Graft Resorption and Remodeling

Graft resorption and remodeling is a normal but complex biologic process. Bone healing occurs by a series of sequential steps that involve an inflammatory phase, with the arrival of inflammatory cells and bone progenitor cells accompanying vascular ingrowth into the graft. This phase is followed by a repair phase in which osteoclasts begin to absorb the graft, while osteoblasts lay down osteoid, and mineralization of the osteoid follows. This repair phase begins the process of new bone formation. The process continues into the remodeling phase as the bone is remodeled into new living mature bone, and necrotic bone is removed by creeping substitution.

Various humoral factors, proteins, growth factors, and mechanical forces mediate this process. Humoral factors include parathyroid hormone, vitamin D, and calcitonin. Proteins and growth factors include a large array of substances such as multiple bone morphogenetic proteins, insulin-like growth factor, transforming growth factor, platelet-derived growth factor, fibroblast growth factor, and nectins.

Mechanical forces are also crucial in this process. As defined by Wolff’s law, bone remodeling (and bone strength) is determined by the load placed on it. This structural adaptation results in bone being formed where stresses engendered by compressive loading or tensile forces occur, and bone is reabsorbed where the stresses do not occur. Bone or bone graft placed in compression is exposed to “bone healing”–enhancing forces as defined by Wolff’s law.

As a result of all of the aforementioned processes, bone grafts first partially resorb before being replaced by new living bone. The process results in subsidence of the bone graft after surgery. This subsidence is part of the normal biology of bone healing and not a pathologic process. It often is not appreciated in noninstrumented constructs. However, in studies that specifically measure construct height, subsidence is routinely observed.^4–6 The amount of subsidence varies with the type of graft used and number of levels fused. Bishop et al.⁴ showed that in uninstrumented ventral cervical discectomies that used iliac crest autograft, the average settling was 1.4 mm for a single-level construct and 1.8 mm for two-level procedures. The amount of settling increased to 2.4 mm for single levels and 3.0 mm for two levels when iliac crest allograft was used.

Graft Collapse

A cervical interbody graft can collapse before its incorporation, also resulting in subsidence. Graft collapse occurs for numerous reasons. If the graft is inadequately sized to handle the loads placed on it, it may collapse. Collapse can occur if the graft is too narrow in width or depth with respect to the adjacent vertebrae. This size mismatch increases the load placed on a graft. A larger graft spreads the load over a larger area, reducing the load per unit of surface area, and the graft may be able to withstand axial loads more effectively. Finally, if there is a good match between the contours of the surface of the vertebrae and the graft (so-called gapless fusion), the weight-bearing forces are evenly distributed over the region of contact, diminishing the chance of graft collapse.

The choice of graft material is also important. The “gold standard” is bone because living structural bone can respond to stresses placed on it and strengthen itself via the aforementioned process and repair itself as needed. Synthetics and allograft bone do not have these properties. Allograft bone preparations are not equal. The techniques of bone handling and processing can affect the structural integrity and strength of bone and its suitability as a grafting material.

Allogeneic bone used for spine surgery must be sterile. The bone can be sterilely harvested. If sterility is maintained, it may not require further processing. If it is not culture negative or is not sterilely harvested and processed, it may be sterilized with gamma irradiation. Low-dose radiation (<1.5–2 megarads) has been shown to diminish graft strength nonsignificantly. However, high-dose radiation (≤4–5 megarads) used by some laboratories causes significant weakening of the bone. Such grafts have a high collapse rate and should be avoided.

The type of bone—cortical or cancellous—is also important. Cortical bone is significantly stronger than cancellous bone, but its density resists vascular ingrowth and the influx of osteoblasts. It is slower to become living bone via true incorporation. It is neither osteoinductive nor osteogenic. As such, osteoclastic activity predominates; this results in progressive weakening of grafts that are primarily of cortical consistency. Cortical bone provides significant early structural support compared with cancellous bone. Because of the aforementioned properties, true bony incorporation is slow, and the weakening of the graft may result in graft collapse or failure before the acquisition of solid fusion. Pseudarthrosis may also result.

Cancellous bone lacks the strength of cortical bone and alone is likely inadequate to support clinical loads. However, cancellous bone has many favorable properties, including significant osteoconductive activity and some osteoinductive capacity. Vascular ingrowth into its loose architecture readily occurs. Cancellous bone incorporates early and more completely. Instrumentation may be used to provide structural support until the cancellous bone fully incorporates and a solid arthrodesis is attained.

An optimal bone graft is one that has the structural integrity of cortical bone and the osteoconductive and osteoinductive properties of cancellous bone. The iliac crest has long been used as a graft source for cervical spine surgery and has both characteristics of an ideal bone graft. It seems to be an excellent choice because it contains a cortical shell that provides structural support and a cancellous core that incorporates quickly. If its size is chosen well, the axial load is shared by the cortical shells of the vertebrae by passing the load onto the cortical walls of the graft. Pistoning into the adjacent vertebrae can be minimized (see subsequent section).

Pistoning (Subsidence)

Pistoning refers to failure of the end plate of the vertebrae with impaction of the graft into the vertebral body above or below the graft. Factors that influence pistoning include graft density, graft/donor size mismatches, and donor site preparation. The density of the bone graft should be ideally matched to the vertebrae. Predominantly cortical implants such as fibula result in a significant graft-to-host density mismatch. Fibula (predominantly cortical graft) placed in an osteoporotic spine results in a significant graft/host density mismatch. This mismatch may result in pistoning of the graft into the vertebra if the stresses placed on the end plates exceed their load tolerance.

The weight or load on the graft is the same whether a small or a large graft is used. Greater forces per unit area of contact are applied to a small bone graft because the forces cannot be dispersed over a larger surface area. This load/surface area mismatch can result in end-plate failure and pistoning. With a larger graft, the load is dispersed over a large surface area, reducing the chance of graft collapse or pistoning. End-plate preparation can also influence this mismatch. Various types of end-plate preparation and various techniques for graft fitting have been described. The goals of end-plate preparation must include removal of the cartilaginous end plate to allow bone graft incorporation and the shaping of the end plate to maximize contact with the graft. Beyond these two considerations, considerable variation in techniques exists.

Some surgeons have attempted to devise methods of interlocking the grafts and vertebrae when fixation is not used. Other surgeons advocate mortising the graft into the vertebrae, that is, countersinking the graft into the cancellous region of the vertebrae beyond the end plates. This approach minimizes the chance of graft migration or expulsion but at the expense of increasing the chance of subsidence. Other surgeons perforate the end plate in various patterns to encourage vascular ingrowth. However, extensive violation of the end plate, especially when used with dense cortical bone, is likely to increase pistoning. Lim et al.⁷ showed that it is important to preserve as much end plate as possible to prevent graft subsidence into the vertebral body. They found that making one central hole to increase vascular access to the graft rather than multiple smaller holes reduced stresses on the end plate.

First-Generation Systems

The need to augment cervical spine stability after spine surgery, trauma, or other destructive pathologic processes led to the development of ventral cervical plating systems. The first generation of these devices included Orozco plates (Synthes, West Chester, PA), which were used primarily in Europe, and Caspar plates (Aesculap, Tuttlingen, Germany), which achieved significant acceptance worldwide. It became evident that such devices offered advantages to cervical spine surgeons and patients.

The use of such plates helps preserve or restore lordosis, reduces graft extrusion, and improves graft union rates. This use also facilitates the performance of more extensive procedures when indicated by the disease process.^8–13 Ventral cervical plating allows the surgeon to use allograft more liberally instead of autograft. Allograft success rates have been shown to equal success rates of autograft when ventral plating was used.¹⁴ Use of allograft helped avoid donor site complications, which can be a significant source of patient morbidity (i.e., pain at the graft harvest site).

The initial ventral cervical implants were not perfect, and implant-related complications were observed. The initial Caspar plating system used parallel slots for the screws, which were not locked or constrained to the plate. This was a truly axially dynamic implant. The implant required bicortical screw placement for optimal stability. Despite bicortical placement, screw backout was observed sometimes. The dynamic nature of the implant permitted axial subsidence. However, this subsidence was not recognized at the time as part of the normal biology of bone healing (as already discussed) and was thought to be undesirable. The plate was modified so that holes replaced one half of the slots (Fig. 145-1). Rather than preventing subsidence, this modification led to screw breakage. The screw breakage virtually always involved the screws placed in the holes, which were subjected to excessive bending moments (Fig. 145-2). Fixed in the cortex dorsally and the hole in the plate ventrally, the screws typically fractured in the middle at the point of both the maximum bending moment application and the maximum stress application (stress = bending moment/strength). The screws placed into the slots were able to toggle or slide along the slots and did not fracture.

FIGURE 145-1 Original Caspar plate (Aesculap) with bilateral slots (A), which was modified to have slots alternating with screw holes (B) after settling was observed and thought to be undesirable.

FIGURE 145-2 Caspar plate as initially placed (A) and with settling (B and C). Solid fusion occurred, but the screw in the holes (white arrows) fractured as settling occurred, as indicated by the plate overlapping the interspace (black arrow).

Second-Generation Systems

In an attempt to prevent the aforementioned complications (presumed pathologic settling, screw backout, and screw fracture) and to avoid the need for bicortical screw purchase, a second generation of cervical fixation systems emerged. These devices featured screws fixed to the implant and permitted screw convergence on placement. The cervical spine locking plate (CSLP) from Synthes was the first such device. The ability to “toe-in” the screws secured the plate in part by triangulation and acted to reduce the chance of screw pull-out (Fig. 145-3). These devices were initially successful, but failures caused by plate or screw fracture, construct pull-out, and delayed union or nonunion were increasingly recognized. Attempts were made to remedy this situation by increasing the strength of the plate; examples are the Orion plate from Sofamor-Danek and the Codman plate (Fig. 145-4). These more rigid implants reduced hardware fracturing but may have increased the incidence of delayed union and pseudarthrosis caused by stress shielding (see later discussion).

FIGURE 145-3 CSLP (Synthes). Triangulation of the locked unicortical screws was used to increase pull-out strength.

FIGURE 145-4 Orion plate (Sofamor-Danek) (A) and Codman plate (B) were made stronger to try to reduce hardware failures.

Despite their shortcomings, second-generation plating systems have achieved better results than noninstrumented fusions, and they have become quite popular. Multiple variations on this basic theme have been developed as other manufacturers entered the market. Despite the popularity and ease of use of these systems, failures continued to occur. At the time (early 1990s), the spine community did not fully appreciate the reason for these nonunions and failures. Because these were often late sequelae, many were not likely recognized or at least were underreported. Figure 145-5