The PROVIT Study-Effects of Multispecies Probiotic Add-on Treatment on Metabolomics in Major Depressive Disorder-A Randomized, Placebo-Controlled Trial

Abstract

The gut-brain axis plays a role in major depressive disorder (MDD). Gut-bacterial metabolites are suspected to reduce low-grade inflammation and influence brain function. Nevertheless, randomized, placebo-controlled probiotic intervention studies investigating metabolomic changes in patients with MDD are scarce. The PROVIT study (registered at clinicaltrials.com NCT03300440) aims to close this scientific gap. PROVIT was conducted as a randomized, single-center, double-blind, placebo-controlled multispecies probiotic intervention study in individuals with MDD (n = 57). In addition to clinical assessments, metabolomics analyses (1H Nuclear Magnetic Resonance Spectroscopy) of stool and serum, and microbiome analyses (16S rRNA sequencing) were performed. After 4 weeks of probiotic add-on therapy, no significant changes in serum samples were observed, whereas the probiotic groups' (n = 28) stool metabolome shifted towards significantly higher concentrations of butyrate, alanine, valine, isoleucine, sarcosine, methylamine, and lysine. Gallic acid was significantly decreased in the probiotic group. In contrast, and as expected, no significant changes resulted in the stool metabolome of the placebo group. Strong correlations between bacterial species and significantly altered stool metabolites were obtained. In summary, the treatment with multispecies probiotics affects the stool metabolomic profile in patients with MDD, which sets the foundation for further elucidation of the mechanistic impact of probiotics on depression.

Keywords: NMR spectroscopy; butyrate; depression; gut–brain-axis; metabolomics; probiotics; randomized controlled trial.

Figure 1

Overview of the PROVIT study design. HAMD = Hamilton Rating Scale for Depression, BDI-II = Beck Depression Inventory-II, M.I.N.I. = Mini International Neuropsychiatric Interview, ICD10 = International Statistical Classification of Diseases and Related Health Problems, MDD = Major Depressive Disorder.

Figure 2

Untargeted metabolomics of the placebo (n = 29) and intervention group (n = 28) in serum. (A) OPLS-DA: T score of 4.8% and Orthogonal T score 2 of 9.4%. Showing minimal clustering Q² of 0.0787 and p-value of 0.02. Time points: (1) Admission t₀, (2) end of the trial t₂ in the placebo group. (B) OPLS-DA: T score of 4.2% and Orthogonal T score 2 of 8.6%, showing no significant clustering Q² of 0.0787 and p-value of 0.02. Time points: (1) Admission t₀, (2) end of the trial t₂ in the placebo group.

Figure 3

Untargeted metabolomics of the placebo (n = 29) and intervention groups (n = 28) in stool. (A) Principal component analysis (PCA), principal component (PC) 1 of 89.4% and PC2 of 9.1%. (B) Untargeted metabolomics of the placebo (n = 29) and intervention group (n = 28) in stool. Partial least-squares–discriminant analysis (sPLS-DA) Scores plot: Component 1 of 18.9% and Component 2 of 11.6%.

Figure 4

Untargeted metabolomics (A) of the placebo group (n = 29) in stool. Features indicating most distinctive metabolites. A heatmap is illustrated on the right side next to the table with the listed features, showing changes in concentration. Time points: (1) Admission t₀, (2) two-week follow up t₁, (3) end of the trial t₂ in the placebo group (n = 28). (B) sPLS-DA scores plot of the placebo group (n = 29) in stool: Component 1 of 3.7% and Component 2 of 14.3%. (C) Untargeted metabolomics of the probiotics group (n = 28) in stool. Features indicating most distinctive metabolites. A heatmap is illustrated on the right side next to the table with the listed features, showing changes in concentration. Time points: (4) Admission t₀, (5) two-week follow-up t₁, (6) end of the trial t₂ in the intervention group (n = 29). (D) sPLS-DA scores plot of the probiotic group (n = 28): Component 1 of 19.9%, Component 2 of 9.1%.

Figure 5

Volcano plot: Probiotics group longitudinal pairwise comparisons of time point t₀ and t₂ in stool. The log₂ fold change (log2 FC) (x-axis) is plotted against the corresponding adjusted p-value (y-axis). Thresholds for significance (adjusted p-value = 0.05; horizontal dashed line) and changes in concentration, vertical dashed lines, are shown. Significantly increased metabolites in the probiotics group (n = 28) are displayed as red dots after 4 weeks compared to the beginning. Insignificantly altered metabolites are shown as gray dots.

Figure 6

Volcano plot: Pairwise comparison placebo vs. probiotics time point t₂ in stool. The log₂ fold change (x-axis) is plotted against the corresponding adjusted p-value (y-axis). Thresholds for significance (adjusted p-value = 0.05; horizontal dashed line) and changes in concentration, vertical dashed lines, are shown. Significantly increased metabolites compared to probiotics at t₂ are displayed as dots in dark red on the right side. Significantly decreased metabolites compared to probiotics (n = 28) are displayed as light blue dots on the left side. Upregulated features are displayed as red dots on the right side, whereas downregulated features are displayed as blue dots on the left side. Insignificantly altered metabolites are shown as gray dots.

Figure 7

Top 25 correlations with the normalized concentrations of (A) valine, (B) butyric acid, (C) capric acid, (D) isoleucine, and 16S rRNA microbiome analysis data and the normalized concentrations of quantitatively assessed metabolites by NMR-Metabolomics.