Association of circulating gene expression signatures with stiffness following total knee arthroplasty for osteoarthritis: a pilot study

Abstract

A subset of patients undergoing total knee arthroplasty (TKA) for knee osteoarthritis develop debilitating knee stiffness (reduced range of motion) for poorly understood reasons. Dysregulated inflammatory and immune responses to surgery correlate with reduced surgical outcomes, but the dysregulated gene signatures in patients with stiffness after TKA are poorly defined. As a consequence, we are limited in our ability to identify patients at risk of developing poor surgical outcomes and develop preventative approaches. In this pilot study we aimed to identify perioperative blood gene signatures in patients undergoing TKA for knee osteoarthritis and its association with early surgical outcomes, specifically knee range of motion. To do this, we integrated clinical outcomes collected at 6 weeks after surgery with transcriptomics analyses in blood samples collected immediately before surgery and at 24 h after surgery. We found that patients with stiffness at 6 weeks after surgery have a more variable and attenuated circulating gene expression response immediately after surgery. Our results suggest that patients with stiffness following TKA may have distinct gene expression signatures detectable in peripheral blood in the immediate postoperative period.

Figure 1

Study workflow. (A) We collected peripheral blood mononuclear cells (PBMCs) and PAXgene blood RNA tubes the day of surgery (DOS) and at 24 h after surgery (POD1) in patients with knee osteoarthritis (OA) undergoing total knee arthroplasty (TKA). At 6 weeks after surgery, patients were classified as cases and controls based on range-of-motion (ROM). Stiffness was defined as ROM ≤ 95° measured by goniometer. (B) RNA isolated from PBMCs was used for NanoString gene expression analyses to compare POD1 and DOS gene expression profiles. (C) RNA isolated from PAXgene blood RNA tubes was used for RNA-seq, comparing POD1 vs. DOS samples from controls and cases. Created with BioRender.com.

Figure 2

NanoString gene expression analyses of the responses to surgery of peripheral blood mononuclear cells. (A) Heatmap representation generated via unsupervised clustering of the normalized gene expression analyses from RNA isolated from peripheral blood mononuclear cells (PBMCs) obtained the day of surgery (DOS) and at 24 h after surgery (POD1) from 5 patients undergoing total knee arthroplasty (TKA) for knee osteoarthritis (OA). Each column represents data from samples collected from one patient on the DOS or at POD1. Each row represents relative gene expression. Orange indicates high expression, and blue indicates low expression. Data were Z-score normalized, scaled to give all genes equal variance. (B) Volcano plot representation of the differentially expressed genes (red, p < 0.05) comparing POD1 to DOS. (C) Representation of the pathway scores for the immune response, inflammation response, response to stress, and response to wounding in PBMCs isolated from DOS and POD1 whole blood. (D) Representation of the linear scores for B-cells, monocytes, and T-cells in PBMCs isolated from DOS and POD1 whole blood. Data are represented as means ± S.D. (error bars). Graphs were created with GraphPad Prism 8 (GraphPad Software, San Diego, CA). *p < 0.05 **p < 0.001, ***p < 0.0001 and ****p < 0.00001 by t-test. ns indicates not significant.

Figure 3

RNA-seq analyses of the responses to total knee arthroplasty in total RNA isolated from PAXgene blood RNA tubes. (A) Principal component analysis (PCA, using top 1,000 genes) of RNA-seq data from POD1 (24 h after surgery) and DOS (day of surgery). RNA-seq analysis is based on 36 samples obtained from 18 patients at DOS and POD1. (B) Volcano plot representing the differentially expressed genes comparing POD1 to DOS samples. Red dots correspond to genes with significant changes greater than 1 log2-fold expression change (FDR < 0.05, LogFC > 1). Selected genes with increased or decreased expression that were also identified by NanoString analyses are highlighted in blue. (C) NanoString normalized counts of selected DEGs, confirming changes identified by RNA-seq. Dotted lines indicate background signal. **p < 0.01, ***p < 0.001, by t-test, calculated using ratio data and the nSolver analysis software. NanoString data are represented as means ± S.D. (error bars) normalized counts. (D) Representation of the QuSAGE pathway analyses in POD1 relative to DOS, showing functional pathways differentially expressed at POD1 (FDR < 0.01). Representative genes associated with the (E) IL1, (F) IFNG signaling, (G) IL6 signaling, and (H) DNA repair pathways are shown, representing changes in gene expression (logFC) in POD1 vs. DOS samples. Graphs were created with GraphPad Prism 8 (GraphPad Software, San Diego, CA).

Figure 4

Transcriptomics analyses of PAXgene blood RNA identifies different gene signatures in cases (stiff knees) and controls following total knee arthroplasty. Volcano plots representing differentially expressed genes at 24 h after surgery (POD1) relative to the day of surgery (DOS) in (A) controls (no stiffness, N = 10) and (B) cases (stiff knees, N = 8). Red dots correspond to genes with significant changes greater than 1 log2-fold expression change (FDR < 0.05, LogFC > 1). Selected DEGs common to cases and controls are highlighted in blue. (C) Venn diagram representation of the DEGs at POD1 compared to DOS that are common to cases and controls (N = 111) or unique to controls (N = 120) and cases (N = 51). Heatmap representation of selected DEGs that are unique to (D) controls and (E) cases is shown. Data were z-score normalized.

Figure 5

Functional analyses of the differentially expressed genes identified by RNA-seq in PAXgene blood RNA in cases (stiff knees) and controls. (A) Representation of the QuSAGE pathway analyses at 24 h after surgery (POD1) relative to the day of surgery (DOS) in control patients (N = 10) showing functional pathways differentially expressed at POD1 vs. DOS (FDR < 0.01). Representative genes associated with the (B) ATF2, (C) INFLAM, (D) canonical NF-κB signaling, and (E) activated TLR4 signaling pathways are shown, representing changes in gene expression (logFC) in POD1 vs. DOS for the controls. (F) Transcription factor regulatory network analysis using differentially expressed genes (logFC > 1, FDR < 0.05) at POD1 compared to DOS in the control group. (G) Representation of the QuSAGE pathway analyses in POD1 relative to DOS in cases (N = 8). Pathways that displayed differential expression (FDR < 0.01) only in the cases are highlighted. Representative genes associated with the (H) matrisome associated and (I) steroid hormone biosynthesis pathways are shown, representing changes in gene expression (logFC) in POD1 vs. DOS for the cases. Graphs were created with GraphPad Prism 8 (GraphPad Software, San Diego, CA).