Gut dysbiosis narrative in psoriasis: matched-pair approach identifies only subtle shifts correlated with elevated fecal calprotectin

Abstract

Many studies have reported gut microbiome alterations in psoriasis patients, suggesting dysbiosis. While evidence for dysbiosis and its link to pathogenesis remains inconclusive, murine models of psoriasis suggest that gut microbiome alterations develop in response to psoriasis-like inflammation. Hence, the dominant narrative about gut microbiome alterations' impact on disease should be evaluated critically with more data and a well-powered approach. In this case-control study, we used deep sequencing of fecal samples from 53 psoriasis patients and 47 healthy donors to reconstruct the strain/species-level content of the gut microbiome. Unlike previous studies, we first identified matched pairs for each patient with healthy donors to adjust for microbiome variability and increase power. We found no evidence for depleted gut community diversity and apparent divergence in structure between patients and healthy individuals. However, our matched-pair approach identified a subtle but systematic increase in select bacteria among patients, e.g., Megasphaera elsdenii and Eubacterium CAG 180. We next showed that these enriched species were correlated with elevated biomarkers of intestinal and systemic inflammation and liver function. Functionally, one of the top species, Megasphaera elsdenii, is a potent lactate utilizer in the context of intestinal lactic acidosis and inflammation. While our findings hardly support overt dysbiosis in the large intestine, the observed microbial changes correlate with moderately elevated calprotectin, albeit at levels not enough to diagnose ongoing inflammation. Hence, the sources of elevated inflammatory markers in patients' intestines remain unclear and warrant further investigation to clarify their cause-and-effect relationship with the disease.

Importance: With sufficient taxonomic resolution and sample size, this study critically evaluates new and published data on the gut microbiome in psoriasis patients. It shows that observed taxonomic changes in patients are modest and do not meet strict criteria for gut dysbiosis, at least in the large intestine. Instead, observed taxonomic changes in psoriasis patients can be explained by the microbial response to possible low-grade inflammation with unknown localization in the intestine and unclear impact on the host. The authors point out that published endoscopic data point to the small intestine as the site of gut inflammation. Therefore, further research focused on the small intestine would be informative to clarify the hypothetical gut-psoriasis link.

Keywords: dysbiosis; gut microbiota; lactate; low-grade Inflammation; psoriasis.

Fig 1

Gut microbiome diversity and composition in patients (n = 53) and healthy controls (n = 47). (A) Alpha diversity (Shannon index) distribution among 53 patients and 47 healthy donors. The box represents the interquartile range, the horizontal line inside shows the median, the whiskers indicate lower to upper quartiles, and points outside the whiskers indicate potential outliers. No significant differences in diversity between the patients and control groups (P = 0.83, Kolmogorov-Smirnov two-sided hypothesis test). (B) Beta diversity (Bray-Curtis dissimilarity) distribution. The box-and-whiskers plot has the same elements as in panel A. No significant divergence between groups (P = 0.285). (C) PCA based on bacterial species proportions. Proportions were CLR transformed. PC1 and PC2 are given with the proportion of total variance explained. (D) ALDEx analysis. Median between-group differences stratified by within-group dispersion. The dashed diagonal line indicates points where within-group differences equal between-group differentiation. None of the between-group differences exceeded within-group dispersion, i.e., no group-level significant differences.

Fig 2

Differentially abundant bacteria in case-control matched pairs. (A) Schematic representation of matched-pair construction procedure and Wilcoxon test for matched pairs. Patients (red circles) and healthy controls (green circles) are paired based on similarity of microbial profiles, i.e., closeness on the Euclidean space based on 11 principal components of microbial abundance. Next, matched pairs are tested to find differentially abundant species using the Wilcoxon test: (B) differentially abundant bacteria in matched pairs (N = 47 matched pairs) ordered by ranks (P ≤ 0.05); (C) differentially abundant bacteria ordered by occurrence in matched pairs (N = 47 matched pairs) (P ≤ 0.05).

Fig 3

Fecal calprotectin levels in the study cohort and associated bacterial taxa. (A) Fecal calprotectin distribution among patients and healthy donors. (B) Bacterial taxa associated with fecal calprotectin levels and diagnosis. Association tests were deemed significant at a 10% FDR. Red and blue shadings indicate the extent of positive and negative association, and the plus sign shows statistically significant findings. (C) Relative abundance (fractions of 1) of bacterial taxa associated with diagnosis and ordered by increasing levels of fecal calprotectin in cases and controls.

Fig 4

Host biomarkers and their association with the top 14 bacterial taxa. (A) Least squares discriminant analysis of patients and controls based on the host biomarkers (blood biomarkers and fecal calprotectin). Informative host biomarkers are summarized in two components. Informative biomarkers are ordered based on their absolute loadings on the first and second components. Calprotectin is black; liver function-related biomarkers are brown; other blood biomarkers are gray. (B) Host biomarkers that correlate with the top psoriasis-associated species (see Fig. 2B). Positive (red) and negative (blue) Kendall’s tau correlation coefficients are shown with circle size proportional to coefficient value. Bacterial taxa that were previously found to be associated (matched-pair Wilcoxon test, see Fig. 2B) with psoriasis are shown in red and green when associated with healthy controls. Bacterial taxa previously found to be significantly associated with calprotectin (see Fig. 3B) are indicated in bold red.