DISTRIBUTION IN KNEE OSTEOARTHRITIS - IMPACT OF DEMOGRAPHIC, RADIOGRAPHIC & MRI STRUCTURAL PATHOLOGY STRATA: DATA FROM THE IMI-APPROACH COHORT

Abstract

INTRODUCTION

Most participants of the IMI-APPROACH knee OA cohort displayed cartilage damage, based on quantitative (segmentation-based) MRI morphometry and MOAKS scoring, predominantly in the medial tibiofemoral compartment. Cartilage surface mapping (CaSM) is a quantitative 3D analytic method that, unconstrained by subregional boundaries, can demonstrate visually how cartilage thickness varies across a joint.

OBJECTIVE

The purpose of this cross-sectional study was to evaluate cartilage thickness distribution in knee OA patients using CaSM, and to analyze how it varies amongst demographic, radiographic, and MRI structural pathology strata.

METHODS

The cohort included 297 participants with clinical knee OA. 1.5T or 3T MRI 3D gradient echo sequences were acquired at baseline and follow-up, with only baseline being used in the current analysis. Semi-automatic segmentation of the femoral and tibial cartilage was performed using Stradview. Segmentations were registered to canonical surfaces using wxRegSurf and analyzed in MATLAB. The relationship between demographic and structural pathology strata on the cartilage thickness distribution was analyzed using statistical parametric mapping (SPM). SPM allows for vertex-wise comparisons and delivers multiple-comparison-corrected F-test statistics, using the Surfstat MATLAB package. p<0.05 was used as the threshold for statistical significance. Sex, age, and BMI were examined (demographic factors). Presence of radiographic OA (ROA; KLG≥2) and degree of medial/lateral JSN were studied as radiographic factors, and MOAKS BMLs and meniscal extrusion (scored on intermediate-weighted fat-suppressed sequences) as MRI structural pathology features, in individual models. Analysis differentiating patients with and without ROA was performed in addition.

RESULTS

287 patients could be analyzed (age 66.4±7.1, BMI 28.0±5.2, 78% female, 55% ROA). Male patients had significantly thicker cartilage across the entire joint, especially the tibiae and trochlea (Figure 1). Age showed a pronounced effect in the trochlea and (central) medial and lateral tibia, with older patients having thinner cartilage (independent of radiographic status); BMI was not significantly associated with cartilage thickness in any region (Figure 1). Patients with ROA showed significantly thinner cartilage in the tibiae and medial femur than those without ROA, but thicker cartilage in the trochlea and lateral femur (Figure 1). Patients with JSN showed opposite effects depending on direction: those with medial JSN displayed significantly thinner cartilage in the medial, and those with lateral JSN in the lateral compartment (Figure 2). Meniscal extrusion results were very similar to JSN results (Figure 2). Presence of BMLs in any subregion of a compartment (lateral/medial FT and PF) was associated with thinner cartilage throughout that entire compartment (Figure 3). While the difference between male and female patients was significant for non-ROA and ROA patients separately, cartilage distribution variations based on structural pathology (JSN, BMLs, meniscal extrusion) were only visible (and significant) for patients with ROA.

CONCLUSION

Both demographic factors (age and sex) and radiographic/MRI structural pathology features (presence of ROA, JSN, BMLs and meniscal extrusion) are significantly associated with variation in cartilage thickness distribution throughout the joint in patients with clinical knee OA, although for structural pathology only in patients with ROA. Future analyses will indicate whether these and other factors are associated with longitudinal cartilage thickness changes.