The winking sign is an indicator for increased femorotibial rotation in patients with recurrent patellar instability

PURPOSE: Rotation of the tibia relative to the femur was recently identified as a contributing risk factor for patellar instability, and correlated with its severity. The hypothesis was that in patellofemoral dysplastic knees, an increase in femorotibial rotation can be reliably detected on anteroposterior (AP) radiographs by an overlap of the lateral femoral condyle over the lateral tibial eminence. METHODS: Sixty patients (77 knees) received low-dose computed tomography (CT) of the lower extremity for assessment of torsional malalignment due to recurrent patellofemoral instability. Three-dimensional (3D) surface models were created to assess femorotibial rotation and its relationship to other morphologic risk factors of patellofemoral instability. On weight-bearing AP knee radiographs, a femoral condyle/lateral tibial eminence superimposition was defined as a positive winking sign. Using digitally reconstructed radiographs of the 3D models, susceptibility of the winking sign to vertical/horizontal AP knee radiograph malrotation was investigated. RESULTS: A positive winking sign was present in 30/77 knees (39.0%) and indicated a 6.3 ± 1.4° increase in femorotibial rotation (p < 0.001). Femoral condyle/tibial eminence superimposition of 1.9 mm detected an increased femorotibial rotation (> 15°) with 43% sensitivity and 90% specificity (AUC = 0.72; p = 0.002). A positive winking sign (with 2 mm overlap) disappeared in case of a 10° horizontally or 15° vertically malrotated radiograph, whereas a 4 mm overlap did not disappear at all, regardless of the quality of the radiograph. In absence of a winking sign, on the other hand, no superimposition resulted within 20° of vertical/horizontal image malrotation. Femorotibial rotation was positively correlated to TT–TG (R(2) = 0.40, p = 0.001) and patellar tilt (R(2) = 0.30, p = 0.001). CONCLUSIONS: The winking sign reliably indicates an increased femorotibial rotation on a weight-bearing AP knee radiograph and could prove useful for day-by-day clinical work. Future research needs to investigate whether femorotibial rotation is not only a prognostic factor but a potential surgical target in patients with patellofemoral disorders. LEVEL OF EVIDENCE: III.