Association between cognitive functioning and microbiota-gut-brain axis mediators in a memory clinic population

Abstract

Introduction: A growing body of evidence recognises the role of signaling molecule of the microbiota-gut-brain axis (MGBA) in cognitive impairment (CI), but data on the link with alterations in specific cognitive domains are limited. We compared the functioning in several cognitive domains (i.e., memory, visuo-constructional, executive, and language) among cognitively unimpaired (CU) subjects, patients with CI due to Alzheimer’s disease (CI-AD) and not due to AD (CI-NAD). Then, we investigated the association of these cognitive domains with the gut microbiota (GM), MGBA mediators, and neurodegeneration-related markers.

Materials and methods: The study included 34 CI-AD, 38 CI-NAD, and 13 CU. Memory, visuo-constructional, executive, and language domains were assessed using composite measures. Faecal GM composition was inferred using 16S rRNA gene sequencing. MGBA mediators included the blood quantification of bacterial products (lipolysaccharide, LPS), cell adhesion molecules indicative of endothelial damage, vascular changes or overexpressed in response to infections, and pro- and anti-inflammatory cytokines. Neurodegeneration-related markers included plasma phosphorylated tau (p-tau181), neurofilament light chain (NfL), and glial fibrillary protein (GFAP).

Results: The CI-NAD and CI-AD groups had significantly lower scores than the CU group for all cognitive domains (p < 0.043). Associations of MGBA modulators with cognitive functioning included pro-inflammatory cytokines, markers of endothelial dysfunction or overexpressed in response to infection in both groups of patients (|ρ| > 0.33, ps < 0.042). In the CU and CI-AD pooled group, lower cognitive functioning was specifically associated with higher abundance of Dialister and Clostridia_UCG-014, higher levels of LPS and with all neurodegeneration markers (|ρ| > 0.32, p < 0.048 for all). In the CU and CI-NAD pooled group, lower cognitive performance was associated with lower abundance of Acetonema, higher abundance of Bifidobacterium, [Eubacterium]_coprostanoligenes_group and Collinsella, and higher levels of vascular changes (|ρ| > 0.30, p < 0.049).

Discussion: These results support the hypothesis that gut dysbiosis and MGBA mediators may have distinct effects on cognitive functioning and different mechanisms of action depending on the disease.

Keywords: Alzheimer’s disease; cognitive function; dementia; gut microbiota; microbiota-gut-brain axis.

Figure 1

Violin plot of the distribution of cognitive domains’ z scores for the three groups (CU, CI-NAD, CI-AD). The plots show the median (indicated by the black horizontal band), the first through the third interquartile range (the vertical band), and an estimator of the density (thin vertical curves) of each cognitive domain functioning in each group. The reported p-values were calculated by using one-way ANOVA with Bonferroni’s correction for normal distributed variables (i.e., visuo-constructional domain) or Kruskall-Wallis test with Bonferroni’s correction for non-normal distributed variables (i.e., memory domain, executive domain, and language domain). CI-AD, patients with cognitive impairment due to AD; CI-NAD, patients with cognitive impairment not due to AD; CU, cognitively unimpaired persons.

Figure 2

GM and MGBA putative mediators of study participants. Bars denote percentage difference in CI-AD and CI-NAD patients versus unimpaired control subjects (CU). The percentage difference has been calculated using control subjects as reference (represented by the threshold line at 0). p-values were calculated by using one-way ANOVA or Kruskall-Wallis test (accordingly with data distribution) with Bonferroni’s test for multiple comparison correction on raw data. Statistical significances is represented by * at p < 0.05, *** at p < 0.001 comparing CI-AD and CI-NAD versus CU and by # at p < 0.05 comparing CI-AD versus CI-NAD. CI-AD, patients with cognitive impairment due to AD; CI-NAD, patients with cognitive impairment not due to AD; CU = cognitively unimpaired persons.

Figure 3

Heatmap of the Spearman’s rho coefficient values (red: positive; green: negative) indicating significant age-adjusted association in CU and CI-NAD or CU and CI-AD (p < 0.05, for exact p-values and confidence intervals please refer to Supplementary Table 2). Asterisks indicate moderate associations ( ρ  > 0.4). For cognition domains, higher values reflected better cognitive performance. CAMs, cell adhesion molecules; CI-AD, patients with cognitive impairment due to AD; CI-NAD, patients with cognitive impairment not due to AD; CU, cognitively unimpaired persons; NMs, neurodegenerative-related markers.

Figure 4

Closest associations (ρ > 0.4) between faecal bacterial genera, microbiota-gut-brain axis mediators, neurodegeneration-related markers and cognitive profile in CU and CI-NAD (A) and CU and CI-AD (B). Black lines indicate common paths for both groups, whereas red lines indicate paths specific to CI-NAD or CI-AD.