Published Data

Molecular profiles of inflammatory myopathies

S.A. Greenberg, D. Sanoudou, J.N. Haslett, I.S. Kohane, L.M. Kunkel, A.H. Begss, A.A. Amato Neurology 59: 1170 - 1182 (2002).
PMID: 12391344 [PubMed - indexed for MEDLINE]

Department of Neurology, Brigham and Women's Hospital, 75 Francis Street, Boston, MA 02115, USA. sagreenberg@partners.org

Abstract

Objective: To describe the use of large-scale gene expression profiles to distinguish broad categories of myopathy and subtypes of inflammatory myopathies (IM) and to provide insight into the pathogenesis of inclusion body myositis (IBM), polymyositis (PM), and dermatomyositis (DM). Methods: Using Affymetrix GeneChip microarrays, we measured the simultaneous expression of approximately 10,000 genes in muscle specimens from 45 patients in 4 major disease categories (dystrophy, congenital myopathy, inflammatory myopathy, and normal). We separately analyzed gene expression in 14 patients limited to the three major subtypes of IM. We used bioinformatics techniques to classify specimens with similar expression profiles based on global patterns of gene expression and to identify genes with significant differential gene expression compared to normal. Results: Ten of 11 patients with IM, all normals and nemaline myopathies, and 10 of 12 patients with Duchenne muscular dystrophy were correctly classified by this approach. The various subtypes of inflammatory myopathies have distinct gene expression signatures. Specific sets of immune-related genes allow for molecular classification of patients with IBM, PM, and DM. Analysis of differential gene expression identifies as relevant to disease pathogenesis previously reported cytokines, major histocompatibility complex class I and II molecules, granzymes, and adhesion molecules as well as newly identified members of these categories. Increased expression of actin cytoskeleton genes is also identified. Conclusions: The molecular profiles of muscle tissue in patients with inflammatory myopathies are distinct and represent molecular signatures from which diagnostic insight may follow. Large numbers of differentially expressed genes are rapidly identified.

Supplemental Data

Supplemental Tables

To download data we recommend using a high-speed connection because of the large file sizes.

Adobe Acrobat Required 
Click here to download

•The microarray data set for 16 experiments (6 IBM, 6 DM, 2 PM, and 2 NORM) can be downloaded as an Excel file from: http://potato.chip.org/~steveng1/Greenberg-et-al-IM-Dataset.xls (9.5 MB).

• Various database analyses performed on this data set can be downloaded as an Access database from: http://potato.chip.org/~steveng1/Greenberg-et-al-IM-Analyses.mdb (9.3 MB).

• Studies on the effect of patient age, gender, and muscle on clustering of expression profiles: Because of their potentially confounding influences, we have studied the effects of patient age, gender, and choice of muscle on molecular profiles. Fifteen normals (mean age 25 years, range 1-72) showed no substantial tendency for an age-related or gender clustering effect (Figure 1), though some effect may be present, particularly for gender. Similar analysis for choice of muscle (e.g., quadriceps, biceps brachii) showed no effect on clustering.

Figure 1. (Above) Effect of age and gender on clustering in 15 normal muscle specimens. Hierarchical clustering across 12,600 genes, filtered for variation, resulting in 2491 genes across 15 normal specimens. No substantial age-related clustering effect is evident, though some effect may be present. The cluster consisting of H207, H239, and H126 may be an age effect or related to the fact that these 3 specimens are the only autopsy specimens from the group. Similarly for gender, among the 4 major clusters present, 2 contain a mixture of male and female specimens. One cluster, S93-3678, Ema2, S93-482, and H029, may be a gender related effect.