Simulations of bone remodeling at the cellular scale

R.F.M. Oers, van

Onderzoeksoutput: ScriptieDissertatie 1 (Onderzoek TU/e / Promotie TU/e)

280 Downloads (Pure)


Bone is adapted to mechanical loading. This is most apparent in cancellous bone, which is found on the inside of our bones, and consists of a porous lattice of branching struts called 'trabeculae'. It has long been recognized that these trabeculae are aligned to the direction of mechanical loading. It also holds for cortical bone, which makes up the dense outer shell of our bones, and contains many tubular structures called 'osteons'. Like trabeculae, osteons are aligned to mechanical loads. Our bones are constantly remodeled by bone-resorbing osteoclasts and bone-forming osteoblasts. These cells frequently cooperate in so-called ‘basic multicellular units’ or BMU's. Osteoclasts dig a resorption cavity, which is then filled with new bone by osteoblasts. In cortical bone osteoclasts dig tunnels through solid bone, in cancellous bone they dig trenches across the trabecular surface. Osteoblasts fill these tunnels and trenches, creating osteons and hemi-osteons, respectively. How mechanical forces guide these cells is still uncertain. It is widely believed that bone remodeling is regulated by osteocytes. Osteocytes are osteoblasts that have become encased in the bone matrix during tissue formation. They are sensitive to mechanical stimuli within the bone tissue and able to send signals to the cells at the bone surface. More than a century ago, when Wolff noted the load-aligned organization of trabecular bone, his contempary Roux pioneered the thought that this architecture resulted from a remodeling process at the cell level based on local tissue strains. Only in the last decades, with the development of computer simulation models, did it become possible to test such hypotheses. Using computer simulations, Huiskes and coworkers demonstrated that local remodeling regulated by mechanosensory osteocytes could indeed produce load-aligned trabeculae. In this model bone formation was stimulated by a strain-induced osteocyte signal, but resorption occured randomly along the bone surface. The architecture adapted when the external loads were changed, aligning the trabeculae with the new loading directions. Reduced loads resulted in reduced trabecular thickness, connectivity and mass, as is seen in disuse osteoporosis. The resorption cavities, however, were no accurate representation of the trenches that osteoclasts dig on trabeculae. To simulate trabecular remodeling at the BMU level, and also to simulate the osteonal tunnels in cortical bone, a new representation of osteoclasts was needed. We therefore extended our model with a cell simulation method, to explicitly represent the osteoclasts. These model osteoclasts could tunnel through the bone, but a mechanism was needed to guide them in the loading direction. Smit and Burger had previously evaluated strains around a BMU resorption cavity and found that strains were concentrated at the lateral sides, away from the loading axis. If strain-induced osteocyte signals from these regions would repel osteoclasts, the osteoclasts would be forced to go in the loading direction. At the same time such signals would recruit osteoblasts to start bone formation. With the osteoclasts thus represented and guided, we were able to simulate cortical BMU's creating load-aligned osteons and cancellous BMU's moving across the surface of trabeculae, in both 2- and 3-dimensional simulations. In these BMU's resorption-formation coupling occurs in response to strains around resorption sites. Thus a simple regulation mechanism, in which strain-induced osteocyte signals inhibit osteoclasts and stimulate osteoblasts, could regulate the steering of and coupling within BMU's. Some of the most intriguing results were unforeseen. We found that the model produced wider osteons at a lower loading magnitude. This could explain the difference in osteon diameter that exists between the endosteal (inner) and periosteal (outer) side of the cortex, with generally wider osteons in the less-strained endosteal side. We also found that a steep gradient in loading magnitude could produce that 'drifted' towards the less-loaded side. In real bone such 'drifting' osteons generally drift towards the less-strained endosteal side of the cortex, a phenomenon that had not yet been explained. We also found that if a region of osteocyte death (therefore lacking osteoclast-repelling signals) was introduced near the path of the BMU, it would redirect its course to resorb this region. This may provide a mechanism for damage removal, because osteocyte death is associated with microdamage. The dead osteocyte region, however, was quite artificial, since no damage algorithms were included in the model. Therefore, we incorporated damage algorithms into the model, to better understand the implications of microdamage-targeted resorption. That remodeling will repair damage is far from trivial. Especially when loading is already intense enough to cause microdamage, adding resorption spaces might just make the problem worse. There are indications that so-called 'fatigue fractures' are preceded by 'runaway resorption'. In our simulations we found that remodeling tends to remove damage under a 'moderate' fatigue regimen, but it exacerbates damage under an 'intense' fatigue regimen. In this light it became of interest to review the effect of osteocyte death on bone formation. In our models we assumed strain-induced stimulatory signals from osteocyte to osteoblast, and osteocyte death would therefore cause a drop in bone formation. However, it was discovered in recent years that osteocytes inhibit bone formation via the protein sclerostin. This appears to be at odds with our stimulatory model, but if sclerostin secretion by the osteocyte decreases with strain, it could fulfill the same function. We used computer simulations to show that a sclerostin-based model is also able to produce a load-aligned trabecular architecture. The interesting difference appears when osteocytes are lost. Whereas formation drops in the stimulatory model, it gets a temporary boost in the inhibitory model, which prevents the loss of trabeculae. This has interesting implications, in particular for microdamage repair.
Originele taal-2Engels
KwalificatieDoctor in de Filosofie
Toekennende instantie
  • Department of Biomedical Engineering
  • Hilbers, Peter A.J. , Promotor
  • Huiskes, Rik, Promotor
  • van Rietbergen, Bert, Co-Promotor
Datum van toekenning9 jun 2010
Plaats van publicatieEindhoven
Gedrukte ISBN's978-90-386-2256-9
StatusGepubliceerd - 2010

Vingerafdruk Duik in de onderzoeksthema's van 'Simulations of bone remodeling at the cellular scale'. Samen vormen ze een unieke vingerafdruk.

Citeer dit