Post-operative fracture risk assessment following tumor curettage in the distal femur: a hybrid in vitro and in silico biomechanical approach

Distal femur is a frequent site for benign bone tumors, and fractures are common after tumor removal or cement augmentation. Current clinical criteria for assessing post-operative fracture risk (e.g., Mirels’ and Harrington’s systems) lack specificity. The lack of a quantitative approach to determine critical defect size leads to reliance on surgeon experience for cementation decisions. This study aims to develop and validate a biomechanical tool for accurate fracture risk assessment. Bone strength is a key fracture risk determinant. Finite element analysis (FEA) based on quantitative computed tomography (QCT) images offers a powerful tool for predicting fracture risk by considering bone geometry, density distribution, and defect size/location. Previous studies using QCT-based FEA have shown promise in predicting bone strength in other areas, but there is a need for a validated biomechanical study specifically addressing post-operative fracture risk in the distal femur's epiphyseal region, which is highly susceptible to primary bone tumors like giant cell tumors (GCTs). The optimal method for defect reconstruction (leaving the defect empty versus bone cement reconstruction) also remains controversial, underscoring the need for this study. This study addresses this gap by using a hybrid in vitro and in silico approach to determine critical defect sizes and identify vulnerable sites in the bone-cement interface.

Several studies have used FEA or rigidity analyses to improve fracture risk prediction and reconstruction method selection following tumor curettage. Li et al. compared stiffness and stress distribution in distal femora with and without cement and internal fixation. A biomechanical fracture risk criterion for tibial defects based on stress distribution was introduced using linear FEA on homogeneous models. A retrospective study on femoral metastases showed that prophylactic stabilization might be unnecessary in some cases identified by Mirels' criteria. A non-linear QCT-based FEA was used to predict pathologic fractures in proximal femora. While some studies have shown positive results for leaving defects empty or filled with bone cement, the decision remains controversial and often based on surgeon's experience. Most previous studies lacked accuracy due to neglecting non-linear effects, non-homogeneous material properties, or lacking experimental validation, highlighting the need for a more accurate and validated biomechanical approach.

Seven pairs of human cadaveric distal femora were used. In one femur from each pair, a defect was created to mimic tumor surgery. The cavity was filled with PMMA bone cement. The contralateral femur served as a control. Specimens were QCT scanned, and mechanical tests were performed using a compression load applied to the medial condyle. 3D FE models were generated from QCT data, assigning heterogeneous material properties based on density. A quad-linear constitutive model was used for bone elements. Homogeneous properties were assigned to the cement. Boundary conditions mimicked the in vitro setup. The FE models were validated using Keyak et al.'s approach. Models were divided into a tuning group (TG) and an evaluation group (EG). In TG, material properties were incrementally reduced until the mean prediction error was not statistically different from zero. The final material properties were then applied to EG. The accuracy and precision of the FE models were evaluated using a paired t-test and linear regression analysis. To investigate the effect of defect size, different defect sizes were incrementally created in a validated model. The fracture load was calculated and compared to that of the intact bone. One-sample t-tests were used to assess the significance of strength reduction. Larger defects were created in both medial and lateral compartments to investigate the effect of location. The mechanics of the bone-cement interface (BCI) were analyzed to identify vulnerable sites for failure.

The FE models accurately predicted bone strength, showing a strong linear relationship with in vitro results (R²=0.95). There was no significant difference between FE predicted and experimentally measured fracture loads (P=0.174). Bone strength reduction was insignificant until the defect size exceeded 65 cc (35% of epiphyseal volume). Beyond this critical size, a sharp reduction occurred (fracture load reduced to 70% of intact bone when defect volume reached 45% of epiphyseal volume). Medially located defects showed a greater reduction in strength compared to laterally located defects of the same size. When the defect exceeded 47% of the epiphyseal volume and invaded the contralateral condyle, a sharper reduction in bone strength was observed. Failure analysis of the bone-cement interface indicated that the proximal end of the cortical window and the most interior wall of the interface were the most vulnerable sites for reconstruction failure.

This study demonstrated the accuracy and precision of QCT-based non-linear FEA in predicting bone strength after tumor curettage, surpassing previous studies by considering bone non-linearity and heterogeneity. The identified critical defect size (35% of epiphyseal volume) aligns well with clinical observations. The finding that lateral defects are less risky than medial ones highlights the importance of defect location, a factor not considered in current clinical criteria. The identification of critical failure sites at the bone-cement interface provides crucial insights for optimizing implant selection and placement during cement augmentation. The results support the use of plates over intramedullary pins for cement augmentation.

This study successfully validated a QCT-based FEA approach for predicting bone strength and identifying critical failure sites in distal femur after tumor curettage. The findings provide valuable insights for developing a more comprehensive criterion for assessing post-operative fracture risk, incorporating defect size, location, and patient-specific factors. Future research should focus on developing an analytical criterion and investigating the effects of different loading conditions and implant types in more detail.

The study assumed isotropic bone material properties, potentially underestimating the influence of bone anisotropy. The tuning process used to account for systematic errors, including anisotropy, is phenomenological and doesn't precisely quantify the contribution of each error source. The effect of different loading conditions (torsional and bending moments) was not investigated. The effect of defect shape variation on strength reduction was not systematically studied. The evaluation of implant efficacy was based solely on bone-cement interface analysis and did not include implant models in the FEA.