INTRODUCTION

Vertebral bone marrow lesions known as Modic changes are specific for axial low back pain. Fibrosis, inflammation, and enhanced myelopoiesis have been described as hallmarks of Modic changes. CD90-positive bone marrow stromal cells (BMSC) regulate inflammatory myelopoiesis and drive fibrosis of the bone marrow in primary myelofibrosis. However, the role of CD90-positive BMSC in Modic change pathophysiology is unknown. In this study, we investigated CD90-positive BMSC in biopsies from MC and their association with inflammatory and fibrotic markers. We hypothesized that presence of CD90-positive BMSC correlates with bone marrow fibrosis and edema.

METHODS

This study was approved by the local Ethics Commission and conducted in accordance with the Declaration of Helsinki. From fourteen patients undergoing lumbar spinal fusion twenty-two bone marrow biopsies were taken through the pedicle screw trajectory before screw insertion. Collected patient demographics included age, sex, size, weight, height, BMI, smoker, 10-point visual analogue score (VAS) for back and leg pain (VAS.back, VAS.leg), and Oswestry-Disability-Index (ODI). Two experienced radiologists independently classified MC main type (Control, MC1, MC2), graded disc degeneration (DD)(Pfirmann grade 0-5), and endplate damage (Rajasekaran grade 0-6). Biopsies were processed for histology and thin sections were stained with hematoxylin/eosin (HE), Masson trichrome (MT) and analysed with immunohistochemistry for alpha-smooth-muscle-actin (αSMA), type I collagen (COL1), type III collagen (COL3), cellular fibronectin (FN), Thy-1-membran-glykoprotein (CD90), and endoglin (CD105). Two pathologists independently scored immunopositivity (0-8) of αSMA, COL1, COL3, FN, CD90, and CD105, and scored inflammatory infiltrates and edema (=interstitial water) on HE slides (0-8). Scores were compared between MC types with Kruskal Wallis test with Bonferroni correction. Scores were (i) correlated to patient demographics and clinical data and (ii) to each other using Kendall-Tau correlation.

RESULTS

There was no difference in patient demographics between MC1, MC2, and control. CD90-positive BMSC were more frequent in MC1 (3.34±2.33, p=0.001) and MC2 (1.88±1.70, p=0.052) compared to control (0.13±0.35) (Fig.1). Edema was not found in control bone marrow but frequently in MC1 (1.13±0.99, p=0.015) and MC2 (0.83±0.75, p=0.044). Scores for inflammatory cell infiltrates above 1.5 were exclusively found in MC1 (1.88±1.69 vs. control 0.38±0.69, p=0.081). Connective tissue in the bone marrow (measured with MT) was increased in MC1 (2.19±1.53, p=0.027) compared to control (0.44±0.82). The discs adjacent to MC1 and MC2 were stronger degenerated, and endplates were more damaged than control discs and endplates (all p<0.001). CD90 correlated with hallmarks of MC1, e.g. edema (tau=0.69, p<0.001), endplate damage (tau=0.65, p<0.001), and DD (tau=0.56, p=0.002). CD90 also correlated with the pro-fibrotic markers MT (tau=0.48, p=0.006) (Fig.2), COL1 (tau=0.44, p=0.011), and FN (tau=0.48, p=0.006). Importantly, correlation of MT with VAS.back (tau=0.51, p=0.068) suggests that bone marrow fibrosis is linked to back pain in Modic changes.

DISCUSSION

Correlation of CD90-positive BMSC with pro-fibrotic markers and edema points at a key role of CD90-positive BMSC in MC pathomechanisms. CD90-positive BMSC might be responsible for bone marrow fibrosis in Modic changes similar as in primary myelofibrosis. Inhibiting proliferation and activation of CD90-positive BMSC could become a novel interesting treatment target for MC-related low back pain.