistent with those reported in previous studies in which serum CRP levels are inversely related to blood levels of DHA and EPA [21], and duration of 6-mo supplementation with ω-3 PUFAs was associated with reduced CRP serum levels [22].Moreover, our findings further supported the notion that ω-3 PUFAs provide anti-inflammatory effects in inflammation-associated disorders through the mechanism counteracting the actions of ω-6 PUFAs [3,23].Our finding of elevated ALT levels in participants taking ω-3 PUFAs is in accordance with previous studies showing the beneficial effects of ω-3 PUFAs on γ-glutamyl transferase but not on other liver function measurements, such as ALT and aspartate aminotransferase [24,25].Surprisingly, the subgroup analysis of our findings showed studies that used RxOME3FAs and performed routine standard AE evaluations were associated with elevated fasting glucose and a higher level of glycated hemoglobin (HbA1c, routine standard AE evaluations only).This is inconsistent with a previous study that found ω-3 PUFAs were able to significantly reduce the risk of T2DM [26] but is consistent with the findings of a meta-analysis showing no beneficial effects of ω-3 PUFAs on glucose control in FIGURE 2. Forest plot of current meta-analysis of adverse events.(A) categorical outcome, (B) continuous outcome.Abbreviations: AC sugar, fasting sugar; Ag, antigen; ALP, alkaline phosphatase; ALT, alanine transaminase; Apo-B, apolipoprotein-B; AST, aspartate transaminase; BP, blood pressure; BUN, blood urea nitrogen; CI, confidence interval; CPK, creatine phosphokinase; Cre, creatinine; CRP, C-reactive protein; DBP, diastolic blood pressure; DM, diabetes mellitus; GI, gastrointestinal; Hb, hemoglobin; HbA1c, glycated hemoglobin; Hct, hematocrit; HDL, high-density lipoprotein; LDL, low-density lipoprotein; MA, meta-analysis; Plt, platelet; PUFA, polyunsaturated fatty acid; SBP, systolic blood pressure; T-Bil, total bilirubin; T-Cho, total cholesterol; TG, triglyceride; tPA, tissue-type plasminogen activator; UTI, urinary tract infection; VLDL, very lowdensity lipoprotein; WBC, white blood cell.findings may be limited.The safety and tolerability profiles may be different among different populations such as the elderly, pregnant women, or patients with other comorbidities.Second, regarding the concern about the bleeding tendency, this cannot be answered by the current study because there were no adequate data either using binary (bleeding events) or continuous (bleeding time) outcomes in this meta-analysis.There is considerable debate as to the relative merits of using RCT data as opposed to observational data in systematic reviews of AEs.
In theory, well-conducted RCTs yield unbiased estimates of treatments and AEs.However, in RCTs, the characteristics of study participants are highly selected for research purpose, and the study sample size is well designed to reach statistical power for estimating the treatment efficacy [28].Therefore, as a meta-analysis of RCTs, the results of the present study may not be able to identify rare or long-term AEs in real-world practice.

Conclusion
In short, our meta-analysis showed that patients receiving all types of ω-3 PUFAs would experience higher rates of some mild AEs.Moreover, although RxOME3FAs may appear to have more AEs than OME3FAs, these differences in the occurrence rates of AEs may be due to the higher dosage and the systematic AE evaluations commonly performed in the trials using RxOME3-FAs.In short, ω-3 PUFAs are not without AE; however, applying routine and standard AE evaluations in trials using RxOME3FAs may help to provide patients with complex medical comorbidities with a safe and tolerable treatment option.

TABLE 1
Meta-analysis of prevalence rate of adverse events Abbreviations: adj.ES, adjusted effect size; CI, confidence interval; CPK, creatine phosphokinase; df, degree of freedom; n/s, not significant; OR, odds ratio; Sig., significant; SOC, System Organ Class.

TABLE 3
Differences (of treatment-emergent adverse effects) between RxOME3FAs and OME3FAs (continued on next page)