Microbiome of Depression

Szabo de Edelenyi et al. (2021) wanted to begin testing the hypothesis that eating more fruits and vegetables would lower the amount of indole produced by gastrointestinal bacteria, resulting in a reduced risk for depression.  They selected a sample of 2,054 women ages 45-65 from within the participants of a larger prospective cohort study called Nutrinet-Santé, and then identified 297 of them who had symptoms of recurrent depression (as measured by a questionnaire given on two occasions about two years apart).  They matched each case with two controls from the same sample, according to some potential confounders.  They assessed indole production by measuring urinary 3-indoxylsulfate.  Multiple 24-hour dietary records were collected.  They found that the control group ate significantly more fruits and vegetables than the case group.  Those who ate less fruits and vegetables had higher urinary 3-indoxylsulfate, but urinary 3-indoxylsulfate was not significantly associated with depression.  The authors conclude by mentioning that other studies have also found low fruit and vegetable intake to be associated with depression, noting that their study design was not able to establish a causal relationship.  They hypothesize that it is still possible that indole could be a mediator of the relationship even though their study was not able to demonstrate this, and propose that future studies use a higher threshold to define depression, suggesting that the association between indole and depression might be found for people with more severe symptoms.  They were not able to examine this possibility in their study because the number of depressive cases they identified with more severe symptoms was not sufficient to provide the statistical power necessary to show the association.

This was a nested case-control study.  The researchers had the benefit of much of the work as far as recruitment and participant management including dietary records and urine collection being taken care of by the larger prospective cohort study.  As an exploratory study on a hypothesis that had not yet received research attention, the design enabled them to attempt to ascertain factors that could influence the success of future more resourced studies.  They were able to use conditional logistic regression to harness some of the statistical power of the size of the full cohort study while only having to measure and analyze the data from their smaller sample.  The study design was not able to establish causality, even if they had found the relationship they were looking for.  For one thing, it did not have the ability to determine that the low fruit and vegetable intake occurred prior to the high indole production or that the high indole production occurred before the depressive symptoms began.  Being a part of the larger cohort study helped to reduce selection bias, which increases the generalizability of the results since the larger study was not even looking at depression.  But making the depression questionnaire voluntary could have been a source of bias; they don’t say what percentage of participants did the questionnaire, but maybe they accounted for this in the statistical analysis.  Because the larger study had collected data on so many potential confounders such as age, smoking, marriage, education, alcohol, fiber, physical activity, and bmi, they were able to adjust for all of these.  The validated methods used to collect the dietary data reduced the potential for information bias. 

This particular study does not give me a way to tell people that eating more vegetables will help with their depression, but isn’t it interesting that these researchers don’t even know whether not eating enough fruits and vegetables caused these women to be depressed, or being depressed caused them to not eat enough fruits and vegetables?

References:

 Szabo de Edelenyi, F., Philippe, C., Druesne-Pecollo, N., Naudon, L., Rabot, S., Hercberg, S., Latino-Martel, P., Kesse-Guyot, E., & Galan, P. (2021). Depressive symptoms, fruit and vegetables consumption and urinary 3-indoxylsulfate concentration: a nested case–control study in the French Nutrinet-Sante cohort. European Journal of Nutrition, 60(2), 1059–1069.