Background/Purpose:

Systemic Sclerosis (SSc) is a complex autoimmune disease with chronic progressive course and high interpatient variability. It is characterized by inflammation, vascular dysfunction and fibrosis. Fibrosis of the skin and visceral organs results in irreversible scarring and ultimately organ failure, accounting for high mortality. There is no approved targeted therapy with disease-modifying potential. Several translational studies were published, which focused on transcriptional analysis of samples from SSc patients in order to understand the heterogeneity of underlying disease. We interrogated these datasets using a novel data analysis methodology that leverages the knowledge of protein interaction networks (STRING) for deriving biological insights and identifying intervention points into this heterogeneous disease.

Methods:

Publically available datasets composed of gene expression analysis of patient skin biopsies were identified. Datasets were subjected to sample size and quality assessment. Selected high-quality datasets were analyzed using a novel computational and statistical method. Namely, each gene was scored for both its differential expression and its known protein interactions with top differentially expressed genes (DEGs). This approach led to identification of Well-Associated Proteins (WAPs); gene products that have a significantly large number of known associations to top DEGs, without choosing a threshold for differential expression. The scoring method corrects for the total number of interactions of each gene and is optimally fast, enabling false discovery rate (FDR) estimates for each WAP via permutation testing, thus taking into account gene co-expression. Furthermore, permutation testing was utilized to identify WAPs that correlated with measurable clinical factors.

Results:

Four publically available datasets were analyzed. Key biology pathways associated with SSc were identified. More importantly, > 100 potential targets/WAPs were identified with FDR score < 0.01%. ~ 75% of the significant WAPs were not perturbed at the mRNA level and would have been missed via standard statistical methods. Their significant connectedness to top DEGs in the datasets suggests biologically relevant role in SSc. Robustness analysis revealed that WAP scores were ~40% reproducible across pairs of datasets, as compared to only ~10% for DEG scores. Of the top 100 WAPs obtained by combining the two largest data sets, 59 were previously targeted providing drug-repurposing opportunities and 41 were novel targets for drug discovery. Additional analysis identified WAPs (e.g. OSM) that significantly correlated with measurable clinical factors such as MRSS (global skin score), diffuse disease and some auto-antibodies.

Conclusion:

A novel data analysis methodology was developed, leveraging protein interaction networks to identify WAPs that represent potentially unique targets for therapeutic intervention in SSc. WAPs are unique as they were reproducibly detected across multiple publicly available SSc datasets. Additionally, patient sub-populations that may benefit from targeted therapeutics against selected WAPs were identified.

« Back to 2017 ACR/ARHP Annual Meeting

ACR Meeting Abstracts - https://acrabstracts.org/abstract/application-of-a-novel-computational-approach-to-identify-new-targets-and-pathways-for-therapeutic-intervention-in-scleroderma/