*Study Limitations*

This was a pilot study with a small sample size. Possibly because of this, most associations lost statistical significance after correcting for multiple testing. However, results were in line with previous research, suggesting that a larger sample size would have retrieved significant results. In addition, the fact that we did not find di fferences in the mMDS adherence between groups may reflect selection bias since this is a sub-study within 400 firefighters participating in the Feeding America's Bravest

trial, and participants willing to participate could potentially be healthier and more health conscious. Moreover, the control group had higher scores of mMDS at baseline and their scores were slightly improved during the intervention; however, the results were consistent in the cross-sectional analysis. We only adjusted for age and sex, since it was a randomized study with no significant di fferences between this sub-study and the parent study for the rest of the variables which suggests the need to perform a study that includes larger metabolites. In addition, we did not analyze other metabolites included in other studies nor at baseline for the parent study. In this pilot trial, Group 2 could be considered as the intervention group but Group 1 was not a pure control group because they already finished an active MedDiet intervention and began their self-sustained MedDiet phase. The changes in biomarkers in Group 1 (between time point 1 and 2) are more likely to reflect both residual e ffects of the MedDiet and continued e ffects from self-sustained diet intervention. Additionally, most of our participants were male and thus generalizability should be explored. In any case, the results in this study should be corroborated in larger clinical studies with longer follow up due to the pilot study nature and with a powerful study design.
