Background
The consequences of patellar dislocation are cartilage injuries to the retropatellar joint and to the medial patellofemoral soft tissue complex with the prediction of subsequent instability [
1‐
9]. Another factor that predisposes to recurrent LPD is trochlear dysplasia with insufficient trochlear depth, which is present in up to 85% of patients with recurrent patellar dislocation [
10]. The incidence of chondral and osteochondral defects after first or recurrent LPD depends on the degree of lesions noted in previous studies. Based on a review of surgical studies, the frequencies of chondral and osteochondral lesions after LPD range between 32% and 96% [
1,
5‐
7]. Similarly, the frequency of cartilage injuries following LPD vary among several MRI studies, ranging from 30% to 75% [
2,
3,
8]. Furthermore, a worsening of the articular cartilage was described at second-look arthroscopy approximately 1.5-2 years after the diagnosis of a LPD was made [
11]. A 7-year non-operative follow-up study demonstrated high frequencies of full-thickness patellar (45%) and trochlear (31%) cartilaginous lesions, which were presumed to be a sign of developing osteoarthritis [
9]. After an average follow-up of 13 years, Mäenpää
et al. [
4] diagnosed patellofemoral osteoarthritis in 22% of the patients, the highest frequency occurring in patients who underwent late surgery for patellofemoral pain or recurrent luxation. With respect to these data, accurate identification and appropriate treatment of cartilaginous lesions appears to be of special interest after patella dislocation. Thus, MR imaging, as a non-invasive method for cartilage assessment, could play an important role in the prevention of subsequent knee disability. This study was performed to investigate, whether MRI provides a reliable diagnostic performance for the assessment of the articular cartilage in patients with LPD. Therefore, cartilage diagnostics on standardized pre-operative MR images was compared to arthroscopic findings performed immediately after a first or recurrent LPD. To our knowledge, this study is the first of its kind to evaluate the diagnostic value of MRI for the cartilage assessment exclusively in a representative sample of patients with first and recurrent LPDs.
Results
There were only 3 patients (8%) without any cartilage disease noted during arthroscopic assessment. The distribution of cartilaginous lesions within the patellofemoral joint is depicted in Table
1. The MRI grading of both reviewers were compared to arthroscopic findings (Table
2), showing an exact agreement in 78% (186 of 240) and 76% (182 of 240) of the joint surfaces. As presented in Table
3, intra- and inter-observer agreement differed markedly between the patient cohorts. Thus, moderate-to-good kappa values were obtained in patients with recurrent LPD, whereas good-to-very good values were yielded in patients with first LPD.
Table 1
Distribution of cartilage disorders within the patellofemoral joint during arthroscopic assessment
Medial facet | 17 (12/5) | 7 (1/6) | 10 (5/5) | 1 (0/1) |
Central dome | 4 (3/1) | 4 (3/1) | 6 (1/5) | 2 (1/1) |
Lateral facet | 1 (0/1) | 1 (1/0) | 0 (0/0) | 1 (0/1) |
Medial trochlear groove | 0 (0/0) | 0 (0/0) | 4 (2/2) | 2 (1/1) |
Central trochlear groove | 1 (1/0) | 0 (0/0) | 4 (2/2) | 1 (0/1) |
Lateral trochlear groove | 5 (2/3) | 2 (0/2) | 7 (5/2) | 2 (0/2) |
Table 2
Comparison between both readers with respect to MRI grading and arthroscopic grading of the cartilage
Grade 0 | 143/138 | 6/10 | 8/8 | 0/1 | 1/1 |
Grade 1 | 8/7 | 1/1 | 0/1 | 0/0 | 0/0 |
Grade 2 | 15/13 | 0/0 | 11/14 | 3/2 | 2/2 |
Grade 3 | 1/2 | 0/0 | 3/3 | 8/7 | 2/2 |
Grade 4 | 1/1 | 0/0 | 3/1 | 3/2 | 23/22 |
Table 3
Weighted kappa values and 95% confidence intervals for both MRI readers, and inter- and intra-observer agreement in all patients with LPD, in patients with a first LPD, and in patients with a recurrent LPD
Reader 1 vs. Reader 2 | 0.75 (0.68-0.83) | 0.82 (0.74-0.91) | 0.67 (0.55-0.80) |
AC* vs. reader 1 | 0.73 (0.65-0.80) | 0.83 (0.75-0.91) | 0.61 (0.48-0.74) |
AC* vs. reader 2 | 0.70 (0.62-0.78) | 0.79 (0.70-0.88) | 0.60 (0.47-0.73) |
Both patient groups were assessed separately for the diagnostic values of each grade of cartilage disease (Table
4). For grade 3 and 4 lesions, the patient group with first LPDs had markedly higher sensitivities and positive predictive values than those with recurrent LPDs. Of 12 patients with osteochondral injuries at the medial patella, 9 had their first LPD. These lesions were all correctly assessed at MRI as grade 4 cartilage defects with ulceration (Figure
4) or fissuring (Figure
5) of the subchondral bone.
Table 4
Diagnostic values of MRI readings (reader 1/reader 2) for each grade of cartilaginous lesion in patients with first and recurrent LPDs
Grade 4
| First LPD | 89/83 | 99/100 | 94/100 | 98/97 |
| Recurrent LPD | 70/70 | 96/95 | 64/58 | 97/97 |
Grade 3
| First LPD | 83/60 | 98/98 | 63/60 | 99/98 |
| Recurrent LPD | 60/44 | 97/97 | 67/57 | 96/95 |
Grade 2
| First LPD | 33/40 | 97/96 | 63/55 | 92/92 |
| Recurrent LPD | 38/50 | 91/91 | 40/47 | 90/92 |
Grade 1
| First LPD | 0/0† | 98/96 | 0/0† | 98/98 |
| Recurrent LPD | 14/14 | 98/97 | 20/11 | 97/97 |
Discussion
During LPD, the medial facet of the patella impacts against the lateral femoral condyle, which can lead to corresponding injuries of the articular surface. Because dislocation is usually transient, the patella recoils back and the corresponding articular surfaces can sustain injury again. This leads to a high incidence and typical locations of cartilaginous defects [
1‐
3,
5,
6,
8]. In agreement with previous studies on LPD, cartilaginous lesions were predominately noted at the medial facet of the patella (55%) and the lateral femoral condyle (25%; Table
1) [
1,
5,
6]. According to clinical follow-up studies, as well as experimental studies, chondral lesions may increase the risk of subsequent patellofemoral joint symptoms and osteoarthritis [
4,
9,
11,
17]. Therefore, accurate identification and appropriate treatment of cartilaginous lesions following LPD play an important role in minimizing knee disability.
Based on the literature, the MRI sequence best suited for cartilage diagnostics is still under debate [
18‐
21]. Cartilage-specific sequences, such as spoiled gradient-recalled echo and fast low-angle shot sequences, provide a high spatial resolution and have therefore been described as being useful in segmenting techniques for quantitative cartilage studies. The disadvantages of these sequences are a high sensitivity to susceptibility artifacts and a limited visualization of the subchondral bone, menisci, and ligaments [
19,
22]. Most experience and good results for the detection of cartilage and subchondral bone disorders were gathered with T2-, intermediate- and PD-weighted fast spin echo sequences [
19,
20,
23‐
25]. In the current study, we used fat-suppressed PD-weighted fast spin-echo sequences with a 3-mm slice thickness in transverse planes and a 4-mm slice thickness in sagittal planes. In previous reports on comparable sequences with and without fat suppression, 1.5 Tesla MRI was described to depict the articular cartilage with an accuracy comparable to that of several cartilage-specific sequence protocols [
18,
20,
21,
24]. Similar results were noticed for T2- and intermediate-weighted fast spin echo sequences, which also yielded comparable results to those of other cartilage-specific sequence protocols [
23,
26,
27]. However, our study demonstrated relatively good inter- and intra-observer agreement (Table
3) in comparison to previous MRI studies on cartilage grading, in which kappa values ranged from 0.60-0.93 [
13,
21,
28,
29].
In patients with first and recurrent LPDs, the diagnostic performance of MRI for cartilaginous lesions was evaluated for each grade of cartilage disease (Table
4). At each grade, the specificities and negative predictive values were relatively high, giving MRI a certain importance for the exclusion of cartilaginous lesions. In agreement with the literature, the sensitivities for the detection of grade 1 and 2 lesions were poor (Tables
2 and
4) [
13,
18,
30]. Thus, reliable MRI differentiation of superficial erosions or fibrillations from intact cartilage appears difficult after LPD.
Regarding grade 3 and 4 lesions, patients with first LPD showed markedly higher diagnostic values compared to those with recurrent dislocation. Reader 1's sensitivity and positive predictive value for grade 4 lesions were 89% and 94% in patients with first LPDs, but only 70% and 64% after LPDs, respectively. Likewise, reader 1's diagnostic values for grade 3 lesions were higher in patients with a first LPD compared to those with a recurrent LPD (Table
4). Similar tendencies existed in reader 2's sensitivities and positive predictive values. Regarding the positive predictive values in patients with recurrent LPDs, the probability that the MRI finding of a grade 3 and 4 defect corresponds exactly to the arthroscopic finding was between 57% and 64%. Therefore, the value of MRI for a detailed assessment and grading of the cartilage should not be overestimated, especially after recurrent LPDs. Likewise, the kappa values for the intra- and inter-observer agreements yielded markedly better results in patients with first LPDs compared to those with recurrent LPDs (Table
3). Regarding the kappa values (Table
3) and the diagnostic values for grade 3 and 4 lesions (Table
4), we assume that MRI is more reliable for the diagnosis of cartilaginous defects in patients with first LPDs, whereas the diagnostic performance is limited after recurrent LPDs.
Better diagnostic values in patients with first LPDs could be explained in part by the higher severity of trauma. Thus, as reported by others, severe cartilaginous lesions with ulceration or fissuring of the subchondral bone were more frequent in patients with first LPDs (48%) compared to patients with recurrent LPDs (16%; Table
1) [
1,
3,
5,
7,
9,
31]. In this context, it has to be mentioned that arthroscopically-detected osteochondral lesions occurring after LPDs were identified with pre-operative x-ray in 29% and 60% of the cases [
31,
32]. Thus, correct identification of osteochondral lesions appears to be limited on standard radiographs. In contrast, all osteochondral defects in our study were correctly assessed at MR imaging as grade 4 cartilage lesions with ulceration (Figure
4) or fissuring (Figure
5) of the subchondral bone. In addition to a role for the detection of osteochondral lesions, MRI could be of practical assistance in planning the surgery. In our patient cohort, visualization of difficulties for a refixation of osteochondral fragments, such as cortical steps (Figure
5) and bone destructions (Figure
4), as well as the visualization of intra-articular loose bodies (Figures
5 and
6), provided additional information before surgery. Therefore, we suggest that MRI is an excellent diagnostic tool for osteochondral lesions in patients with LPD.
A limitation of this study was the use of the Outerbridge classification for cartilage assessment. Recent reports describe quantitative, semi-quantitative, and whole organ approaches for MRI assessment of the cartilage as reliable scoring and research tools, especially in patients with osteoarthritis [
22,
29]. Furthermore, the use of arthroscopic grading as a reference standard should be regarded with caution. In the literature, inter-observer agreement at arthroscopy demonstrates sufficient reproducibility [
33], but poor results for cartilage grading [
34]. On the other hand, a study by Bachmann
et al. [
35] yielded an exact agreement between arthroscopic and histopathologic grading in 287 of 300 cases. Thus, the arthroscopic method is a valuable tool in clinical research to score chondropathies, even if inspection and palpation with the hook probe cannot detect all changes of the cartilage as a histomorphologic evaluation.
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
LVvE, GS, AD, BB, PH, and TKL conceived and designed the study. LVvE, MR, PH, and AD were involved in the execution of the study. In addition, LVvE and MR performed MRI grading and LVvE performed the statistical analysis. LVvE, GS, MR, PH, AD, BB, and TKL were involved in the writing and proofreading of this manuscript. All authors read and approved the final manuscript.