Responsiveness of quantitative cartilage measures over one year in knee osteoarthritis: comparison of radiography and MRI assessments.

PURPOSE: To directly compare the responsiveness of quantitative imaging measures of disease progression in knee osteoarthritis (OA). In the medial compartment of the knee comparison was made between: 1) radiographic joint space narrowing (JSN); 2) global quantitative magnetic resonance imaging (qMRI) of cartilage volume; 3) regional qMRI of cartilage thickness; and 4) regional analysis using an ordered value (OV) methodology. MATERIALS AND METHODS: 3T MRI and weight-bearing radiography of the knees were performed at baseline and 1-year timepoints in 23 subjects (mean age 63 years) with symptomatic knee OA. Standardized response means (SRM) were calculated for each measure. Statistical analysis to determine significance of change between timepoints was performed with a two-tailed Student's t-test (JSN, global, regional analysis) and nonparametric Mann-Whitney test (ordered values). RESULTS: At 1 year, global cartilage volume losses of 2.3% (SRM -0.44) in the medial tibia and 6.9% in the medial femur (SRM -0.74) were recorded. SRM for JSN was -0.46. Regional analysis revealed largest reductions in cartilage thickness in the external (SRM -0.84) weight-bearing subregion of the medial femur and in the posterior subregion of the medial tibia (SRM -0.79). OV analysis in the medial compartment revealed areas of cartilage thinning (four ranked OV) and cartilage thickening (two ranked OV). CONCLUSION: The MRI OV approach proved to be a superior analysis tool for detecting changes in cartilage morphology over a 1-year period. Radiographically defined JSN was found to be the least responsive measurement method of knee OA disease progression.

Use of 3T MRI and an unspoiled 3D fast gradient echo sequence for porcine knee cartilage volumetry: preliminary findings.

PURPOSE: To assess the utility of knee cartilage volumetry using an unspoiled fat-suppressed 3D fast gradient echo (FGRE) sequence at 3T. MATERIALS AND METHODS: Sagittal magnetic resonance (MR) images were obtained with an unspoiled fat-suppressed 3D FGRE sequence in eight porcine knees. Manual segmentation was used to derive the cartilage volume. This volume was compared to a volume measurement of cartilage scraping specimens obtained by water displacement. Imaging was repeated five times in four of the knees to assess interscan volume measurement reproducibility and calculate precision error. A single 3D dataset was manually segmented five times at weekly intervals to assess intraobserver volume measurement reproducibility. RESULTS: Total cartilage volume obtained from MRI and water displacement correlated well (r = 0.75). The interscan reproducibility of total volume measurements, expressed as the coefficient of variation (CV), was 4.2%, and the precision error (root mean square [RMS] CV) was 4.1%. The CV of intraobserver estimates of total cartilage volume by MRI was 3.6%. CONCLUSION: Interscan reproducibility of quantification of total cartilage volume and reproducibility of the manual segmentation technique were both high (>95%). Accurate and reproducible cartilage volumetry can be obtained by using a clinical unspoiled fat-suppressed 3D FGRE acquired at 3T MRI.