Baseline knee joint effusion and medial femoral bone marrow edema, in addition to MRI-based T2 relaxation time and texture measurements of knee cartilage, can help predict incident total knee arthroplasty 4–7 years later: data from the Osteoarthritis Initiative

• Published in: Skeletal radiology 2019, Vol.48 (1), p.89-101
• Main Authors: Heilmeier, Ursula, Wamba, John Mbapte, Joseph, Gabby B., Darakananda, Karin, Callan, Jennifer, Neumann, Jan, Link, Thomas M.
• Format: Article
• Language: English
• Subjects: Arthritis; Arthroplasty (knee); Biocompatibility; Bone marrow; Cartilage; Cartilage diseases; Dropsy; Edema; Effusion; Femur; Imaging; Joint surgery; Knee; Magnetic resonance imaging; Measurement; Medical research; Medicine; Medicine & Public Health; Medicine, Experimental; NMR; Nuclear magnetic resonance; Nuclear Medicine; Orthopedics; Osteoarthritis; Parameters; Pathology; Radiology; Regression analysis; Relaxation time; Scientific Article; Statistical analysis; Surgical implants; Texture; Tibia; Variance analysis
• ISSN: 0364-2348; 1432-2161
• DOI: 10.1007/s00256-018-2995-4

Objective To evaluate if baseline pathological knee conditions as assessed via single features of the MR-based Whole-Organ Magnetic Resonance Imaging Scoring (WORMS), standard T2, and T2 gray-level co-occurrence matrix (GLCM) texture parameters of knee cartilage can serve as potential long-term radiological predictors of incident total knee arthroplasty (TKA) 4–7 years later. Materials and methods Baseline 3-T knee MRIs of 309 subjects from the Osteoarthritis Initiative ( n  = 81 TKA cases, with right-knee TKA 4–7 years after enrolment, and n  = 228 TKA-free matched controls) were evaluated for the presence and severity of pathological knee conditions via modified WORMS. Knee cartilage was segmented and standard T2 cartilage and T2 GLCM texture measures (contrast, variance) were computed. Statistical analysis employed conditional logistic regression. Results We found that a one-point increase on the joint effusion scale, the bone marrow edema scale or on the cartilage lesion scale at baseline predicted incident TKA (ORs: 2.45, 1.65, and 1.37 respectively ( p  ≤ 0.003)). For T2 cartilage measurements, we observed that in the lateral femur, a 1-SD increase in T2 relaxation time yielded a 28% increase in the odds of TKA (1.28 [1.09–1.643], p  = 0.046). When looking at cartilage texture, we similarly noted that a 1-SD increase in the cartilage texture parameter “contrast” was associated with a 33–40% increased risk of incident TKA in the lateral femur and tibia (0.003 ≤  p  ≤ 0.021), as was a 1-SD increase in the texture parameter “variance” in the lateral femur ( p  = 0.002). Conclusion Radiological evaluation of standard knee MR images via single WORMS features and T2 standard and texture analysis at baseline can help predict the patient’s individual risk for an incident TKA 4–7 years later.