Arthritis & Rheumatism, Volume 62,
November 2010 Abstract Supplement
Abstracts of the American College of
Rheumatology/Association of Rheumatology Health Professionals
Annual Scientific Meeting
Atlanta, Georgia November 6-11, 2010.
Stratification of Rheumatoid Arthritis Patients into Disease Subtypes with Clinical Implications.
Dennis, Glynn, Holweg, Cecile, Hackney, Jason, Glibert, Houston, Yu Lin, Wei, Diehl, Lauri, Endres, Judith
The study aimed to identify molecular subtypes of rheumatoid arthritis (RA) that differ in their biological composition and response to therapy.
Genome-wide transcriptional profiles for 81 synovial tissue samples obtained by surgery from 50 active RA patients fulfilling the 1987 ACR criteria were stratified by bootstrapped hierarchical clustering. Statistically supported sample subtypes were further characterized by significance analysis of microarrays and pathway analysis. Platform independent validation of subtypes was performed by immunohistochemistry, fluorescence-based cell sorting and semi-quantitative polymerase chain reaction. Subtype-intrinsic gene signatures were tested for their ability to classify external data using linear discriminant analysis. Protein biomarker assays measured the association between serum proteins and response to rituximab.
Multi-scale bootstrapped hierarchical clustering of 81 samples inferred four molecular subtypes of RA. Each subtype represented a discrete transcriptional program that reflected differences in tissue cellularity and biological composition. Diffusely infiltrated tissues were assigned to one of two subtypes characterized by fibroblast-like transcriptional programs that differed slightly, but significantly, in their pathway composition. Tissues with substantial immune infiltration were assigned to one of two immunoregulatory subtypes distinguishable by either a potent B cell signature or a broad inflammatory signature reflecting various myeloid and T cell transcriptional programs. Predictive gene signatures for each subtype accurately classified institutionally independent microarray profiles of histologically similar samples. Protein-based assays of B cell signature genes in a large placebo-controlled study of rituximab in RA revealed a combinatorial relationship between low levels of certain B cell biomarkers in serum and a poor clinical response to B cell depleting therapy.
We reduced RA tissue heterogeneity down to 4 discrete molecular subtypes and provided subtype-intrinsic gene signatures that were able to accurately classify independently derived external data. The unique cellular and biological composition of each subtype suggests that molecular stratification of RA patients has the potential to improve clinical trial design and therapeutic intervention by identifying those patients that are most likely to benefit from a given therapy.
To cite this abstract, please use the following information:
Dennis, Glynn, Holweg, Cecile, Hackney, Jason, Glibert, Houston, Yu Lin, Wei, Diehl, Lauri, et al; Stratification of Rheumatoid Arthritis Patients into Disease Subtypes with Clinical Implications. [abstract]. Arthritis Rheum 2010;62 Suppl 10 :1609