Earliest stages of cartilage degeneration were associated with decrease in glycosaminoglycan concentration (GAG). It is hypothesized that cartilage degeneration might be reversible in this early stages, thus a noninvasive evaluation of the GAG content is highly desirable. Since the GAG molecules in cartilage are in equilibrium with sodium (23Na) ions, in vitro experiments with controlled cartilage degradation showed very high sensitivity of 23Na-MRI to even small changes in GAG concentration.(1) Although 23Na-MRI was used for the evaluation of GAG content in cartilage repair tissue(2) and in osteoarthritic cartilage(3), this method has not yet been employed for the evaluation of untreated posttraumatic low-grade chondral defects. Thus, the aim of this 7T study was: i) to evaluate the reproducibility of 23Na-MRI by comparing data obtained at baseline and at 8-days follow-up in patients with chondral defects; ii) to compare 23Na-values between defects and normal-appearing femoral cartilage in weight and non-weight bearing region.

Institutional Review Board approval and informed consent from all patients were obtained. Six patients (4 females, 2 males; mean age, 51.5±5.0 years) with untreated chondral defects of ICRS grade 1 or 2 in the femoral cartilage (4 right and 2 left knees; 1 trochlea, 3 lateral and 2 medial condyles) were included in this study. All MRI scans were acquired at 7T whole body system (Magnetom, Siemens Healthcare, Germany). T2-weighted 3D DESS images with water excitation (resolution= 0.5×0.5×0.5 mm³; TR/TE= 8.68/2.55 ms; measurement time= 3:58 minutes) (Fig.1) were acquired using a 28-channel knee array coil (Quality Electrodynamics, OH). 23Na-images were acquired using a spoiled gradient echo sequence with variable echo time scheme (vTE-GRE)(4) optimized for 23Na-MRI of cartilage (resolution= 1.6×1.6×3.0 mm³; TR= 9.2 ms; TEmin/TEmax= 1.22/1.82 ms; flip angle= 51 degrees; bandwidth= 100 Hz/pixel; measurement time= 25 minutes) (Fig.1) using a 15-channel 23Na-only knee array coil (Quality Electrodynamics, OH).

A radiologist with 24 years of experience in musculoskeletal MRI identified one ICRS grade 1-2 cartilage defect for each patient on morphological images from a routine 3T exam and selected corresponding DESS images that were used for manual segmentation of cartilage and for measurements of cartilage thickness (Fig.1). Corresponding 23Na-images were corrected for the spatially variable sensitivity of the knee coil using measurements of a homogeneous phantom and Matlab scripts (Mathworks, MA). All regions-of-interest (ROI) were drawn on DESS images and transferred to the corresponding 23Na-images. In each patient, ROI analyses were performed in a chondral defect, in weight-bearing and in non-weight bearing regions of normal-appearing femoral cartilage on three consecutive 23Na-slices using the JiveX viewer (VISUS GmbH, Germany). Mean 23Na-signal intensities were normalized using a factor derived from the signal intensity of the reference sample attached to knee coil and measured with each patient. The corrected signal intensities (cSI) were calculated by dividing the normalized signal intensities (nSI) with correction factors for partial volume effects (PVEcf) obtained by fitting the cartilage thickness in ROI to a function describing the signal attenuation in 23Na-image with a 2 mm resolution (Fig.2).(5) Intra-class correlation coefficients and paired t-tests were calculated in the IBM SPSS statistic software.

The mean cartilage thickness was not significantly different in the defect (3.1±0.5mm), weight-bearing (2.4±0.1mm), and non-weight-bearing ROIs (2.7±0.3mm) (all p-values >0.087). The mean and standard deviation of 23Na-nSI and 23Na-cSI values from weight-bearing, non-weight-bearing and defect ROIs are shown in Table1. The intra-class correlation coefficient was 0.95 for both, 23Na-nSI and 23Na-cSI values. No significant differences were observed in 23Na-nSI values between defect, weight-bearing, and non-weight-bearing ROIs at baseline and at 8-days follow-up (all p>0.224) (Fig.3). However, 23Na-cSI values showed significant differences between weight-bearing zone and defect at baseline (p=0.047) and at 8-days follow-up (p=0.011) (Fig.4). Other differences in 23Na-cSI values were not statistically significant.Our preliminary results demonstrate the ability of 23Na-MRI to differentiate between normally appearing cartilage and low-grade chondral defects (ICRS grade 1 or 2). This underlines high sensitivity of 23Na-MRI to small changes in GAG content that was shown in previous in vitro studies.(1) However, due to low resolution of sodium images, correction for partial volume effects is necessary for visualization of GAG differences. Our data from baseline and 8-days follow-up showed that 23Na-MRI at 7T can offer very robust and repeatable results. This is an ongoing study that aims to measure up to 20 patients with additional follow-up measurement after three months.23Na-MRI can be used for noninvasive follow-up of cartilage degeneration based on monitoring of changes in GAG content in patients with low-grade cartilage defects. 23Na-MRI data may serve as a helpful biomarker for the evaluation of cartilage regenerating therapies.