Skin microbiota signature distinguishes IBD patients and reflects skin adverse events during anti-TNF therapy

Abstract

Crohn's disease (CD) and ulcerative colitis (UC) are two forms of inflammatory bowel disease (IBD), where the role of gut but not skin dysbiosis is well recognized. Inhibitors of TNF have been successful in IBD treatment, but up to a quarter of patients suffer from unpredictable skin adverse events (SkAE). For this purpose, we analyzed temporal dynamics of skin microbiota and serum markers of inflammation and epithelial barrier integrity during anti-TNF therapy and SkAE manifestation in IBD patients. We observed that the skin microbiota signature of IBD patients differs markedly from healthy subjects. In particular, the skin microbiota of CD patients differs significantly from that of UC patients and healthy subjects, mainly in the retroauricular crease. In addition, we showed that anti-TNF-related SkAE are associated with specific shifts in skin microbiota profile and with a decrease in serum levels of L-FABP and I-FABP in IBD patients. For the first time, we showed that shifts in microbial composition in IBD patients are not limited to the gut and that skin microbiota and serum markers of the epithelium barrier may be suitable markers of SkAE during anti-TNF therapy.

Keywords: 16S RNA sequencing; IBD; TNF-alpha antagonist; serum biomarker; skin adverse events; skin microbiota.

Figure 1

Skin microbiota diversity differs between CD, UC, and HC at the retroauricular crease. (A) Comparison of alpha diversity indices between CD, UC, and HC. Each point represents one sample. The differences were analyzed by Linear mixed effect models (LMM) and subsequent Tukey post hoc test. (B) Principal coordinate analysis of the Bray-Curtis distance between CD, UC, and HC. Each point represents one sample. Groups were compared by PERMANOVA. (C) Relative abundance of the top 10 most abundant taxa of skin microbiota at the genus and species level. In all panels: **p<0.01; ***p<0.001. ASVs, amplicon sequence variants; CD, Crohn’s disease; UC, ulcerative colitis; HC, healthy controls.

Figure 2

Specific microbiota pattern at the retroauricular crease distinguishes CD, UC and healthy controls. (A) Differential abundance analysis (DAA) of ASVs between CD and UC versus HC. Model estimates and 95% confidence intervals are shown. Positive and negative values indicate an increase and decrease in abundance, respectively, in the IBD groups compared with HC. Green color indicates ASVs abundance changes in CD compared to HC and blue color abundance changes between UC and HC. (B) Significant differences in taxa between CD and UC assessed by Differential abundance analysis. The taxa with significantly higher abundances in CD over UC patients are shown as negative estimates (color-coded in green), and the taxa with significantly higher abundances in UC over CD patients are shown as positive estimates (color-coded in blue). ASVs with relative abundance of more than 0.01% in all samples and detected in at least 10% samples were used in DAA analysis. CD, Crohn’s disease; UC, ulcerative colitis; HC, healthy controls.

Figure 3

Characteristics of the IBD patients suffering from SkAE and their microbiota of retroauricular crease. (A) Pie charts showing patient’s distribution into groups. (B) Timeline of Shannon diversity changes in patients suffering from SkAE. The particular time of SkAE manifestation is color-coded in red. Graphs show a shift of Shannon entropy in particular patient’s visits. (C) Principal coordinate analysis of the Bray-Curtis distance between samples from patients with SkAE. All samples from one patient are represented by the same symbol, and the order of sampling is indicated by a label in italics, i.e., 1 is the first sampling. SkAE manifestation is color-coded in red. Groups were compared by PERMANOVA. (D) Relative abundances of the most abundant taxa of skin microbiota at the genus and species level in patients with SkAE. Visit number, response to treatment, and the disease index for CD (HBI) or UC (pMayo) are shown. P7, P21, P27, P53, P6, P20, P25, and P26 represent the patient’s code designations. CD, Crohn’s disease; UC, ulcerative colitis. Response: B, baseline; P, partial; F, full; N, no response. HBI, Harvey-Bradshaw index. pMayo, partial Mayo score.

Figure 4

The analysis of I-FABP, L-FABP, and E-FABP in serum of patients with SkAE. (A) Comparison of L-FABP, I-FABP, and E-FABP levels in patients (n=7) at baseline, during the manifestation of SkAE and at the endpoint. For all panels: (^∗ p < 0.05; paired t-test). (B) Correlation heatmap showing the Spearman’s correlation coefficient of pairwise comparison between clinical parameters and biomarkers at baseline, during SkAE incidence, and at the study endpoint (burgundy, positive correlation; turquoise, negative correlation). The heatmap was constructed in GraphPad Prism, version 8.4.3.