Skin microbiome signatures associated with psoriasis and seborrheic dermatitis
Rong Tao,Ruoyu Li,Zhe Wan,Yan Wu,Ruojun Wang
DOI: https://doi.org/10.1111/exd.14618
2022-01-01
Experimental Dermatology
Abstract:Dear Editor, Psoriasis and seborrheic dermatitis (SD) are common skin inflammatory diseases, which were difficult to discriminate due to similar clinical features, especially when the scalp is the only area involved. The pathogenesis also has some common immunological and biomolecular pathways as these two conditions are considered to be driven by the Th17 response and related to dysbiosis of the skin microbiome.1-3 In this study, we aim to characterize and compare the skin microbial communities of the two diseases. We recruited 45 patients with SD, 37 patients with psoriasis and 30 healthy controls. The study protocol was approved by the Institutional Review Board of Peking University First Hospital. All participants provided written informed consent prior to participation. Patients above 21 years and diagnosed with psoriasis or SD involving the scalp were eligible for this study. The diagnosis of psoriasis and SD was based on characteristic clinical and dermoscopic signs of the affected skin, and histopathological examination was performed if necessary. None of the participants had used topical or systemic antifungals or antibiotics in the past 3 months. Skin swabs were collected from the scalps at lesional and nonlesional sites. Amplicon sequencing and analyses were performed as previously described.4 No significant differences were detected in the age and sex ratio among groups (Table 1). Analysis on bacterial diversity showed the Shannon diversity was significantly increased in psoriasis lesions compared with SD lesions and controls. No significant differences in fungal diversity were observed among groups (Figure 1A,B). Principal coordinates analyses did not show clear separations among the disease groups; however, healthy bacterial communities appear to be more similar among samples in comparison with psoriasis and SD communities (Figure 1C,D). ANOSIM was calculated showing significantly higher dissimilarities between groups compared with dissimilarities within groups (R = 0.077, p < 0.001, Figure S1). The fungal taxonomic profiles revealed that Malassezia was the predominant fungal genus in all groups. The relative abundance of Malassezia was slightly higher in lesions of SD than psoriasis, though no statistical significance was detected. Increased relative abundances of Fusarium and Phoma and decreased relative abundance of Chaetomidium were observed in the lesions of psoriasis compared with SD (Figure 1H). Species level analysis showed an increasing trend in the relative abundance of unidentified Malassezia/M. globosa ratio in the lesions of both diseases compared with controls (Figure 1I). Candida tropicalis was significantly higher in psoriasis lesions than in SD lesions (Figure 1J). Analysis of the bacterial microbiome showed higher relative abundance of Staphylococcus and lower relative abundance Cutibacterium in the lesions of both diseases than controls and their nonlesional sites (Figure 1E–G). Compared with SD lesions, significantly higher relative abundances of Streptococcus, Corynebacterium and Psychrobacter were observed in psoriasis lesions. Next, we used receiver operating characteristic curve analysis to discriminate SD and psoriasis. Based on a random forest algorithm, we could accurately differentiate the two diseases using the top 10 most abundant bacterial and fungal genera (area under the curve: 0.98, sensitivity: 0.96, specificity: 0.72) or the most abundant fungal species (area under the curve: 0.97, sensitivity: 0.72, specificity: 1.0) (Figure 1K,L, Figure S2). The disease and healthy states could also be accurately discriminated using a random forest algorithm (Figure S3). In this cross-sectional study, we have shown that both SD and psoriasis are characterized by alterations in the bacterial and fungal community. Comparison of the skin microbiome between psoriasis and SD reveal distinct pattern of microbial community alterations, especially in the level of Streptococcus, Corynebacterium, Psychrobacter and Candida. Candida has been shown to be involved in the pathogenesis of psoriasis by promoting tissue inflammation in a Candida-specific Th7/Langerhans cell-dependent manner.5 Streptococcus and Corynebacterium, whose levels are found elevated in psoriatic lesions, have been demonstrated to be associated with the exacerbation of psoriasis by direct binding of their superantigens to human cells, or inducing higher intensities of IL-23, IL-1β and dermal gdT17 cells.6, 7 In contrast to psoriasis, SD is more likely to be associated with disequilibrium of Malassezia species, though controversy exists between studies.3 Overall, our study shows that lesions of SD have microbiome signatures distinct from those of psoriatic lesions. Corynebacterium, Streptococcus and Candida could be potential biomarkers for identifying the two diseases. However, whether microbial shifts are a cause or consequence of the pathogenic process is unknown. Future studies focusing on specific microorganisms and the inflammatory responses may provide new ideas for explaining the pathogenesis of the two diseases. R.T, R.W., R.L. and Y.W designed the research study. R.T. involved in writing—original draft preparation. R.W. involved in writing— review and editing. R.T and R.W. analysed the data and performed the research; Z.W contributed essential experimental tools; R.L. and Y.W involved in supervision. All authors have read and agreed to the published version of the manuscript. This work was supported by the National Natural Science Foundation of China (81903225). The authors Rong Tao, Ruoyu Li, Zhe Wan, Yan Wu and Ruojun Wang declare no conflicts of interest. Figure S1 ANOSIM (ANalysis of Similarities) showing dissimilarities between groups compared to dissimilarities within groups. Figure S2 (a) Receiver operating characteristic (ROC) curve using the top 10 abundant fungal species based on a machine learning random forest algorithm to discriminate seborrheic dermatitis and psoriasis. (b) Mean decrease accuracy of taxas showing the predictive features to classify psoriasis and SD. SD: seborrheic dermatitis. Figure S3 (a) Receiver operating characteristic (ROC) curve using the top 10 abundant bacterial and fungal genera based on a machine learning random forest algorithm to discriminate seborrheic dermatitis and healthy controls. (b) ROC curve to discriminate psoriasis and healthy controls. HC: healthy control; SD: seborrheic dermatitis. Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.