Characteristics of Eligible Studies
Literature search
A total of 8964 publications was collected from the four databases, of which 1058 were reviewed through titles and, or abstracts, 958 articles were excluded because they were titles, abstracts, meta-analyses, case reports, systematic reviews and irrelevant articles. 100full-text articles were screened, and 36 articles fulfilled the inclusion criteria. Finally, 13 studies were selected for the meta-analysis, and the reasons for excluding 23 studies from the remaining articles were as follows: 19 studies combined astaxanthin with other drugs, 4 studies hadn't control group. The flow diagram is shown in Figure 1.
Study characteristics
The Study characteristics show in Table 1. They were all prospective cohort studies. Eight studies were conducted in Asia (Korea, n=1; Japan, n=5; Iran, n=1; China, n=4), one study was conducted in Europe (Finland, n=1), and one study was conducted in America (Canada, n=1). The other information of the included studies was age range, sex, astaxanthin dose, placebo dose and follow-up period.
Quality assessment
The results of the quality assessment with Cochrane criteria are summarized in Table 3. The quality about the included studies was varied, 46 items of information were low-risk bias, and 43 items of information were unclear bias risk. Only 1 item was high-risk bias, but it didn't affect our findings. Therefore, the included articles were high-quality studies.
Blood lipid
Thirteen studies on the relationship between astaxanthin and the concentration of cholesterol were shown in Figure 2.1. The astaxanthin could remarkably recover the plasma concentration of total cholesterol (WND: -0.310, 95%CI: -0.541 to -0.168, P = 0.000, Z = -4.294); HDL-C (WND: -0.294, 95%CI:-0515 to -0.072, P = 0.009, Z = 2.594); LDL-C (WND: -0.207, 95%CI: -0.396 to -0.019, P = 0.031, Z = -2.153); triglyceride (WND:-0.603, 95%CI:-1.028 to -0.179, P = 0.005, Z = -2.787). The analyses for total cholesterol, HDL-C and LDL-C showed in Table 2 that the heterogeneity was not statistically significant (Total cholesterol: I2 = 0.000, Q = 13.598, P = 0.755; HDL-C: I2 = 37.794, Q = 27.329, P = 0.053: LDL-C: I2 = 0.000, Q = 3.210, P = 0.999). However, there were publication biases of total cholesterol, HDL-C and LDL-C in Table 2. (Total cholesterol: pBegg = 0.008, pEgger = 0.024; HDL-C: pBegg = 0.014, pEgger = 0.083: LDL-C: pBegg = 0.002, pEgger = 0.008). According to the trim and fill analysis, these studies were filled in the right of the mean to make the funnel symmetrical, and the adjusted estimated effect based on the random-effects model, the result emerged the adjusted values was consistent with the observed values in Figure 5. (Total cholesterol: lower limit -0.451 to upper limit -0.168, Q =13.598; HDL-C: lower limit 0.07178 to upper limit 0.51532, Q=27.329; LDL-C: lower limit -0.39587 to upper limit 0.01854, Q=3.210). In analyses for Triglyceride, no publication bias was detected (Triglyceride: pBegg = 0.06250, pEgger = 0.20944). There was significant heterogeneity (Triglyceride: I2 = 84.083, Q = 87.956, P = 0.000). They were shown in Table 2. To determine whether the comprehensive assessments would have been markedly affected by single research, the sensitivity analysis was conducted. The results in Figure 3 showed that substantial heterogeneity was not observed in this analysis (all of P-Value>0.00).
Plasma glucose
Four studies focused on the effect of astaxanthin on plasma glucose concentrations, as shown in Figure 2.2. The results suggested that astaxanthin significantly reduced the blood glucose of patients (WND: -0.353, 95%CI: -0.596 to -0.110, P=0.004, Z=-2.848). The meta-analysis in Table 2 suggested that the heterogeneity was not statistically significant (I2 = 20.991, Q = 6.328, P = 0.276). These result sindicated evidence of no publication bias (pBegg = 0.42549, pEgger = 0.24654).
Blood pressure
Six studies investigated the role of astaxanthin in controlling blood pressure, as shown in Figure 2.3. The results didn't show the potential role of astaxanthin in DBP (WND: -0.186, 95%CI: -0.374 to 0.002, P = 0.052, Z = -1.943), while astaxanthin was significant for reducing SBP (WND: -0.236, 95%CI:-0.425 to -0.048, P=0.014, Z = -2.460). No significant publication bias was observed in this analysis (DBP: pBegg = 0.105, pEgger = 0.103; SBP: pBegg = 0.266, pEgger = 0.216). The heterogeneity wasn't significant in this study (DBP: I2 = 0.000, Q = 1.235, p = 0.996; SBP: I2 = 0.000, Q = 3.636, p = 0.888). These were shown in Table 2.
BMI
As shown in Figure 2.4, five studies evaluated the association between astaxanthin and body mass index. The results indicated the null effect of astaxanthin on BMI (WND: 0.095, 95%CI: -0.164 to 0.354, p = 0.473). Heterogeneity was not found in the analysis (I2 = 0.000, Q = 0.592, P = 0.999). However, the publication bias existed in Table 2 (pBegg=0.013, pEgger = 0.042). Then, the trim and fill analysis were implemented, and the adjusted estimated effect was based on the random-effects model. The result in Figure 5 emerged that the adjusted values were consistent with the observed values (BMI: lower limit -0.164 to upper limit 0.354, Q = 0.592 ).