Diagnostic value of the lateral femoral notch sign and kissing contusion in patients with anterior cruciate ligament injuries: a case-control study.

Abstract

The lateral femoral notch sign (LFNS) and the kissing contusion (KC) are two indirect signs of anterior cruciate ligament (ACL) injuries. They can be used to diagnose ACL injuries. A total of 1000 patients were enrolled in this study, including 500 patients with ACL injuries who assigned to experimental group and 500 patients with meniscal tear (MT) who allocated to control group. All the patients underwent magnetic resonance imaging (MRI) preoperatively, and the diagnosis was confirmed with the aid of arthroscopy. The depth of LFNS and the presence of KC were determined on MRI findings. The relationship and characteristics between these two indicators was explored. The notch depth of lateral femoral condyle in the experimental group (0.99 ± 0.56mm) was significantly greater than that in the control group (0.49 ± 0.28mm) (P < 0.05). The positive rate of KC in the experimental group (183/500) was markedly higher than that in the control group (3/500) (P < 0.05). The values of notch depth in patients who had ACL rupture concomitant lateral MT injuries and medial collateral ligament (MCL) injuries were 1.12 ± 0.64 and 1.23 ± 0.74mm, respectively, which were significantly higher than those in patients with only ACL injury (0.89 ± 0.49mm) (P < 0.05). It also was revealed that when the optimal cut-off point of LFNS was 0.72mm (area under the curve (AUC) = 81%), the values of specificity and sensitivity were 67% and 84%, respectively. For KC, the corresponding values were 36.6% and 99.4%, respectively. The diagnostic outcome of LFNS was not in agreement with that of KC, as there was a poor coincidence according to the Kappa coefficient (Kappa = 0.155 < 0.4, P = 0.035). The LFNS and KC have strong clinical significance in the diagnosis of ACL injuries. A deeper notch often indicates a more complex knee injury. Notch depth equal to 0.72mm can be basically considered as the optimal cut-off point for LFNS in statistics.