Virtual mechanical tests out‐perform morphometric measures for assessment of mechanical stability of fracture healing in vivo

Abstract

Finite element analysis with models derived from computed tomography (CT) scans is potentially powerful as a translational research tool because it can achieve what animal studies and cadaver biomechanics cannot-low-risk, noninvasive, objective assessment of outcomes in living humans who have actually experienced the injury, or treatment being studied. The purpose of this study was to assess the validity of CT-based virtual mechanical testing with respect to physical biomechanical tests in a large animal model. Three different tibial osteotomy models were performed on 44 sheep. Data from 33 operated limbs and 20 intact limbs was retrospectively analyzed. Radiographic union scoring was performed on the operated limbs and physical torsional tests were performed on all limbs. Morphometric measures and finite element models were developed from CT scans and virtual torsional tests were performed to assess healing with four material assignment techniques. In correlation analysis, morphometric measures and radiographic scores were unreliable predictors of biomechanical rigidity, while the virtual torsion test results were strongly and significantly correlated with measured biomechanical test data, with high absolute agreement. Overall, the results validated the use of virtual mechanical testing as a reliable in vivo assessment of structural bone healing. This method is readily translatable to clinical evaluation for noninvasive assessment of the healing progress of fractures with minimal risk. Clinical significance: virtual mechanical testing can be used to reliably and noninvasively assess the rigidity of a healing fracture using clinical-resolution CT scans and that this measure is superior to morphometric and radiographic measures.

Biomechanical duality of fracture healing captured using virtual modeling and validated in ovine bones

“…As previously reported, the virtual mechanical tests with single-zone material assignment reliably recapitulated the physical mechanical tests for the intact ovine tibiae, with a strong correlation and good absolute agreement (Fig. 3A; R 2 = 0.697, RMSE = 0.106) (12). Applying the singlezone material model in the operated limbs also produced a strong correlation, but with VTR clearly over-predicting GJ (Fig.…”

“…The use of linear elastic models for bone is widespread common practice in finite-element modeling of bone, but the soft tissues present in early fracture healing can have viscoelastic or poroelastic behavior (34)(35)(36)(37). The mechanical testing data for this study used a constant loading rate (5°/min) and produced linear torque-angle response curves (12). This observation lends support to the theory that the elastic bone contribution within the callus dominates the overall structural mechanics, at least at this stage of healing.…”

“…5A). Detailed animal information for these studies has been previously reported (12). In brief, there were a total of 33 animals that had both postmortem biomechanical testing data and CT scans available for analysis.…”

Biomechanical duality of fracture healing captured using virtual mechanical testing and validated in ovine bones

“…First, we developed a methodology for processing computed tomography (CT) scans to develop subject-specific 3D finite element models of healing bones 10 , 11 . In a recent ovine study, we demonstrated that a virtual torsion test outperforms subjective methods like radiographic scoring for predicting the progress of healing and that it is a reliable surrogate for postmortem physical torsion testing in intact tibiae 12 . In clinical pilot testing, out virtual torsion tests have been able to detect clinically significant delayed fracture healing in patients with comorbidities such as smoking and diabetes 13 , 14 .…”

“…This naturally adaptive structure develops in response to biological and mechanical cues and is comprised of hard callus (woven bone, newly mineralized or approaching the density of the intact cortex) and soft callus (fibrous and/or cartilaginous interstitial tissues) 19 , 20 . Our previous ovine study found that simply extending a density-modulus scaling law developed for cortical bone to include regions of callus produced strong correlations between virtual and physical biomechanical tests, but that the virtual tests over-predicted the measured torsional rigidity in osteotomized specimens by an average of 58% 12 . This over-prediction of rigidity when callus is present suggests that a constitutive material model derived for cortical bone does not completely capture the mechanical behavior of callus and that more work is needed to virtually replicate the in vivo biomechanics of healing fractures.…”