| |
Mediterranean/Low Carb Diet Superior in Reducing Hepatic
Fat for Fatty Liver/NAFLD & Non-NAFLD too
|
| |
| |
Download the PDF here
• After controlling for VAT changes, MED/LC diet tended to decrease %HFC (hepatic, liver fat, content-more than LF (low-fat) diet after 6 months [MED/LC: -7.3±9.2% vs. LF: -5.8±7.2% (absolute units), p=0.079 between diets], a differential effect between the dietary intervention groups that became significant at 18 months of intervention [MED/LC: -4.2±7.1% vs. LF: -3.8±6.7% (absolute units), p=0.036 between diets]. Furthermore, the advantageous effect of MED/LC on HFC reduction over LF diet was significant even in non-NAFLD patients (HFC≤5%, p=0.037), as in patients with NAFLD (HFC>5%, p=0.014). No significant differences were observed between the PA -exercise- groups (p=0.32) for HFC changes after 18 months, with or without adjustment for VAT changes.
• The beneficial effect of Mediterranean diet over the low-fat diet is mainly mediated by decreases in hepatic fat rather than visceral fat loss.
• The relative reductions of %HFC induced by the intervention were higher in males and in patients with BMI≥30 or VAT≥30%, even after controlling for 18-month VAT changes. Interestingly.
• Decreased %HFC directly correlated with loss in all three layers of abdominal sub-depots after 6 and 18 months, when the models were adjusted for age and sex (p<0.001 for all). However, when the models were further adjusted for weight loss, decreased %HFC remained associated only with reduction of VAT at 6 (β=0.232; 95%CI 0.13-0.34) and 18 months (β=0.155; 95%CI 0.04-0.31), but not with deep-SAT or superficial-SAT changes at 6 and 18 months (p>0.54 for all).
• Compared to the LF diet, MED/LC diet induced a greater increase in HDL-c (3.3±7.5 vs. 5.6±7.1 mg/dl), and a more pronounced decrease in diastolic BP (1.2±10.1 vs. -1.9±7.5 mmHg), triglycerides (-3.4±43.7 vs. -10.8±28.0 mg/dl), TG/HDL ratio (-0.15±0.4 vs. -0.23±0.4) and cardiovascular risk
• Previous studies have demonstrated the inter-relationship between VAT and hepatic fat content (HFC), and indeed, increases in HFC were associated with similar metabolic abnormalities as observed for increases in VAT.
• In the present sub-study we hypothesized that low-fat (LF) and Mediterranean/low-carbohydrate (MED/LC), similarly-hypocaloric diets, differ in their capacity to induce HFC loss, which mediates the improvements in cardiometabolic parameters independently of the impact of accompanying decreases in VAT.
• In an 18-month weight-loss trial, 278 participants with abdominal obesity/dyslipidemia were randomized to low-fat (LF) or Mediterranean/low-carbohydrate (MED/LC+28g walnuts/day) diets with/without moderate physical activity (PA). HFC and abdominal fat-depots were measured using magnetic-resonance-imaging at baseline, after 6 (sub-study, n=158) and 18-months.
• At Baseline, participants (mean age=48years, 89% men, BMI=30.8±3.8kg/m2) had 10.2% HFC (median=6.38%), widely ranging between 0.01% and 50.4%. Of the 278 participants, 53% had NAFLD (HFC above 5%), 40% met the criteria for the Metabolic Syndrome, 75% had abnormal WC and 11% were diabetic. Few participants used medications chronically (anti-platelet=7%, anti-hypertensive=8%, lipid-lowering=12%, oral glycemic-control=3% and insulin treatment=1%), with minor changes during the intervention, that were similar between groups. Characteristics of the CENTRAL study population across intervention groups are shown in Table 1. There were no significant differences at baseline between the intervention groups in demographic variables, consumption of energy intake and macronutrients, blood markers, HFC, or abdominal fat sub-depots, but only in VAT area in female.
• %HFC substantially decreased after 6 [-6.6% absolute-units (-41% relatively)] and 18-months [-4.0% absolute-units (-29% relatively);p<0.001 vs. baseline]. Reduction of HFC associated with decreases in VAT beyond weight loss. After controlling for VAT loss, decreased %HFC remained independently associated with reductions in serum gamma-glutamyl-transferase and alanine-aminotransferase, circulating chemerin, and HbA1c (p<0.05). While reduction of HFC was similar between PA groups, compared to LF diet, MED/LC induced a greater %HFC decrease (p=0.036) and greater improvements in cardiometabolic risk parameters (p<0.05), even after controlling for VAT changes. Yet, the greater decreases induced by MED/LC compared to LF diets in triglycerides, TG/HDL ratio and cardiovascular risk score were all markedly attenuated when controlling for HFC changes.
• "In this CENTRAL 18-months intervention trial, Mediterranean/low-carbohydrate diet induced a greater decrease in hepatic fat content than low-fat diet, and the beneficial health effects were beyond the favorable effects of visceral fat loss.”
• In the entire cohort, HFC substantially decreased after 6 [-6.6% absolute-units (-41% relatively)] and 18 months [-4.0% absolute-units (-29% relatively)] (p<0.001 vs. baseline), along with moderate body weight loss (-5.8% and -3.1% after 6 and 18 months, respectively). 18m retention rate was 86.3%. Decreased %HFC directly correlated with loss in all three layers of abdominal sub-depots after 6 and 18 months, when the models were adjusted for age and sex (p<0.001 for all). However, when the models were further adjusted for weight loss, decreased %HFC remained associated only with reduction of VAT at 6 (β=0.232; 95%CI 0.13-0.34) and 18 months (β=0.155; 95%CI 0.04-0.31), but not with deep-SAT or superficial-SAT changes at 6 and 18 months (p>0.54 for all).
• After controlling for VAT changes, MED/LC diet tended to decrease %HFC more than LF diet after 6 months [MED/LC: -7.3±9.2% vs. LF: -5.8±7.2% (absolute units), p=0.079 between diets], a differential effect between the dietary intervention groups that became significant at 18 months of intervention [MED/LC: -4.2±7.1% vs. LF: -3.8±6.7% (absolute units), p=0.036 between diets]. Furthermore, the advantageous effect of MED/LC on HFC reduction over LF diet was significant even in non-NAFLD patients (HFC≤5%, p=0.037), as in patients with NAFLD (HFC>5%, p=0.014). No significant differences were observed between the PA groups (p=0.32) for HFC changes after 18 months, with or without adjustment for VAT changes.
• Compared to the LF diet, MED/LC diet induced a greater increase in HDL-c (3.3±7.5 vs. 5.6±7.1 mg/dl), and a more pronounced decrease in diastolic BP (1.2±10.1 vs. -1.9±7.5 mmHg), triglycerides (-3.4±43.7 vs. -10.8±28.0 mg/dl), TG/HDL ratio (-0.15±0.4 vs. -0.23±0.4) and cardiovascular risk by the three different scores: Framingham (-0.27±2.2 vs. -0.81±1.9), SCORE (-0.16±1.4 vs. -0.50±1.2), and ACC/AHA score (-0.39±2.7 vs. -1.13±2.5), (p<0.05 for all, Figure 4- white and black bars). These differences remained significant when body weight and VAT changes were added to the multivariate model (Figure 4- colored bars). However, after adjustment for HFC changes, differences between diets were significantly attenuated, particularly the changes in triglycerides, TG/HDL ratio and in the cardiovascular risk scores. When data was stratified for within and above normal HFC at baseline, the beneficial effect of the MED/LC diet over the LF in reducing cardiovascular risk scores became insignificant, possibly due to the lower power of the analysis. Nevertheless, similar trends were noted in both the normal HFC and abnormal HFC sub-groups (Supp 2a and 2b, respectively). There was no significant effect by the PA intervention on improvement of cardiometabolic parameters or cardiovascular risk.
• The changes in HFC over 18-month of intervention across different subgroups of the cohort are shown in Figure 1. Higher HFC at baseline was found, as expected, in males (10.7% vs. 5.8%, p=0.001), in participants with BMI≥30 (12.7% vs. 7.0%, p<0.001) and in those with VAT>30% at baseline (11.5% vs. 8.0%, p=0.007). The relative reductions of %HFC induced by the intervention were higher in males and in patients with BMI≥30 or VAT≥30%, even after controlling for 18-month VAT changes. Interestingly, in a model adjusted for weight loss, the beneficial effect of MED/LC diet over the LF diet was more apparent among males (p=0.016) and in participants with VAT over 30% at baseline (p=0.018), but similarly in both BMI groups. We further performed sensitivity analyses among the participants that completed both 6- and 18-months MRI-body fat measurements (i.e., 6 month sub-study), and a similar pattern was observed (data not shown). Similar results were also found when excluding participants using insulin.
• These mechanisms - see just below - may also explain the superiority of the MED/LC diet, including a daily intake of walnuts, over the LF diet, regarding the reduction in HFC. Thus, our study highlights the specific potential of MED/LC as a particular dietary strategy to treat NAFLD.
Discussion
In this long-term lifestyle intervention trial, the MED/LC induced a significantly greater decrease in HFC than the LF diet, even after accounting for the differences in VAT loss. The impact of HFC reduction is highlighted by associated improvements in GGT, ALT, chemerin and HbA1c, which remained significant after adjustment for total weight loss or VAT change. In addition, the MED/LC diet was superior to the LF diet in decreasing cardiometabolic risk, a difference that was attenuated when adjusting for the decrease in HFC, but not following adjustment for weight or VAT.
In summary, this sub-study demonstrates how different weight loss strategies may induce favorable dynamics of HFC and consequently improve cardio-metabolic risk. We suggest that improvements in specific easily-tracked blood biomarkers and cardiovascular risk associated with a decrease in HFC, beyond the loss of VAT. Thus, rather than focusing on weight loss only, our findings suggest that LC/MED dietary intervention may be used as a specific approach for the management of NAFLD.
Our analyses suggest that HFC changes, rather than VAT changes, may play a particular role in mediating the greater beneficial effects of MED/LC over the LF dietary intervention. After adjusting for HFC changes, the differences in the association of the diets with improvements in lipid profile and in the cardiovascular risk scores became statistically insignificant. We did not observe such attenuation when controlling for weight or VAT loss. Recent long-term dietary interventions and meta-analyses have shed light on the ability of low-carbohydrate and Mediterranean [14, 15, 49, 50] diets to serve as alternatives to traditional low-fat diets in inducing weight loss and improved cardio-metabolic profile. A previous cross-sectional analysis has also suggested a stronger association of HFC than of VAT with obesity-related cardio-metabolic risk [8]. Moreover, another study that matched subgroups of patients with similar VAT but different HFC, suggested that increased HFC was cross-sectionally associated with insulin resistance [7]. Our current randomized trial results strengthen the notion, and provide evidence, supporting the unique impact of HFC on such risk, showing that these associations occur in response to intervention and are not merely cross-sectional observations. Our finding are also in line with results from mechanistic, fat transplantation studies in mice, in which mesenteric (portally-drained), but not parietal peritoneal (systemically-drained via the vena cava) transplantation induced worse metabolic outcome.[51, 52] However, in humans, conflicting results were obtained on the putative metabolic benefit of omentectomy (surgical VAT reduction) during bariatric surgery.[53] Thus, although HFC partially reflects a downstream consequence of increased VAT, our results strengthen the notion that HFC mechanistically contributes to cariometabolic risk independently of VAT. Moreover, they highlight the potential value of interventions specifically targeting the hepatic manifestations of obesity, such as LC/MED diet, in diminishing health risks associated with obesity.
The amount of HFC accumulation depends, among other things, on an interaction between hepatic FA uptake, derived from plasma free fatty acid (FFA) released from TG hydrolysis in adipose tissue and circulating TG, and de novo lipogenesis (DNL) [54]. It has been demonstrated that LF, high-carbohydrate diet increased hepatic DNL significantly compared to an isocaloric high-fat, low-carbohydrate diet [55]. Moreover, it is well-established that excessive consumption of sugar, and fructose in particular, leads to dietary carbons channeling directly to the liver, supporting DNL [56]. These mechanisms may also explain the superiority of the MED/LC diet, including a daily intake of walnuts, over the LF diet, regarding the reduction in HFC. Thus, our study highlights the specific potential of MED/LC as a particular dietary strategy to treat NAFLD.
Physical Activity Intervention
At the second randomization (after 6 months of dietary intervention), participants assigned to added PA received a free supervised gym membership for the following 12 months. The exercise intervention included monthly educational workshops, and one hour of exercise, three times/week. Participants were guided to engage in 45 minutes of aerobic training at 80% of maximum heart rate and 15 minutes of resistance training at 80% of the one-repetition maximum (1RM) of the weight.
Association between %HFC loss and nutritional intake changes
Overall, during the intervention, participants significantly decreased their energy intake after 6 and 18 months (p<0.001 vs. baseline), but similarly across diet groups [total calorie intake (-26% vs. -22% in the MED/LC diet vs LF diet, respectively, p=0.18)]. Changes in the intake of marco- and micro-nutrients compared to baseline are shown in Figure 2. After 18 months of intervention the MED/LC diet greatly decreased intake of carbohydrate and trans fat, while the LF diet decreased more the intake of total fat, monounsaturated fat and cholesterol, and tended to decrease more polyunsaturated and saturated fats (p<0.05 for all, Figure 2a). In addition, the MED/LC diet increased nuts consumption (p<0.05, Figure 2b). Decrease of HFC after 18 months correlated with decreased carbohydrate intake (r=0.175, p=0.009), and with increased fat intake (r=-0.217, p=0.001), as proportions of total calorie intake.
To compare the impact of losses of HFC, VAT and total weight on improvement in cardiometabolic parameters induced by MED/LC versus LF, we determined how adjustment for those parameters attenuated the differences between the two dietary interventions. Compared to the LF diet, MED/LC diet induced a greater increase in HDL-c (3.3±7.5 vs. 5.6±7.1 mg/dl), and a more pronounced decrease in diastolic BP (1.2±10.1 vs. -1.9±7.5 mmHg), triglycerides (-3.4±43.7 vs. -10.8±28.0 mg/dl), TG/HDL ratio (-0.15±0.4 vs. -0.23±0.4) and cardiovascular risk by the three different scores: Framingham (-0.27±2.2 vs. -0.81±1.9), SCORE (-0.16±1.4 vs. -0.50±1.2), and ACC/AHA score (-0.39±2.7 vs. -1.13±2.5), (p<0.05 for all, Figure 4- white and black bars). These differences remained significant when body weight and VAT changes were added to the multivariate model (Figure 4- colored bars). However, after adjustment for HFC changes, differences between diets were significantly attenuated, particularly the changes in triglycerides, TG/HDL ratio and in the cardiovascular risk scores. When data was stratified for within and above normal HFC at baseline, the beneficial effect of the MED/LC diet over the LF in reducing cardiovascular risk scores became insignificant, possibly due to the lower power of the analysis. Nevertheless, similar trends were noted in both the normal HFC and abnormal HFC sub-groups (Supp 2a and 2b, respectively). There was no significant effect by the PA intervention on improvement of cardiometabolic parameters or cardiovascular risk.
-----------------------------
The Beneficial effects of Mediterranean diet over low-fat diet may be mediated by decreasing hepatic fat content
Highlights
• Mediterranean and low carbohydrate diet greater decrease hepatic fat than the LF diet, beyond visceral fat changes.
• Decrease in hepatic fat is independently associated with specific improved parameters.
• The beneficial effect of Mediterranean diet over the low-fat diet is mainly mediated by decreases in hepatic fat rather than visceral fat loss.
Dynamics of HFC throughout the intervention
In the entire cohort, HFC substantially decreased after 6 [-6.6% absolute-units (-41% relatively)] and 18 months [-4.0% absolute-units (-29% relatively)] (p<0.001 vs. baseline), along with moderate body weight loss (-5.8% and -3.1% after 6 and 18 months, respectively).
18m retention rate was 86.3%.
Decreased %HFC directly correlated with loss in all three layers of abdominal sub-depots after 6 and 18 months, when the models were adjusted for age and sex (p<0.001 for all).
However, when the models were further adjusted for weight loss, decreased %HFC remained associated only with reduction of VAT at 6 (β=0.232; 95%CI 0.13-0.34) and 18 months (β=0.155; 95%CI 0.04-0.31), but not with deep-SAT or superficial-SAT changes at 6 and 18 months (p>0.54 for all).
After controlling for VAT changes, MED/LC diet tended to decrease %HFC more than LF diet after 6 months [MED/LC: -7.3±9.2% vs. LF: -5.8±7.2% (absolute units), p=0.079 between diets], a differential effect between the dietary intervention groups that became significant at 18 months of intervention [MED/LC: -4.2±7.1% vs. LF: -3.8±6.7% (absolute units), p=0.036 between diets].
Furthermore, the advantageous effect of MED/LC on HFC reduction over LF diet was significant even in non-NAFLD patients (HFC≤5%, p=0.037), as in patients with NAFLD (HFC>5%, p=0.014).
No significant differences were observed between the PA groups (p=0.32) for HFC changes after 18 months, with or without adjustment for VAT changes. The changes in HFC over 18-month of intervention across different subgroups of the cohort are shown in Figure 1.
Higher HFC at baseline was found, as expected, in males (10.7% vs. 5.8%, p=0.001), in participants with BMI≥30 (12.7% vs. 7.0%, p<0.001) and in those with VAT>30% at baseline (11.5% vs. 8.0%, p=0.007).
The relative reductions of %HFC induced by the intervention were higher in males and in patients with BMI≥30 or VAT≥30%, even after controlling for 18-month VAT changes.
Interestingly, in a model adjusted for weight loss, the beneficial effect of MED/LC diet over the LF diet was more apparent among males (p=0.016) and in participants with VAT over 30% at baseline (p=0.018), but similarly in both BMI groups. We further performed sensitivity analyses among the participants that completed both 6- and 18-months MRI-body fat measurements (i.e., 6 month sub-study), and a similar pattern was observed (data not shown). Similar results were also found when excluding participants using insulin.
Abstract
Background and Aim
It is unclear if reduction in hepatic fat content (HFC) is a major mediator of the cardiometabolic benefit of lifestyle intervention, and whether it holds prognostic information beyond visceral adipose tissue (VAT) loss. In the present sub-study, we hypothesized that HFC loss in response to dietary interventions induces specific beneficial effects independent of VAT changes.
Methods
In an 18-month weight-loss trial, 278 participants with abdominal obesity/dyslipidemia were randomized to low-fat (LF) or Mediterranean/low-carbohydrate (MED/LC+28g walnuts/day) diets with/without moderate physical activity (PA). HFC and abdominal fat-depots were measured using magnetic-resonance-imaging at baseline, after 6 (sub-study, n=158) and 18-months.
Results
Of 278 participants [age=48yr;88% men; body-mass-index=30.8kg/m2; mean HFC =10.2%,(range:0.01%-50.4%)], retention rate was 86.3%. %HFC substantially decreased after 6 [-6.6% absolute-units (-41% relatively)] and 18-months [-4.0% absolute-units (-29% relatively);p<0.001 vs. baseline]. Reduction of HFC associated with decreases in VAT beyond weight loss. After controlling for VAT loss, decreased %HFC remained independently associated with reductions in serum gamma-glutamyl-transferase and alanine-aminotransferase, circulating chemerin, and HbA1c (p<0.05). While reduction of HFC was similar between PA groups, compared to LF diet, MED/LC induced a greater %HFC decrease (p=0.036) and greater improvements in cardiometabolic risk parameters (p<0.05), even after controlling for VAT changes. Yet, the greater decreases induced by MED/LC compared to LF diets in triglycerides, TG/HDL ratio and cardiovascular risk score were all markedly attenuated when controlling for HFC changes.
Conclusions
%HFC is substantially reduced by diet-induced moderate weight loss, more effectively by MED/LC diet, beyond VAT changes. HFC loss is associated with specific improved parameters. Beneficial effects of MED/LC diet may be largely mediated by decreases in %HFC rather than VAT loss.
Lay Summary
High hepatic fat content is associated with metabolic syndrome, type two diabetes mellitus, and coronary heart disease. In this CENTRAL 18-months intervention trial, Mediterranean/low-carbohydrate diet induced a greater decrease in hepatic fat content than low-fat diet, and the beneficial health effects were beyond the favorable effects of visceral fat loss.
Results
Baseline characteristics
At Baseline, participants (mean age=48years, 89% men, BMI=30.8±3.8kg/m2) had 10.2% HFC (median=6.38%), widely ranging between 0.01% and 50.4%. Of the 278 participants, 53% had NAFLD (HFC above 5%), 40% met the criteria for the Metabolic Syndrome, 75% had abnormal WC and 11% were diabetic. Few participants used medications chronically (anti-platelet=7%, anti-hypertensive=8%, lipid-lowering=12%, oral glycemic-control=3% and insulin treatment=1%), with minor changes during the intervention, that were similar between groups. Characteristics of the CENTRAL study population across intervention groups are shown in Table 1. There were no significant differences at baseline between the intervention groups in demographic variables, consumption of energy intake and macronutrients, blood markers, HFC, or abdominal fat sub-depots, but only in VAT area in female.
Introduction
Beyond total body fat content, fat distribution, both within adipose tissue depots and in ectopic fat deposits, is increasingly shown to determine obesity-related health impact [1, 2]. Visceral adipose tissue (VAT), due to its unique anatomical location, releases free fatty acids (FFA) and adipokines to the liver via the portal vein. Previous studies have demonstrated the inter-relationship between VAT and hepatic fat content (HFC), and indeed, increases in HFC were associated with similar metabolic abnormalities as observed for increases in VAT [3, 4]. In addition, reduction in VAT and HFC are increasingly thought to mediate beneficial cardiometabolic outcomes of weight loss [1, 5]. Though closely associated with HFC, VAT and HFC may uniquely associate with specific effects and independently be linked with cardiometabolic disease risk factors [6]. Interestingly, data from recent studies found that HFC was more strongly associated with obesity’s metabolic complications than VAT [7], including the deterioration of glucose tolerance [8], possibly by mediating the link between obesity and metabolic dysfunction [9, 10]. Most recently, the decrease in HFC was associated with diabetes remission [11] .
Diet plays an important role in the accumulation of HFC and VAT [12]. Several short [6, 13] and long-term [14, 15] dietary interventions have suggested that Mediterranean and low-carb diets had favorable effects on their accumulation, but also on glycemic status and lipid biomarkers. Others found no differences between HFC changes induced by diets with different amounts of carbohydrate [16]. The effect of long-term specific lifestyle interventions on HFC and its association with the dynamics of cardiometabolic risk, beyond VAT loss, remains unclear. Notably, recent guidelines for decreasing HFC do not suggest a particular lifestyle strategy, but only endorse weight loss as a general recommendation [17].
In the present sub-study we hypothesized that low-fat (LF) and Mediterranean/low-carbohydrate (MED/LC), similarly-hypocaloric diets, differ in their capacity to induce HFC loss, which mediates the improvements in cardiometabolic parameters independently of the impact of accompanying decreases in VAT.
|
|
| |
| |
|
|
|
