Impact of a Mediterranean Diet on Hepatic and Metabolic Outcomes in Non-alcoholic Fatty Liver Disease

The Medina Randomised Controlled Trial

Elena S. George; Anjana Reddy; Amanda J. Nicoll; Marno C. Ryan; Catherine Itsiopoulos; Gavin Abbott; Nathan A. Johnson; Siddharth Sood; Stuart K. Roberts; Audrey C. Tierney


Liver International. 2022;42(6):1308-1322. 

In This Article


Forty-two participants were included and 39 completed the study. A flow diagram of participation is shown in Figure 2. Of these 42 participants, 60% were female and the mean age was 52.3 ± 12.6 years. Eighteen participants had diabetes n = 7 in the MedDiet group and n = 11 in the LFD group. The cohort was multiethnic with a mean BMI of 32.2 ± 6.2 kg/m2 and 43% of participants had type 2 diabetes mellitus. Participants were recruited and randomised to the MedDiet (n = 19) or LFD (n = 23). The baseline demographic, clinical and biochemistry characteristics are shown in Table 1. At baseline visceral fat, heart rate and glucose were substantially higher in the LFD group compared to the MedDiet group.

Figure 2.

MEDINA participant flow chart

Liver Outcomes

Following 12 weeks of dietary intervention, there was a modest and statistically significant improvement in IHL in the LFD group representing approximately a 17% reduction on the log scale, (p = .02). In comparison, there was a non-significant reduction in the MedDiet group (−8%, p = .07). There was no significant difference between-groups even after adjusting for baseline values, diabetes status and visceral fat (p = .865). (Table 2). There was also no significant difference for LSM between-groups (p = .58) while within-group changes were also not significant (MedDiet: 7.8 ± 4.0 to 7.6 ± 5.2, p = .43; LFD: 11.8 ± 14.3 to 10.8 ± 10.2, p = .99). Liver enzymes including serum ALT, AST and GGT levels were not statistically significant following the MedDiet, however, significant reductions were noted in the LFD group, and the changes observed between-groups were significant (Table 3).

Insulin Resistance

There was a significant one-unit reduction of HOMA-IR following the LFD (6.5 ± 5.6 to 5.5 ± 5.5, p < .01) and a non-significant 0.5 unit reduction in the MedDiet group (4.4 ± 3.2 to 3.9 ± 2.3, p = .07) (Table 2). However, the baseline values for HOMA-IR were notably lower in the MedDiet group. The differences at the end of the intervention for HOMA-IR were not significant between-groups (p = .58).


Within both dietary intervention groups, weight, BMI and waist circumference were not significantly different from pre- to post-intervention. However, the weight reduction in the LFD group almost reached 5% (89.8 ± 24.5 kg to 85.8 ± 18.14 kg, p = .382); whereas in those in the MedDiet group gained 1.6 kg (87.7 ± 21.1 to 89.3 ± 22.8, p = .63). Visceral fat determined using BIA was reduced significantly in both groups; LFD ([log scale] -76%, p = <.0005), MedDiet (−61%, p = <.0005) although there were no between-group differences.

Dietary Intervention and Compliance

The MEDAS score used to assess adherence to the MedDiet and the equivalent score for the LFD were applied to each group's respective food diaries. Compliance with the MedDiet improved by 2.7 units (6.5 ± 2.0 to 9.2 ± 1.9, out of a maximum possible score of 14) (p < .0005). In the LFD group compliance with the prescribed diet improved by 1.0 unit (5.4 ± 2.0 to 6.4 ± 2.3, out of a maximum possible score of 9) (p = .035). In the MedDiet at the macronutrient level, there was a non-significant reduction in energy consumption from 9.2 to 8.4 MJ, this was accompanied by a significant reduction in carbohydrate intake, displaced with an increase in total fat (NS) and a significant increase in MUFAs. Conversely in the LFD, at the macronutrient level, there was a significant reduction in total fat, and while not significant the reduction in energy is of note (8.1 to 7 MJ). However, an in-depth dietary assessment highlighted further details with respect to compliance with key diet principles. Firstly, there were no significant differences with regard to energy intake following the intervention for either of the dietary arms, as expected given the focus was not on caloric deficit. Wholegrains, fruits and vegetables did not improve within or between-groups across the intervention period. Baseline intakes for these food groups, as shown in Table 4, were better than expected, compared to the consumption of the general Australian population.[34] The most prominent changes between-groups pertained to fat consumption whereby total fat, mono and polyunsaturated fatty acids were significantly increased in the MedDiet group, aligning more closely to dietary intervention targets. This was also affirmed with significant between-group increases in the MedDiet group at the end of the intervention in long-chain omega 3's (p = .035) and oil equivalents (predominantly because of an increase in extra virgin olive oil) (p = .01) in line with MedDiet recommendations.

Sensitivity Analysis

Fourteen of the between-group models were refitted with a single potentially influential observation (not always the same participant) removed from each. In these models, the results were largely unchanged in terms of overall interpretation, however, in contrast to the primary analysis there was statistically significantly higher HOMA-IR values (B = 0.28 [95%% CI: 0.05, 0.51], p = .015) for MedDiet compared to LFD at end-intervention.