Increased interleukin-1β gene expression in peripheral blood leukocytes is associated with increased pain and predicts risk for progression of symptomatic knee osteoarthritis

Abstract

Objective: To evaluate whether gene expression profiles could serve as biomarkers of symptomatic knee osteoarthritis (OA) by examining gene expression profiles in peripheral blood leukocytes (PBLs) from patients with OA compared with those from non-OA controls, and to determine whether candidate genomic biomarkers (PBL expression of inflammatory genes) predict an increased risk of disease progression in patients with symptomatic radiographic knee OA.

Methods: Three independent cohorts of patients with knee OA and non-OA control subjects were studied. Two cohorts (a learning cohort and a validation cohort) were recruited at New York University Hospital for Joint Diseases (NYUHJD), and 1 cohort (a validation cohort) was recruited at Duke University Medical Center. PBL gene expression was assessed using Affymetrix microarray and was confirmed by quantitative polymerase chain reaction (qPCR). Radiographic progression at 2 years was assessed in 86 patients.

Results: We identified 173 genes that were significantly up-regulated or down-regulated (≥1.5-fold change) in OA PBLs, at a false discovery rate of 5%. Cluster analysis revealed 2 distinct subgroups among the patients with OA: those in whom the expression of interleukin-1β (IL-1β) was increased ≥2-fold compared with controls, and those in whom the expression of IL-1β was comparable with that in controls. Overexpression of IL-1β in these OA subclasses was validated using qPCR in all 3 cohorts. Patients with the inflammatory "IL-1β signature" had higher pain scores and decreased function and were at higher risk of radiographic progression of OA.

Conclusion: PBLs from patients with symptomatic knee OA display a characteristic transcriptome profile. Moreover, increased expression of IL-1β identifies a subset of patients with OA who have increased pain and are at higher risk of radiographic progression of OA.

Figure 1

Osteoarthritis peripheral blood leukocyte (OA PBMC) gene expression study strategy. Three cohorts were studied. The original study cohort (NYUHJD Learning Cohort) recruited 45 OA patients and 25 non-OA controls. Five microgram of total RNA isolated from PBMC was used for biotin cRNA synthesis and hybridized against Affymetrix human U133A chip; microarray assessment was accomplished for 41 OA patients and 25 non-OA controls. Using Significance Analysis of Microarrays (SAM), we identified 173 genes significantly up- or down-regulated (≥1.5-fold change) in the NYUHJD Learning Cohort OA PBMC, at a False Discovery Rate (FDR) of 5%. Two additional cohorts, NYUHJD Validation Cohort and Duke Validation Cohort, were used for validation of IL-1β expression by QPCR.

Figure 2

Hierarchical clustering of OA samples based on gene expression of 21 selected inflammatory genes. A, Hierarchical clustering identified two subclasses of OA based on gene expression of 21 selected cytokines: cytokine overexpressors (OA^IL-1) and cytokine underexpressors (OA^nl). The cytokines are IL-1β, IL-8, COX-2 and chemokines GRO2, cytokine A3 (MIP-1α) and cytokine A4 (MIP-1β). B, Centroids of OA subclasses. The horizontal bars represent the average gene expression for each of the three classes: OA^IL-1, OA^nl and non-OA controls (NC). The red bars indicate positive average expression while the green bars indicate negative average expression. Abbreviations: CCL3L1, chemokine (C-C) motif ligand 3-like protein 1; CXCL2, chemokine C-X-C motif ligand 2; IL, interleukin; PTGS2, prostaglandin-endoperoxide synthase 2; TNF, tumor necrosis factor; TNFLSF, TNF ligand superfamily; TNFRSF, TNF receptor superfamily.

Figure 3

Upregulation of inflammatory genes in the NYUHJD Learning Cohort and NYUHJD and Duke Validation Cohorts. A, Real Time PCR analysis: relative expression levels of inflammatory gene (IL-1β, TNFα, IL-8, IL-6, COX-1 and COX-2) expression in OA PBMCs (NYUHJD Learning Cohort). The raw QPCR data were normalized against housekeeping gene GAPDH, and relative expression levels were calculated using the delta-delta C_T method (22). These data show that five of these genes are differentially over-expressed in the OA^IL-1 group as compared to non-OA controls (NC) and OA^nl. B and C, distributions of IL-1β and TNFα expression levels by group in two validation cohorts. Real Time PCR analysis: relative expression levels of IL-1β in (B) the NYUHJD Validation Ccohort and (C) Duke Validation Cohort by group. These data show that the IL-1β gene is differentially overexpressed in the OA^IL-1group as compared to non-OA controls (NC) and OA^nl. Box-Cox transformations were applied to normalize variables prior to regression modeling.