Patient selection
Approval of the local ethics committee was obtained before starting the study. The study population was composed of knee pain and trauma patients referred to our hospital. This retrospective study included 450 consecutive patients who were examined between November 2014 and February 2015. Among them, 133 patients were excluded because of the presence of major trauma, anterior cruciate ligament rupture, multiple ligament injuries, femoral fracture, bipartite patella, previous knee surgery, and/or widespread artifacts. The final analysis included 317 patients.
MRI techniques
A 1.5-T MRI unit (Signa HDxt; GE Medical Systems, Carrollton, TX, USA) and an extremity coil were used. Sagittal T1-weighted fast spin echo (TR/TE 750/10, matrix size 256 × 256, field of view 18 cm, slice thickness 4 mm, number of excitations 2) and axial proton density (PD) fat-suppressed (TR/TE 4000/40, matrix size 288 × 256, field of view 18 cm, slice thickness 3 mm, number of excitations 2) sequences were used for measurements.
Evaluation of the images
The frequency of patellar malalignment, trochlear dysplasia, supratrochlear spurs, and patellar height were investigated in patients with patellar malalignment and those with a normal patellofemoral joint. We also studied the types of trochlear dysplasia based on the Dejour classification.
Patellar malalignment
Detecting patellar malalignment was performed using the qualitative method of Shellock et al. (1989), which is based on the relation between the mediolateral edges of the patella and the femoral trochlear mediolateral sides. In addition, patellar tilt was defined as the angulation between the posterior femoral condylar line and the largest diameter of the patella.
Sulcus angle and trochlear typing
Axial plane images >3 cm from the knee joint were used. The sulcus angle was measured from the highest lateral corner on the anterior surface to the deepest sulcus point and then to the highest medial corner. A trochlear angle of 137° ± 8° was accepted as normal (Fig. 1).
The Dejour classification was used to classify trochlear dysplasia. Dejour et al. (1990, 1994) classified trochlear dysplasia based on the trochlear angle and configuration. Dejour suggested the following morphological classification for trochlear dysplasia (Dejour et al. 1990).
Type A: sulcus angle >145° but with normal shape (Fig. 2)
Type B: flattened trochlear surface and a supratrochlear spur (Fig. 3a, b)
Type C: asymmetric trochlear surface; hypoplastic medial facet and convex lateral facet (Fig. 4)
Type D: humped shape; asymmetric trochlear surface with a supratrochlear spur (Fig. 5a)
The supratrochlear spur can be described as a ventral trochlear prominence (Pfirrmann et al. 2000). On a mid-sagittal image, the spur is seen as the distance between the anterior femoral cortical surface and the most prominent point of the trochlear surface (Fig. 5b). Measurements of >3 mm are accepted as indicative of a spur.
Patellar height
The Insall and Salvati method (Insall and Salvati 1971) was used to measure the patella alta and patella baja. On MR imaging, the patellar and patellar tendon lengths of 0.8 and 1.3, respectively are considered normal on mid-sagittal images. The values for patella alta and patella baja were >1.3 and <0.8, respectively.
Evaluation of the MPFLL/LPR ratio
Axial sections passing through the center of the patella were used to determine the MPFLL/LPR ratio. The MPFLL ligament was measured between the patellar insertion and the femoral adductor tubercle. The LPR was measured between the patellar insertion of the retinaculum and the lateral epicondyle of the femur (Fig. 6a–c). Both retinacula exhibited a wide, fan-shaped extension from the patellar insertion region and distributed laterally among the muscle planes. The thickest parts of the ligament at the femoral and patellar insertion points were used for the measurements. This part of the study was conducted as an inter-observer study, and two blinded radiologists calculated the MPFLL/LPR ratio separately.
Statistical analysis
A pilot study was first conducted as a power analysis. We predicted that we needed a minimum of 317 patients based on a 60 % frequency rate of trochlear dysplasia and 10 % margin of error, with an alpha error of 0.05 and a beta error of 0.05. The Shapiro–Wilk and single-sample Kolmogorov–Smirnov tests were used to test the normal distribution, and a histogram was drawn. Data are given as means and standard deviations; median, minimum, and maximum values; frequencies; and percentages based on their characteristics. Age and angle relations were tested using Spearman’s correlation test. Nominal variables were compared using the χ2 test with Yates correction and Fisher’s probability test. The odds ratio (OR) of trochlea types were obtained according to the “0” value.
For the MPFLL/LPR ratio, normality tests were conducted using the one-sample Kolmogorov–Smirnov test, histograms, box plots, and Q–Q (where Q = quantile) graphs. The correlation between the radiologists was evaluated using Pearson’s correlation test (for quantitative measurement values). Separate receiver operating characteristic (ROC) analyses were performed for the results of the radiologists. The comparison between the results according to the cutoff values found by the radiologists was assessed by the Z test. The concordance of these specialists with one another and with the gold standard was evaluated using the κ test. The specificity, sensitivity, positive predictive value (PPV), and negative predictive value (NPV) were also calculated for the two radiologists.
Non-parametric data were compared using the Mann–Whitney U test. The two-tailed significance level was adjusted to P < 0.05. All statistical analyses were conducted using NCSS10 software (www.ncss.com) and MedCalc 10.2 (medcalc.software.informer.com).