Bones are subjected to a variety of loads during a lifetime. Already in the 19th century it was suggested that the microstructure of bones is adapted to these loads and that this is achieved by continuous modelling and remodelling in which cells add tissue where needed and resorb it where not needed. This process ultimately leads to a mechanically optimised structure with a uniform tissue loading distribution and is nowadays commonly accepted and referred to as ‘Wolff’s law’. Assuming that bone strives for a uniform tissue loading, we here hypothesise that it might be possible to derive a bone’s loading history from its microstructure by finding a set of external forces that results in the most uniform tissue loading. This would be the inverse of ‘Wolff’s law’. If deriving loading histories from bone microstructure would be indeed possible, this would enable estimations of bone loading where it is difficult or impossible to measure them in any other way, such as for fossil bones or bones in vivo. The former would enable, for example, making inferences about the locomotion of extinct animals from the derived loading histories. The latter would enable to determine in vivo loading conditions that are required for patient-specific bone remodelling studies. In such remodelling studies, load-driven bone remodelling simulations could be used to study the effects of diseases and treatments on the bone microstructure of patients. However, in vivo loading is usually not known and it has not yet been shown that load-driven remodelling can be assumed in human bones. In my doctoral thesis, I therefore explored the concept of deriving loading histories from bone microstructure and investigated the clinically important application of estimating realistic in vivo loading conditions for patient-specific bone remodelling simulations.

To do so, a novel approach to estimate bone loading histories from bone microstructure was first developed. The method can be summarized in three steps:​ First, unit load cases are defined and the resultant tissue loading is calculated using micro-finite element analysis. Second, the predefined load cases are scaled until the most uniform tissue loading is achieved when the results of each load case are added. Third, the final loading history is determined from the scaling factors of the unit load cases.

Using this approach, validation and feasibility studies were carried out to investigate the accuracy and robustness of the algorithm. In a first study, compressive forces that were applied to murine caudal vertebrae during an animal experiment were estimated from the developed bone microstructures measured in vivo by micro-CT. The algorithm successfully derived the magnitude of this simple load case. In order to further validate the load estimation algorithm for more complex loading situations and to control the state of mechanical adaptation, we conducted a study in which the loading history from a set of synthetic bone structures was derived. These structures were generated by performing bone remodelling simulations where the loading history as well as the degree of mechanical adaptation can be controlled exactly. It was found that differences between the estimates based on the adapted structures and the actually applied loading in the simulations were less than 4.4%. To challenge the load estimation algorithm even more and to validate it with human bones, forces working in the human distal radius were estimated based on high-resolution peripheral CT images and compared to literature data in a third study. It was found that these estimates well compared to values reported in experimental studies. Finally, as a proof-of-concept of this approach to decipher realistic more complex loading patterns as they occur at the hip joint, forces acting at the human and canine femoral head were estimated from the bone microstructure as obtained from micro-CT scans. Here also, the estimated forces were in good agreement with direct in vivo measurements reported in the literature and reflected the loading conditions during walking and thus a realistic loading pattern.

To investigate the clinically important application of the load estimation algorithm – to determine in vivo loading conditions needed for patient-specific bone remodelling studies – two more studies were performed. In a first study, the load estimation algorithm was used to derive patient-specific loading conditions from bone biopsies based on which bone remodelling simulations of hypoparathyroidism were performed. It was found that the changes in the bone structure predicted by the simulation compared successfully to those seen in the patients. In a second study, bone loading was estimated from high-resolution in vivo peripheral CT images of the human distal tibia. Additionally, bone remodelling sites were quantified by comparing baseline and follow-up scans. It was found that the local bone tissue loading conditions were good predictors for bone loss and gain as quantified from the images. This result thus confirms the assumption of load-driven bone remodelling. Furthermore, we developed a prototype for clinical bone remodelling simulations. Healthy remodelling was simulated based on high-resolution in vivo peripheral CT images of the distal radius and resultant microstructures were compared to follow-up measurements. Morphometric parameters of baseline did not differ much from 6- months follow-up and simulated bone microstructures, indicating a realistic prediction of bone remodelling. It also confirms the estimation of realistic in vivo loading conditions for such simulations because otherwise the bone microstructure would have changed due to mechanical adaptation.

These results are in agreement with our hypothesis that there is a strong correlation between bone microstructure and functional use and that this form-function relationship allows deriving loading histories from bone microstructure, which can be used for patient-specific bone remodelling simulations. We conclude that the results obtained so far support the idea that deciphering the secret message within bone microstructure is possible.