Background: Previous studies indicate that isolated posterior cruciate ligament injuries demonstrate magnetic resonance imaging (MRI) and clinical evidence of healing when treated nonoperatively; however, the authors are unaware of any other study that has looked at whether initial MRI can predict posterior cruciate ligament stability at the time of surgery in patients with knee dislocation.

Hypothesis: An MRI grading system will predict laxity on posterior drawer testing at the time of surgery in patients with knee dislocations.

Study Design: Cohort study (prognosis); Level of evidence, 2.

Methods: Forty-two consecutive patients with knee dislocation or multiple-ligament knee injury evaluated by MRI were included in the study. An assignment of grade 0 (intact), grade I (injured/fibers intact), grade II (partial tearing of ligament), or grade III (complete tear) was made after each reading on 2 separate occasions by 3 surgeons. Posterior laxity of the knee was graded by the magnitude of excursion on the posterior drawer test by the senior author at the time of surgery. Interobserver and intraobserver reliability of the MRI grading scheme expressed by the kappa statistic κ, as well as the predictive value of MRI grade in determining stability of the posterior cruciate ligament at the time of surgery, was assessed.

Results: The posterior cruciate ligament injury grading scheme tested demonstrated moderate to substantial intraobserver agreement (κ = 0.66, κ = 0.53, and κ = 0.52, respectively, for all raters). Interobserver reliability demonstrated only moderate agreement (κ = 0.49). If the grading scheme was changed to group both grades 0 and I (intact) and grades II and III (disrupted), intraobserver reliability demonstrated substantial to almost perfect agreement (κ = 0.83, κ = 0.80, and κ = 0.75), and interobserver reliability demonstrated substantial agreement (κ = 0.70). If the posterior cruciate ligament was classified as intact (grade 0 [intact] or grade I [injured]) on initial MRI, the injured knee was judged clinically stable (tibia anterior to or flush with the femoral condyles on posterior drawer testing) at the time of surgery 98.5% (95% confidence interval, 93%-100%) of the time. When the posterior cruciate ligament was classified as disrupted (grade II [partial tear] or grade III [complete tear]), the injured knee was judged unstable (tibia posterior to the femoral condyles on posterior drawer testing) 57.5% (95% confidence interval, 40%-73%) of the time.

Conclusion: The presented system of grading posterior cruciate ligament injury in patients with knee dislocation on initial MRI demonstrates moderate to substantial interobserver and intraobserver reliability that increases if the grading scheme is modified. An initial MRI scan read as grade I may predict stability to posterior drawer at the time of surgery. Even with MRI evidence of disruption in the posterior cruciate ligament (grade II and grade III injuries), posterior cruciate ligament reconstruction may not be clinically indicated at the time of reconstruction and/or repair of other associated injuries.