Literature search and study characteristics
Figure 1 shows the details of the literature search and study selection. Our search strategy initially identified 404 papers. Duplicate removal and screening through article title and abstract review identified 28 studies. Nineteen studies were excluded after the full text was reviewed. Finally, 9 studies involving 951 participants were included in the meta-analysis[8, 14-21]. Among them, 2 studies[17, 21] had more than one intervention groups, therefore each type of intervention was compared with control group and analyzed. Table 1 outlines the characteristics of the included studies. The majority of included studies reported both male and female NAFLD patients except one involving only male patients[19]. All studies involved adults NAFLD patients. The intervention duration ranged from 8 to 48 weeks and the median was 16 weeks (4 months).
Methodological quality
Figure 2 presents the methodological quality of the included studies. Most studies had low risk of bias in random sequence generation and all four studies reported information in allocation concealment had low risk of selection bias; however, three studies had high risk in blinding of participants and personnel and only one study reported the use of blinding in outcome assessment. All studies had low risk of bias in incomplete outcome data and selective reporting.
Effect of physical activity on hepatic enzyme parameters
Nine, eight and five studies reported data for alanine aminotransferase (ALT), aspartate aminotransferase (AST), and γ-glutamyl transferase (GGT) respectively. The combined results suggested that physical activity alone can significantly improve all three hepatic enzyme parameters: ALT (SMD -0.17, 95% CI: -0.30 to -0.05), AST (SMD -0.25, 95% CI: -0.38, -0.13) and GGT (SMD -0.22, 95% CI: -0.36, -0.08), and the heterogeneity among studies were all insignificant ( I²<50%). But the effect size is generally small with marginal confidence interval. (Fig. 3)
Compared with no physical activity, subgroup analysis showed that aerobic exercise alone can significantly reduce AST (SMD -0.26, 95% CI: -0.43, -0.10). Regarding to ALT and GGT, the effects were insignificant (SMD -0.16, 95% CI: -0.32 to 0.00 and SMD -0.19, 95% CI: -0.40 to 0.01 respectively). Resistance exercise alone had similar small improvement effect on these two hepatic enzyme parameters: AST (SMD -0.23, 95% CI: -0.43, -0.03) and GGT (SMD -0.24, 95% CI: -0.44, -0.03). But regarding to the combination of aerobic and resistance exercise, results showed no improvements on all three parameters. Heterogeneity between subgroups were insignificant for ALT (p=0.32), AST (p=0.11) and GGT (p=0.88). (Fig. 3)
Further categorizing studies according to the intervention duration, subgroup analysis showed that, regardless of the type of physical activity, keeping regular physical activity for more than 4 months can significantly improve hepatic enzyme level, while less than 4 months’ physical activity had no significant effect. (Fig. 4)
Effect of physical activity on serum lipid parameters
Eight, seven, seven, and eight studies had sufficient data for inclusion in meta analyses for total cholesterol (TC), triglycerides (TG), low-density lipoprotein cholesterol (LDL-C), and high-density lipoprotein cholesterol (HDL-C), respectively. There was no significant heterogeneity among these studies for all these four parameters (I²<50%). The pooled result showed that, compared with control group, participants who had regular physical activities were more likely to have slightly lower TC (SMD = -0.22, 95% CI: -0.34, -0.09) , TG (SMD = -0.18, 95% CI: -0.31, -0.06) and LDL-C (SMD = -0.26, 95% CI: -0.39, -0.13). Regarding to HDL-C, physical activity tended to increase the level of it compared with control, but the effect was insignificant (SMD = 0.07, 95% CI: -0.06, 0.19). (Fig. 5). In the subgroup analysis, taking resistance exercise alone can significantly reduce TC but with a small effect size (SMD = -0.31, 95% CI: -0.51, -0.10), while neither aerobic exercises alone nor combination of it with resistance exercise can improve TC. Heterogeneity between subgroups were not significant (p=0.71). As for TG, only resistance exercise alone can significantly reduce the level of it, with small effect size (SMD = -0.32, 95% CI: -0.52 to -0.11) . The heterogeneity between subgroups for TG was significant (p=0.09). The subgroup analyses also suggested that aerobic exercises alone and resistance exercise alone can significantly reduce the level of LDL-C, with SMD -0.21, 95% CI: -0.37 to -0.04 and SMD -0.35, 95% CI: -0.56 to -0.15 respectively, but combination of aerobic and resistance exercise had no significant effect on LDL-C, which may due to small sample size as only 1 study was included in this subgroup. The heterogeneity between subgroups was insignificant (p=0.66). Regarding HDL-C, all three subgroups showed insignificant results and there was no difference between subgroups (p=0.94). (Fig. 5).
As for duration of physical activity, both ≥4 months and <4 months groups had significant effect on TC regardless of the type of physical activity (SMD -0.19, 95% CI -0.32 to -0.06 and SMD -0.60, 95% CI -1.10 to -0.10 respectively). Only ≥4 months intervention significantly improved TG (SMD -0.16, 95% CI: -0.30 to -0.03) and LDL-C (SMD -0.25, 95% CI: -0.38 to -0.12). Regarding HDL-C, both subgroups showed insignificant effect. (Fig. 6)
Effect of physical activity on glucose metabolism parameters
Seven, four and four studies had sufficient data for inclusion in analyses of fasting glucose, fasting insulin, and homeostasis model assessment of insulin resistance (HOMA-IR), respectively. The heterogeneity among studies for these three glucose metabolism parameters was all significant (I2>50%). The random-effect model showed that there was no significant effect of physical activity on improving these glucose metabolism parameters.
In the subgroup analysis, compared with control, only aerobic exercise alone can significantly reduce HOMA-IR (SMD = -0.42, 95% CI: -0.63 to -0.22). Heterogeneities between subgroups for HOMA-IR was significant (p=0.04). Combination of aerobic exercise and resistance exercise can improve fasting insulin (SMD = -0.80, 95% CI: -1.59 to -0.01), but this subgroup only included 1 study with a small sample size of 15. As for fasting glucose, all three types of physical activity had no significant improvement effect. (Fig. 7).
In the subgroup analysis according to duration of intervention, significant improvement was found in ≥4 months group for fasting glucose (SMD = -0.27, 95% CI: -0.48, -0.07) and HOMA-IR (SMD = -0.44, 95% CI: -0.60, -0.29) . There were no significant effects in subgroups for fasting insulin. (Fig. 8)
Effect of physical activity on intra-hepatic lipid content
Four studies had sufficient data to be included in the analysis of effect of physical activity on intra-hepatic lipid content. The heterogeneity among studies for this parameter was not significant (I2<50%). The fixed-effect model showed that physical activity can significantly reduce NAFLD patient’s liver fat content (SMD = -0.21, 95% CI: -0.36 to -0.06). In the subgroup analysis of different intervention types, neither aerobic exercise nor resistance exercise showed significant improvement on intra-hepatic lipid content, but only one study had an intervention arm of resistance exercise with 153 participants (Fig.9). Because all four included studies had intervention durations longer than 4 months, the effect of physical activity on intra-hepatic lipid cannot be assessed in subgroup analysis.
Sensitivity analysis and publication bias
Sensitivity analysis was conducted by removing each individual study, except for fasting insulin, HOMA-IR, and intra-hepatic lipid content where only 4 studies were included. The results suggested that the pooled effect were unlikely to be substantially altered (Fig. 10).As there were less than 10 included studies, publication bias analysis was not conducted.
Adverse events
Three studies[16, 20, 21] mentioned adverse events during the intervention of physical activity. Among 951 participants, there were 2 knee pain, 1 shoulder pain, 1 back pain and 2 bone fractures which did not occur during exercise sessions reported. Overall, the incidence of adverse events was quite low, which suggesting that physical activity intervention was well-tolerated. However, only a few studies had reported on adverse events and the duration of trials ranged from 2 to 13 months. Therefore, the long-term safety of physical activity intervention needs further studies to prove.