The interaction among clinical features, imaging findings, and outcome measures of OA development is complicated. In this longitudinal study, we assessed the evolution of joint pre-structural and structural features, cartilage volume and thickness, and cartilage biochemical composition as determined by knee MRI at one year after initiation of intra-articular injection in patients with mildly to moderately symptomatic OA. Our findings indicate that quantitatively defined baseline MFTJ cartilage thickness and semi-quantitatively assessed MFTJ and total joint cartilage loss over the 12 months are associated with changes in clinical outcome. These results confirm the important role of cartilage in knee OA progression.
MRI is a useful, non-invasive imaging tool for evaluation of OA in longitudinal clinical trials. There have been conflicting reports regarding the association between MRI OA features and symptomatic progression [11]. Sayre et al. suggested that only presence of osteophytes was significantly related to pain progression over a 7.5-year longitudinal early OA study period [13]. Magnusson et al. found that meniscal extrusion, full thickness cartilage loss, and presence of osteophytes were associated with increased knee pain in a six-year study of pre-radiographic OA patients [10]. Eckstein et al. demonstrated that reduced MFTJ cartilage thickness over two years had a strong association with radiographic progression but a weak association with pain progression [14]. In addition, some studies have demonstrated that increased severity of synovitis was related to increasing pain severity [15]. However, our study showed that clinical improvement was significantly correlated with cyst progression and synovitis. These results imply that intra-articular injection provided some beneficial effect for advanced lesions in OA patients.
We found inconsistent results between quantitative and semiquantitative evaluations of cartilage. At baseline, quantitative measurements of cartilage thickness and volume correlated with changes in clinical outcome, but semiquantitative measurements did not. In contrast, when evaluating cartilage change over time, the semiquantitative method showed a significant correlation with changes in clinical outcome, but the quantitative method did not. However, previous studies have shown that quantitative measurement was more sensitive than semiquantitative scoring for detection of cartilage morphology change [16, 17]. In contrast, Reichenbach et al. reported that semiquantitative assessment could be more sensitive to detect cartilage damage in mild OA because semiquantitative scores captured focal cartilage erosions that might have been missed by quantitative measurements that summarize morphology over a broad area [18]. We suggest that minor changes in the local area undetected by quantitative measurement can be captured by semiquantitative assessment.
T1ρ and T2 mapping methods are the most clinically applicable MRI techniques evaluating biochemical composition before morphologic change. T1ρ may be more sensitive than T2 for early cartilage degeneration [19]. Interestingly, our analysis of the association between T1ρ and clinical outcome indicated a trend opposite to our expectation. Previous researchers provided incompatible findings that baseline T1ρ and T2 parameters are potential predictors of OA progression. Baseline MRI compositional markers may not provide better discrimination between knees with OA progression and those without significant progression than simple radiographic measures [20, 21]. Edd et al. investigated longitudinal change in femoral cartilage T2 value and thickness in progressive OA and found that T2 values increased in the early stage and cartilage thickness decreased mainly in the later stage [22]. These findings support the ability of the cartilage imaging biomarker to predict OA severity and progression dependent on the disease stage. We assumed that inclusion of patients with moderate OA of K-L grade III in the target group affected these results.
The general belief is that joint symptoms as well as imaging findings of OA increase with age [23]. In this study, however, changes in clinical outcome were found to be more favorable with increasing age. This may be explained by the relatively lower physical demands of older patients and higher expectations of younger patients. Also, previous studies demonstrated that patients over the age of 60 years with moderate OA are more likely to have a positive response to intra-articular HA administration [24].
There are several limitations in this study. First, our study included a small sample size and heterogeneous population of KL grade from I to III. KL II and KL III knees, in particular, cover a wide spectrum of structural disease that varies spatially [25]. The heterogeneity of KL II and III patients should be considered when selecting target subjects for clinical trials of disease progression. Therefore, further investigation using larger cohorts and homogeneous subject groups are needed to confirm our results. Second, we did not include the entire joint for quantitative analysis of cartilage; we only analyzed the MFTJ. Although OA is a disease involving the entire joint, the medial femorotibial compartment is the most sensitive region as medial regions experience greater loads that lead to greater cartilage loss [26]. Cartilage manual segmentation was time-consuming and labor-intensive and was performed only for the medial compartment. Third, the obtained T1ρ and T2 measurements may be inappropriate owing to contamination of the cartilage signal by synovial fluid. Mismatch related to different 2D and 3D image positions or knee movement may also affect these measurements. T1ρ and T2 values greater than 200 msec, more than double the highest expected mean T2 value, were considered outliers and were excluded from statistical analysis.