*3.2. Fish Consumption and Cardiovascular Disease Mortality Risk*

Eighteen studies, involving 1,267,951 participants and 51,628 CVD deaths, investigated the association between the fish intake and the CVD mortality risk [13–18,23–33,40]. The pooled RR (95% CI) was 0.91 (0.85–0.98) for the highest versus the lowest fish consumption category (I2 = 70.0%) (Figure 2). Sensitivity analysis did not change the protective effects of fish on CVD mortality (Figure S1). Subgroup analysis suggested that there was a significant negative association between the fish intake and the CVD mortality risk among the subgroups with nine years or more follow-up duration (Table 3). No publication bias was found (Egger's test: *p* = 0.919; funnel plot: Figure S2).

Figure 3a showed the linear and non-linear dose-response analyses between the fish intake and the CVD mortality risk. Ten prospective cohort studies met the requirements for dose-response analysis [13,15–18,23,27,29,33,40], and the curvilinear correlation presented a downward trend for the adjusted RR of CVD deaths with the increase of fish consumption from zero to 40 g/d (*p* non-linearity < 0.001). The adjusted RR reached a steady value when fish consumption increased beyond 40 g/d. In the linear dose-response analysis, the summary RR (95% CI) for a 20 g/d increment was 0.96 (0.94–0.99) for CVD mortality risk (*p* trend = 0.002).

**Figure 2.** Forest plot of the highest versus lowest fish intake category and CVD mortality risk. Plot demonstrates decreased risk of CVD mortality risk with fish intake by the random-effects model (RR = 0.91; 95% CI, 0.85–0.98). CVD, cardiovascular disease; RR, relevant risk; CI, confidence intervals.

#### *3.3. Marine n-3 PUFA and Cardiovascular Disease Mortality Risk*

Ten eligible studies with 1,337,660 participants and 76,537 CVD deaths explored the association of marine n-3 PUFA intake with CVD mortality risk [11,18,25,27,34–39]. The pooled RR (95% CI) for the highest versus the lowest marine n-3 PUFA consumption category was 0.87 (0.85–0.89), with a low heterogeneity (I2 = 37.8%) (Figure 4). Sensitivity analysis suggested a great impact on one article with high quality (Figure S3) [35]. The negative association between marine n-3 PUFA and the risk of CVD mortality was altered from 0.87 (0.85–0.89) to 0.84 (0.81–0.87) by deleting this study. Subgroup analyses displayed a significant negative association among the Americas, and Asian and European countries compared with Oceania countries (Table 3). No publication bias was found (Egger's test: *p* = 0.722; funnel plot: Figure S4). Figure 3b showed the linear and non-linear dose-response analysis between marine n-3 PUFA intake and CVD mortality risk. Eight prospective cohort studies met the requirements of dose-response analysis [18,25,27,34,36–39], and the curvilinear correlation presented a downward trend of CVD deaths with the increase of n-3 PUFA intake (*p* non-linearity < 0.001). Linear dose-response analysis suggested that an increase of 80 milligrams of n-3 PUFA per day was associated with a 4% lower risk of CVD mortality (95% CI: 0.94–0.98; *p* trend < 0.001).

**Figure 3.** Dose-response association: (**a**) fish and CVD mortality (*n* = 10, *p* non-linearity < 0.001; *p* trend = 0.002); the risk of CVD mortality was decreased by 4% with an increase of 20 g of fish intake (RR = 0.96; 95% CI: 0.94–0.99) per day. (**b**) marine n-3 PUFA and CVD mortality (*n* = 8, *p* non-linearity < 0.001; *p* trend < 0.001); the risk of CVD mortality was decreased by 4% with an increase of 80 milligrams of marine n-3 PUFA intake (RR = 0.96; 95% CI: 0.94–0.98) per day. CVD, cardiovascular disease; n-3 PUFA, n-3 polyunsaturated fatty acids; RR, relevant risk; CI, confidence intervals; g/d, grams per day; mg/d, milligrams per day.


**Figure 4.** Forest plot of the highest versus lowest marine n-3 PUFA intake category and CVD mortality risk. Plot demonstrates decreased risk of CVD mortality risk with n-3 PUFA intake by the fixed-effects model (RR = 0.87; 95% CI, 0.85–0.89). CVD, cardiovascular disease; n-3 PUFA, n-3 polyunsaturated fatty acids; RR, relevant risk; CI, confidence intervals.

**Table 3.** Subgroup and meta-regression analyses for the association between fish, n-3 PUFA intake and CVD mortality.



**Table 3.** *Cont.*

N † Number of included studies; *p* <sup>a</sup> for heterogeneity; *p* <sup>b</sup> for significance test; *p* <sup>c</sup> for meta-regression analysis. Others \* All patients were diabetic or not diabetic. n-3 PUFA, n-3 polyunsaturated fatty acid; CVD, cardiovascular disease.

#### **4. Discussion**

To our knowledge, the current work is the first meta-analysis of prospective observational studies for associating marine n-3 PUFA intake and CVD mortality risk. This study showed a significant inverse association between fish, marine n-3 PUFA intake and CVD mortality risk. Nonlinear dose-response relationship found that an increase of 20 g of fish intake or 80 milligrams of marine n-3 PUFA intake per day was associated with a 4% reduction in risk of CVD mortality.

In accordance with the previous study, the fish consumption was inversely associated with the CVD mortality risk in the current meta-analysis [12]. Bechthold et al.'s study also suggested a negative association between fish consumption and the risk of CVD [41]. Several studies showed no association between the fish intake and the risk of CVD [42,43]. Differences in preparation and type of fish might explain the observed difference. The progress of frying deteriorates oils through oxidation and hydrogenation, leading to an increase of trans fatty acids [44]. Trans fatty acids can aggravate inflammation and endothelial dysfunction, increasing the risk of CVD mortality [45]. Fish high in salt during cooking can increase the risk of CVD through increasing production of reactive oxygen species and oxidative stress, which contribute to impaired vascular function [46,47]. Fish can be divided into lean, medium-fatty or fatty fish with less than 2 g, 2–8 g and more than 8 g fat per 100 g in its body tissue [48]. Fatty fish diets significantly decreased the serum concentrations of triacylglycerol, apolipoprotein B, apolipoprotein CII and apolipoprotein CIII, which were known CVD risk markers [49]. Fishes also contain vitamin D, proteins, minerals and taurine which may decrease markers of inflammation and improve vascular function by increasing adiponectin levels [50]. In the subgroup of adjustment for diabetes, fish intake was associated with a reduction in the rate of major CVD mortality that approached significance (RR = 0.93; 95% CI: 0.85–1.01). Previous study has showed that supplementation of fish can decrease the CVD mortality risk in a diabetic population [51], the possible reason being that diabetes is a significant risk factor for CVD mortality [52]. EPA and DHA derived from fish can activate the G protein–coupled receptor 120 to reverse insulin resistance [53]. n-3 PUFA supplementation can protect against CVD in patients with diabetes [54].

In most studies where fish exits as an exposure variable, the observed benefits could often be attributed to the presence of fatty acids [55,56]. The long chain n-3 PUFA—namely, EPA and DHA—are naturally presented not only in fatty fish, but also in lean fish [57,58]. n-3 PUFA supplementation can decrease the risk of CVD [59,60]. The plasma level of EPA and DHA in humans may increase after intake of fish to improve the composition of lipoprotein cholesterol as cardiovascular markers affecting the risk of CVD [61,62]. However, previous study showed that low-dose supplementation with EPA and DHA did not significantly

reduce the rate of CVD events [63]. This possible reason may be related to presence or absence of a history of CVD. The patients in the trial were all myocardial infarction patients for 4 years before enrollment. 85% of the patients were receiving statins. Patients with CVD who are receiving good clinical treatment showed low risk of future cardiovascular events [64]. Therefore, we wanted to observe the effect of the long chain n-3 PUFA on CVD mortality through the long-term duration.

In this meta-analysis, we also found a negative association between the marine n-3 PUFA intake and the CVD mortality risk. In previous studies, the results were not consistent [65]. A randomized controlled trial (RCT) showed that n-3 PUFA supplementation (866 mg/d) for 3.5 years could reduce CVD mortality risk [66]. In contrast, the RCT with one-year n-3 PUFA supplementation (850 mg/d) suggested no association [67]. Although some randomized controlled trials (RCTs) had been published, the follow-up duration were short with most studies ranged from 1–5 years [66–68]. Hoverer, the cohort studies included in this meta-analysis have longer follow-up duration ranged from 5–29 years. CVD is a chronic disease with a long disease course. Longer follow-up duration was more in line with the nature of the CVD disease. The possible mechanisms were as follows. First, the plasma n-3 PUFA increased with the frequency and the amount of dietary n-3 PUFA intake [69,70]. A higher circulating n-3 PUFA may alter the cell membrane fluidity which modulates protein function and signaling. The dimerization and recruitment of toll-like receptor-4 may be disrupted to down-regulate the expression of nuclear factor-kappaB reducing the inflammatory responses, with the enrichment of n-3 PUFA [71]. Second, n-3 PUFA may inhibit oxidative stress through the nuclear factor E2-related factor 2/heme oxygenase-1 signaling pathway. 4-hydroxy-2E-hexenal, the product of n-3 PUFA peroxidation, will dissociate Nrf2 from Keap1 and react with the cysteine residues of Keap1 [72]. Then, Nrf2 can translocate into the nucleus and bind to antioxidant responsive element to increase the expression of HO-1 [73]. HO-1 is a representative antioxidant enzyme that can confer cytoprotection on a wide variety of cells against oxidative damage [72]. Third, n-3 PUFA may reduce the hepatic very low-density lipoprotein production rate to decrease the plasma triglyceride levels through affecting fatty acid desaturases, fatty acid elongases and peroxisomal β- gene expression and fatty acid beta-oxidation [74,75]. In addition, long-chain n-3 PUFA may play an important role in improving the endothelial function, lowering circulating markers of endothelial dysfunction, such as E-selectin, vascular cell adhesion molecule-1 and intercellular adhesion molecule-1 [76–78].

The dose–response analysis showed that the risk of CVD mortality decreased with the increase of fish consumption from zero to 40 g/d. The adjusted RR reached a steady value when fish consumption increased beyond 40 g/d. Therefore, we believe that 40 g/d is the ideal dose for preventing CVD mortality. This is basically consistent with the average fish intake of the population of Europe and America [23,30]. However, the average intake of people in Japan is higher than this level [13].

This study has several strengths. First, compared with the previous meta-analysis [12], this study included additional 11 studies to investigate the association between the fish consumption and the CVD mortality risk, which may have a higher statistical power. Second, this meta-analysis was first to investigate the association between marine n-3 PUFA intake and CVD mortality risk with prospective cohort studies. Third, most studies had a long follow-up duration (9–30 years). CVD is a chronic disease and longer followup duration can better explain the association between fish, marine n-3 PUFA and CVD mortality risk.

The limitations should be acknowledged. First, several deep-sea fishes may be contaminated, while only one article reported whether fishes had pollutants or not [28]. Second, it is hard to standardize the fish and marine n-3 PUFA consumption due to the details of measurement methods not being available. Thus, we chose RR (95% CI) of the highest versus lowest fish and marine n-3 PUFA intake category and CVD mortality risk.
