Histologic Components

On a gross level, all bones are composed of two basic components: cortical (compact) bone and cancellous (trabecular) bone. Cortical bone is a dense, solid mass, except for its microscopic channels, and contains parallel stacks of curved sheets called lamellae, which are separated by bands of interlamellar cement. Regularly spaced throughout the lamellae are small cavities, or lacunae. Lacunae are interconnected by thin, tubular channels called canaliculi. Entrapped bone cells (osteocytes) are located in the lacunae, and their long, cytoplasmic processes occupy canaliculi. The cell processes within canaliculi communicate by gap junctions, with processes of osteocytes lying in adjacent lacunae. Canaliculi open to extracellular fluid at bone surfaces, thus forming an anastomosing network for the nutrition and metabolic activity of the osteocytes. Cortical bone possesses a volume fraction of pores less than 30% and has an apparent density of up to about 2 g/mL. Its compressive strength is approximately 10-fold that for a similar volume of cancellous bone.

Cancellous bone is porous and appears as a lattice of rods, plates, and arches individually known as trabeculae. It has a greater surface area and can be readily influenced by adjacent bone marrow cells. Because of this structural difference, cancellous bone has a higher metabolic activity and responds more readily to changes in mechanical loads.⁹

Cortical and cancellous bone may consist of woven (primary) or lamellar (secondary) bone. Woven bone forms the embryonic skeleton and is then resorbed and replaced by mature bone as the skeleton develops.¹⁰ In the adult, woven bone is found only in pathologic conditions, such as fracture healing and in tumors. Woven and lamellar bones differ in formation, composition, organization, and mechanical properties. Woven bone has an irregular pattern of collagen fibers, contains approximately four times as many osteocytes per unit volume, and has a rapid rate of deposition and turnover. The osteocytes of woven bone vary in orientation, and the mineralization of woven bone follows an irregular pattern in which mineral deposits vary in size and in their relationship to collagen fibrils. In contrast, the osteocytes of lamellar bone are relatively uniform, with their principle axis oriented parallel to that of other cells and to the collagen fibrils of the matrix. The collagen fibrils of lamellar bone lie in tightly organized, parallel sheets, with uniform distribution of mineral within the matrix.^11,12

The irregular structure of woven bone makes it more flexible, more easily deformed, and weaker than lamellar bone.⁹ For these reasons the restoration of normal mechanical properties to bone tissue at the site of a healing fracture requires eventual replacement of the woven bone of the fracture callus with mature lamellar bone.¹¹

Biomechanical Properties of Graft Material

In vivo, the mechanical performance of a bone graft is a function of the intrinsic property of the graft and the properties of the graft-host interface.¹³ Intrinsic properties of a graft are related to its geometry and composition and include its fracture toughness, yield strength, and elastic modulus.⁸

In a clinical setting, where the graft has geometric and mechanical properties similar to the host bone, it may function almost immediately.¹⁴ Nevertheless, in the case of inferior bone graft mechanical properties, the construct should be designed with additional graft material or incorporate internal fixation until remodeling occurs and the graft can provide adequate load-bearing function.¹⁴ A graft’s load-bearing capacity is achieved after complete biologic incorporation by the host, which is related to the mechanical and biologic properties of the graft-host interface.

Iliac crest wedges are the most commonly used graft material. The percentages of cortical and cancellous bone remain constant at 41% and 59%, respectively, regardless of the total cross-sectional area of the wedge. Donor age also does not affect this physical parameter.¹⁵

To reduce the immune response and also as methods of preservation and sterilization, allografts undergo certain modifications. These modifications have a profound effect on the biomechanical properties of the graft. Freezing has minimal effects compared with freeze-drying, which significantly reduces both the yield strength and stiffness of the bone graft.¹⁶ Autoclaving produces a dose-dependent decrease in strength and stiffness.¹⁷ The relationship between gamma radiation and mechanical properties has yet to be established at doses between 0 and 25 kGy (standard dose). But it becomes dose-dependent at 25 kGy for cortical bone or 60 kGy for cancellous bone.¹⁸ Complete demineralization of the bone graft results in loss of almost all of its mechanical properties. Comparison testing of various graft materials shows allograft or fresh-frozen cancellous bone to be the weakest, failing at 863 N of compression. Air-dried, ethylene oxide-sterilized, tricortical bone failed at an average load of 2308 N, and fresh-frozen, tricortical allograft bone failed at an average load of 2257 N. Rehydrated iliac crest wedges are more deformable than freeze-dried wedges.¹⁹ During loading, freeze-dried wedges fail dramatically, fracturing into many small pieces; this occurrence is secondary to its brittle nature. Rehydrated wedges fail with a circumferential fracture along the side of the wedge where the cortical bone is thinnest. It has been recommended that freeze-dried wedges be rehydrated in a vacuum before clinical use.¹⁹ When water or saline is added to the vacuum-sealed container holding the wedge, the wedges gain 100% of their wet weight within 5 minutes of addition of the fluid. Graft collapse occurred more frequently with freeze-dried allografts (30%) than with autografts in anterior cervical fusions.

The loads at the lumbar spine have been well documented in various positions and levels of activity.²⁰ Either autograft or allograft iliac crest wedges are biomechanically sound in an interbody fusion of the lumbar spine, since such fusions would provide load-bearing capacities approximately fourfold greater than would be applied in vivo. Specimens from the anterosuperior iliac spine could bear substantially greater axial loads (average 3230 N) compared with specimens from the posterosuperior iliac spine (average 1458 N).²¹ Fibular strut grafts are the strongest and have been shown to have a compressive strength of 5070 N.²² However, their cross-sectional area, which is important in preventing telescoping of the graft, is much smaller. In interbody fusion, the cross-sectional area of the graft should be substantially greater than 30% of the end plate to provide a margin of safety.²³

Incorporation of Bone Graft

Bone graft incorporation is a prolonged process with a sequence of complicated steps involving the interrelationship of the graft and host. This ultimately leads to the envelopment of a complex of necrotic old bone with viable new bone.²⁴ The complex develops through resorption of the necrotic old bone with viable new bone being laid down. The incorporation of the bone graft is a dynamic process involving the following processes: osteoinduction, osteoconduction along with the availability of osteogenic cells, and the structural integrity, which provides mechanical support.^14,19,25,26 This ultimately leads to the replacement of the graft by host bone in a predictable pattern under the influences of load bearing.^14,27

At the beginning, the inflammatory response at the host-graft interface results in migration of inflammatory cells and fibroblasts into the bone graft. In addition, the developing hematoma enhances the release of both cytokines and growth factors. Osteoinduction is the process whereby a tissue is influenced to form osteogenic elements through chemotaxis, mitosis, and differentiation of the host osteoprogenitor cells. Induction requires an inducing stimulus, such as a piece of bone or an osteogenic cell, and an environment favorable for osteogenesis. Osteoconduction is the process by which capillaries, perivascular tissue, and osteoprogenitor cells from the recipient bed grow into the graft. It can occur within a framework of nonbiologic materials or nonviable biologic materials. In viable bone grafts, osteoconduction is facilitated by osteoinductive processes and therefore occurs more rapidly than in nonviable or nonbiologic materials.²⁸ Ultimately, this process results in the resorption of the original graft tissue and replacement with new host bone. Remodeling is a response to weight bearing.

Biomechanics of Graft Incorporation

Porosity is a dominant factor in determining the material properties of bone. It is directly related to the stiffness of the tissue and yield of strength.^13,14,32 Therefore, any change in porosity result in important effects on the bone graft material properties. Cortical bone grafts initially may have as little as 5% to 10% porosity, whereas cancellous grafts may be as high as 70% to 80%. This explains the material strength of cancellous graft, which is roughly equivalent to 4% of that of cortical bone.¹³

Cancellous grafts are incorporated by an early appositional phase. New bone formation onto the necrotic trabeculae of the graft tissue leads to an early increase in graft strength. It has been shown that necrotic bone maintains its mechanical strength.³⁰ Cancellous grafts therefore initially strengthen with the addition of new bone. As the necrotic cores are resorbed, the mechanical strength of the graft area normalizes.

Cortical bone grafts first undergo osteoclastic bone resorption, which significantly increases graft porosity and thus decreases the graft strength. In the canine model of autogenous cortical transplant, the greatest compromise in mechanical strength occurs at 12 weeks³⁰ (Fig. 17-1). The strength returns to normal between 1 and 2 years after transplantation. Human data suggest that cortical grafts lose approximately half their biomechanical strength during the first 6 months—a decline that persists for another 6 months.³³ This process is related to osteoclastic graft resorption and is slowly reversed during the second year after implantation. These observations correlate with the highest incidence of mechanical graft failure between 6 and 8 months after transplantation. If the graft is allogenic, this process is further prolonged. Hence it is important to protect segmental grafts during the critical phase when the resorptive phase outstrips the appositional phase. This is usually accomplished by load sharing with spinal instrumentation or a spinal orthosis.

FIGURE 17-1 Graph illustrating the quantitative temporal interrelationships between the physical integrity and the biologic processes of repair within a segmental autogenous cortical bone transplant. The initial persistence of strength (0–4 weeks after transplantation) indicates the subsequent loss was caused by reparative processes rather than any intrinsic weakness in the material. The sudden loss in strength at 6 weeks is caused by the increased internal porosity. From 6 to 12 weeks, the decrease in mechanical strength is reduced by 50%. The level of porosity continues to increase until week 12 because of the temporal lag in the apposition of new bone formation. At 24 weeks, there is no significant improvement in strength, despite the beginning reduction in the porosity of the transplant and maturation of the callus. At 48 weeks, however, the physical integrity of the transplant has returned toward normal, primarily as the result of decreased material porosity, since the amount of callus has not increased. By 2 years, the physical integrity of the transplant has returned toward normal, primarily as a result of decreased material porosity, since the amount of callus has not increased. By 2 years, the physical integrity of the transplant and the internal porosity of the remaining transplanted material are normal. The biologic completeness of repair (i.e., approximately 50% of the graft is viable) is not significant, because mechanical strength has been retained. The admixture of necrotic and viable bone remains for the life of the individual’s skeletal metabolic activity.

(From Burchardt H: Biology of cortical bone graft incorporation. In Friedlaender GE, Mankin HJ, Sell KW, editors: Osteochondral allografts: biology, banking, and clinical applications, Boston, 1983, Little, Brown, p 55.)

Temporal Profile of Graft Incorporation

During the first week after grafting, both cancellous and cortical grafts have similar histologic features. Both are surrounded by coagulated blood, and the graft is the focus of a tissue response characterized by vascular buds infiltrating the grafted bed. By the second week, fibrous granulation tissue becomes increasingly dominant in the graft bed, the number of inflammatory cells decrease, and osteoclastic activity increases. Within the confines of the graft, osteocytic autolysis proceeds, resulting from anoxia and injury by surgery, with necrosis delineated by vacant lacunae. Some cells, however, survive by diffusion of nutrients from surrounding host tissues. Creeping substitution of cortical bone grafts progresses transversely and parallel to the long axis of the transplanted segment. Thus the repair is found to be greater at the graft-host junctions.³⁴

A study done in rabbits³⁵ showed the sequence of events during the process of dorsolateral intertransverse fusion. Three phases were identified. Phase 1 represents the early reparative phase (1–3 weeks). It consists of hematoma formation and granulation tissue. There is minimal ossification. Phase 2 represents the middle reparative phase (4–5 weeks), when the fusion solidifies. Finally, phase 3 represents the late remodeling phase (6–10 weeks).³⁵

Both membranous and enchondral ossification play a role in the fusion process. Membranous ossification is the predominant mechanism that begins at the termini of the fusion mass and emanates from the decorticated transverse process. The central portion of the fusion mass, where the vascular supply is poorer and movement is greater, heals by cartilage formation and enchondral ossification.

Host Response and Incorporation of Autograft and Allograft

Autograft remains the gold standard in most fusion applications. In certain situations in which available autologous bone is insufficient or when large structural grafts are needed, allograft fusion rates can approach or equal those of autograft rates, without donor site morbidity. A successful spine fusion requires a sufficient area of decorticated host bone, ample graft material, minimal motion at the fusion site, and a rich vascular supply.³⁶

Histocompatibility matching has an important influence on the process of incorporation. Allograft that is mismatched for major histocompatibility complex antigens functions poorly compared with autogenous grafts.^37,38 Bone cells display class I and class II histocompatibility antigens, and there are both cellular and humoral responses to bone allografts.³⁹ Syngeneic grafts are the most successful. Grafts with major histocompatibility mismatch have delayed and incomplete revascularization, compared with syngeneic grafts. In addition, marked resorption of bone often occurs, resulting in almost complete loss of graft.³⁸ Freezing the graft, followed by thawing, disrupts and kills the cells. It mutes the antigenicity in major mismatches and thus enhances incorporation of such grafts. However, the killing of cells also diminishes the biologic activity of the graft. It is the osteoinductive component that is mainly affected. The function of an allograft as an osteoconductive system seems virtually unimpaired.

In fresh cancellous allografts, the initial phase consisting of hemorrhage and necrosis is identical to that of the autograft. The fibrin clot and the same inflammatory response develop. However, in the allograft the fibrin clot breaks down, and the granulation tissue, which provides nutrition to the repair site, is invaded by chronic inflammatory cells rather than fibroblasts and blood vessel elements. The major portion of the delay appears to occur in osteoclastic resorption and new bone formation. Final graft incorporation remains incomplete.

In cortical allografts the length of time of creeping substitution is greatly prolonged. The invasion by host vessels and recruitment and differentiation of cellular elements to become osteoblastic and osteoclastic cells are greatly diminished. The proportion of necrotic graft bone to viable host bone is much greater in allogeneic grafts. In fact, the active process of graft substitution may last several years.

Modeling and Remodeling Associated with Spine Fusion

The bone modeling associated with spine fusion is extremely complex. Variables that may affect bone remodeling after graft insertion include (1) the design of implant, materials used, and methods of fixation; (2) the local bone, including its density and shape; and (3) the patient characteristics, including age, gender, hormonal balance, and activity.⁴⁰ Osteoblasts and osteoclasts are influenced by the magnitude and state of strain imposed on them by load applied to the bone. Stresses or strains within a given range seem to be required to maintain a steady state remodeling of bone in which the rate of bone formation equals the rate of resorption. Stresses below the optimum are often associated with stress shielding, leading to bone resorption. Stresses and strains exceeding upper limits can also produce resorption of bone as a result of pressure necrosis. Cyclic stresses are required to maintain osseous homeostasis. Constant loads, even when within the desired range, can result in insufficient stimulus to maintain bone mass. Observations of strain-related electric potentials in bone, biopotentials, and electrical stimulation of osteogenesis suggest that bioelectric phenomena function as the regulators of adaptive remodeling of bone.

Growth Factors and Cytokines in Regulating Bone Remodeling

Bone cells carry out diverse functions and are mainly derived from two cell lines: mesenchymal and hematopoietic. The mesenchymal stem cell line consists of undifferentiated cells, or preosteoblasts, that differentiate into osteoblasts, bone-lining cells, and osteocytes. The hematopoietic stem cell line consists of circulating marrow monocytes that differentiate into preosteoclasts and osteoclasts. These cells are regulated by various cytokines.

Bone formation in spine fusion is a complex and regulated process. The cellular events involved in bone formation include chemotaxis of osteoblast precursors, proliferation of committed osteoblast precursors, and differentiation and expression of regulatory factors and structural proteins of bone and mineralization.⁴¹ These processes require tight regulatory control. They may be modulated by systemic hormones, such as parathyroid hormone, but predominant control is by local factors or cytokines. Cytokines are small proteins that serve as signaling agents for cells. Cytokines are classified based on their cellular origin and principal biologic activities.⁴² The main families include interleukins, tumor necrosis factors, growth factors, colony-stimulating factors, interferons, and chemokines.

Bone morphogenetic protein (BMP) is a member of the transforming growth factor beta (TGF-β) superfamily. The BMP constitutes a growth family of more than 12 proteins, 9 of which have been shown individually to induce ectopic bone formation.⁴³ They are water soluble, noncollagenous substances found in the bone matrix with osteoinductive activity. BMPs 2, 4, and 7 are specially increased in the primitive mesenchymal and osteoprogenitor cells, fibroblasts, and proliferating chondrocytes present at the fracture site.^18,44,45 During the phases of healing, the expression of BMPs 2, 4, and 7 is strongly present in undifferentiated mesenchymal cells during the inflammatory phase. During intramembranous ossification, these BMPs are strongly present in the proliferating osteoblasts. During chondrogenesis and endochondral ossification, BMPs 2 and 4 are found in proliferating chondrocytes and strongly in osteoblasts near the endochondral ossification front. BMP 7 is found in later stages of healing in proliferating chondrocytes and weakly in mature chondrocytes.^18,45 BMPs also affect the expression of other growth factors that may function to mediate the effects of BMPs on bone formation.

They are the most widely investigated osteoinductive growth factor in spine fusion.^43,46 Several animal studies have shown that recombinant human (rh) BMPs 2, 4, and 7 induce bone formation at an orthotopic site at which the integration with the preexisting bone is structurally sound. It has also been shown that BMP plus marrow yields the highest union rates (100%) and is three times better than autogenous cancellous graft.^46–54

However, other factors may play a role in the biologic response to rhBMP in vivo, including the time course activity of rhBMP on the bone formation process, its interaction with other growth factors, and the influence of delivery vehicles.

Biomechanics of the Fusion Mass

In dorsal and dorsolateral fusion, healing occurs through callus formation. As ossification proceeds, the callus is converted from a low-stiffness, rubbery quality to a hard tissue type of resiliency. The mineralization of the callus progressively increases its tensile strength. The fusion site during all stages of the reparative process is highly susceptible to mechanical forces directly related to the amount of motion between the graft fragment and host surface. The amount of relative motion determines the morphologic patterns of fracture repair. As healing proceeds, the amount of motion decreases. When mechanical stability is compromised, more cartilage always forms and, occasionally, an exuberant callus. Frequently, with excessive motion the fusion mass is incomplete and a pseudarthrosis develops. Rigid internal fixation has been demonstrated to reduce pseudarthrosis rates in most clinical applications.

Positioning of Bone Graft

As White and Panjabi²⁰ describe, “The placement of a fusion mass at the maximum distance from the instantaneous axes of rotation will be more effective in preventing the movement around those axes” (Fig. 17-2). The instantaneous axis of rotation (IAR) is defined as the point in the body, or some hypothetical extension of it, that does not move when a rigid body moves in a plane. An axis perpendicular to the plane of motion and passing through the point is the IAR for that motion at that instant (Fig. 17-3). It can be defined more simply as the axis around which the vertebral body rotates. It is like a fulcrum. Usually, but not always, the IAR passes through the confines of the vertebral body. With isolated destruction of columns of the spine, the IAR migrates to the remaining intact structures, as shown in Figure 17-4.

FIGURE 17-2 A, To prevent the opening of the blades of the scissors by holding them together, it is distinctly easier to pinch the blades together at the tips (distance B) rather than at the midpoint of the blade (distance A). Because distance B is farther from the instantaneous axis of the rotation (IAR), there is greater leverage. The same concepts apply to the vertebral functional spinal unit. Flexion, separation, or opening of the spinous processes is more readily prevented by placing the fingers at the tips of the spinous processes (distance B) rather than at the facet joints (distance A). Thus, with regard to a flexion movement, a healed bone graft at distance B, at the tips of the spinous processes, is more effective than one closer to the IAR, other factors being constant. These concepts partially explain the efficacy of the rather delicate interspinous and supraspinous ligaments. B, The concept of leverage is shown again here. The anterior bone graft A is a short distance (analogous to L) from the IAR and therefore provides less leverage than bone graft B, which is a greater distance (analogous to 2L) from the instantaneous axis of rotation.

(From White AA, Panjabi MM: Clinical biomechanics of the spine, Philadelphia, 1990, JB Lippincott, p 533.)

FIGURE 17-3 Instantaneous axis of rotation (IAR). A construction for determining the IAR is shown. A₁–A₂ and B₁–B₂ are translation vectors of points A and B.

(From White AA, Panjabi MM: Clinical biomechanics of the spine, Philadelphia, 1990, JB Lippincott, p 660. Reprinted with permission.)

FIGURE 17-4 Location of the axis. The location of the IAR for the intact and compromised specimens. A, Facet joints compromised. B, Intact spine. C, Facet joints and anulus compromised.

(From Haher TR, O’Brien M, Felmly WT, et al: Instantaneous axis of rotation as a function of the three columns of the spine. Spine [Phila Pa 1976] 1992:17[6S]:S153. Reprinted with permission.)

The greater the distance of the fusion mass from the IAR, the greater the leverage in preventing motion around those axes of rotation. Examples include dorsal, dorsolateral, and intertransverse lumbar fusion, in which the fusion mass is located at a distance from the IAR (which is located in the region of the vertebral body). A ventrally placed graft is closer to the IAR and applies less leverage, but can still be extremely effective, especially in cases of anterior column deficiency.

The concept of rigidity is also important. A fusion mass that involves all the dorsal elements and transverse processes provides more rigidity than a fusion that only involves the spinous process. In some situations it can be disadvantageous to place the graft at a distance from the IAR. For example, after a dorsal fusion to treat discogenic pain, motion may still occur at the disc interspace, even when all dorsal elements except the pedicle are fixed.^55,56 In such situations an interbody fusion may be considered (Fig. 17-5).⁵⁷

FIGURE 17-5 Illustration of the position of a bone graft (B), which can provide maximum rigidity by eliminating interbody motion.

(From White AA, Panjabi MM: Clinical biomechanics of the spine, Philadelphia, 1990, JB Lippincott, p 535. Reprinted with permission.)

The production of biomechanical changes as a function of different types of lumbar fusion has been studied.^58,59 The three types of fusion evaluated included dorsal, bilateral lateral, and ventral. All types of fusion increased bending and axial stiffness. There is increased stress on the adjoining segments that were not fused, especially the facet joints. Overall, bilateral intertransverse fusion is a superior method because it provides good stabilization to the fused segments and has less effect on adjacent, unfused segments, especially the facet joints. Dorsal (intraosseous) fusion is the least beneficial, producing the highest amount of stress in adjoining segments and allowing superficial motion in the disc space.

The spine experiences compressive forces on the concave side and tensile forces on the convex side of a curve. In the lumbar spine, if the graft material is placed in the intervertebral disc space, it is subjected to compressive loading. It is believed that compressive forces acting on the graft will promote fusion by stimulating the osteoconductive healing process. In contrast, a graft placed in a dorsal location experiences only tensile forces and will not be stimulated in a similar manner (Fig. 17-6).

FIGURE 17-6 A, The spinal column may be analyzed by regarding it as similar to a beam. There is tension on the convex side of the curve and compression on the concave side. The dashed line is the neutral axis, and there is neither tension nor compression along this line. B, Bone grafts inserted at the various points tend to behave as follows. The graft on the convex side of the curve is mainly under tension and cannot resist deforming forces until fully attached at both interfaces. The graft at the dashed line (neutral axis) provides little or no resistance to bending in the plane of the page. C, In the spine, the graft at position 1 is well away from the neutral axis and when biologically fused at both interfaces can offer effective tensile resistance against progressive kyphosis. The graft at position 2 can do the same, but is less effective because it is closer to the neutral axis. The graft at position 3 is not likely to be as effective as those at position 1 or 2 in preventing progression of deformity because it is even closer to the neutral axis. The graft at position 4 is effective because it immediately begins to resist compressive forces and therefore tends to prevent additional deformity and angulation at that point. The graft is also some distance away from the neutral axis, giving it a mechanical advantage.

(From White AA, Panjabi MM, Thomas CL: The clinical biomechanics of kyphotic deformities. Clin Orthop Relat Res 8:128, 1977.)

Kyphotic Deformity and Bone Graft Positioning

At the IAR, there is neither compression nor tension. The farther instrumentation or bone graft is placed from the IAR, the greater the stress. For instance, in a kyphotic deformity, dorsal instrumentation is subjected to severe tensile stress. To reduce stress on a dorsal implant, some structural graft should be placed as ventrally as possible, away from the IAR. This counteracts the tensile stress dorsally. At times, with severe kyphotic deformity, multiple ventral grafts may be required (Fig. 17-7). It has been demonstrated that ventral and dorsal fusions are associated with a better correction and maintenance of correction than the dorsal group, but only with congenital kyphosis.⁶⁰

FIGURE 17-7 Illustration of the various locations of ventral bone grafts for kyphotic deformity. The biomechanical considerations involved in choosing graft A, B, or C are discussed in the text.

(From White AA, Panjabi MM, Thomas CL: The clinical biomechanics of kyphotic deformities. Clin Orthop Relat Res 8:128, 1977.)

Load Sharing

Denis⁶¹ introduced the three-column theory of the spine to classify and assist with the management of thoracolumbar spine injuries. Of these three columns, the anterior and posterior columns are the principle support structures.⁶² The anterior column resists compression and axial loading, and the posterior column maintains the tension. To maintain an erect posture, all forces and movements must be balanced about the IAR. The IAR is located dorsal to the anulus fibrosus in the intact spine.⁶³

Deficiencies in the anterior or posterior column in the thoracolumbar spine usually lead to kyphosis.⁶⁴ Kyphosis is corrected by lengthening the anterior column or shortening the posterior column. If the anterior or middle column is destroyed, alignment can be restored by a ventral structural graft and the resulting fusion. In this situation the axial load is shared by both anterior and posterior columns. When deciding on whether to perform a ventral or dorsal fusion, or a combination of both, the principles of load sharing should be considered. If both the ventral and dorsal elements are involved, both columns usually must be instrumented and fused. For example, a burst fracture will be compromised if the dorsal elements frequently require both ventral and dorsal spine reconstruction. With persistent posttraumatic kyphosis after a dorsal instrumentation procedure to treat a cervical or thoracolumbar fracture, anterior column load sharing is eliminated. Instrumentation such as a pedicle screw implant is exposed to high cantilever bending loads and may therefore fail.⁶⁵ With correction of a kyphotic deformity, ventral surgery may not be necessary if the weight-bearing line is shifted behind the axis of rotation.⁶⁶ By shifting the center of gravity dorsally, the anterior column does not have to support as much axial load. The prerequisites for such a strategy include (1) correction or overcorrection, if surgically feasible; (2) intact dorsal elements; and (3) good osteogenic potential. If sagittal correction is not accomplished, the load on the anterior column is high, and anterior column reconstruction is needed to prevent dorsal instrumentation failure.

Ventral instrumentation, without structural bone grafting, usually fails. A strong structural graft is required to resist axial loading and flexion.⁶⁷ Tricortical ilium, fibula, humerus, or titanium cages packed with autogenous graft provide excellent anterior column support. However, single-rib grafts do not provide adequate structural support. Load sharing, in this case, implies a balance between ventral structural bone grafts and ventral or dorsal instrumentation. As the fixation length of ventral and dorsal constructs is reduced, load sharing with the anterior column has become increasingly important in reducing the incidence of failure of the shorter devices. The conditions frequently requiring both ventral and dorsal reconstruction include tumors involving both anterior and posterior columns, fractures involving all three columns, and postlaminectomy kyphosis.⁶⁴

In dorsolateral spine fusion, instrumentation adds to the stability of the fusion by significant load sharing. In a human spine model where bilateral facetectomies were performed and transpedicular screws were used to restore stability, the spinal instrumentation provided 68% of the load sharing, along with the anterior and middle columns.⁶⁸ As the fusion mass develops in vivo the load-sharing component of the instrumentation decreases. If an adequate fusion mass does not develop, the cyclical stresses placed on the instrumentation will lead to hardware failure (Fig. 17-8).

FIGURE 17-8 The relationship between bone fusion acquisition and spinal implant integrity changes with time after surgery.

(From Benzel EC: Biomechanics of spine stabilization, New York, 1994, McGraw-Hill, p 104.)

Stress Shielding

In a canine model, dorsolateral fusion without instrumentation resulted in fusion in only 57%, compared with a 100% fusion rate with pedicle screw fixation and a 71% fusion rate with Luque rods. Histologic evaluation of the vertebral body at the level of the fusion demonstrated osteoporosis in animals that had received instrumentation. This has been corroborated in humans.⁶⁹ Patients who had undergone instrumented dorsolateral lumbar fusion were found to have decreased vertebral body mineral density at the level of fusion, compared with matched controls. This phenomenon has been termed stress shielding. However, in animal models the spine fusions that had been instrumented demonstrated increased areas of bone incorporation and biomechanical stability,^70–72 and for any preexisting osteoporosis, compensation was more than adequate. In general, rigid fixation results in better union. Ventral interbody fusions are more prone than dorsal-only fusions to the negative effects of stress shielding.