Dissecting the heterogeneity of skin gene expression patterns in systemic sclerosis

Objective To examine the heterogeneity of global transcriptome patterns in systemic sclerosis (SSc) skin in a large sample of patients with SSc and control subjects. Methods Skin biopsy specimens obtained from 61 patients enrolled in the Genetics versus Environment in Scleroderma Outcome Study (GENISOS) cohort and 36 unaffected control subjects with a similar demographic background were examined by Illumina HumanHT-12 bead arrays. Followup experiments using quantitative polymerase chain reaction and immunohistochemical analysis were also performed. Results We identified 2,754 differentially expressed transcripts in SSc patients compared with controls. Clustering analysis revealed 2 prominent transcriptomes in SSc patients: the keratin and fibroinflammatory signatures. Higher keratin transcript scores were associated with shorter disease duration and interstitial lung disease, while higher fibroinflammatory scores were associated with diffuse cutaneous involvement, a higher skin score at the biopsy site, and a higher modified Rodnan skin thickness score. A subgroup of patients with significantly longer disease duration had a normal-like transcript pattern. Analysis of cell type-specific signature scores revealed remarkable heterogeneity across patients. Significantly higher scores were calculated for fibroblasts (72% of patients), microvascular cells (61%), macrophages (54%), and dendritic cells (DCs) (49%). The majority of samples with significantly higher fibroblast scores (35 of 44 [80%]) had significantly increased macrophage and/or DC scores. Further analysis and immunohistochemical staining indicated that the keratin signature was not a general marker of keratinocyte activation but was in fact associated with an activation pattern in hair and adnexal structures. Conclusion Prominent fibroinflammatory and keratin signatures are present in SSc skin. Expression profiles of SSc skin show significant heterogeneity, and this finding might be useful for stratifying patients for targeted therapies or predicting the response to immunosuppression.