Reliability of current classification systems for periprosthetic distal femur fractures

Abstract

Introduction: This study aims to determine which Periprosthetic Distal Femur Fracture (PDFF) classification system is the most reliable. The secondary aim was to determine which classification system correlated most accurately with the surgical management recommended and delivered.

Methods: Between 2011 and 2019, 83 patients with 83 PDFFs that extended to the femoral component of a total knee arthroplasty (TKA) were retrospectively identified from a trauma database. Minimum follow-up was 1 year. Age, BMI, time from TKA, operative management, and Nottingham Hip Fracture Scores were collected, and AP and lateral radiographs used to classify all fractures using seven established classification systems by two observers blinded to management. In patients treated operatively (n = 69), preoperative radiographs were reviewed by two surgeons with expertise in trauma and knee revision who recommended fixation or distal femoral replacement (DFR) requirement.

Results: Mean age was 80.7 years (SD9.4) and 50 (84.7%) were female. PDFFs occurred at a mean 9.5 years (SD5.2) after primary TKA. Mean follow-up was 3.8 years (SD2.9). Management was fixation in 47, DFR in 22 and non-operative for 14. The Fakler classification demonstrated highest interobserver reliability (ICC=0.948), followed by the Rorabeck (ICC=0.903), UCS (ICC=0.850) and Chen (ICC=0.906). The Neer classification demonstrated weakest agreement (ICC=0.633). Overall accuracy of predicting DFR requirement (as determined by two experts) was highest for the Fakler system (83.9%). Compared with actual management delivered the Rorabeck system was most accurate (94.1%). Multivariate regression demonstrated that the ultimate need for DFR (n = 22) was independently associated with medial comminution (HR 2.66 (1.12-6.35 95%CI), p = 0.027) and fractures distal to the anterior flange and posterior condyle of the femoral component (HR 2.45 (1.13-5.31), p = 0.024).

Conclusion: The Fakler classification showed highest interobserver agreement and was most accurately predictive of the management recommended by two experts. No classification system accurately predicted the fractures that required DFR, and none included medial comminution which was independently associated with DFR requirement. There remains a need for a PDFF classification system that reliably guides operative management of PDFFs.

Keywords: Knee Replacement; Periprosthetic; Predictors; Prognosis; Total Knee arthroplasty.