Combined meta-analysis of systemic effects of allogeneic stem cell transplantation and systemic sclerosis

Abstract

Abstract

Background Chronic graft-versus-host disease (cGVHD) is a major factor of morbidity and mortality for allogeneic stem cell transplantation (aSCT). The skin and internal organ involvement is the most common systemic complication of cGVHD and closely resembles systemic sclerosis (SSc). Circulating lymphocytes characterize the autoimmune nature of both conditions. Therefore we hypothesized that the common clinical manifestation (systemic organ and skin injury) and the common underlying players (lymphocytes) justify the combined meta-analysis of these diseases.

Results The aSCT and SSc datasets were uploaded from Gene Expression Omnibus (GEO), a public functional genomics data repository. The available microarray studies of peripheral blood mononuclear cells (PBMCs) and isolated lymphocytes were limited to well established microarray platforms (Affymetrix, Agilent, Canvac, and Illumina) and experimental settings with ≥10 patients per group. The resulting pools of data were merged by unique gene identifier and analyzed by the expression genome-wide association studies (eGWAS) coupled with the subtraction of the cGVHD+ and cGVHD− molecular signatures. The eGWAS was applied to 47 and 50 lymphocyte profiles from aSCT and SSc patients, respectively. The identified 35 candidates were represented by 8 known cGVHD genes (including CXCR4, LTBR and PML) and 28 new candidate genes (including SEPX1 and DNJGB1). The further mutual subtraction of cGVHD+ and cGVHD− candidates and pathway analysis identified a list of 25 genes. Seven of these genes belong to the fibroblast development and function pathway, consisting of the well known cGVHD genes CCND1, JUN, and FOS, and the new molecular targets MMP2, FOSB, TNFAIP8, and DUSP1. These genes become primary candidates for a potential link of systemic effects of cGVHD and SSc.

Conclusions We designed a new approach for meta-analysis by combining data from different diseases using common clinical manifestation as a linker. This allowed us to power up the insufficient standalone meta-analysis of aSCT microarray studies, by adding SSc samples to the data pool. This new method has successfully identified novel molecular targets for systemic effects of both aSCT and SSc. We believe that this approach is generalizable and can be applied to an array of diseases with common clinical manifestations.