Higher expression of TNFα-induced genes in the synovium of patients with early rheumatoid arthritis correlates with disease activity, and predicts absence of response to first line therapy

Abstract

Background: IL6-related T cell activation and TNFα-dependent cell proliferation are major targets of therapy in the RA synovium. We investigated whether expression of these pathways in RA synovial biopsies is associated with disease activity and response to therapy.

Method: Correlation and gene set enrichment studies were performed using gene expression profiles from RA synovial biopsies. Immunostaining experiments of GADD45B and PDE4D were performed on independent additional sets of early untreated RA samples, obtained in two different centers by needle-arthroscopy or US-guided biopsies.

Results: In 65 RA synovial biopsies, transcripts correlating with disease activity were strongly enriched in TNFα-induced genes. Out of the individual variables used in disease-activity scores, tender joint count, swollen joint count and physician's global assessment, but not CRP or patient's global assessment displayed a similar correlation with the expression of TNFα-dependent genes. In addition, TNFα-induced genes were also significantly enriched in transcripts over-expressed in synovial biopsy samples obtained from poor-responders to methotrexate or tocilizumab, prior to initiation of therapy. GADD45B (induced by TNFα in monocytes) and PDE4D (induced by TNFα in FLS) immunostaining was significantly higher in overall poor-responders to therapy in 46 independent baseline samples obtained from early untreated RA patients prior to initiation of therapy. GADD45B (but not PDE4D) immunostaining was significantly higher in the sub-group of patients with poor-response to methotrexate therapy, and this was confirmed in another population of methotrexate-treated patients.

Conclusion: Higher expression of TNFα-induced transcripts in early RA synovitis is associated with higher disease activity, and predicts poor response to first-line therapy. That over-expression of TNFα-induced genes predicts poor-response to therapy regardless of the drug administered, indicates that this molecular signature is associated with disease severity, rather than with specific pathways of escape to therapy.

Figures

Fig. 1

Correlations between clinical disease activity indices and the expression of 51,452 probe sets in 65 synovial biopsy samples from untreated and treated patients with rheumatoid arthritis (RA). a Distribution of Pearson r correlation coefficients between disease activity score in 28 joints assessed by C-reactive protein level (DAS28-CRP)/simplified disease activity index (SDAI)/clinical disease activity index (CDAI) and synovial gene expression levels. b Overlap between probe sets displaying a Pearson r correlation coefficient >0.5 with DAS28-CRP, SDAI and CDAI. c, d Gene set enrichment analyses and eigenvalues calculations using probe sets displaying a Pearson r correlation coefficient >0.5 with DAS28-CRP indicate that the majority of them are downregulated by tocilizumab in synovial tissue in RA (c), and a large proportion are induced by TNFα in cultured fibroblast-like synoviocytes (FLS) or monocytes (d). e Pathway analyses (http://david.abcc.ncifcrf.gov) indicate that transcripts displaying a Pearson r correlation coefficient >0.5 with DAS28-CRP are significantly enriched in genes associated with T cell activation, inflammatory responses, antigen presentation and lysosomes. Probe sets belonging to these pathways are downregulated by tocilizumab in RA synovial tissue

Fig. 2

Correlations between clinical disease activity indices and the expression of 51,452 probe sets in 21 synovial biopsy samples from untreated patients with early rheumatoid arthritis (RA). a Distribution of Pearson r correlation coefficients between disease activity score in 28 joints assessed by C-reactive protein level (DAS28-CRP)/simplified disease activity index (SDAI)/clinical disease activity index (CDAI) and synovial gene expression levels. b Overlap between probe sets displaying a Pearson r correlation coefficient >0.5 with DAS28-CRP, SDAI and CDAI. c, d Gene set enrichment analyses and eigenvalues calculations using probe sets displaying a Pearson r correlation coefficient >0.5 with DAS28-CRP indicate that a subset of these transcripts is downregulated by tocilizumab in synovial tissue in RA (c), and a large proportion is induced by TNFα in cultured fibroblast-like synoviocytes (FLS) or monocytes (d). e Pathway analyses (http://david.abcc.ncifcrf.gov) indicate that transcripts displaying a Pearson r correlation coefficient >0.5 with DAS28-CRP are significantly enriched in genes associated with inflammation, mitosis, DNA remodeling and lysosomes. The radar plots indicate how the transcripts belonging to these pathways are influenced by tocilizumab in synovial biopsies, and by TNFα in cultured monocytes and FLS. INV inverse (a negative effect of tocilizumab appears as positive and suggests a positive effect of IL6)

Fig. 3

Higher synovial expression of TNFα-induced transcripts is associated with overall absence of response to therapy in early rheumatoid arthritis (RA). a Baseline synovial gene expression profiles were compared in poor versus good responders to either tocilizumab or methotrexate therapy at 3 months (4 non-responders versus 9 good responders), and at 6 months (4 non-responders versus 15 good responders). The radar plots indicate how transcripts over-expressed in poor responders (p <0.05, >1.5-fold change in gene expression) are influenced by tocilizumab in synovial biopsies, and by TNFα in cultured monocytes and fibroblast-like synoviocytes (FLS). INV inverse (a negative effect of tocilizumab appears as positive and suggests a positive effect of IL6). b Fold changes in the expression of GADD45B and PDE4D encoding probe sets obtained from HGU133 Plus2.0 transcriptomic data generated in TNFα-stimulated versus non-stimulated monocytes (n = 3) and FLS (n = 3). Box plots represent minimum, maximum and median log2 (fold changes) compared to non-stimulated cells. c Typical GADD45B and PDE4D immunostaining pictures (original magnification × 400) obtained in early RA synovial biopsies, illustrative of marked inter-individual variations in protein expression

Fig. 4

Quantification of GADD45B and PDE4D immunostaining in baseline synovial biopsies from patients with early rheumatoid arthritis (RA), osteoarthritis and psoriatic arthritis. a Needle-arthroscopic synovial biopsies from untreated patients with early RA (n = 46), psoriatic arthritis (n = 9) and osteoarthritis (n = 12) were stained with polyclonal antibodies directed at GADD45B and PDE4D. Data are represented as the ratio of GADD45B or PDE4D surface staining to staining of the nuclei. Horizontal bars represent the median values. P values were calculated using Kruskal–Wallis test. b GADD45B and PDE4D immunostaining in needle-arthroscopic synovial biopsies obtained in patients with early RA prior to initiation of therapy. Patients are distributed as good, moderate and poor responders to therapy according to European League Against Rheumatism (EULAR) response criteria at 6 months. P values were calculated using Kruskal–Wallis test. c GADD45B and PDE4D immunostaining in needle-arthroscopic (sub-group of population displayed in a and b) and ultrasound-guided biopsies obtained in patients with early RA prior to initiation of methotrexate therapy. Patients are distributed as responders or non-responders according to EULAR response criteria at 6 months. P values were calculated using Mann–Whitney test