Knee torsion predicts the preoperative J-sign grade in patients with recurrent patellar dislocation : a prospective study.

Abstract

Aims The aim of this study was to determine the values for knee torsion in healthy knees, and explore predictive thresholds of torsion in grades of the J-sign in patients with recurrent dislocation of the patella. Methods A five-year prospective study was conducted involving healthy volunteers and patients with recurrent dislocation of the patella. The knees of the patients were categorized into four groups (J-sign -, J-sign 1+, J-sign 2+, and J-sign 3+) based on visual examination of the patellar maltracking. The torsion of the knee and other bony abnormalities were obtained from routine hip-knee-ankle CT scans. The ranges of the thresholds were determined from the 95% CI of the mean for each group. The predictive thresholds for each grade of J-sign were identified from receiver operating characteristic (ROC) and predictive probability curves. Results The study included 44 knees (in 22 healthy volunteers) and 255 knees in 234 patients with recurrent dislocation of the patella. In the J-sign -/1+, J-sign 2+, and J-sign 3+ groups, there were 87 (29.1%), 85 (28.4%), and 83 (27.8%) knees, respectively. Age, sex, and laterality were comparable among the groups. There were significant differences in the mean torsion between the groups: healthy knees 1.4° (-8.8° to 5.4°); J-sign 1+ 6.9° (SD 4.3°); J-sign 2+ 9.2° (SD 4.8°) and J-sign 3+ knees 13.9° (SD 5.4°) (p < 0.001). The ranges of the thresholds for the four groups based on the 95% CI of the means were: < 5.5°, 7.2° to 8.1°, and 10.2° to 12.8°, respectively. Ordered regression revealed a 25.4% increased probability of advancing to a higher J-sign grade for every 1° increase in torsion (adjusted odds ratio (OR) 1.254 (95% CI 1.182 to 1.331); p < 0.001). It was shown in the ROC curves that the presence of torsion effectively differentiated between normal and positive J-sign knees (area under curve (AUC) 0.92; sensitivity 0.76; specificity 1; cut-off 5.5°). Analysis of the ROC and predictive probability curves identified threshold values for the three abnormal J-signs at 7° and 12°, showing an excellent discriminatory performance (pall post-hoc < 0.01). Conclusion A threshold of 5° to 6° of torsion in the knee is associated with patellar maltracking in extension. In patients with recurrent dislocation of the patella, there is a one-grade increase in J-sign for approximately every 4° increase in torsion. The 7° and 12° of torsion represent the thresholds for the preoperative J-sign -/1+, J-sign 2+, and J sign 3+, respectively. Routine bilateral lower limb CT scans are therefore required if surgery is to be considered in the knees of patients with recurrent dislocation of the patella who have a positive J-sign.