Fecal microbiome alterations in treatment-naive de novo Parkinson's disease

Abstract

Gut microbiota alterations in Parkinson's disease (PD) have been found in several studies and are suggested to contribute to the pathogenesis of PD. However, previous results could not be adequately adjusted for a potential confounding effect of PD medication and disease duration, as almost all PD participants were already using dopaminergic medication and were included several years after diagnosis. Here, the gut microbiome composition of treatment-naive de novo PD subjects was assessed compared to healthy controls (HC) in two large independent case-control cohorts (n = 136 and 56 PD, n = 85 and 87 HC), using 16S-sequencing of fecal samples. Relevant variables such as technical batches, diet and constipation were assessed for their potential effects. Overall gut microbiome composition differed between PD and HC in both cohorts, suggesting gut microbiome alterations are already present in de novo PD subjects at the time of diagnosis, without the possible confounding effect of dopaminergic medication. Although no differentially abundant taxon could be replicated in both cohorts, multiple short chain fatty acids (SCFA) producing taxa were decreased in PD in both cohorts. In particular, several taxa belonging to the family Lachnospiraceae were decreased in abundance. Fewer taxonomic differences were found compared to previous studies, indicating smaller effect sizes in de novo PD.

Fig. 1. Intra-sample (alpha diversity) and inter-sample differences in microbial community structure between PD and HC.

a Alpha diversity indices indicate increased intra-sample diversity in PD in the NL cohort, whereas in b intra-sample diversity in PD is reduced in the FIN cohort. Each box represents the first quartile, median and third quartile at the lower, middle and upper boundaries, with the whiskers representing points within 1.5 times the interquartile range and the red diamond representing the mean. Each point represents one sample. Univariable, uncorrected (Mann-Whitney U test) p-values are 0.016, 0.016, 0.060 and 0.56 for the NL cohort and 4.3E-03, 4.2E-03, 7.8E-04, and 3.9E-03 for the FIN cohort for, respectively, observed richness, Chao1, Shannon and Inverse Simpson. c, d Inter-sample differences in microbial community structure were visualized using a principal component analysis of the Aitchinson distance with each red point representing one HC sample and each blue point representing one PD sample. A statistically significant difference was found between PD and HC with p = 6.4E-03 in the NL cohort c, and p = 4.0E-04 in the FIN cohort d. Sample sizes: 136 PD and 85 HC for the NL cohort and 56 PD and 87 HC in the FIN cohort. p < 0.1; *p < 0.05; **p < 0.01; ***p < 0.001; NL cohort, Dutch case-control cohort; FIN cohort, Finnish case-control cohort; PD, Parkinson’s disease; HC, healthy control.

Fig. 2. Relative abundances of genera and families identified as differentially abundant in at least one of the two cohorts.

Relative abundances of genera identified with DESeq2 and/or ANCOM in at least one of the two cohorts are depicted in a for the NL cohort and b for the FIN cohort. Relative abundances of families identified with DESeq2 and/or ANCOM in at least one of the two cohorts are depicted in c for the NL cohort and d for the FIN cohort. Each box represents the first quartile, median and third quartile at the lower, middle, and upper boundaries, with the whiskers representing points within 1.5 times the interquartile range and the X representing the mean for HC (green) and PD (gray). Sample sizes: 131 PD and 84 HC subjects for the NL cohort and 55 PD and 87 HC subjects for the FIN cohort. NL cohort, Dutch case-control cohort; FIN cohort, Finnish case-control cohort; PD, Parkinson’s disease; HC, healthy control; DESeq2, Differential Expression analysis for Sequence count data; ANCOM, ANalysis of Composition of Microbiomes.