Pathophysiology, Classifications, and Natural History of Perthes Disease

Since the original reports of Legg-Calvé-Perthes disease (LCPD), much research effort has been undertaken to improve understanding of this idiopathic hip disorder. This article focuses on the current knowledge of the pathophysiology, classifications, and natural history of LCPD. Although the cause of LCPD remains largely unknown, some insight has been gained on its pathophysiology through experimental studies using animal models of ischemic necrosis. The few available clinical studies on the natural history of LCPD suggest that femoral head deformity is well tolerated in short and intermediate terms, but 50% of patients develop disabling arthritis in the sixth decade of life.


It is generally accepted that a disruption of blood supply to the femoral head is a key pathogenic event in Legg-Calvé-Perthes disease (LCPD). Necropsy and biopsy studies of patients with LCPD show evidence of tissue necrosis consistent with ischemic injury. Various imaging studies also show evidence of disruption of blood flow to the affected femoral head. Furthermore, the disruption of the blood supply to the femoral head in large animal models (porcine and canine models) produced histopathological and radiographic changes resembling LCPD.

The histopathological features of LCPD have been reported in several studies; however, these studies are limited by a small sample size. This limitation underscores one of the major obstacles in trying to understand the pathophysiology of a condition that seems to have so many variables. From the studies that are available, it can be summarized that the pathologic processes in LCPD affect the articular cartilage and the bony epiphysis, and in some patients the metaphysis and the physis.

Pathologic changes in the articular cartilage are mainly observed in the deep layer of the cartilage, which functions as a growth cartilage responsible for the circumferential growth of the bony epiphysis. The damage produces a cessation of endochondral ossification at the cartilage-subchondral bone junction. Other changes, including separation of cartilage from underlying subchondral bone, vascular invasion of the cartilage, and new accessory ossification, have been observed in various stages of LCPD. In the bony epiphysis, the necrosis of the marrow space and the trabecular bone, compression fracture of trabeculae, osteoclastic resorption, fibrovascular granulation tissue invasion of the necrotic head, and thickened trabeculae have been reported in various stages of LCPD. The physeal changes are most often seen in the anterior aspect of the femoral head, with areas of growth plate cartilage extending below the endochondral ossification line. Premature growth arrest is seen in fewer than 30% of patients with LCPD, suggesting that in a majority of the patients the growth plate continues to function. Metaphyseal changes are commonly seen during the early stages of LCPD, usually found below the growth plate. The mechanisms responsible for the appearance of these lesions are unclear. Various tissue types have been reported, including columns of normal or degenerated cartilage extending down to the metaphysis, fibrocartilage, fat necrosis, vascular proliferation, and focal fibrosis. Some have found an association between the presence of radiolucent metaphyseal changes and poor prognosis whereas others have not.

Experimental Studies

Further insight into the pathophysiology of ischemic necrosis in the immature femoral head and also the temporal evolution of histopathological changes over time has been gained using an experimental model of ischemic necrosis (piglet model) ( Fig. 1 ). After the induction of ischemic necrosis in immature pigs by ligating the femoral neck blood vessels, the earliest histologic changes are seen in the marrow space with diffuse cell death, disorganization of the marrow stroma, and the loss of osteoblasts lining the trabecular bone. The mechanisms of cell death appear to involve apoptosis and necrosis. Empty lacunae, a classic feature of osteonecrosis, are seen in the necrotic bone within a few weeks.

Fig. 1

Pathologic changes in the immature femoral head following ischemic necrosis. Ischemic injury produces extensive cell death in the bony epiphysis (osteonecrosis) and the deep layer of the articular cartilage (chondronecrosis). The deep layer of the cartilage represents the growth cartilage responsible for the circumferential growth of the secondary center of ossification. The ischemic damage to the deep layer of the cartilage produces a growth arrest of the secondary center. Ischemic necrosis is also associated with an increased calcium content of the necrotic bone, which is thought to make the bone more brittle and more prone to microdamage accumulation with hip joint loading. Revascularization of the infarcted femoral head is associated with a predominance of osteoclast-mediated resorption and a delayed bone formation, which further contribute to the development of the femoral head deformity.

( Courtesy of Texas Scottish Rite Hospital for Children; with permission.)

Along with the necrosis of the bony epiphysis, cell death is observed in the deep layer the articular cartilage, which is similar to the findings in cartilage from histopathological studies of LCPD. The reason for the cell death is that the articular cartilage of the immature femoral head is thicker than that of the adult articular cartilage, and its deep layer is dependent on the subchondral vascularity for nutritional support. Because of this dependence, a loss of blood supply to the immature femoral head not only produces a necrotic damage to the bony epiphysis but also damage to the deep layer of the articular cartilage surrounding the secondary center of ossification. The consequence is a growth arrest of the secondary center, which is an early radiographic feature of the piglet model and LCPD. One significant implication of this damage is a potential for growth disturbance of the secondary center. Unless the restoration of growth of the secondary center is symmetric, restoring a spherical growth, a further deformity of the femoral head can ensue with asymmetric restoration of the growth. This process occurs in addition to the potential growth disturbance of the metaphyseal physis, which is responsible for the length and the alignment of the femoral neck (coxa breva, coxa vara, or coxa valga). In the piglet model, disruption of blood flow to the epiphysis produced limited damage to the metaphyseal physis. In most of the animals the growth plate continued to function, albeit at a slower rate than the growth plate on the normal side, indicating that femoral head ischemia does not necessarily produce a growth arrest of the metaphyseal physis.

Revascularization and healing of the necrotic femoral head in immature pigs come in the form of fibrovascular tissue invasion of the necrotic marrow space and resorption of the necrotic bone. The new vessels arise from the existing vessels on the femoral neck, which invade through the periphery of the cartilage to reach the necrotic secondary center. In general, the invading vessels do not cross the metaphyseal physis. The fibrovascular tissue consists of inflammatory, mesenchymal (fibroblast-like in appearance), osteoclasts, and endothelial cells. In the revascularized regions of the bone, increased osteoclast-mediated resorption is observed with a net loss of bone, due to the imbalance of bone resorption and formation. The necrotic bone is replaced by a fibrovascular tissue, and bone formation is noticeably absent in these areas. Radiographic and histologic findings in the piglet model at this stage of healing resemble the fragmentation stage in LCPD, at which resorptive changes in the necrotic bone predominate while reossification of the resorbed areas is delayed for many months.

Pathogenesis of Femoral Head Deformity

The pathogenesis of the femoral head deformity in the immature femoral head is complex. From a mechanical perspective, the infarcted femoral head begins to deform when the forces applied to the femoral head due to loading are greater than its ability to resist deformation ( Fig. 2 ). While it is clear that osteonecrosis induces mechanical weakening of the femoral head, the mechanisms involved with the process are only beginning to be elucidated. Experimental studies in a large animal model of ischemic necrosis found a significant decrease in the mechanical properties of the infarcted femoral head and its components (articular cartilage and bone) from an early stage of the model. These studies, along with recent study on the alteration of calcium content of the necrotic bone, suggest that the mechanical properties of the infarcted femoral head are compromised by various mechanisms at different stages of the disease. In the early stage (avascular stage), before the initiation of revascularization, increased calcium content of the necrotic bone is thought to increase the brittleness of the necrotic bone and make it more prone to microdamage. In the absence of bone cells, such as osteoblasts, osteocytes, and osteoclasts, due to necrosis, the microfractures remain undetected and unrepaired, and accumulate. It is proposed that the accumulation of microdamage incurred with normal activities compromise the mechanical properties of the femoral head in the early stages of LCPD, and lead to the development of a subchondral fracture and compression fracture of the superior region of the bony epiphysis.

Fig. 2

A line graph representing proposed mechanical changes in the necrotic femoral head. The extent of head involvement, the degree of imbalance between bone resorption and formation, the duration of healing, and the level of hip loading will likely affect the deformity. The potential to remodel the deformed head, as seen in young patients, will offset the deformity produced at the acute phase of the disease.

( From Kim HK. Legg-Calvé-Perthes Disease. J Am Acad Orthop Surg 2010;18:676–86; with permission. Copyright © 2010 American Academy of Orthopaedic Surgeons.)

Vascular invasion and subsequent resorption of the necrotic bone seen at the vascular stage of healing further compromise the mechanical properties of the infarcted head. Experimental studies show that the imbalance of bone resorption and formation with a predominance of resorption contributes significantly to the pathogenesis of the femoral head deformity. Experimental studies also show that inhibition of bone resorption using bisphosphonates or a RANKL (receptor activator of nuclear factor kappa-B ligand) inhibitor can improve the preservation of the trabecular bone and the femoral head shape. These findings support the hypothesis that osteoclast-mediated bone resorption is a significant contributor to the pathogenesis of femoral head deformity. Although there are a few studies on the use of bisphosphonate to treat femoral head osteonecrosis in children, the efficacy of this treatment for LCPD has not been investigated.

Because the hip joint is a major load-bearing joint and children with LCPD tend to be active, it is important to consider the potential role of hip joint loading on the development of the femoral head deformity. This area of LCPD research is one that has a paucity of data. Basic data such as the hip contact pressures associated with various activities of daily living in children are not available. In adults, a femoral head prosthesis equipped with a strain gauge and telemetric transmission capability has allowed a real-time collection of femoral head loading data after total hip replacement while the patients were performing various activities and positioning of the affected leg. The measurements indicated that substantial forces act on the femoral head with weight-bearing activities. Walking was associated with the hip contact pressure reaching about 2.5 times the body weight with each step. Running on a treadmill at a rate of 8 km per hour increased the contact pressure to about 4.5 times the body weight with each stride. Greater pressures are generated with faster speeds of walking or running. Some supine and prone activities were also associated with elevated hip contact pressures above the body weight. Direct measurements of the hip contact pressures are not available in children; however, children in general are more active and take more steps, taking on average 7500 steps per day. In a disease where femoral head deformity is produced because of mechanical weakening, avoidance of activities that generate high hip contact pressures would seem reasonable. While some retrospective studies support the role of non–weight bearing on protecting the femoral head from the deforming forces, the compliance and the true efficacy of this type of treatment remain unclear.

In contrast to the factors such as femoral head weakening and hip joint loading that may promote the development of femoral head deformity, the healing or remodeling potential related to the age of the patient seems to offset the deformity. Clinical studies show a better outcome in younger patients with LCPD (onset of disease before age 6 years) compared with older patients (onset of disease after age 8 years). The mechanisms responsible for the difference in healing are unclear. One important age-related factor to consider is that LCPD affects a wide age range of children (preschool to teenage years), which represents a growth period when significant changes in the femoral head anatomy, size, and vasculature are occurring. The bony epiphysis increases in size while the articular cartilage thickness and the growth potential of the bony epiphysis decrease with age. Other changes, such as the regression of cartilage canals (blood vessels within cartilage) and the changes in the vascular anatomy of the proximal femur, are also occurring. As these changes are taking place, the onset of the disease at different ages implies that the disease is affecting a femoral head that may have significantly different growth and remodeling potentials that could affect the outcome of the femoral head. Preliminary experimental studies suggest that the potential for femoral head healing may be greater in the younger animals than in older, immature animals, and that one of the mechanisms may be related to greater hypoxic and angiogenic repair responses generated in the femoral head cartilage of the younger animals.


The classifications for LCPD can be divided into the one that defines the stage of the disease and the ones used to prognosticate outcome. Waldenström’s radiographic classification defines 4 radiographic stages of LCPD during the active phase of the disease, termed the initial stage, fragmentation stage, reossification stage, and residual stage, according to the characteristic radiographic features of each stage ( Table 1 ). The duration of each stage is variable from one patient to another. What determines the duration of each stage and the total duration of the active phase is unknown. In general, older patients appear to have a longer duration than younger patients. According to one study, the fragmentation staging lasts about 1 year while the reossification stage lasts from 3 to 5 years.

Table 1

Waldenstrom’s radiographic stages of LCPD

Stages of LCPD Radiographic Features
Initial stage (stage of increased radiodensity) Smaller ossific nucleus
Increased radiodensity of the ossific nucleus
Widening of medial joint space
Subchondral fracture in some
Metaphyseal cyst in some
Mild flattening of the ossific nucleus
Fragmentation stage (resorptive stage) Appearance of radiolucencies in the ossific nucleus
Fragmented appearance of the ossific nucleus
Further flattening of the ossific nucleus
Further lateralization of the head
Demarcation of central radiodense fragment (“sequestrum”) from the medial and lateral pillars of the head in some versus whole head flattening in others
Some show minimal fragmentation and flattening
Reossification stage (healing stage) Appearance of new bone in the medial and lateral aspects of the femoral head
Central and anterior aspects of the head are last to reossify
Disappearance of radiodense fragment
Some show improvement of head shape while few show worsening of head shape
Residual stage (healed stage) Normal radiodensity of the femoral head
Shape of the femoral head may change until skeletal maturity

Three radiographic classification systems, namely the Catterall, Salter-Thompson, and lateral pillar, have been developed as prognosticators of outcome that are to be applied at the stage of fragmentation. The Salter-Thompson classification is a two-category system (group A or group B) based on the extent of subchondral fracture (crescent sign). In group A, less than half of the femoral head is involved whereas in group B, more than half of the femoral head is involved. Because the crescent sign can be observed at the initial or early fragmentation stage, it has the advantage of being applicable at an earlier time point than the other 2 systems, which are applicable at the stage of maximal fragmentation. Its application, however, is restricted by the absence of the crescent sign in many patients at the time of presentation and subsequent follow-up. The presence of the sign may also be transient and have a narrow window for detection.

The Catterall classification system is a 4-category system (groups I–IV), and the first to emphasize the relationship between the extent of head involvement and the outcome. The Catterall groups I, II, III, and IV represent 25%, 50%, 75%, and total head involvement, respectively. The system was developed to be applied during the fragmentation stage when the necrotic sequestrum becomes well demarcated from the viable segment of the femoral head. Along with his classification system, Catterall also described head-at-risk signs associated with a poor outcome discussed earlier. The major criticism for the Catterall classification system has been its poor interobserver reliability. Recently, a modified Catterall classification system with two categories (the groups I and II combined and the groups III and IV combined) has been shown to have a better interobserver reliability. Because the groups I and II are often associated with good outcome and the groups III and IV with poor outcome, the simplification seems reasonable.

The lateral pillar classification was originally designed as a 3-category system (groups A, B, or C) with a recent addition of group B/C border, making this a 4-category system ( Fig. 3 ). The system is based on the height of the lateral pillar, defined as the lateral 15% to 30% of the epiphysis. Group A represents no loss of the lateral pillar height, whereas group B represents less than 50% loss and group C more than 50% loss of the lateral pillar height. Because the lateral aspect of the femoral head is a site of new ossification, it does pose an uncertainty of what is actually being assessed anatomically in some cases: the collapsed lateral pillar or the new ossification in the lateral aspect of the femoral head. Regardless of this uncertainty, the classification system does reflect the extent of the femoral head deformity, and the 3-category lateral pillar classification has been shown to have better interobserver reliability than the Salter-Thompson and Catterall classification systems. It has also been reported to be a better predictor of a Stulberg radiographic outcome than the Catterall classification.

Fig. 3

Examples of lateral pillar classification. The upper left radiograph shows bilateral femoral head involvement with the right femoral head ( white arrow ) classified as a lateral pillar group A.

Because the Catterall and lateral pillar classification systems are applicable at the stage of maximal fragmentation when the femoral head deformity is at its peak, the timing poses a dilemma for those patients seen at the initial stage or at the early fragmentation stage when the femoral head cannot be correctly classified. Assignment of the lateral pillar classification based on the initial presenting radiographs was found to be unreliable in 92 of 275 hips (33%), as the hips showed worsening of the lateral pillar height over time ( Fig. 4 ). One treatment approach for the patients presenting in the early stages of the disease has been to wait and observe until the patient can be classified into one of the guarded prognostic groups, either Catterall groups III or IV or lateral pillar groups B, B/C, or C, before instituting a specific treatment. The idea of “wait-and-classify” is a concern for older patients (>8 years old), who are known to have a limited potential to remodel the deformed femoral head, as a significant deformity can develop during this wait and see period. An argument for the “wait-and-classify” approach is that it may prevent having to operate on patients who would not have needed surgery (Catterall group I or II or lateral pillar group A) or who would not have benefited from it (lateral pillar C). An argument against this approach is that the treatment should be instituted early in the older patients rather than waiting for the head to deform, because they do not have as much remodeling potential as the younger patients. This controversy highlights the limitations of these radiographic classification systems, which cannot be applied in the early stages of LCPD prior to the development of the femoral head deformity. There is clearly a need to develop an earlier prognostic indicator based on a more sensitive imaging modality, such as a magnetic resonance image, which can be applied before the deformity develops to guide treatment from the initial stage of the disease. An additional limitation of the current classification systems is that they may not be relevant for older patients (>12 years), who tend to have poor prognosis regardless of the extent of head or lateral pillar involvement.

Fig. 4

Serial radiographs of a 7-year-old patient obtained at the time of presentation ( A ), at 3 months’ follow-up ( B ), and at 6 months’ follow-up ( C ). Application of the lateral pillar classification too early (at the initial or early fragmentation stage) would have led to an incorrect assignment of the lateral pillar group for this patient as B or B/C border at the 3-month follow-up ( B ). A subsequent follow-up at maximal fragmentation stage ( C ) showed that the affected hip was in the lateral pillar group C.

Only gold members can continue reading. Log In or Register to continue

Oct 6, 2017 | Posted by in ORTHOPEDIC | Comments Off on Pathophysiology, Classifications, and Natural History of Perthes Disease
Premium Wordpress Themes by UFO Themes