A T cell gene expression panel for the diagnosis and monitoring of disease activity in patients with systemic lupus erythematosus

Abstract

Systemic Lupus Erythematosus (SLE) remains a challenging disease to diagnose and follow, as no reliable biomarkers are known to date. We designed a gene expression panel with 40 genes known to play a role in SLE pathogenesis. We found that the combined expression of these genes in SLE T cells can accurately differentiate SLE from healthy individuals and patients with other autoimmune diseases. The accuracy of the test increased further (83%) when only three out of the initial genes (OAS2, CD70 and IL10) were used. A T cell score, calculated from the combined expression levels of these genes, correlated positively with various SLE activity markers in a cross-sectional cohort and in a few patients that were followed prospectively. These data showcase the usefulness of measuring mRNA levels of key molecules in diagnosing and following patients with SLE.

Keywords: Biomarkers; CD70; Diagnostic; Flare; IL10; OAS2.

Figure 1

Real time PCR amplification products visualised in agarose gel. Real-time PCR amplification products for all genes included in the array were electrophoresed in agarose gel and visualised under UV light. In all cases single band products were seen and their size agreed with the predicted one (shown in brackets, bp: base pairs).

Figure 2

Accuracy of the genetic panel in predicting SLE from control status. a) Genes were ranked according to their ability to differentiate SLE status from control using comparative marker selection method. For each gene a test statistic score was calculated: the highest the score, the higher the potential of the gene to be used as a biomarker. Gene targets with a positive score were those upregulated in SLE, while those with a negative, in healthy controls. OAS2, CD70 and IL10 were the genes that ranked the highest scores. b) Reducing the gene panel to only include the top differentially expressed genes improves its overall classification accuracy. Measurement of only three genes, OAS2, CD70 and IL10 correctly differentiates SLE samples from control in 83% of cases. Classification accuracy was assessed using CART (classification and regression trees).

Figure 3

The three-gene panel differentiates SLE samples from healthy control and rheumatoid arthritis (RA) samples. a) Heatmap of expression levels of OAS2, CD70 and IL10 in 65 SLE, 27 control and 16 RA samples. Control and RA samples (blue and green bars respectively, top of the heatmap) grouped together, and separately from SLE samples (red bars) in hierarchical clustering. b) Three dimensional plotting of SLE, RA and control samples based on their OAS2, CD70 and IL10 expression levels (SLE samples in red, control samples in blue and RA samples in green). Plotting on expression levels, control and RA samples were seen to cluster separately from SLE samples. c) Accuracy of the three-gene panel was higher in steroid-free (77%), compared to steroid-treated (64%) patients, and in those with a longer (>1 year) history of disease (72%). Patients having nephritis were the ones most accurately diagnosed (74%), followed by those with cytopenia (66%), arthritis (58%), serositis (57%) and mucocutaneous manifestations (50%). d) Expression levels of OAS2, CD70 and IL10 were combined to create a ‘T cell score’. T cell score levels were significantly higher in the SLE than the healthy control group and correlated significantly with dsDNA, PGA, SLEDAI and ESR. Platelet and creatinine levels also correlated positively while C3, C4 levels and lymphocyte count negatively with the T cell score.

Figure 4

Evaluation of the three-gene panel in PBMCs. a) RNA extracted from PBMCs was used to measure expression levels of OAS2, CD70 and IL10 in 20 SLE and 9 control samples. After converting into copy numbers, expression levels of these 3 genes were combined into a single score. Cumulative expression was significantly higher in SLE patients than in controls and correlated significantly with PGA (b). A positive correlation was also noted for SLEDAI and ESR; and a negative for lymphocyte count and C3 levels.

Figure 5

T cell score in longitudinally followed up patients. Seven SLE patients were monitored longitudinally on 2 month intervals. Blood was drawn on each visit to determine T cell expression levels of OAS2, CD70 and IL10 and measure standard laboratory markers. Fifty two time points were recorded in total over a period of two and a half years. The T cell score correlated positively with SLEDAI in six patients. Three examples are shown (left panel: r = 0.71 (1) 12.6, 16 (2) −1.9, 12 (3) −2.8, 4 (4) −1.5, 4 (5) −2.6, 4 (6) 1.8, 2(7) −1.4, 2 (8) −2, 4; middle panel: r = 0.35 (1) 7.2, 5 (2) −0.02, 4 (3) 0.12, 4 (4) −1.2, 4 (5) −1.3, 2 (6) 0.02, 2(7) −2.4, 2 (8) −1, 8 (9) −1.7, 2 (10) −2, 2; right panel: r = 0.64 (1) 1.87, 16 (2) −0.8, 2 (3) 5.4, 8 (4) −0.3, 0 (5) −2.7, 0 (6) −0.5, 0; correlation coefficient, T cell and SLEDAI scores in each patient visit).