Machine learning integration of scleroderma histology and gene expression identifies fibroblast polarisation as a hallmark of clinical severity and improvement

Abstract

ObjectiveWe sought to determine histologic and gene expression features of clinical improvement in early diffuse cutaneous systemic sclerosis (dcSSc; scleroderma).MethodsFifty-eight forearm biopsies were evaluated from 26 individuals with dcSSc in two clinical trials. Histologic/immunophenotypic assessments of global severity, alpha-smooth muscle actin (aSMA), CD34, collagen, inflammatory infiltrate, follicles and thickness were compared with gene expression and clinical data. Support vector machine learning was performed using scleroderma gene expression subset (normal-like, fibroproliferative, inflammatory) as classifiers and histology scores as inputs. Comparison of w-vector mean absolute weights was used to identify histologic features most predictive of gene expression subset. We then tested for differential gene expression according to histologic severity and compared those with clinical improvement (according to the Combined Response Index in Systemic Sclerosis).ResultsaSMA was highest and CD34 lowest in samples with highest local Modified Rodnan Skin Score. CD34 and aSMA changed significantly from baseline to 52 weeks in clinical improvers. CD34 and aSMA were the strongest predictors of gene expression subset, with highest CD34 staining in the normal-like subset (p<0.001) and highest aSMA staining in the inflammatory subset (p=0.016). Analysis of gene expression according to CD34 and aSMA binarised scores identified a 47-gene fibroblast polarisation signature that decreases over time only in improvers (vs non-improvers). Pathway analysis of these genes identified gene expression signatures of inflammatory fibroblasts.ConclusionCD34 and aSMA stains describe distinct fibroblast polarisation states, are associated with gene expression subsets and clinical assessments, and may be useful biomarkers of clinical severity and improvement in dcSSc.

Correspondence on ‘Machine learning integration of scleroderma histology and gene expression identifies fibroblast polarisation as a hallmark of clinical severity and improvement’

“…Elegantly, the authors have shown that CD34 staining decreases with worsening clinical severity and subsequently increases with clinical improvement, while the opposite was observed for α-SMA myofibroblast marker staining. 1 Moreover, CD34 and α-SMA immunostaining intensity appeared to be highly predictive of scleroderma gene expression subsets, with highest CD34 staining and lowest α-SMA staining being detectable in normal-like compared with fibroproliferative and inflammatory dcSSc skin samples. 1 Further analyses revealed that clinically severe dcSSc skin exhibiting the α-SMA high /CD34 low immunophenotype associates with a high inflammatory gene expression signature that can reverse over time in patients with clinical improvement, but not in those who do not improve.…”

“…1 Moreover, CD34 and α-SMA immunostaining intensity appeared to be highly predictive of scleroderma gene expression subsets, with highest CD34 staining and lowest α-SMA staining being detectable in normal-like compared with fibroproliferative and inflammatory dcSSc skin samples. 1 Further analyses revealed that clinically severe dcSSc skin exhibiting the α-SMA high /CD34 low immunophenotype associates with a high inflammatory gene expression signature that can reverse over time in patients with clinical improvement, but not in those who do not improve. 1 Mechanistically, the authors have suggested that differential CD34 and α-SMA stains may reflect distinct fibroblast polarisation states, with the α-SMA high /CD34 low immunophenotype characteristic of clinically severe dcSSc skin samples reflecting a transition from CD34 + normal fibroblasts to α-SMA + inflammatory fibroblasts/myofibroblasts.…”

“…1 Further analyses revealed that clinically severe dcSSc skin exhibiting the α-SMA high /CD34 low immunophenotype associates with a high inflammatory gene expression signature that can reverse over time in patients with clinical improvement, but not in those who do not improve. 1 Mechanistically, the authors have suggested that differential CD34 and α-SMA stains may reflect distinct fibroblast polarisation states, with the α-SMA high /CD34 low immunophenotype characteristic of clinically severe dcSSc skin samples reflecting a transition from CD34 + normal fibroblasts to α-SMA + inflammatory fibroblasts/myofibroblasts. 1 However, it should be considered that gene expression analyses were carried out on whole skin biopsy specimens, which does not allow the identification of the specific cell types expressing relevant target genes, and that double α-SMA/CD34 stain would be essential in an attempt to identify transitional fibroblasts expressing both markers.…”

“…Correspondence on 'Machine learning integration of scleroderma histology and gene expression identifies fibroblast polarisation as a hallmark of clinical severity and improvement' I read with great interest the article by Showalter et al 1 who, through an unbiased machine learning approach integrating skin biopsy histology and gene expression, describe the relevance of CD34 and α-smooth muscle actin (α-SMA) stains as biomarkers reflecting clinical severity and improvement in patients with diffuse cutaneous systemic sclerosis (dcSSc; scleroderma). Elegantly, the authors have shown that CD34 staining decreases with worsening clinical severity and subsequently increases with clinical improvement, while the opposite was observed for α-SMA myofibroblast marker staining.…”

“…1 Mechanistically, the authors have suggested that differential CD34 and α-SMA stains may reflect distinct fibroblast polarisation states, with the α-SMA high /CD34 low immunophenotype characteristic of clinically severe dcSSc skin samples reflecting a transition from CD34 + normal fibroblasts to α-SMA + inflammatory fibroblasts/myofibroblasts. 1 However, it should be considered that gene expression analyses were carried out on whole skin biopsy specimens, which does not allow the identification of the specific cell types expressing relevant target genes, and that double α-SMA/CD34 stain would be essential in an attempt to identify transitional fibroblasts expressing both markers. In support of their hypothesis, the authors further discuss that a previous study showed that scleroderma skin displays an inverse staining pattern between CD34 and two other fibroblast markers, namely podoplanin and CD90/ Thy1, and that the proposed fibroblast transition could be induced in vitro in response to proinflammatory cytokines.…”