In this cross-sectional study, we observed that NAFLD was associated with higher intake levels of red meat. Significant associations of serum levels of GGT, ALT, AST, TG and HDL-C with meat subtypes intakes were found as well. Additionally, no significant interactions between meat consumptions and potential confounders for NAFLD were detected.
Significant associations between high meat consumptions and NAFLD were demonstrated in a few studies [7, 9, 10, 26]. Our results are in accordance with the previous studies, indicating a positive association between high red meat intake and NAFLD. Two cross-sectional studies have presented that red meat was significantly correlated with NAFLD [7, 10]. In another cross-sectional study, high intakes of total meat, especially red meat and/or processed meat were positively linked to NAFLD and insulin resistance, while processed meat alone was only relevant to insulin resistance. This is mainly due to a relatively low level of processed meat consumption in their research set [9]. In addition, a nested case-control study also showed that high consumptions of red meat, processed red meat and poultry were positively associated with NAFLD [26]. Due to the better accuracy of the continuous ORs [27], the dose-response analysis we have employed can better measure the overall trends of the ORs for meat intakes. The 95% CI for red meat intake beyond 200 g/day was slightly wider, because that less than 2% of participants had red meat intake > 200 g/day, that was, the tendency of red meat intake within the range 0 ~ 200 g/day was relatively reliable and stable.
There are several plausible mechanisms by which meat intake is related to NAFLD. NAFLD was reported to be closely linked to hepatic insulin resistance, which had strong correlation with liver biomarkers such as ALT, AST and GGT [28]. GGT and ALT had been considered as biomarkers of hepatic fat accumulation, which can lead to hepatic insulin resistance and increase the contribution of gluconeogenesis to total endogenous glucose production [29]. A cross-sectional study of 2198 European reported a significant positive association between red meat and GGT. As GGT is also a potential nonspecific marker of oxidative stress, the author suggested that oxidative stress may plays a vital role underlying the development of chronic diseases with red meat intake [30]. Another cross-sectional study indicated that TG/HDL-C was independently relevant to the risk of NAFLD. The author attributed this result to insulin resistance [31]. Positive associations of serum levels of GGT, ALT and TG with red meat intakes were found in this study. Inversely, serum HDL-C concentration was negatively relevant to red meat and processed meat intakes. Hence, it’s plausible that increased hepatic lipid accumulation and insulin resistance play a substantial role in the relationship of meat intakes with the development of NAFLD. In addition, a study by Avila et al. found that red meat was positively relevant to serum ferritin [32], which can increase the risk of NAFLD [33, 34]. Fried food will produce some hazardous chemicals, such as advanced glycation end products and trans fatty acids [35, 36], which were reported to play a critical role in NAFLD [37, 38].
Significant associations were found in several subgroups. A positive association was observed between red meat and NAFLD in males, perhaps it is because males had a higher meat intakes and a higher prevalence of fatty liver than females [39]. Moreover, several studies had found significant associations of high meat intakes with obesity, type 2 diabetes and hypertension [14, 17, 40], which were considered as risk factors in the development of NAFLD [19, 41, 42]. In our study, positive association of red meat intake with NAFLD was found in people with BMI≥24 kg/m2 and people with hypertension.
This study, however, had several limitations. Firstly, because of the cross-sectional study design, casual inference was not allowed. Secondly, measurement error was unavoidable for self-reported diet and other data. Nevertheless, since all participants and researchers in this study were blinded to the results of abdominal ultrasonography and blood test, a reporting bias without differences is likely only to attenuate our observed association. Thirdly, since the study subjects were middle-aged and elders, it should be cautious in generalizing our findings to the wider population. Lastly, although a comprehensive set of confounders were considered, as an observational study, the presence of unmeasured confounders is possible.