gms | German Medical Science

Objectives: Optimal fracture clinical care is based on accurate assessment of bone healing progress. However, there are no objective measures currently being used in the clinics. New techniques have been developed to measure structural bone healing: virtual biomechanical testing and in vivo implantable sensors. This study aimed to correlate continuous data from an implantable sensor assessing healing status through implant load monitoring with computer tomography (CT) based longitudinal finite element (FE) simulations in a large animal model.

Methods: This study used a previous preclinical study of a tibial osteotomy with gap sized ranging from 0.6 to 30 mm. Each specimen had in vivo implanted strain-based sensor data and CT scans for eight sheep. Sensor signal was measured continuously, and CT scans were performed monthly. FE models were derived from the CT scans at each timepoint in a specimen-specific manner with heterogeneous material properties for the bone and fracture callus. Implants were modelled with metal properties. Two types of FE-based virtual mechanical tests were performed (Figure 1B-C [Fig. 1]): I) virtual torsional rigidity (VTR) test with 1° rotation applied on bone and callus without the implant and II) axial bending of the bone-implant construct and sensor with 500 N axial load measuring strain value at the sensor location and calculating relative implant load (RIL) relative to the max value as in the in vivo sensor data. A time point of healing was defined as a signal drop below 21% of the max value with a <5% change versus the previous month sustained for 6 weeks.

Results and conclusion: The diagnostic healing simulations agreed closely with the in vivo sensor data for all eight specimens. Six animals healed in the observed time and two had a delayed- or non-union (Figure 1E [Fig. 1]). Linear regression analysis showed strong correlations with the in vivo data for both FE-based VTR (coefficient of determination: R² = 0.70) and RIL (R² = 0.80).

This study highlighted the potential of both virtual biomechanical testing and in vivo sensor as methods to address the lack of objective fracture healing assessment by cross validating subject-specific FE models to in vivo sensors for the first time. The results showed that FE diagnostic simulations were strongly correlated to in vivo sensor data, and both predicted similar healing times. This study demonstrates consistency between these two emerging technologies as promising clinically applicable methods in objectifying the assessment of healing and allow for better diagnosis and early reaction to complications.