Primary Bone Healing

General Aspects

Bone has a breaking strength similar to that of medium steel. Yet it is elastic and relatively light, representing only 10% of body weight.1 A fracture occurs when a bone is subjected to forces greater than its ultimate failure load, resulting in a loss of continuity. In these instances, the main goal of orthopedic surgeons is to palliate the biomechanical functions of bone until its structural integrity has been regained. Bone assumes other vital functions such as storage of phosphorus and of approximately 99% of the body’s calcium [1]. Its medullary cavity is a site of hematopoiesis and, in adults, a storage area for fat. However, the consequences of fractures on these functions are typically negligible compared with their impact on the biomechanical functions of bones. Indeed, structural integrity of the skeleton is essential to maintain locomotion and protection of vital organs. Bone serves as a frame for the origin and insertion of surrounding muscles and allows transmission of loads across the weight-bearing axis. Osteosynthesis was developed to restore these roles and promote early return to function of the patient.

The lack of callus formation between two bone fragments apposed under a rigid plate was first noted by Lane in 1914 [2]. Krompecher later reported that a neutral biomechanical environment was required to allow primary angiogenic bone formation in embryonic intact bone. The lack of callus formation as seen on postoperative radiographs of fractures repaired with compression plates was documented by Danis in 1949 [3]. He termed this mode of repair "internal welding," referring to direct filling of the fracture site with bone, without formation of mechanically relevant periosteal or endosteal callus. Schenk and Willenegger later confirmed that healing under these conditions occurred by direct osteonal proliferation.4 Stable interdigitation of bone fragments is clinically achieved by application of rigid, nongliding implants, such as compression bone plates or lag screws. Precise reduction and rigid fixation appear to eliminate the biological signals that are known to attract osteoprogenitor cells from surrounding soft tissues and contribute to callus formation in secondary healing [5]. Even under these conditions, the microenvironment differs within the fracture site and influences the process by which bone is laid. Indeed, full congruency between the fragment ends is never achieved, even after meticulous reposition. Instead, contact and compression are obtained in circumscribed zones (contact points), separated by areas where fragment ends are separated by small gaps [6].

Primary Healing of a Diaphyseal Fracture or Osteotomy

In the presence of normal blood supply and cell functions, accurate apposition and compression between bone fragments are prerequisites for primary healing. This may be achieved in clinical cases after anatomic reduction of a simple fracture and fixation with a compression plate. However, the application of a bone plate tends to create different biomechanical microenvironments within the fracture site. A compression plate applied across a transverse osteotomy generates high pressure and, therefore, improves contact in the cortex located directly under the plate (Fig. 85-1)). On the other hand, the far cortex becomes a tension side and is predisposed to gap healing [7,8]. Both gap and contact healing differ from secondary healing by their lack of resorption of the fracture ends, even if a high magnitude of compression is applied across the fracture site.

Contact Healing

Contact healing occurs between apposed cortical surfaces, when the defect between bone ends measures less than 0.01 mm and interfragmentary strain is less than 2% [9,10]. These conditions allow the defect to be filled by primary osteonal reconstruction. Lamellar bone is directly deposited in the normal axial direction of the bone [8,10]. The process is initiated by the formation of cutting cones at the ends of osteons located closest to the fracture site [11].

Osteoclasts line the spearhead of the cutting cone, while osteoblasts form the rear of the cutting cones, so that bony union and Haversian remodeling occur simultaneously [7,11]. Osteoclasts advance across the fracture site, forming longitudinally oriented resorption cavities [7]. Perivascular osteoblastic precursors accompanying the capillary loops within these cavities differentiate into osteoblasts and produce osteoid [10]. Crossing osteons appeared at osteotomy sites 3 weeks after canine radii had been experimentally transected and stabilized with a plate [6,7]. Based on tetracycline labeling, the daily progress of these cutting cones across the fracture site has been calculated to vary between 70 and 100 μm, which is about 3 times the size of each individual osteoclast [12]. Crossing osteons mature by filling with osteonal lamellar bone and become the “spot welds” that unite the two fragments, without formation of periosteal callus. The lamellar bone formed is immediately aligned parallel to the long axis of the bone, with Haversian canals deviating from this axis by less than 10° [13]. However, the newly formed bone is less dense than the intact cortex during the first few months; therefore, the fracture area remains visible on radiographs until complete remodeling, which takes a few months up to a few years, depending on the species [8].

Gap Healing

A different process of primary bone healing has been observed in experimental defects, and later confirmed to occur across small gaps of rigidly fixed fractures. If interfragmentary deformation remains less than 2%, bone directly forms in gaps measuring less than 800 μm to 1 mm. This “gap healing” differs from the healing in contact areas as bony union and Haversian remodeling become separate sequential steps [7]. The fracture site fills directly by intramembranous bone formation, but the newly formed lamellar bone is oriented perpendicular to the long axis and later undergoes secondary osteonal reconstruction. A vascular loop from the medullary vasculature grows into the gap, and loose connective tissue fills the site [9]. Osteoprogenitor cells accompany the vascular loop and differentiate into osteoblasts. After two weeks, the blood supply is well established and osteoblasts deposit layer upon layer of lamellar bone on each surface of the gap, until the two fragment ends are united [11]. Woven bone may form initially in larger gaps, to subdivide the area into smaller compartments, subsequently filled with lamellar bone [7]. Although the fragment ends are united by lamellar bone, this area remains mechanically weak, as the bone is oriented perpendicular to the long axis and poorly connected to adjacent intact cortex. Haversian remodeling starts between three and eight weeks, when osteoclasts form longitudinally oriented resorption cavities (Fig. 85-1) [11]. These cutting cones are formed by new osteons within the fracture gap and by osteons originating from the surrounding intact bone. They advance across the fracture plane to unite the new lamellar bone deposited in the gap to each fragment end. The resorption cavities mature into longitudinally oriented lamellar bone, so that with time, the anatomical and mechanical integrity of the cortex is reestablished.

Figure 85-1. Primary healing after anatomic reduction of a transverse diaphyseal fracture and application of a compression plate. The cis-cortex undergoes contact healing, whereas gap healing occurs in the trans-cortex.

Primary Healing of Cancellous Bone

Cancellous bone has a lower volume-to-surface ratio, which makes it 25 to 100 times weaker than cortical bone [14]. In a long bone, this relative weakness is palliated by the greater cross-sectional area of metaphyseal or epiphyseal bone, compared with the diaphysis. This characteristic improves the stability of metaphyseal bone in bending and torsion and increases its tolerance to strain, compared with diaphyseal bone. Indeed, any deformation applied to the bone will be distributed over a larger area in the metaphysis than in the diaphysis. In addition, metaphyseal fractures often result in impaction of cancellous bone, thereby creating a mechanical environment minimizing interfragmentary motion [15]. Cancellous bone may consequently undergo primary healing, with new bone being deposited along the surfaces of the trabeculae.

Healing of cancellous bone has also been described in the context of bone grafting techniques. In dogs, the largest amounts of autogenous bone can be collected from the iliac crest, followed by the proximal humerus and medial proximal tibia [16]. Ribs and proximal femur have been used less commonly as donor sites [17,18]. Metaphyseal cancellous bone is typically harvested via a round hole created through the cortex with a pin or a trephine. The healing of these circular defects has been studied in the proximal tibia of dogs [19]. By two weeks, a hematoma and fibrovascular tissue fill the bone defect. An endosteal callus forms by four weeks, with foci of cartilage and woven bone later replaced with lamellar bone. The normal structural arrangement of lamellar bone and hematopoietic marrow was reestablished in the marrow cavity at 12 weeks. Healing of donor sites was later found to vary somewhat with location, but a second collection can be obtained from the same metaphyseal site 12 weeks after the first harvest [16,17]. Defects in the proximal tibia heal slower than do those located in the humerus and fill with more fibrous tissue [17].

Primary Healing of Circular Bone Defects

The healing of circular bone defects occurs in a stable environment because the surrounding cortex remains continuous. However, the healing process varies with the diameter of the defect.

The healing of unicortical transverse burr holes has been described as similar to fracture gap healing by Schenk [6]. In this experiment, unicortical 200 μm diameter holes created in the tibia of rabbits initially contained well vascularized granulation tissue. Osteogenic cells in the periosteum and endosteum rapidly proliferated and differentiated, initiating a healing process along the walls of the defect. Osteoblasts subsequently formed a continuous layer and deposited lamellar bone in a concentric manner. This bone was gradually replaced by longitudinally oriented osteons over the next months. Larger holes, up to 1 mm in diameter, are initially filled with a primary scaffold of woven bone. This scaffold consists of randomly oriented trabeculae, creating 200 μm diameter compartments filled with granulation tissue. Osteoblasts form a film over these trabeculae and begin to fill intertrabecular spaces with lamellar bone. Within a month, these compartments have narrowed down to the size of cortical bone vascular channels. Remodeling then proceeds to restore the original microscopic appearance of the cortex within a couple of months. Schenk drew four important conclusions from these observations; 1) small bone defects heal by primary intention, under stable conditions; 2) bone is always deposited on a solid surface, consisting of the wall of the defect or the surfaces of trabeculae within the defect; 3) osteoclasts do not appear within the defect until 3 to 4 weeks, once remodeling is initiated; 4) the absence of osteoclasts within the first week after creation of small circular defects would imply that all signaling factors involved at this stage must influence osteprogenitor cells.

Filling of larger holes, such as screw holes, takes longer and may not be complete. Fibrous tissue is first formed along the surface of the walls (Fig. 85-2) [20]. Lamellar bone is then deposited into the fibrous network, proceeding from the periphery toward the center of the defect. Although remodeling gradually restores the orientation of the bone, the thickness of the cortex may not be regained. The prolonged radiographic visibility of these defects reflects both their decreased cortical thickness and the presence of less mineralized new bone. Increasing the diameter of the defect eventually reaches a critical point, beyond which the regenerative capacities of the bone are exceeded and the defect persists permanently. The dimension of these critical-size defects vary with the location, species, and other environmental conditions. These defects are especially relevant in the skull and mandible, where they provide nonunion models for evaluation of new agents stimulating bone healing. In the diaphysis of long bones, the size of a bicortical circular defect, such as a screw hole, is essentially limited by its stress-rising effect and potential for iatrogenic fractures. For example, defects measuring 20% of the diameter of the femur decreased the torsional strength of the bone by 34% in sheep [21].

Figure 85-2. Healing of a 5-mm diameter unicortical circular defect in the femoral diaphysis of a dog (x1.2, H&E). The defect initially fills with connective tissue (A). Bone forms along the edges of the defect and progresses toward the center (B). Woven bone fills the entire defect and remodeling starts along the periphery (C).

Evaluation of Bone Healing

Clinical assessment of fracture healing has traditionally been based on physical examination and radiographs. Current research focuses on improving existing imaging techniques and developing new tools for early detection of complications (such as infection and delayed healing) and objective guidance of postoperative fracture management (return to function and implant removal) [22-24]. For example, the ultrasonographic appearance of healing has recently been described after biologic fracture fixation of long bones (e.g., external coaptation, external fixators, or interlocking nails) in dogs and cats. Ultrasonographic signs of bony union included a hyperechoic continuous line in three views and disappearance of intramedullary implants. In this clinical study, secondary fracture healing was diagnosed 20 days earlier via ultrasonography than it was radiographically [25]. Correlating these findings with a biomechanical evaluation of the callus would strengthen recommendations for implant destabilization or removal based on early ultrasonographic signs of healing. Indeed, fracture healing should intuitively be evaluated by the return of prefracture stiffness and ultimate strength, because the main goal of fracture management is to restore the structural functions of bone. Ultimate strength and stiffness of the bone defect are commonly measured to quantify healing in experimental fracture models used to study bone repair or evaluate fracture treatment modalities. For example, no refracture nor angulation were diagnosed in an experiment where an external fixator was removed once the fracture gap had regained a stiffness equal to 10 Nm/degree [26]. Although measurement of fracture stiffness recently allowed early detection of human patients at risk for tibial nonunions, biomechanical testing of clinical cases remains impractical [27]. Instead, clinical assessments of fracture healing are based primarily on radiographic criteria. Experimental evaluation of fracture healing strives for added precision and objectivity, and often includes densitometry and/or histologic techniques.

Radiographic Evaluation

Danis was the first to characterize primary fracture healing based on its radiographic appearance [3]. The gradual disappearance of a fracture line without formation of an external callus defines direct healing on serial radiographs (Fig. 85-3). Although no resorption of the fragment ends occurs with contact healing, the progression of cutting cones across the gap decreases the radioopacity of the zone around the fracture [8,28].

Figure 85-3. Radiographic appearance of primary healing. A. Postoperative lateral radiograph of transverse fractures of the distal diaphysis of the radius and ulna in a 6-month-old dog. The radial fracture has been managed with direct reduction and plate fixation. B. Similar projection obtained 8 weeks later. Cortical apposition and rigid fixation lead to bony union without formation of callus.

Complete remodeling of the fracture varies with the location and species from a few months up to a few years, during which the fracture site remains radiolucent compared with intact cortex [8]. Primary healing displays few radiographic signs compared with secondary bone formation, and the monitoring of fracture healing under these circumstances is indirectly based on the absence of adverse clinical and radiologic signs [29]. Radiographic evidence of callus formation and bone resorption after “rigid” fixation of a fracture is consistent with secondary bone healing. These signs are interpreted as evidence that the stability and/or reduction achieved did not match those intended.

Dual Energy X-Ray Absorptiometry

Dual energy X-ray absorptiometry (DEXA) quantifies the bone mineral content and density of the entire skeleton (whole body analysis) or of a specific region of interest within a bone. The first mode of analysis is most commonly used to diagnose and monitor human osteoporosis. The second mode of analysis has been used experimentally in animals to objectively evaluate healing of bone defects [30-33]. Based on the area of interest selected, the computer software allows analysis of another region of similar dimensions, to act as a control. This control may consist of an area of intact bone or the same region evaluated serially. Because the software measures areas rather than volumes, the thickness of samples should be standardized for meaningful comparison. DEXA can be repeated several times in small animals with no morbidity other than that of routine sedation. This lack of invasiveness makes DEXA especially relevant in longitudinal studies of unicortical bone defects and experimental osteotomies. DEXA measurements have been found to correlate with the torsional properties of healing canine tibial osteotomies [30]. These findings would support the use of DEXA as an objective guide for timing fracture fixation removal, recommendations related to exercise regimen, and prediction of abnormal fracture healing patterns. However, metal implants create artifacts preventing the evaluation of regions in contact with plates. Therefore, osteotomies and fractures suitable for DEXA include those that do not require fixation (single ulnar osteotomy for example) or can be managed by external fixation. In the future, DEXA may also be indicated in veterinary patients for early detection of bone pathology, such as sclerosis (as a marker for elbow dysplasia), neoplasia, and infection. Clinical trials may also support its use, to monitor local bone metabolism and response to therapy in patients with osteomyelitis or neoplasia. In the meantime, clinical applications of DEXA in veterinary medicine remain limited by the lack of indications and equipment cost.

Histology

Histologic evaluation of fracture sites has improved our understanding of the cellular mechanisms involved in primary healing. For example, gap and contact healing are essentially differentiated based on their microscopic appearance. Histology remains especially relevant when testing new therapies, where precise evaluation of the fracture site is required. Studies may focus on potential adjuncts to current fracture treatments, in which case, the agent must be found to accelerate the normal healing process of an uncomplicated fracture or osteotomy model. Experiments focusing on the treatment of fracture complications, such as nonunions, test whether novel agents promote healing of critical defects that would otherwise not fill with new bone. In both instances, histologic evaluation provides valuable information regarding the type and extent of healing occurring within the defect. Biocompatibility of new grafting materials may be determined by the degree of inflammation shortly after implantation in bone defects. However, serial histologic studies are required to evaluate the osteoconductive properties of an agent, its degree of incorporation, its rate of degradation, and cellular mechanisms involved in these processes [34]. Histologic evaluation of bone defect healing may be subjectively scored on 5-µm thick decalcified sections, stained with hematoxylin and eosin [3,20]. However the microscopic appearance of primary bone healing is best evaluated on thin sections of undecalcified bone [14]. These 50- to 80-µm thick sections may be obtained on fresh bone or bone defatted in xylene and embedded in methylmethacrylate [20,35]. Histomorphometric analysis of these sections generates quantitative data that can serve as a basis for objective evaluation of bone healing. Digital images of sections are captured with a digital camera and analyzed with image analysis software to quantify each type of tissue present within the defect. Nomenclature and calculations for bone histomorphometry have previously been standardized by the American Society of Bone and Mineral Research [13]. Four types of histomorphometric data may be directly measured on these images: area, length (such as a perimeter), distance (between points or lines), and number (of cells, for example). These measurements are indices of the amount of tissue examined and can only be compared between subjects when related to a common referent, such as a clearly defined area or perimeter within a section. For example, we have used the total area of standardized cortical defects as a referent to measure the percentage of this area filled with implant, fibrous tissue, and new bone [20,36]. These measurements were then used to compare the degradation rate and osteoproductive properties of bone graft substitutes. Two-dimensional histologic sections actually display profiles of three-dimensional bone defects. Histomorphometric criteria, therefore, may also be reported as three-dimensional terms, as long as consistency is maintained throughout the study. Three-dimensional data commonly derived from measurements of areas on histologic sections include bone volume, osteoid volume, and fibrosis volume. Bone and osteoid surfaces are calculated based on two-dimensional bone and osteoid perimeters, respectively. The fourth type of histomorphometric data, number measurements, can only be reported in two-dimensional analyses, thereby requiring a referent.

The origin of dynamic bone histomorphometry dates back to 1958, when Milch discovered the localization of tetracycline in bone [37]. A few years later, Frost developed the methodology to study the tetracycline-based histology of bone remodeling [38]. Since then, other fluorescent labels such as xylenol orange, calcein blue, and calcein green have been used instead of, or in addition to, tetracycline (Fig. 85-4). Combining labels of different fluorescence within the same study facilitates the identification of individual lines and corresponding time of administration. The mineral apposition rate (MAR) is then calculated as the distance between the midpoints, or between corresponding edges of two consecutive labels, divided by the time between the midpoints of the labeling periods [13]. The MARs for resting lamellar bone in mammals range from 0.83 to 2.7 µm/day [6,39]. Histologic techniques based on fluorescent bone labeling have played a key role in improving our understanding of bone biology. These techniques provide crucial insights into the dynamic processes of activation, resorption, and formation (ARF) of the basic multicellular unit (BMU) responsible for the remodeling of cancellous and cortical bone [40]. Dynamic bone histomorphometry has contributed to numerous studies exploring the relationship between bone and various biomechanical or biologic factors. The findings of these studies supported mechanistic theories about the regulation of bone formation and maintenance, eventually leading to the Utah paradigm of bone physiology [41]. Dynamic histomorphometry is now commonly used to evaluate new therapies stimulating bone formation [36,39]. For example, the MAR of woven and lamellar bone nearly doubled during distraction osteogenesis of an experimental osteotomy in goats [39]. We have previously measured MARs in an ovine metaphyseal defect model designed to evaluate the biologic properties of impacted bone graft substitutes [20]. In this study, the MARs measured within circular defects were 50% greater than remodeling rates 4 weeks after surgery, and returned to resting levels by 12 weeks. The values obtained at the center of the defects at 4 weeks were also greater than those obtained at the periphery of the defects at the same time period. Measurements obtained along the edges of these defects appear to reflect a more mature phase of healing, with bone formation proceeding from the periphery toward the center of the defects. While these findings reflect the progression of healing in unicortical bone defects, they also stress the importance of standardizing techniques for histomorphometry. Measurements should be obtained on a representative number of sections and at standardized locations within bone defects to allow meaningful comparison of bone healing between treatment groups.

Figure 85-4. Fluorescence labeling for evaluation of bone healing in sheep. Bone labeling consisted of oxytetracycline (30 mg/kg) on days 23 and 35, and alizarin complexone (30 mg/kg) 84 days after creation of a metaphyseal bone defect [36].

Advantages and Disadvantages of Rigid Internal Fixation

Primary fracture healing skips the intermediate steps of tissue differentiation and resorption of the fragment ends, leading directly, although not necessarily faster, to the final remodeling of Haversian canals. Direct healing is rarely a goal in itself but rather a product of maintained absolute stability.29 Rigid fixation is one of the concepts laid out by the association for the study of internal fixation or AO (Arbeitsgemeinschaft fur Osteosynthesefragen), in order to promote early return to mobility and function of the fracture patient. In 1958, the AO recognized the advantages of anatomic fracture reduction, stable internal fixation, and preservation of the blood supply. At that time, the treatment of fractures involved mostly immobilization in plaster or by traction, often leading to prolonged healing and loss of function. In comparison, the pursuit of absolute stability, originally proposed for most fractures, immediately restored limb length and joint alignment. Rigid implants palliated the biomechanical function of bones and allowed early mobilization of joints adjacent to the fracture. Although AO principles still stand as fundamentals, their interpretation and clinical application have been adjusted in response to the knowledge emerging from experimental and clinical studies. Anatomic reduction, compression of articular fragments, and rigid immobilization remain golden standards in the management of joint fractures (Table 85-1). Eliminating gaps or steps in the articular surface and preventing callus formation facilitate cartilage healing and minimize postoperative joint disease. However, the emphasis in extraarticular fracture fixation has evolved in recent decades from mechanical to biologic priorities. This change has mainly been prompted by a better understanding of the effects of plating on the underlying bone and the influence of micromotion on closely apposed bone fragments.

Traditional plating techniques affect the blood supply of a fractured bone not only by the design and nature of fixation, but also by the soft tissue trauma associated with their placement. The surgical approach, elevation of soft tissues, and manipulation of bone fragments required to achieve anatomic reduction and plate fixation add to the traumatic event initially responsible for the fracture. Rigid immobilization of the plate and bone construct then relies on frictional hold. The extent of this interface and the rigidity of the plate correlate with the degree of osteoporosis of the underlying cortex [42]. This ischemic osteoporosis has been attributed to a blockade of the centrifugal flow of blood [43]. The vascular impairment is believed to be short-lived, with revascularization occurring four weeks after application of a plate and markedly increasing by eight weeks. Other studies failed to correlate bone necrosis of vascular origin with cortical porosis 8 and 24 weeks following plate application to intact canine femurs [44]. In fact, porosis was greater in the inner endosteal layer, away from the bone-plate interface, and in the absence of necrosis. If rigid fixation allows direct formation of bone, stress shielding develops when two components of different elastic moduli form one mechanical system. Adaptational osteopenia has therefore been suggested as another mechanism to explain the osteoporosis of the bone in contact with a plate. Although the structural changes of bone after plate fixation have been well characterized, the relative contribution of decreased cortical perfusion and stress redistribution to these changes remains controversial. The biphasic appearance of these changes is currently attributed to an initial osteonecrosis 8 to 12 weeks after plate fixation, resulting from cortical vascular disturbances, followed by osteopenia at 24 to 36 weeks, secondary to the mechanical environment [45]. These studies support the design of new implants minimizing contact with the fractured bone and/or providing a “less-than-rigid” fixation. The concept of “biologic fracture fixation” or “indirect fracture repair” has consequently gained acceptance among surgeons, affecting primarily the reduction techniques, method of fixation, and postoperative management of extraarticular comminuted fractures of long bones [46].