All patients undergoing MRI examinations in our hospital from June 2018 to June 2020 were selected. First, the two keywords of “patella dislocation” and “no obvious abnormality of knee joint” were used to search MR reports in the picture archiving and communications system (PACS) workstation (Centricity, GE Healthcare, St. Gilles, United Kingdom) to initially screen patients. Then, on the basis of both their medical histories and previous medical records. Two head doctors with more than three years of work experience in joint and sports medicine screened the subjects according to the following inclusion and exclusion criteria. A total of fifty-four patients with recurrent LPD (seventy-five knees) and seventy controls (seventy-five knees) were enrolled. Moreover, age and sex were matched as closely as possible between the two groups.
The inclusion and exclusion criteria for the patients with recurrent LPD were as follows:
Inclusion criteria: ① Recurrent LPD was diagnosed by two senior doctors in the joint and sports medicine department according to the patient's history, physical examination findings and MRI findings. ② The patient was not previously treated in the rehabilitation department or receive any special training related to strengthening the quadriceps muscles. ③ MRI images were taken within 10 days after the recurrence of LPD.
Exclusion criteria: ①Patients with primary patellar dislocation. ② Traumatic patellar dislocation occurred as a result of direct trauma to the medial patella or a fall onto the knee joint with concomitant patellar dislocation. ③ Patients with any preexisting knee disorders, previously underwent knee surgery, had a fracture of the distal femur or tibial head, or had a multi-ligament injury. ④ Patients with history of a neuromuscular disease (e.g., polio). ⑤ Patients with obvious effusion of the knee joint.
The inclusion and exclusion criteria for the control group were as follows: MRI examination of the knee was performed for people to exclude diseases because of knee discomfort and no significant structural damage (e.g., fractures) or anatomical abnormalities (e.g., osteoarthritis) was reported.
Sagittal, coronal, and transverse MR images were obtained in all patients. Two doctors in joint and sports medicine measured the following five parameters related to the VMO (elevation on sagittal plane and coronal plane, craniocaudal extent, muscle-fibre angulation, cross-sectional area ratio) and two patella tilt parameters (patella tilt angle, bisect offset ratio) in both groups. The type of femoral trochlear dysplasia present in each patient was recorded according to the classification system reported by Dejour et al. (18) and Lippacher et al. (19) on axial MR images: type A, shallow trochlea and a subchondral sulcus angle >145 degrees; type B, flat or convex trochlea; type C, asymmetry of trochlear facets with a hypoplastic medial facet; type D, asymmetry of trochlear facets or cliff pattern, it was further categorized as normal or low-grade (type A), or high-grade dysplasia (type B, C, or D). Moreover, the diagnosis of the VMO injury was recorded according to the criteria reported by Elias et al. (20). Except for the cross-sectional area, which was calculated by ImageJ freeware, the parameters were measured by the PACS workstation. All the parameters were repeatedly measured within an interval of two weeks. The MRI (Philips MR Systems Ingenia 3.0T, Andover, Massachusetts) protocols used in our hospital were described in our previous study (21). All patients were in a supine position, with a standard knee coil center level against the lower edge of the patella. The knee and hip joint naturally extended, and the feet were braced to prevent any movement. Our MRI protocol includes: ① coronal proton density weighted spectral attenuated inversion recovery (PDW-SPAIR) MR images [repetition time msec (TR)/echo time msec (TE) 1,940/30, field of view (FOV) 220mm ×179mm, matrix 368 × 245, slice thickness 3 mm, sections per slab 21]; ② transverse PDW-SPAIR MR images (TR/TE 2,036/30, FOV 169mm × 189 mm, matrix 344 × 264, slice thickness 4 mm, act slice gap 0.4mm, sections per slab 24); ③ sagittal T1- weighted aTSE (turbo spin-echo) MR images (TR/TE 694/12, FOV 160 mm × 160 mm, matrix 308 × 240, slice thickness 3 mm, act slice gap 0.3mm, sections per slab 24); ④ sagittal proton density weighted spectral inversion recovery (PDW-SPIR) MR images (TR/TE 1,554/30, FOV 160 × 160 mm, matrix 292 × 231, slice thickness 3 mm, act slice gap 0.3mm, sections per slab 24).
MR measurements
The measurement of VMO elevation
The VMO elevation was measured in the sagittal and coronal planes according to Zhang et al.'s (22) measurement method. In brief, the transverse slice in which the adductor tubercle could clearly be seen was defined as the optimally measurable slice, as indicated by a blue line. In this transverse image, the corresponding sagittal and coronal planes were identified (Fig. 1).
On the selected sagittal slice, the apex of the anterosuperior border of the bone cortex of the adductor tubercle was set as the starting point. VMO elevation was defined as the shortest distance from the starting point extending obliquely to the inferior edge of the muscle belly. On the selected coronal slice, the apex of the medial superior border of the adductor tubercle was set as the starting point. VMO elevation in the coronal plane was defined as the vertical distance from the starting point to the inferior margin of the VMO muscle (Fig. 1).
The measurement of muscle-fibre angulation and craniocaudal extent of the VMO
First, the “Roman arch” was most obvious in the axial plane, and the corresponding sagittal slice was selected (Fig. 2 a, d). Two concentric circles were drawn on the proximal and distal sides of the femur, and the line passing through two centers was taken as the longitudinal axis of the femoral shaft. Second, the adductor tubercle was found on the transverse slice, and the corresponding sagittal slice was located to determine the lowest point of the VMO. The VMO muscle-fibre angulation, the angle between the VMO muscle-fibre and the longitudinal axis of femoral shaft, was measured in the sagittal plane (Fig. 2 b, e). The lowest point of VMO was located in this plane, and the corresponding horizontal line was established in the sagittal plane central to the patella longitudinal axis. The craniocaudal extent of the VMO was defined as the vertical distance from this horizontal line to the proximal patellar pole (Fig. 2 c, f).
The measurement of the cross-sectional area ratio of the VMO
According to the method introduced by Balcarek et al. (23). First, the longitudinal axis of the patella was established in the central sagittal plane. In this sagittal image, the corresponding transverse slice located at the proximal patellar pole and the adjacent slice located above and below this reference slice were identified. Then the cross-sectional area of the VMO and the whole thigh were calculated on these three slices respectively (Fig. 3). The cross-sectional area ratio of the VMO was defined as the ratio between the cross-sectional area of the VMO and the whole thigh. Finally, the mean cross-sectional area ratio among the three slices was obtained.
The measurement of the patella tilt angle and patella offset index
The transverse plane, which allows the visualization of the intact “Roman arch” and posterior femoral condyles, was selected. The posterior condylar reference line was drawn tangent to the posterior femoral condyles. The patella tilt angle was formed by the line along the width of the maximal patella and the line along the posterior femoral condyle (Fig. 4).
According to the method described by Christopher et al. (24) and Callaghan et al. (25), a line was drawn through the deepest portion of the trochlear groove and perpendicular to the posterior condylar reference line. The intersection of this line and the line along the width of the maximal patella was defined as point O. In the transverse plane of the widest layer of the patella, the innermost point of the patella was defined as point A, and the outermost point was defined as point B (Fig. 4). The ratio of OB / AB was defined as the bisect offset ratio.
Statistical analysis
SPSS 22.0 (IBM Corp. Released 2013. IBM SPSS Statistics for Windows. Armonk, NY: IBM Corp) was used to assess the relevant data. All parameters are presented as the mean ± standard deviation. The continuous and categorical variables were compared between the two groups were analyzed by the independent-samples t test and chi-square test, respectively. P <.05 was considered statistically significant. Moreover, the intraclass correlation coefficient (ICC) was also analysed for duplicated measurements taken by two observers.