Fracture healing

Healing time is a difficult concept

If we imagine that there is a standard fracture of some kind and that there is a good way to measure the healing time; what would be the distribution of healing times?  Owing to biological diversity the healing time would be shorter for some and longer for others, but it is unlikely that it would have a normal distribution. Biological variables that are dependent on a number of normally distributed factors tend instead, for purely mathematical reasons, to have a lognormal distribution. This might mean, for example, that the fractures of 10% of the patients are healed within 6 weeks, 50% after 9 weeks, but for 90% to be healed we would have to wait as long as 15 weeks. The “most common” healing time is 9 weeks; however, healing could take considerably longer time purely by chance. This skewed distribution may mean that we all too easily misinterpret the situation as pathologically delayed healing and intervene unnecessarily.

Shaft fracture healing is a long chain of events

Most fractures occur in predominantly spongy bone, such as vertebral fractures and hip fractures. Nevertheless, nearly all the literature about fracture healing deals with the shaft of long bones. This may partly be because there have only been animal models for these types of fractures, partly because patients with long bone fractures are often younger and more appealing to work with.

Around a long bone fracture, a hematoma is formed at first, and this is quickly transformed into granulation tissue. An inflammatory reaction takes place in this granulation tissue that involves a variety of immune cells. Gradually, connective tissue cells start to dominate, and the hematoma becomes a clump of connective tissue, a callus. A large number of the processes that form an integral part of fetal bone development occur again here. However, these connective tissue type (mesenchymal) cells must first be recruited to the site. If the bone membrane (periosteum) has survived, it can contribute a great deal, as it contains dormant cells that are already disposed to form bone. In animal models at least, the mesenchymal stem cells that are recruited via blood play a large role, as do cells from surrounding muscle. How are these cells attracted to the fracture area? This is a key question. If we knew the answer, then we would be able to control fracture healing much better than we can now. However, it seems clear that the initial inflammation is a determining factor, probably mainly for recruitment of competent mesenchymal cells. If the inflammation is reduced, by giving NSAIDs or cortisone, then healing is markedly impaired.

In Uppsala during the 1930s it had already become clear that cells were enticed into the callus from outside and it was thought that substances released from the fracture surfaces were behind this. These theories gave rise to speculation about a “Bone Morphogenetic Protein”, BMP, which was then actually found to exist. It is thought that BMP, which is stored in bone, is exposed on the fracture surfaces. Mesenchymal cells that come into contact with BMP then produce this protein themselves so a chain reaction starts with the release of BMP in the whole of the callus. BMP has the ability to adapt the genetic machinery of the cells so that the cells differentiate to cartilage or bone instead of forming scar tissue, which they would otherwise have done. Mice whose bones lack BMP-2, cannot heal fractures.

There is much to suggest that whether a fracture will heal or not is decided during the inflammatory phase of healing, which lasts for one or two weeks after the injury.

The continued development of the callus depends to a large extent on the mechanical situation. In the parts where the deformation is small but the pressure variations large, the recruited cells will predominantly form cartilage. If the deformations are larger (involving shearing) then we instead get formation of woven bone. If the deformation is too large, then mostly scar tissue is formed. All these processes usually occur simultaneously in different parts of the callus. Even if it was known how to optimise the mechanical environment locally, this cannot be achieved in practice due to the, often complicated, geometry in the callus (bones seldom break cleanly), which means that the effects of loading will be beneficial in some, and detrimental in other parts.

The cartilage that has formed in parts of the callus is then replaced with bone by endochondral bone formation, which is a closely regulated process that occurs also during longitudinal bone growth. The bone in the callus is initially of low quality and strength, but the callus is large so it can anyway hold quite well. Now a phase of extensive remodelling begins, where the first generation of bone is replaced with higher quality lamellar bone. The large callus is no longer required and during remodelling it reduces in size so that the end result can be almost the same as the original bone, so long as the fracture has not healed crookedly.

The importance of the blood circulation for fracture healing is not clear. It has often been thought that increased blood circulation stimulates healing. Other data, however, indicate that the callus acquires the vascularisation that it needs, and that this is seldom a limiting factor, except in special cases where the anatomy prevents vascular invasion (fractures of the neck of femur or the navicular bone).

Healing of spongy bone may be a very different process

So far we have discussed the long bones. Sometimes in these bones, albeit rarely, healing does not occur, probably because the initial cell recruitment was too disturbed. However, metaphyseal fractures, primarily of spongy bone, almost always heal. Clearly there is no problem with the recruitment of competent cells for healing here. This is not surprising, since the bone marrow here already contains an abundance of various mesenchymal stem cells. If the main function of the inflammatory phase of healing is to attract competent cells, then it is not so important in fractures where these cells are already there. It would then be less harmful to inhibit this phase with, for example, NSAIDs. Experience and a number of animal experiments also suggest that this is the case.

Fracture healing can be influenced pharmacologically. The inhibiting effects of NSAIDs and cortisone have already been mentioned. At the beginning of the 21st century, there were high hopes for signalling molecules of the BMP family. These have evident beneficial effects in animal experiments, but in clinical studies the effects have been more difficult to demonstrate, whilst side effects have been relatively common. Another possibility is parathyroid hormone. Here, the results from animal experiments are also hopeful, as is some clinical data, but completely convincing studies are lacking.

During bone remodelling, the work of the osteoclasts and the osteoblasts is coordinated, “linked” within the remodelling unit, so that if the osteoclasts’ activity is inhibited then so is that of the osteoblasts. This does not apply however to fracture healing, at least not at the beginning. This means that osteoclast activity can be inhibited with bisphosphonates, while the osteoblasts tirelessly continue to form bone. Since both types of cells work simultaneously, this means that the bisphosphonates can indirectly increase the amount of bone formed. Once again, we know from animal experiments that this is how it works and there is also some data from humans. Theoretically, this beneficial effect should be seen mainly in the healing of spongy bone fractures. Convincing studies are lacking however, except in connection with implants, where there are a number of double blind randomised studies that show that bisphosphonates improve fixation of joint prostheses and other implants.