- Email: [email protected]
Pathophysiology and natural history of avascular necrosis of bone
Joint Bone Spine 73 (2006) 500–507 http://france.elsevier.com/direct/BONSOI/
Review
Pathophysiology and natural history of avascular necrosis of bone Pierre Lafforgue Service de Rhumatologie, CHU la Conception, 147, boulevard Baille, 13385 Marseille cedex 5, France Received 23 August 2005; accepted 30 January 2006 Available online 08 August 2006
Abstract Avascular necrosis of bone (AVN) occurs as two main variants, local and systemic. Local AVN is usually caused by trauma or microtrauma; examples include primary osteonecrosis of the medial condyle, vertebral osteonecrosis, necrosis after meniscectomy, and osteonecrosis of the mandible or small bones. Systemic AVN manifests as epiphyseal necrosis or bone infarction, which is often multifocal. Little is known about the factors that trigger AVN. One possible mechanism is intraluminal obliteration of blood vessels by microscopic fat emboli, sickle cells, nitrogen bubbles (caisson disease), or focal clotting due to procoagulant abnormalities. Extraluminal obliteration may result from elevated marrow pressure or increased marrow fat. Cytotoxicity and genetic factors may be involved also. Many factors are probably capable of inducing AVN, and combinations of factors may be required, although the final mechanism is always critical ischemia. The natural history of AVN is better understood than the early triggering factors. AVN becomes detectable 1–6 months after exposure to an easily identifiable risk factor such as high-dose glucocorticoid therapy or femoral neck fracture. Later on, AVN is uncommon even when the patient remains exposed to the risk factor. The turning point in the natural history of AVN is subchondral plate fracture, which leads to collapse of the necrotic segment of the epiphysis, usually within the first 2 years. The risk of collapse depends chiefly on the initial size and location of the necrotic segment, which can be determined accurately by magnetic resonance imaging (MRI). This natural history has obvious clinical implications. © 2006 Elsevier SAS. All rights reserved. Keywords: Aseptic necrosis; Avascular necrosis of bone
1. Introduction
2. Definition of avascular necrosis of bone
Knowledge of the events that contribute to the genesis and progression of avascular necrosis of bone (AVN) has direct therapeutic implications. Although epidemiological studies and the introduction of magnetic resonance imaging (MRI) have produced valuable insights, a number of uncertainties remain, most notably regarding the earliest pathogenic factors. The absence of animal models that replicate the human disease is a major obstacle to pathogenic studies. The objectives of this review of the pathophysiology of AVN are to define the nosological boundaries, to identify causative factors, and to describe the natural history of the disease.
The term “avascular necrosis of bone” is often used improperly to designate any bone lesion that contains some histological evidence of necrosis or that is misinterpreted as exhibiting imaging features of AVN. The images shown in several publications are highly suggestive of stress fractures. Elsewhere, AVN exists but is strictly local. This confusion has hindered the collection of sound scientific data.
E-mail address: [email protected] (P. Lafforgue). 1297-319X/$ - see front matter © 2006 Elsevier SAS. All rights reserved. doi:10.1016/j.jbspin.2006.01.025
2.1. The bone necrosis concept Bone tissue necrosis is a nonspecific abnormality that can occur whenever a disease process causes major cell stress. Thus, some degree of bone necrosis can be seen in severe osteoarthritis, fractures, tumors, infections, and other bone disorders. AVN, in contrast, is defined as massive necrosis of bone and bone marrow occurring as the only, or largely predominant, abnormality.
P. Lafforgue / Joint Bone Spine 73 (2006) 500–507
2.2. Local osteonecrozers versus systemic osteonecroses Bone necrosis may result from factors that cause local ischemia, such as trauma (fracture) or microtrauma. Although histology shows bone necrosis, the clinical setting and imaging study findings differ from those seen in systemic AVN. An example is idiopathic AVN of the medial condyle, which occurs in older individuals, probably as a complication of stress fracture. The imaging features are readily differentiated from those seen in systemic AVN. Similarly, AVN of a vertebral body probably results from mobility at the site of a severe fracture. Again, the imaging study findings and patient-related factors differ from those seen in systemic AVN. Although histology may show necrotic bone tissue, intravertebral vacuum phenomenon is the most characteristic finding [1] and is a less controversial concept than AVN of the vertebral body. AVN of the small carpal and tarsal bones or of the metacarpals is chiefly related to local ischemia caused by trauma or microtrauma. Osteochondritis is a term used to designate a heterogeneous group of disorders that involve local ischemia. However, osteochondritis cannot be equated with AVN. AVN of the mandible has been described recently after dental interventions in patients taking bisphosphonates [2]. Bone necrosis of unclear pathogenesis has been described after arthroscopic mechanical or laser meniscectomy. Finally, radiation-induced necrosis is a well-individualized disorder that can affect any tissue, including the bone. The term “avascular necrosis” generally refers to bone necrosis related to systemic factors, of which the paradigm is femoral head AVN. Features that contribute to make systemic AVN a well-individualized entity include the frequently multifocal distribution, risk factors (Table 1), imaging study findings, and outcomes. Although the underlying bone disease is systemic, the lesions develop in vulnerable areas such as longbone epiphyses (most notably the femoral heads, humeral heads, femoral condyles, and distal end of the tibia), longTable 1 Known risk factors for systemic avascular necrosis of bone Causes Glucocorticoid therapy Endogenous hypercorticism Organ transplant Systemic lupus erythematosus Antiphospholipid antibodies
Comments High doses (> 0.5 mg/kg) Rare GCs mediate much or all of the effect GCs mediate much or all of the effect Controversial++, may be mediated by severity of SLE and by GCs Alcohol abuse – Pregnancy Rare Dyslipidemia Hypertriglyceridemia (often combined with other risk factors) – Caisson disease – Sickle cell disease Homozygous sickle cell disease or sickle cell disease plus thalassemia Gaucher disease – HIV infection Unclear mechanism, role for concomitant risk factors? Idiopathic About 1/4 cases, males ++ GCs: glucocorticoids, HIV: human immunodeficiency virus infection; SLE: systemic lupus erythematosus.
501
bone metaphyses (classical bone infarction), and more rarely, short bones (chiefly the tarsal bones). AVN due to systemic factors is the focus of this review. 3. Early events: the genesis of avascular necrosis of bone (Fig. 1) Early events are incompletely understood. AVN is often diagnosed late, occasionally by routine follow-up investigations; and bone tissue is not readily accessible to sampling. As a result, pathogenic studies rely chiefly on epidemiological data and animal models. Although epidemiological studies may detect factors associated with AVN, they cannot prove causality. AVN is difficult to induce in animals. Aggressive methods must be used, such as total devascularization or massive glucocorticoid doses. Even then, the potential for healing is far greater in animals than in humans. None of the current hypotheses has been confidently established, although some are more plausible than others. Furthermore, several factors may act in combination to produce AVN [3]. The final mechanism is ischemia, which may be related to direct blood vessel injury (posttraumatic necrosis), intraluminal obliteration (vasculopathy), or extraluminal obliteration in the bone marrow. Direct toxic effects on bone marrow and bone cells may also contribute to AVN. 3.1. Mechanisms leading to intraluminal obliteration In fat embolism, a shower of microscopic fat emboli leads to critical ischemia, either directly or by triggering intravascular coagulation. The result is ischemic necrosis of vulnerable regions such as the epiphyses [4]. Fat emboli have been seen in the neighborhood of necrotic lesions. Glucocorticoid therapy and dyslipidemia may promote fat embolus formation. However, fat emboli are found also in healthy bone, and fat or cholesterol embolus formation does not consistently lead to AVN. In sickle cell crisis, sickle cell formation may lead to intraluminal vascular obliteration and bone necrosis. Dysbaric bone necrosis may be related to intravascular obliteration by nitrogen bubbles or microthrombi or to extraluminal obliteration by bone marrow gas or edema [5]. Hypercoagulability has been identified recently as a possible contributor to AVN. Procoagulant abnormalities have been found in patients with AVN or Legg Perthes disease [6,7]. A broad array of factors have been incriminated, including antiphospholipid antibodies (APL), abnormal plasminogen activator inhibitor activity, elevated homocysteine or lipoprotein(a), decreased protein S, resistance to activated protein C, and factor V Leiden mutations. However, few studies included adequate control groups. Furthermore, the well-known thrombophilias have not been identified as risk factors for AVN, and no increase in the thrombosis risk has been reported in patients with AVN. Epidemiological studies have shown that the association between APL and AVN is ascribable to the high rate of systemic lupus erythematosus and glucocorticoid therapy in
502
P. Lafforgue / Joint Bone Spine 73 (2006) 500–507
Fig. 1. Diagram of the pathophysiology of systemic avascular necrosis of bone.
patients with APL. Some studies failed to identify procoagulant factors in patients with AVN [8,9]. Finally, arterial abnormalities have been described, affecting the extraosseous [10] and above all intraosseous arteries [11, 12]. The role for these abnormalities is unclear.
patients with AVN [19–22] Thus, critical marrow pressure elevation may result from adipocyte hypertrophy (related to glucocorticoid therapy, alcohol abuse, or dyslipidemia), edema, nitrogen bubbles (caisson disease), proliferation of histiocytes in overload disorders (Gaucher disease), or bleeding within the bone marrow [23].
3.2. Mechanisms leading to extraluminal obliteration 3.3. Direct cell toxicity Because the bone is inelastic, compartment syndrome may develop when pressure increases within the bone marrow, leading to impaired blood flow in the intraosseous blood vessels. This hypothesis was developed based on a 1972 report by Arlet et al. [13] that pressure within the bone marrow was markedly elevated in patients with femoral head AVN. Subsequent studies confirmed that marrow pressure was elevated [14], even when measured before necrosis was detectable [15]. Further support for the compartment syndrome hypothesis comes from the intriguing results obtained by Wang et al. [16–18] in studies of glucocorticoid-induced experimental AVN. In rabbits and chickens, glucocorticoid administration leads to increased adipocyte size with a proportionate decrease in intraosseous blood flow, which can be corrected by core decompression or by fibrate or statin therapy. Fat conversion of the marrow has been documented in the proximal femur of
Agents that exert direct toxic effects on cells may contribute to AVN. Kawai et al. [24] reported gradual lipid accumulation within osteocytes followed by cell death in rabbits exposed to glucocorticoids. Increased osteocyte apoptosis was recently demonstrated in humans with AVN related to glucocorticoid therapy or alcohol abuse [25,26]. Similarly, decreases have been reported in the osteoblastic differentiation of marrow stem cells [27] and in the division potential of osteoblasts near AVN lesions [28]. 3.4. Genetic factors Recent evidence supports a role for genetic factors in some forms of AVN. In a study of 136 renal transplant recipients, the risk of AVN was closely dependent on the multidrug resistance
P. Lafforgue / Joint Bone Spine 73 (2006) 500–507
(MDR) gene, of which the ABCB1 C3435 TT haplotype was strongly protective (odds ratio, 0.10) [29]. This haplotype was associated with decreased intracellular concentrations of glucocorticoids. The ADH2*1 allele of hepatic alcohol dehydrogenase may diminish the risk of AVN related to alcohol abuse [30]. An abnormality in the gene encoding collagen Type I was identified in three families who had AVN with autosomal dominant inheritance [31]. Although none of these hypotheses is clearly superior over the others, the evidence is strongest for marrow fat conversion and adipocyte hypertrophy. Several mechanisms may be capable of leading to AVN and may be combined in the same patient. Thus, only 5% of rabbits exposed to glucocorticoids experience AVN when given warfarin and a lipid-lowering agent simultaneously, compared to 33% with warfarin alone, 38% with a lipid-lowering agent alone, and 70% with a placebo [32]. Improved knowledge of the early events involved in AVN may open up new avenues for preventive treatment. 4. Natural history of avascular necrosis of bone Knowledge of the natural history of AVN and the factors that influence it is valuable for several reasons: to serve as a basis for comparison when studying the effects of therapeutic interventions in nonrandomized studies; to inform patients of expected outcomes; and to determine which treatment strategy is best, most notably whether surgery is appropriate and which procedure is most likely to succeed. 4.1. The overall scenario AVN affects all the cell types in the bone and marrow (osteocytes, hematopoietic cells, and adipocytes). Except in sickle cell disease and Gaucher disease, necrosis develops at sites where the marrow is composed predominantly of adipocytes (yellow marrow), such as the femoral head. Marrow adipocytes have small nuclei and scant cytoplasm, being composed mainly of huge lipid vacuoles in which triglycerides predominate. Cell death has little effect on the vacuoles, which remain unchanged for some time, so that the MRI signal of necrotic bone is normal initially. MRI signal changes do not occur until the cell membranes rupture and the intravacuolar lipids undergo denaturation [33]. The cancellous trabecular network remains intact initially, and as a result radiographs are normal at first (Ficat and Arlet stages 0 and 1). After a few days or weeks, vascular connective tissue (granulation tissue) develops at the interface between healthy bone and necrotic bone and secondarily undergoes calcification. This active interface allows the early diagnosis by producing a hot spot on scintigrams, a characteristic linear MRI signal and, at a later stage, a rim of radiographic sclerosis (radiological stage 2). Although granulation tissue development is a physiological response that contains the lesion and promotes healing, it has limited effectiveness. Within the necrotic segment, the lipid vacuoles burst and their contents undergo sapo-
503
nification and calcification, leading to an MRI signal decrease in all sequences [33]. Neovascularization is inadequate and bone remodeling is both extremely limited and ill-adapted to local constraints [34]. The final result is persistence of a necrotic segment that has poor biomechanical properties. When mechanical constraints exceed bone resistance, stress fracture occurs, causing incapacitating pain and irreversible epiphyseal deformity (radiological stage 3, which may be preceded by a phase of subchondral dissection). This complication is not inevitable, however. Alteration of the spherical shape of the femoral head due to collapse of the necrotic segment induces undue mechanical stress on the cartilage, which almost consistently results in hip osteoarthritis (radiological stage 4). 4.2. Time to necrosis development The date of AVN development is usually unknown, since there is no pain initially. However, the exact date of onset can be determined when the cause is trauma, barotrauma, or glucocorticoid exposure. AVN of the femoral head is usually considered a late complication. However, “late-diagnosis complication” is a more appropriate designation, since necrosis occurs immediately or shortly after exposure to the risk factor. Prospective MRI studies of patients with femoral neck fractures showed evidence of femoral head AVN in 39% of cases (whereas radiographic changes occurred in 25% of cases), all of which were diagnosed 1–6 months after the fracture [35,36]. In caisson disease, most cases of AVN develop at sites where acute manifestations (the bends) occurred at decompression [5]. AVN can develop after high-dose glucocorticoid therapy, however brief, whereas low-dose long-term glucocorticoid therapy does not cause AVN [37]. Fleeting joint pain may occur at the initiation of high-dose glucocorticoid therapy at sites where necrosis is diagnosed later on [38]. Available data indicate that AVN can be triggered, often at multiple sites, by a brief course of high-dose glucocorticoid therapy. In prospective MRI studies of patients given glucocorticoid therapy for connective tissue disease or renal transplantation [39–44], the diagnosis was made within 2–6 months after treatment onset in the overwhelming majority of cases and within 1 month in some patients. A single case diagnosed after 1 year has been reported [43]; no other cases are known to have been diagnosed after 6 months. Early serial MRI follow-up of 57 renal transplant recipients showed femoral head AVN in 12 patients. The typical lucent line was visible within 6–12 weeks on conventional spin-echo images and even earlier on STIR sequences [44]. The earliest change was the classic rim around the necrotic segment. Edema of the underlying marrow became visible later on. Similarly, in animal models AVN was detected about 6–12 weeks postexposure with conventional MRI and after 72 hours with gadolinium-enhanced MRI [45]. Taken in concert, available data indicate that AVN occurs very early after high-dose glucocorticoid treatment. On the other hand, when AVN is not detected after 6 months, the risk of later detection is extremely small, even when glucocorticoid therapy is continued. This time pattern suggests a key
504
P. Lafforgue / Joint Bone Spine 73 (2006) 500–507
role for individual susceptibility factors. It should be kept in mind that studies of AVN onset used prospective serial MRI. Only a minority of the cases detected using this method become symptomatic or manifest as radiographic alterations during follow-up. 4.3. Can avascular necrosis of bone recur? Nontrauma-related AVN is bilateral at diagnosis in about 50–60% of cases overall and in up to 75–95% of cases induced by glucocorticoid treatment. However, when the lesion is unilateral at diagnosis, the other side remains unaffected, even when risk factor exposure persists [46]. A histological study of 606 necrotic femoral heads removed during surgery showed only 2 (0.3%) cases with lesions of different ages [47]. However, three cases each with evidence of two attacks of AVN have been reported [48]. In sum, as a rule, all sites of AVN develop within 6 months after risk factor exposure. Subsequent recurrences are exceedingly rare, even in patients taking long-term glucocorticoid therapy. 4.4. Can established avascular necrosis of bone decrease or increase in size? This point remains controversial. Spontaneous improvement of radiologically visible AVN is exceedingly rare. Few studies have evaluated changes in the size of the necrotic segment over time. In patients with femoral head AVN, MRI showed a moderate decrease in size (not due to femoral head collapse) in 23– 45% of cases [39,40,42]. In a cohort of 104 renal transplant recipients, Kopecky et al. [41] noted MRI signs of femoral head AVN in 25 hips (14 patients); during follow-up, MRI abnormalities improved in seven hips, returning to normal in six hips. This finding flies in the face of clinical experience and of the classic concept that necrotic tissue remains unchanged over time. In other studies, the size of the necrosis remained unchanged [49,50]. None of these studies were specifically designed to evaluate changes in size over time. Furthermore, the reports contain limited details about the method used to determine lesion size, no information on reproducibility, and no numerical data on measured values or variation magnitude. We are not aware of reports describing an increase in the size of the necrotic segment. Taken in concert, available data indicate that lesion size is usually stable over time, although a modest decrease may occur spontaneously. Although complete resolution of imaging study abnormalities may indicate regression of the lesion, misdiagnosis is a more likely explanation. Subchondral stress fractures of the femoral head occur predominantly in transplant recipients and patients with glucocorticoid-induced osteoporosis, two populations also at risk for AVN. These recently described fractures are extremely difficult to distinguish from femoral head AVN.
4.5. Risk and timing of radiological progression Disruption of the femoral head contour by collapse of the necrotic segment (Ficat and Arlet stage 3, Steinberg stage 4) is considered the turning point in the natural history of femoral head AVN. Collapse is irreversible, strongly associated with pain, and the major determinant of progression to hip osteoarthritis. In addition, collapse is an easily assessable objective sign, in contrast to pain. In most of the studies of outcomes in patients with femoral head AVN, femoral head collapse was the endpoint used to indicate a poor outcome. In early case-series, the diagnosis of femoral head AVN was based on strongly suggestive radiographic changes or, in patients with normal radiographs, on radionuclide bone scan results, pressure measurements in the bone marrow (which may have led to inclusion of patients with other disorders), or histology of core decompression specimens (although in this situation outcomes do not reflect the natural history of the disease). Since 1990, several prospective studies have provided information on the natural history of femoral head AVN seen at an early stage and diagnosed by MRI [49–54]. The rate of femoral head collapse varied across studies from 32% to 79%, probably because of differences in study populations; the overall average was about 50%. All the patients were free of radiological deformities at inclusion; and most of them had stage 0 diseases, with no symptoms or radiological abnormalities at baseline. These characteristics explain the improved outcomes compared to earlier studies. In a systematic review that mainly included pre-MRI-era studies, Mont and Hungerford [55] noted that the spontaneous outcome was unfavorable in 78% of cases. There is general agreement that the time to femoral head collapse is usually less than 2 years after the diagnosis of AVN. Beyond 3 years, the risk of radiological deterioration is virtually nonexistent. However, in a long-term follow-up study of patients with necrosis involving less than 10% of the femoral head, Hernigou et al. [56] found high rates of delayed pain and collapse, after a mean follow-up of 92 months. Should these data be confirmed, then collapse would appear to be inevitable in the long-term. Three factors may influence the natural history of femoral head AVN. First, the outcome is more likely to be favorable in patients with no radiological abnormalities at baseline than in those with stage 2 disease. Second, there is a consensus that the size of the necrotic segment is the main determinant of collapse. However, methods for measuring the size of the necrosis are nearly as numerous as papers on the topic, so that some doubt remains about the cutoffs that can be used to predict the prognosis. Roughly, involvement of less than 10% of the femoral head carries an excellent prognosis and involvement of more than 25% a poor prognosis. Third, the proportion of the weight-bearing area affected by necrosis may play a key role in the outcome, most notably when the necrosis is small or intermediate in size. The most widely used measurement method consists in determining the proportion of the acetabular roof
P. Lafforgue / Joint Bone Spine 73 (2006) 500–507
that is located opposite the necrotic segment in the coronal plane: less than 1/3 defines stage a, 1/3 to 2/3 stage b, and more than 2/3 stage c. The rate of progression to radiological stage 3 increases from 0% in stage a disease to 70% in stage c disease. A number of other factors may influence the outcome of femoral head AVN. Intuitively, pain would seem to predict a poor outcome, even when the radiological abnormalities are minimal. The prognostic significance of pain has not been established, however. MRI evidence of marrow edema under the necrotic segment and/or joint effusion is very significantly associated with pain and worse radiological stages [57]. These findings may also predict femoral head collapse. Finally, AVN related to glucocorticoid or alcohol exposure carries a poor prognosis, which is chiefly related to the often multiple and extensive foci. 4.6. After collapse occurs, is progression to pain and functional disability inevitable? Flattening of the femoral head related to collapse of the necrotic segment (Ficat and Arlet stage 3, Steinberg stage 4) is almost consistently followed by progressive joint damage that usually requires total hip replacement. However, the spontaneous clinical outcome may be acceptable. High rates of clinical and radiological stabilization have been reported in patients with limited collapse (less than 2 mm) or limited necrosis [58,59]. 5. Practical implications Genetic studies provide hope that tools for identifying highrisk patients will be available in the future. Such tools may prove useful for evaluating patients before starting high-dose glucocorticoid therapy. At present, caution is in order when considering glucocorticoid therapy or determining whether to allow deep sea diving in patients who have additional risk factors such as dyslipidemia or alcohol abuse. On the opposite, the risk of AVN is virtually nil after 6–12 months on glucocorticoid therapy (proving a posteriori that the patient had a low risk of AVN). Data on the role for lipid metabolism disorders or clotting disorders in the genesis of AVN may lead to preventive strategies. Animal studies conducted by Wang et al. [16–18] suggest a protective effect of lipid-lowering agents. Of 284 patients who were given glucocorticoid therapy while on long-term statin therapy, only 1% experienced femoral head AVN during the 7.5-year follow-up [60]. This rate was lower than expected, although there was no control group. Thus, statins may decrease the risk of glucocorticoid-induced AVN. In one study, long-term therapy with low-molecular-weight heparin decreased the risk of progression of stage 1 femoral head necrosis in patients with idiopathic AVN associated with clotting disorders [61]. This finding requires confirmation by further studies.
505
In medicolegal evaluations of patients with AVN, absence of MRI evidence of necrosis 6 months after exposure to a risk factor (e.g. trauma, diving, or glucocorticoid therapy) rules out a role for the risk factor. When MRI documentation is not available, an interval longer than 3 years between risk factor exposure and femoral head collapse militates against a relation between the two events. Importantly, data are now available for predicting the outcome of early-stage AVN. AVN involving more than 25% of the femoral head or more than 2/3 of the weight-bearing portion carries a very high-risk of collapse. In this situation, core decompression is consistently ineffective and primary hip replacement should be considered. Necrosis involving less than 10% of the femoral head or less than 1/3 of the weightbearing portion is likely to have a favorable outcome without treatment. In this situation, the appropriateness of any therapeutic intervention is debatable. The optimal management strategy is more difficult to determine in intermediate situations, which may constitute the best indication for core decompression. Finally, except when AVN is caused by local factors (Table 1), the disease potentially involves the entire skeleton. When a focus of AVN is identified, radiographs, radionuclide scanning, or MRI should be used to investigate other high-risk sites (hips, shoulders, and knees). References [1] Armingeat T, Pham T, Legre V, Lafforgue P. Coexistence of intravertebral vacuum and intradiscal vacuum. Joint Bone Spine 2006;73:428–32. [2] Naveau A, Naveau B. Osteonecrosis of the jaw in patients taking bisphosphonates. Joint Bone Spine 2006;73:7–9. [3] Kenzora JE, Glimcher MJ. Accumulative cell stress: the multifactorial etiology of idiopathic osteonecrosis. Orthop Clin North Am 1985;16: 669–79. [4] Jones Jr. JP. Fat embolism and osteonecrosis. Orthop Clin North Am 1985;16:595–633. [5] Hutter CD. Dysbaric osteonecrosis: a reassessment and hypothesis. Med Hypotheses 2000;54:585–90. [6] Jones LC, Mont MA, Le TB, Petri M, Hungerford DS, Wang P, et al. Procoagulants and osteonecrosis. J Rheumatol 2003;30:783–91. [7] Glueck CJ, Freiberg RA, Fontaine RN, Tracy T, Wang P. Hypofibrinolysis, thrombophilia, osteonecrosis. Clin Orthop Relat Res 2001;386:19– 33. [8] Lee JS, Koo KH, Ha YC, Koh KK, Kim SJ, Kim JR, et al. Role of thrombotic and fibrinolytic disorders in osteonecrosis of the femoral head. Clin Orthop Relat Res 2003;417:270–6. [9] Koo KH, Song HR, Ha YC, Kim JR, Kim SJ, Kim KI, et al. Role of thrombotic and fibrinolytic disorders in the etiology of Perthes’ disease. Clin Orthop Relat Res 2002;399:162–7. [10] Atsumi T, Kuroki Y, Yamano K. A microangiographic study of idiopathic osteonecrosis of the femoral head. Clin Orthop Relat Res 1989; 246:186–94. [11] Ohzono K, Takaoka K, Saito S, Saito M, Matsui M, Ono K. Intraosseous arterial architecture in nontraumatic avascular necrosis of the femoral head. Microangiographic and histologic study. Clin Orthop Relat Res 1992;277:79–88. [12] Arlet J, Laroche M, Soler R, Thiechart M, Pieraggi MT, Mazieres B. Histopathology of the vessels of the femoral heads in specimens of osteonecrosis, osteoarthritis and algodystrophy. Clin Rheumatol 1993;12:162– 5.
506
P. Lafforgue / Joint Bone Spine 73 (2006) 500–507
[13] Arlet J, Ficat P, Lartigue G, Tran MA. Recherches cliniques sur la pression intra-osseuse dans la metaphyse et l’epiphyse femorales superieures chez l’homme. Application du diagnostic des ischemies et necroses. Rev Rhum Mal Osteoartic 1972;39:717–23. [14] Ficat RP. Idiopathic bone necrosis of the femoral head. Early diagnosis and treatment. J Bone Joint Surg Br 1985;67:3–9. [15] Zizic TM, Lewis CG, Marcoux C, Hungerford DS. The predictive value of hemodynamic studies in preclinical ischemic necrosis of bone. J Rheumatol 1989;16:1559–64. [16] Wang GJ, Dughman SS, Reger SI, Stamp WG. The effect of core decompression on femoral head blood flow in steroid-induced avascular necrosis of the femoral head. J Bone Joint Surg Am 1985;67:121–4. [17] Cui Q, Wang GJ, Su CC, Balian G. The Otto Aufranc Award. Lovastatin prevents steroid induced adipogenesis and osteonecrosis. Clin Orthop Relat Res 1997;344:8–19. [18] Wang GJ, Moga DB, Richemer WG, Sweet DE, Reger SI, Thompson RC. Cortisone induced bone changes and its response to lipid clearing agents. Clin Orthop Relat Res 1978;130:81–5. [19] Mitchell DG, Rao VM, Dalinka M, Spritzer CE, Axel L, Gefter W, et al. Hematopoietic and fatty bone marrow distribution in the normal and ischemic hip: new observations with 1.5-T MR imaging. Radiology 1986;161:199–202. [20] Koo KH, Dussault RG, Kaplan PA, Ahn IO, Kim R, Devine MJ, et al. Fatty marrow conversion of the proximal femoral metaphysis in osteonecrotic hips. Clin Orthop Relat Res 1999;361:159–67. [21] Vande Berg BC, Malghem J, Lecouvet FE, Devogelaer JP, Maldague B, Houssiau FA. Fat conversion of femoral marrow in glucocorticoid-treated patients: a cross-sectional and longitudinal study with magnetic resonance imaging. Arthritis Rheum 1999;42:1405–11. [22] Motomura G, Yamamoto T, Miyanishi K, Yamashita A, Sueishi K, Iwamoto Y. Bone marrow fat-cell enlargement in early steroid-induced osteonecrosis—a histomorphometric study of autopsy cases. Pathol Res Pract 2005;200:807–11. [23] Saito S, Inoue A, Ono K. Intramedullary haemorrhage as a possible cause of avascular necrosis of the femoral head. The histology of 16 femoral heads at the silent stage. J Bone Joint Surg Br 1987;69:346–51. [24] Kawai K, Tamaki A, Hirohata K. Steroid-induced accumulation of lipid in the osteocytes of the rabbit femoral head. A histochemical and electron microscopic study. J Bone Joint Surg Am 1985;67:755–63. [25] Weinstein RS, Nicholas RW, Manolagas SC. Apoptosis of osteocytes in glucocorticoid-induced osteonecrosis of the hip. J Clin Endocrinol Metab 2000;85:2907–12. [26] Calder JD, Buttery L, Revell PA, Pearse M, Polak JM. Apoptosis—a significant cause of bone cell death in osteonecrosis of the femoral head. J Bone Joint Surg Br 2004;86:1209–13. [27] Suh KT, Kim SW, Roh HL, Youn MS, Jung JS. Decreased osteogenic differentiation of mesenchymal stem cells in alcohol-induced osteonecrosis. Clin Orthop Relat Res 2005;431:220–5. [28] Gangji V, Hauzeur JP, Schoutens A, Hinsenkamp M, Appelboom T, Egrise D. Abnormalities in the replicative capacity of osteoblastic cells in the proximal femur of patients with osteonecrosis of the femoral head. J Rheumatol 2003;30:348–51. [29] Asano T, Takahashi KA, Fujioka M, Inoue S, Okamoto M, Sugioka N, et al. ABCB1 C3435T and G2677T/A polymorphism decreased the risk for steroid-induced osteonecrosis of the femoral head after kidney transplantation. Pharmacogenetics 2003;13:675–82. [30] Chao YC, Wang SJ, Chu HC, Chang WK, Hsieh TY. Investigation of alcohol metabolizing enzyme genes in Chinese alcoholics with avascular necrosis of hip joint, pancreatitis and cirrhosis of the liver. Alcohol Alcohol 2003;38:431–6. [31] Liu YF, Chen WM, Lin YF, Yang RC, Lin MW, Li LH, et al. Type II collagen gene variants and inherited osteonecrosis of the femoral head. N Engl J Med 2005;352:2294–301. [32] Motomura G, Yamamoto T, Miyanishi K, Jingushi S, Iwamoto Y. Combined effects of an anticoagulant and a lipid-lowering agent on the prevention of steroid-induced osteonecrosis in rabbits. Arthritis Rheum 2004;50:3387–91.
[33] Hernigou P, Voisin MC, Marichez M, Despres E, Goutallier D. Confrontation de l’imagerie par resonance magnétique nucleaire et de l’histologie dans les necroses des tetes femorales. Rev Rhum Mal Osteoartic 1989; 56:741–4. [34] Glimcher MJ, Kenzora JE. The biology of osteonecrosis of the human femoral head and its clinical implications: II. The pathological changes in the femoral head as an organ and in the hip joint. Clin Orthop Relat Res 1979;139:283–312. [35] Kawasaki M, Hasegawa Y, Sakano S, Sugiyama H, Tajima T, Iwasada S, et al. Prediction of osteonecrosis by magnetic resonance imaging after femoral neck fractures. Clin Orthop Relat Res 2001;385:157–64. [36] Sugano N, Masuhara K, Nakamura N, Ochi T, Hirooka A, Hayami Y. MRI of early osteonecrosis of the femoral head after transcervical fracture. J Bone Joint Surg Br 1996;78:253–7. [37] Taylor LJ. Multifocal avascular necrosis after short-term high-dose steroid therapy. A report of three cases. J Bone Joint Surg Br 1984;66:431– 3. [38] Paolaggi JB, Le Parc JM, Arfi S. Severe arthralgias after pulse corticosteroid therapy in transplant patients. J Rheumatol 1987;14:1077–8. [39] Sakamoto M, Shimizu K, Iida S, Akita T, Moriya H, Nawata Y. Osteonecrosis of the femoral head: a prospective study with MRI. J Bone Joint Surg Br 1997;79:213–9. [40] Oinuma K, Harada Y, Nawata Y, Takabayashi K, Abe I, Kamikawa K, et al. Osteonecrosis in patients with systemic lupus erythematosus develops very early after starting high dose corticosteroid treatment. Ann Rheum Dis 2001;60:1145–8. [41] Kopecky KK, Braunstein EM, Brandt KD, Filo RS, Leapman SB, Capello WN, et al. Apparent avascular necrosis of the hip: appearance and spontaneous resolution of MR findings in renal allograft recipients. Radiology 1991;179:523–7. [42] Le Parc JM, Andre T, Helenon O, Benoit J, Paolaggi JB, Kreis H. Osteonecrosis of the hip in renal transplant recipients. Changes in functional status and magnetic resonance imaging findings over 3 years in three hundred five patients. Rev Rhum 1996;63:413–20 [Engl Ed]. [43] Kubo T, Yamazoe S, Sugano N, Fujioka M, Naruse S, Yoshimura N, et al. Initial MRI findings of non-traumatic osteonecrosis of the femoral head in renal allograft recipients. Magn Reson Imaging 1997;15:1017– 23. [44] Fujioka M, Kubo T, Nakamura F, Shibatani M, Ueshima K, Hamaguchi H, et al. Initial changes of non-traumatic osteonecrosis of femoral head in fat suppression images: bone marrow edema was not found before the appearance of band patterns. Magn Reson Imaging 2001;19:985–91. [45] Sakaia T, Sugano N, Tsuji T, Nishii T, Yoshikawa H, Ohzono K. Serial magnetic resonance imaging in a non-traumatic rabbit osteonecrosis model: an experimental longitudinal study. Magn Reson Imaging 2000; 18:897–905. [46] Sugano N, Nishii T, Shibuya T, Nakata K, Masuhara K, Takaoka K. Contralateral hip in patients with unilateral nontraumatic osteonecrosis of the femoral head. Clin Orthop Relat Res 1997;334:85–90. [47] Yamamoto T, DiCarlo EF, Bullough PG. The prevalence and clinicopathological appearance of extension of osteonecrosis in the femoral head. J Bone Joint Surg Br 1999;81:328–32. [48] Kim YM, Rhyu KH, Lee SH, Kim HJ. Can osteonecrosis of the femoral head be recurrent? Clin Orthop Relat Res 2003;406:123–8. [49] Hernigou P, Lambotte JC. Bilateral hip osteonecrosis: influence of hip size on outcome. Ann Rheum Dis 2000;59:817–21. [50] Shimizu K, Moriya H, Akita T, Sakamoto M, Suguro T. Prediction of collapse with magnetic resonance imaging of avascular necrosis of the femoral head. J Bone Joint Surg Am 1994;76:215–23. [51] Koo KH, Kim R, Ko GH, Song HR, Jeong ST, Cho SH. Preventing collapse in early osteonecrosis of the femoral head. A randomised clinical trial of core decompression. J Bone Joint Surg Br 1995;77:870–4. [52] Lafforgue P, Dahan E, Chagnaud C, Schiano A, Kasbarian M, Acquaviva PC. Early-stage avascular necrosis of the femoral head: MR imaging for prognosis in 31 cases with at least 2 years of follow-up. Radiology 1993;187:199–204.
P. Lafforgue / Joint Bone Spine 73 (2006) 500–507 [53] Nishii T, Sugano N, Ohzono K, Sakai T, Sato Y, Yoshikawa H. Significance of lesion size and location in the prediction of collapse of osteonecrosis of the femoral head: a new three-dimensional quantification using magnetic resonance imaging. J Orthop Res 2002;20:130–6. [54] Takatori Y, Kokubo T, Ninomiya S, Nakamura S, Morimoto S, Kusaba I. Avascular necrosis of the femoral head. Natural history and magnetic resonance imaging. J Bone Joint Surg Br 1993;75:217–21. [55] Mont MA, Hungerford DS. Non-traumatic avascular necrosis of the femoral head. J Bone Joint Surg Am 1995;77:459–74. [56] Hernigou P, Poignard A, Nogier A, Manicom O. Fate of very small asymptomatic stage-I osteonecrotic lesions of the hip. J Bone Joint Surg Am 2004;86-A:2589–93. [57] Iida S, Harada Y, Shimizu K, Sakamoto M, Ikenoue S, Akita T, et al. Correlation between bone marrow edema and collapse of the femoral
507
head in steroid-induced osteonecrosis. AJR Am J Roentgenol 2000;174: 735–43. [58] Churchill MA, Spencer JD. End-stage avascular necrosis of bone in renal transplant patients. The natural history. J Bone Joint Surg Br 1991;73: 618–20. [59] Nishii T, Sugano N, Ohzono K, Sakai T, Haraguchi K, Yoshikawa H. Progression and cessation of collapse in osteonecrosis of the femoral head. Clin Orthop Relat Res 2002;400:149–57. [60] Pritchett JW. Statin therapy decreases the risk of osteonecrosis in patients receiving steroids. Clin Orthop Relat Res 2001;386:173–8. [61] Glueck CJ, Freiberg RA, Sieve L, Wang P. Enoxaparin prevents progression of stages I and II osteonecrosis of the hip. Clin Orthop Relat Res 2005;435:164–70.
Review
Pathophysiology and natural history of avascular necrosis of bone Pierre Lafforgue Service de Rhumatologie, CHU la Conception, 147, boulevard Baille, 13385 Marseille cedex 5, France Received 23 August 2005; accepted 30 January 2006 Available online 08 August 2006
Abstract Avascular necrosis of bone (AVN) occurs as two main variants, local and systemic. Local AVN is usually caused by trauma or microtrauma; examples include primary osteonecrosis of the medial condyle, vertebral osteonecrosis, necrosis after meniscectomy, and osteonecrosis of the mandible or small bones. Systemic AVN manifests as epiphyseal necrosis or bone infarction, which is often multifocal. Little is known about the factors that trigger AVN. One possible mechanism is intraluminal obliteration of blood vessels by microscopic fat emboli, sickle cells, nitrogen bubbles (caisson disease), or focal clotting due to procoagulant abnormalities. Extraluminal obliteration may result from elevated marrow pressure or increased marrow fat. Cytotoxicity and genetic factors may be involved also. Many factors are probably capable of inducing AVN, and combinations of factors may be required, although the final mechanism is always critical ischemia. The natural history of AVN is better understood than the early triggering factors. AVN becomes detectable 1–6 months after exposure to an easily identifiable risk factor such as high-dose glucocorticoid therapy or femoral neck fracture. Later on, AVN is uncommon even when the patient remains exposed to the risk factor. The turning point in the natural history of AVN is subchondral plate fracture, which leads to collapse of the necrotic segment of the epiphysis, usually within the first 2 years. The risk of collapse depends chiefly on the initial size and location of the necrotic segment, which can be determined accurately by magnetic resonance imaging (MRI). This natural history has obvious clinical implications. © 2006 Elsevier SAS. All rights reserved. Keywords: Aseptic necrosis; Avascular necrosis of bone
1. Introduction
2. Definition of avascular necrosis of bone
Knowledge of the events that contribute to the genesis and progression of avascular necrosis of bone (AVN) has direct therapeutic implications. Although epidemiological studies and the introduction of magnetic resonance imaging (MRI) have produced valuable insights, a number of uncertainties remain, most notably regarding the earliest pathogenic factors. The absence of animal models that replicate the human disease is a major obstacle to pathogenic studies. The objectives of this review of the pathophysiology of AVN are to define the nosological boundaries, to identify causative factors, and to describe the natural history of the disease.
The term “avascular necrosis of bone” is often used improperly to designate any bone lesion that contains some histological evidence of necrosis or that is misinterpreted as exhibiting imaging features of AVN. The images shown in several publications are highly suggestive of stress fractures. Elsewhere, AVN exists but is strictly local. This confusion has hindered the collection of sound scientific data.
E-mail address: [email protected] (P. Lafforgue). 1297-319X/$ - see front matter © 2006 Elsevier SAS. All rights reserved. doi:10.1016/j.jbspin.2006.01.025
2.1. The bone necrosis concept Bone tissue necrosis is a nonspecific abnormality that can occur whenever a disease process causes major cell stress. Thus, some degree of bone necrosis can be seen in severe osteoarthritis, fractures, tumors, infections, and other bone disorders. AVN, in contrast, is defined as massive necrosis of bone and bone marrow occurring as the only, or largely predominant, abnormality.
P. Lafforgue / Joint Bone Spine 73 (2006) 500–507
2.2. Local osteonecrozers versus systemic osteonecroses Bone necrosis may result from factors that cause local ischemia, such as trauma (fracture) or microtrauma. Although histology shows bone necrosis, the clinical setting and imaging study findings differ from those seen in systemic AVN. An example is idiopathic AVN of the medial condyle, which occurs in older individuals, probably as a complication of stress fracture. The imaging features are readily differentiated from those seen in systemic AVN. Similarly, AVN of a vertebral body probably results from mobility at the site of a severe fracture. Again, the imaging study findings and patient-related factors differ from those seen in systemic AVN. Although histology may show necrotic bone tissue, intravertebral vacuum phenomenon is the most characteristic finding [1] and is a less controversial concept than AVN of the vertebral body. AVN of the small carpal and tarsal bones or of the metacarpals is chiefly related to local ischemia caused by trauma or microtrauma. Osteochondritis is a term used to designate a heterogeneous group of disorders that involve local ischemia. However, osteochondritis cannot be equated with AVN. AVN of the mandible has been described recently after dental interventions in patients taking bisphosphonates [2]. Bone necrosis of unclear pathogenesis has been described after arthroscopic mechanical or laser meniscectomy. Finally, radiation-induced necrosis is a well-individualized disorder that can affect any tissue, including the bone. The term “avascular necrosis” generally refers to bone necrosis related to systemic factors, of which the paradigm is femoral head AVN. Features that contribute to make systemic AVN a well-individualized entity include the frequently multifocal distribution, risk factors (Table 1), imaging study findings, and outcomes. Although the underlying bone disease is systemic, the lesions develop in vulnerable areas such as longbone epiphyses (most notably the femoral heads, humeral heads, femoral condyles, and distal end of the tibia), longTable 1 Known risk factors for systemic avascular necrosis of bone Causes Glucocorticoid therapy Endogenous hypercorticism Organ transplant Systemic lupus erythematosus Antiphospholipid antibodies
Comments High doses (> 0.5 mg/kg) Rare GCs mediate much or all of the effect GCs mediate much or all of the effect Controversial++, may be mediated by severity of SLE and by GCs Alcohol abuse – Pregnancy Rare Dyslipidemia Hypertriglyceridemia (often combined with other risk factors) – Caisson disease – Sickle cell disease Homozygous sickle cell disease or sickle cell disease plus thalassemia Gaucher disease – HIV infection Unclear mechanism, role for concomitant risk factors? Idiopathic About 1/4 cases, males ++ GCs: glucocorticoids, HIV: human immunodeficiency virus infection; SLE: systemic lupus erythematosus.
501
bone metaphyses (classical bone infarction), and more rarely, short bones (chiefly the tarsal bones). AVN due to systemic factors is the focus of this review. 3. Early events: the genesis of avascular necrosis of bone (Fig. 1) Early events are incompletely understood. AVN is often diagnosed late, occasionally by routine follow-up investigations; and bone tissue is not readily accessible to sampling. As a result, pathogenic studies rely chiefly on epidemiological data and animal models. Although epidemiological studies may detect factors associated with AVN, they cannot prove causality. AVN is difficult to induce in animals. Aggressive methods must be used, such as total devascularization or massive glucocorticoid doses. Even then, the potential for healing is far greater in animals than in humans. None of the current hypotheses has been confidently established, although some are more plausible than others. Furthermore, several factors may act in combination to produce AVN [3]. The final mechanism is ischemia, which may be related to direct blood vessel injury (posttraumatic necrosis), intraluminal obliteration (vasculopathy), or extraluminal obliteration in the bone marrow. Direct toxic effects on bone marrow and bone cells may also contribute to AVN. 3.1. Mechanisms leading to intraluminal obliteration In fat embolism, a shower of microscopic fat emboli leads to critical ischemia, either directly or by triggering intravascular coagulation. The result is ischemic necrosis of vulnerable regions such as the epiphyses [4]. Fat emboli have been seen in the neighborhood of necrotic lesions. Glucocorticoid therapy and dyslipidemia may promote fat embolus formation. However, fat emboli are found also in healthy bone, and fat or cholesterol embolus formation does not consistently lead to AVN. In sickle cell crisis, sickle cell formation may lead to intraluminal vascular obliteration and bone necrosis. Dysbaric bone necrosis may be related to intravascular obliteration by nitrogen bubbles or microthrombi or to extraluminal obliteration by bone marrow gas or edema [5]. Hypercoagulability has been identified recently as a possible contributor to AVN. Procoagulant abnormalities have been found in patients with AVN or Legg Perthes disease [6,7]. A broad array of factors have been incriminated, including antiphospholipid antibodies (APL), abnormal plasminogen activator inhibitor activity, elevated homocysteine or lipoprotein(a), decreased protein S, resistance to activated protein C, and factor V Leiden mutations. However, few studies included adequate control groups. Furthermore, the well-known thrombophilias have not been identified as risk factors for AVN, and no increase in the thrombosis risk has been reported in patients with AVN. Epidemiological studies have shown that the association between APL and AVN is ascribable to the high rate of systemic lupus erythematosus and glucocorticoid therapy in
502
P. Lafforgue / Joint Bone Spine 73 (2006) 500–507
Fig. 1. Diagram of the pathophysiology of systemic avascular necrosis of bone.
patients with APL. Some studies failed to identify procoagulant factors in patients with AVN [8,9]. Finally, arterial abnormalities have been described, affecting the extraosseous [10] and above all intraosseous arteries [11, 12]. The role for these abnormalities is unclear.
patients with AVN [19–22] Thus, critical marrow pressure elevation may result from adipocyte hypertrophy (related to glucocorticoid therapy, alcohol abuse, or dyslipidemia), edema, nitrogen bubbles (caisson disease), proliferation of histiocytes in overload disorders (Gaucher disease), or bleeding within the bone marrow [23].
3.2. Mechanisms leading to extraluminal obliteration 3.3. Direct cell toxicity Because the bone is inelastic, compartment syndrome may develop when pressure increases within the bone marrow, leading to impaired blood flow in the intraosseous blood vessels. This hypothesis was developed based on a 1972 report by Arlet et al. [13] that pressure within the bone marrow was markedly elevated in patients with femoral head AVN. Subsequent studies confirmed that marrow pressure was elevated [14], even when measured before necrosis was detectable [15]. Further support for the compartment syndrome hypothesis comes from the intriguing results obtained by Wang et al. [16–18] in studies of glucocorticoid-induced experimental AVN. In rabbits and chickens, glucocorticoid administration leads to increased adipocyte size with a proportionate decrease in intraosseous blood flow, which can be corrected by core decompression or by fibrate or statin therapy. Fat conversion of the marrow has been documented in the proximal femur of
Agents that exert direct toxic effects on cells may contribute to AVN. Kawai et al. [24] reported gradual lipid accumulation within osteocytes followed by cell death in rabbits exposed to glucocorticoids. Increased osteocyte apoptosis was recently demonstrated in humans with AVN related to glucocorticoid therapy or alcohol abuse [25,26]. Similarly, decreases have been reported in the osteoblastic differentiation of marrow stem cells [27] and in the division potential of osteoblasts near AVN lesions [28]. 3.4. Genetic factors Recent evidence supports a role for genetic factors in some forms of AVN. In a study of 136 renal transplant recipients, the risk of AVN was closely dependent on the multidrug resistance
P. Lafforgue / Joint Bone Spine 73 (2006) 500–507
(MDR) gene, of which the ABCB1 C3435 TT haplotype was strongly protective (odds ratio, 0.10) [29]. This haplotype was associated with decreased intracellular concentrations of glucocorticoids. The ADH2*1 allele of hepatic alcohol dehydrogenase may diminish the risk of AVN related to alcohol abuse [30]. An abnormality in the gene encoding collagen Type I was identified in three families who had AVN with autosomal dominant inheritance [31]. Although none of these hypotheses is clearly superior over the others, the evidence is strongest for marrow fat conversion and adipocyte hypertrophy. Several mechanisms may be capable of leading to AVN and may be combined in the same patient. Thus, only 5% of rabbits exposed to glucocorticoids experience AVN when given warfarin and a lipid-lowering agent simultaneously, compared to 33% with warfarin alone, 38% with a lipid-lowering agent alone, and 70% with a placebo [32]. Improved knowledge of the early events involved in AVN may open up new avenues for preventive treatment. 4. Natural history of avascular necrosis of bone Knowledge of the natural history of AVN and the factors that influence it is valuable for several reasons: to serve as a basis for comparison when studying the effects of therapeutic interventions in nonrandomized studies; to inform patients of expected outcomes; and to determine which treatment strategy is best, most notably whether surgery is appropriate and which procedure is most likely to succeed. 4.1. The overall scenario AVN affects all the cell types in the bone and marrow (osteocytes, hematopoietic cells, and adipocytes). Except in sickle cell disease and Gaucher disease, necrosis develops at sites where the marrow is composed predominantly of adipocytes (yellow marrow), such as the femoral head. Marrow adipocytes have small nuclei and scant cytoplasm, being composed mainly of huge lipid vacuoles in which triglycerides predominate. Cell death has little effect on the vacuoles, which remain unchanged for some time, so that the MRI signal of necrotic bone is normal initially. MRI signal changes do not occur until the cell membranes rupture and the intravacuolar lipids undergo denaturation [33]. The cancellous trabecular network remains intact initially, and as a result radiographs are normal at first (Ficat and Arlet stages 0 and 1). After a few days or weeks, vascular connective tissue (granulation tissue) develops at the interface between healthy bone and necrotic bone and secondarily undergoes calcification. This active interface allows the early diagnosis by producing a hot spot on scintigrams, a characteristic linear MRI signal and, at a later stage, a rim of radiographic sclerosis (radiological stage 2). Although granulation tissue development is a physiological response that contains the lesion and promotes healing, it has limited effectiveness. Within the necrotic segment, the lipid vacuoles burst and their contents undergo sapo-
503
nification and calcification, leading to an MRI signal decrease in all sequences [33]. Neovascularization is inadequate and bone remodeling is both extremely limited and ill-adapted to local constraints [34]. The final result is persistence of a necrotic segment that has poor biomechanical properties. When mechanical constraints exceed bone resistance, stress fracture occurs, causing incapacitating pain and irreversible epiphyseal deformity (radiological stage 3, which may be preceded by a phase of subchondral dissection). This complication is not inevitable, however. Alteration of the spherical shape of the femoral head due to collapse of the necrotic segment induces undue mechanical stress on the cartilage, which almost consistently results in hip osteoarthritis (radiological stage 4). 4.2. Time to necrosis development The date of AVN development is usually unknown, since there is no pain initially. However, the exact date of onset can be determined when the cause is trauma, barotrauma, or glucocorticoid exposure. AVN of the femoral head is usually considered a late complication. However, “late-diagnosis complication” is a more appropriate designation, since necrosis occurs immediately or shortly after exposure to the risk factor. Prospective MRI studies of patients with femoral neck fractures showed evidence of femoral head AVN in 39% of cases (whereas radiographic changes occurred in 25% of cases), all of which were diagnosed 1–6 months after the fracture [35,36]. In caisson disease, most cases of AVN develop at sites where acute manifestations (the bends) occurred at decompression [5]. AVN can develop after high-dose glucocorticoid therapy, however brief, whereas low-dose long-term glucocorticoid therapy does not cause AVN [37]. Fleeting joint pain may occur at the initiation of high-dose glucocorticoid therapy at sites where necrosis is diagnosed later on [38]. Available data indicate that AVN can be triggered, often at multiple sites, by a brief course of high-dose glucocorticoid therapy. In prospective MRI studies of patients given glucocorticoid therapy for connective tissue disease or renal transplantation [39–44], the diagnosis was made within 2–6 months after treatment onset in the overwhelming majority of cases and within 1 month in some patients. A single case diagnosed after 1 year has been reported [43]; no other cases are known to have been diagnosed after 6 months. Early serial MRI follow-up of 57 renal transplant recipients showed femoral head AVN in 12 patients. The typical lucent line was visible within 6–12 weeks on conventional spin-echo images and even earlier on STIR sequences [44]. The earliest change was the classic rim around the necrotic segment. Edema of the underlying marrow became visible later on. Similarly, in animal models AVN was detected about 6–12 weeks postexposure with conventional MRI and after 72 hours with gadolinium-enhanced MRI [45]. Taken in concert, available data indicate that AVN occurs very early after high-dose glucocorticoid treatment. On the other hand, when AVN is not detected after 6 months, the risk of later detection is extremely small, even when glucocorticoid therapy is continued. This time pattern suggests a key
504
P. Lafforgue / Joint Bone Spine 73 (2006) 500–507
role for individual susceptibility factors. It should be kept in mind that studies of AVN onset used prospective serial MRI. Only a minority of the cases detected using this method become symptomatic or manifest as radiographic alterations during follow-up. 4.3. Can avascular necrosis of bone recur? Nontrauma-related AVN is bilateral at diagnosis in about 50–60% of cases overall and in up to 75–95% of cases induced by glucocorticoid treatment. However, when the lesion is unilateral at diagnosis, the other side remains unaffected, even when risk factor exposure persists [46]. A histological study of 606 necrotic femoral heads removed during surgery showed only 2 (0.3%) cases with lesions of different ages [47]. However, three cases each with evidence of two attacks of AVN have been reported [48]. In sum, as a rule, all sites of AVN develop within 6 months after risk factor exposure. Subsequent recurrences are exceedingly rare, even in patients taking long-term glucocorticoid therapy. 4.4. Can established avascular necrosis of bone decrease or increase in size? This point remains controversial. Spontaneous improvement of radiologically visible AVN is exceedingly rare. Few studies have evaluated changes in the size of the necrotic segment over time. In patients with femoral head AVN, MRI showed a moderate decrease in size (not due to femoral head collapse) in 23– 45% of cases [39,40,42]. In a cohort of 104 renal transplant recipients, Kopecky et al. [41] noted MRI signs of femoral head AVN in 25 hips (14 patients); during follow-up, MRI abnormalities improved in seven hips, returning to normal in six hips. This finding flies in the face of clinical experience and of the classic concept that necrotic tissue remains unchanged over time. In other studies, the size of the necrosis remained unchanged [49,50]. None of these studies were specifically designed to evaluate changes in size over time. Furthermore, the reports contain limited details about the method used to determine lesion size, no information on reproducibility, and no numerical data on measured values or variation magnitude. We are not aware of reports describing an increase in the size of the necrotic segment. Taken in concert, available data indicate that lesion size is usually stable over time, although a modest decrease may occur spontaneously. Although complete resolution of imaging study abnormalities may indicate regression of the lesion, misdiagnosis is a more likely explanation. Subchondral stress fractures of the femoral head occur predominantly in transplant recipients and patients with glucocorticoid-induced osteoporosis, two populations also at risk for AVN. These recently described fractures are extremely difficult to distinguish from femoral head AVN.
4.5. Risk and timing of radiological progression Disruption of the femoral head contour by collapse of the necrotic segment (Ficat and Arlet stage 3, Steinberg stage 4) is considered the turning point in the natural history of femoral head AVN. Collapse is irreversible, strongly associated with pain, and the major determinant of progression to hip osteoarthritis. In addition, collapse is an easily assessable objective sign, in contrast to pain. In most of the studies of outcomes in patients with femoral head AVN, femoral head collapse was the endpoint used to indicate a poor outcome. In early case-series, the diagnosis of femoral head AVN was based on strongly suggestive radiographic changes or, in patients with normal radiographs, on radionuclide bone scan results, pressure measurements in the bone marrow (which may have led to inclusion of patients with other disorders), or histology of core decompression specimens (although in this situation outcomes do not reflect the natural history of the disease). Since 1990, several prospective studies have provided information on the natural history of femoral head AVN seen at an early stage and diagnosed by MRI [49–54]. The rate of femoral head collapse varied across studies from 32% to 79%, probably because of differences in study populations; the overall average was about 50%. All the patients were free of radiological deformities at inclusion; and most of them had stage 0 diseases, with no symptoms or radiological abnormalities at baseline. These characteristics explain the improved outcomes compared to earlier studies. In a systematic review that mainly included pre-MRI-era studies, Mont and Hungerford [55] noted that the spontaneous outcome was unfavorable in 78% of cases. There is general agreement that the time to femoral head collapse is usually less than 2 years after the diagnosis of AVN. Beyond 3 years, the risk of radiological deterioration is virtually nonexistent. However, in a long-term follow-up study of patients with necrosis involving less than 10% of the femoral head, Hernigou et al. [56] found high rates of delayed pain and collapse, after a mean follow-up of 92 months. Should these data be confirmed, then collapse would appear to be inevitable in the long-term. Three factors may influence the natural history of femoral head AVN. First, the outcome is more likely to be favorable in patients with no radiological abnormalities at baseline than in those with stage 2 disease. Second, there is a consensus that the size of the necrotic segment is the main determinant of collapse. However, methods for measuring the size of the necrosis are nearly as numerous as papers on the topic, so that some doubt remains about the cutoffs that can be used to predict the prognosis. Roughly, involvement of less than 10% of the femoral head carries an excellent prognosis and involvement of more than 25% a poor prognosis. Third, the proportion of the weight-bearing area affected by necrosis may play a key role in the outcome, most notably when the necrosis is small or intermediate in size. The most widely used measurement method consists in determining the proportion of the acetabular roof
P. Lafforgue / Joint Bone Spine 73 (2006) 500–507
that is located opposite the necrotic segment in the coronal plane: less than 1/3 defines stage a, 1/3 to 2/3 stage b, and more than 2/3 stage c. The rate of progression to radiological stage 3 increases from 0% in stage a disease to 70% in stage c disease. A number of other factors may influence the outcome of femoral head AVN. Intuitively, pain would seem to predict a poor outcome, even when the radiological abnormalities are minimal. The prognostic significance of pain has not been established, however. MRI evidence of marrow edema under the necrotic segment and/or joint effusion is very significantly associated with pain and worse radiological stages [57]. These findings may also predict femoral head collapse. Finally, AVN related to glucocorticoid or alcohol exposure carries a poor prognosis, which is chiefly related to the often multiple and extensive foci. 4.6. After collapse occurs, is progression to pain and functional disability inevitable? Flattening of the femoral head related to collapse of the necrotic segment (Ficat and Arlet stage 3, Steinberg stage 4) is almost consistently followed by progressive joint damage that usually requires total hip replacement. However, the spontaneous clinical outcome may be acceptable. High rates of clinical and radiological stabilization have been reported in patients with limited collapse (less than 2 mm) or limited necrosis [58,59]. 5. Practical implications Genetic studies provide hope that tools for identifying highrisk patients will be available in the future. Such tools may prove useful for evaluating patients before starting high-dose glucocorticoid therapy. At present, caution is in order when considering glucocorticoid therapy or determining whether to allow deep sea diving in patients who have additional risk factors such as dyslipidemia or alcohol abuse. On the opposite, the risk of AVN is virtually nil after 6–12 months on glucocorticoid therapy (proving a posteriori that the patient had a low risk of AVN). Data on the role for lipid metabolism disorders or clotting disorders in the genesis of AVN may lead to preventive strategies. Animal studies conducted by Wang et al. [16–18] suggest a protective effect of lipid-lowering agents. Of 284 patients who were given glucocorticoid therapy while on long-term statin therapy, only 1% experienced femoral head AVN during the 7.5-year follow-up [60]. This rate was lower than expected, although there was no control group. Thus, statins may decrease the risk of glucocorticoid-induced AVN. In one study, long-term therapy with low-molecular-weight heparin decreased the risk of progression of stage 1 femoral head necrosis in patients with idiopathic AVN associated with clotting disorders [61]. This finding requires confirmation by further studies.
505
In medicolegal evaluations of patients with AVN, absence of MRI evidence of necrosis 6 months after exposure to a risk factor (e.g. trauma, diving, or glucocorticoid therapy) rules out a role for the risk factor. When MRI documentation is not available, an interval longer than 3 years between risk factor exposure and femoral head collapse militates against a relation between the two events. Importantly, data are now available for predicting the outcome of early-stage AVN. AVN involving more than 25% of the femoral head or more than 2/3 of the weight-bearing portion carries a very high-risk of collapse. In this situation, core decompression is consistently ineffective and primary hip replacement should be considered. Necrosis involving less than 10% of the femoral head or less than 1/3 of the weightbearing portion is likely to have a favorable outcome without treatment. In this situation, the appropriateness of any therapeutic intervention is debatable. The optimal management strategy is more difficult to determine in intermediate situations, which may constitute the best indication for core decompression. Finally, except when AVN is caused by local factors (Table 1), the disease potentially involves the entire skeleton. When a focus of AVN is identified, radiographs, radionuclide scanning, or MRI should be used to investigate other high-risk sites (hips, shoulders, and knees). References [1] Armingeat T, Pham T, Legre V, Lafforgue P. Coexistence of intravertebral vacuum and intradiscal vacuum. Joint Bone Spine 2006;73:428–32. [2] Naveau A, Naveau B. Osteonecrosis of the jaw in patients taking bisphosphonates. Joint Bone Spine 2006;73:7–9. [3] Kenzora JE, Glimcher MJ. Accumulative cell stress: the multifactorial etiology of idiopathic osteonecrosis. Orthop Clin North Am 1985;16: 669–79. [4] Jones Jr. JP. Fat embolism and osteonecrosis. Orthop Clin North Am 1985;16:595–633. [5] Hutter CD. Dysbaric osteonecrosis: a reassessment and hypothesis. Med Hypotheses 2000;54:585–90. [6] Jones LC, Mont MA, Le TB, Petri M, Hungerford DS, Wang P, et al. Procoagulants and osteonecrosis. J Rheumatol 2003;30:783–91. [7] Glueck CJ, Freiberg RA, Fontaine RN, Tracy T, Wang P. Hypofibrinolysis, thrombophilia, osteonecrosis. Clin Orthop Relat Res 2001;386:19– 33. [8] Lee JS, Koo KH, Ha YC, Koh KK, Kim SJ, Kim JR, et al. Role of thrombotic and fibrinolytic disorders in osteonecrosis of the femoral head. Clin Orthop Relat Res 2003;417:270–6. [9] Koo KH, Song HR, Ha YC, Kim JR, Kim SJ, Kim KI, et al. Role of thrombotic and fibrinolytic disorders in the etiology of Perthes’ disease. Clin Orthop Relat Res 2002;399:162–7. [10] Atsumi T, Kuroki Y, Yamano K. A microangiographic study of idiopathic osteonecrosis of the femoral head. Clin Orthop Relat Res 1989; 246:186–94. [11] Ohzono K, Takaoka K, Saito S, Saito M, Matsui M, Ono K. Intraosseous arterial architecture in nontraumatic avascular necrosis of the femoral head. Microangiographic and histologic study. Clin Orthop Relat Res 1992;277:79–88. [12] Arlet J, Laroche M, Soler R, Thiechart M, Pieraggi MT, Mazieres B. Histopathology of the vessels of the femoral heads in specimens of osteonecrosis, osteoarthritis and algodystrophy. Clin Rheumatol 1993;12:162– 5.
506
P. Lafforgue / Joint Bone Spine 73 (2006) 500–507
[13] Arlet J, Ficat P, Lartigue G, Tran MA. Recherches cliniques sur la pression intra-osseuse dans la metaphyse et l’epiphyse femorales superieures chez l’homme. Application du diagnostic des ischemies et necroses. Rev Rhum Mal Osteoartic 1972;39:717–23. [14] Ficat RP. Idiopathic bone necrosis of the femoral head. Early diagnosis and treatment. J Bone Joint Surg Br 1985;67:3–9. [15] Zizic TM, Lewis CG, Marcoux C, Hungerford DS. The predictive value of hemodynamic studies in preclinical ischemic necrosis of bone. J Rheumatol 1989;16:1559–64. [16] Wang GJ, Dughman SS, Reger SI, Stamp WG. The effect of core decompression on femoral head blood flow in steroid-induced avascular necrosis of the femoral head. J Bone Joint Surg Am 1985;67:121–4. [17] Cui Q, Wang GJ, Su CC, Balian G. The Otto Aufranc Award. Lovastatin prevents steroid induced adipogenesis and osteonecrosis. Clin Orthop Relat Res 1997;344:8–19. [18] Wang GJ, Moga DB, Richemer WG, Sweet DE, Reger SI, Thompson RC. Cortisone induced bone changes and its response to lipid clearing agents. Clin Orthop Relat Res 1978;130:81–5. [19] Mitchell DG, Rao VM, Dalinka M, Spritzer CE, Axel L, Gefter W, et al. Hematopoietic and fatty bone marrow distribution in the normal and ischemic hip: new observations with 1.5-T MR imaging. Radiology 1986;161:199–202. [20] Koo KH, Dussault RG, Kaplan PA, Ahn IO, Kim R, Devine MJ, et al. Fatty marrow conversion of the proximal femoral metaphysis in osteonecrotic hips. Clin Orthop Relat Res 1999;361:159–67. [21] Vande Berg BC, Malghem J, Lecouvet FE, Devogelaer JP, Maldague B, Houssiau FA. Fat conversion of femoral marrow in glucocorticoid-treated patients: a cross-sectional and longitudinal study with magnetic resonance imaging. Arthritis Rheum 1999;42:1405–11. [22] Motomura G, Yamamoto T, Miyanishi K, Yamashita A, Sueishi K, Iwamoto Y. Bone marrow fat-cell enlargement in early steroid-induced osteonecrosis—a histomorphometric study of autopsy cases. Pathol Res Pract 2005;200:807–11. [23] Saito S, Inoue A, Ono K. Intramedullary haemorrhage as a possible cause of avascular necrosis of the femoral head. The histology of 16 femoral heads at the silent stage. J Bone Joint Surg Br 1987;69:346–51. [24] Kawai K, Tamaki A, Hirohata K. Steroid-induced accumulation of lipid in the osteocytes of the rabbit femoral head. A histochemical and electron microscopic study. J Bone Joint Surg Am 1985;67:755–63. [25] Weinstein RS, Nicholas RW, Manolagas SC. Apoptosis of osteocytes in glucocorticoid-induced osteonecrosis of the hip. J Clin Endocrinol Metab 2000;85:2907–12. [26] Calder JD, Buttery L, Revell PA, Pearse M, Polak JM. Apoptosis—a significant cause of bone cell death in osteonecrosis of the femoral head. J Bone Joint Surg Br 2004;86:1209–13. [27] Suh KT, Kim SW, Roh HL, Youn MS, Jung JS. Decreased osteogenic differentiation of mesenchymal stem cells in alcohol-induced osteonecrosis. Clin Orthop Relat Res 2005;431:220–5. [28] Gangji V, Hauzeur JP, Schoutens A, Hinsenkamp M, Appelboom T, Egrise D. Abnormalities in the replicative capacity of osteoblastic cells in the proximal femur of patients with osteonecrosis of the femoral head. J Rheumatol 2003;30:348–51. [29] Asano T, Takahashi KA, Fujioka M, Inoue S, Okamoto M, Sugioka N, et al. ABCB1 C3435T and G2677T/A polymorphism decreased the risk for steroid-induced osteonecrosis of the femoral head after kidney transplantation. Pharmacogenetics 2003;13:675–82. [30] Chao YC, Wang SJ, Chu HC, Chang WK, Hsieh TY. Investigation of alcohol metabolizing enzyme genes in Chinese alcoholics with avascular necrosis of hip joint, pancreatitis and cirrhosis of the liver. Alcohol Alcohol 2003;38:431–6. [31] Liu YF, Chen WM, Lin YF, Yang RC, Lin MW, Li LH, et al. Type II collagen gene variants and inherited osteonecrosis of the femoral head. N Engl J Med 2005;352:2294–301. [32] Motomura G, Yamamoto T, Miyanishi K, Jingushi S, Iwamoto Y. Combined effects of an anticoagulant and a lipid-lowering agent on the prevention of steroid-induced osteonecrosis in rabbits. Arthritis Rheum 2004;50:3387–91.
[33] Hernigou P, Voisin MC, Marichez M, Despres E, Goutallier D. Confrontation de l’imagerie par resonance magnétique nucleaire et de l’histologie dans les necroses des tetes femorales. Rev Rhum Mal Osteoartic 1989; 56:741–4. [34] Glimcher MJ, Kenzora JE. The biology of osteonecrosis of the human femoral head and its clinical implications: II. The pathological changes in the femoral head as an organ and in the hip joint. Clin Orthop Relat Res 1979;139:283–312. [35] Kawasaki M, Hasegawa Y, Sakano S, Sugiyama H, Tajima T, Iwasada S, et al. Prediction of osteonecrosis by magnetic resonance imaging after femoral neck fractures. Clin Orthop Relat Res 2001;385:157–64. [36] Sugano N, Masuhara K, Nakamura N, Ochi T, Hirooka A, Hayami Y. MRI of early osteonecrosis of the femoral head after transcervical fracture. J Bone Joint Surg Br 1996;78:253–7. [37] Taylor LJ. Multifocal avascular necrosis after short-term high-dose steroid therapy. A report of three cases. J Bone Joint Surg Br 1984;66:431– 3. [38] Paolaggi JB, Le Parc JM, Arfi S. Severe arthralgias after pulse corticosteroid therapy in transplant patients. J Rheumatol 1987;14:1077–8. [39] Sakamoto M, Shimizu K, Iida S, Akita T, Moriya H, Nawata Y. Osteonecrosis of the femoral head: a prospective study with MRI. J Bone Joint Surg Br 1997;79:213–9. [40] Oinuma K, Harada Y, Nawata Y, Takabayashi K, Abe I, Kamikawa K, et al. Osteonecrosis in patients with systemic lupus erythematosus develops very early after starting high dose corticosteroid treatment. Ann Rheum Dis 2001;60:1145–8. [41] Kopecky KK, Braunstein EM, Brandt KD, Filo RS, Leapman SB, Capello WN, et al. Apparent avascular necrosis of the hip: appearance and spontaneous resolution of MR findings in renal allograft recipients. Radiology 1991;179:523–7. [42] Le Parc JM, Andre T, Helenon O, Benoit J, Paolaggi JB, Kreis H. Osteonecrosis of the hip in renal transplant recipients. Changes in functional status and magnetic resonance imaging findings over 3 years in three hundred five patients. Rev Rhum 1996;63:413–20 [Engl Ed]. [43] Kubo T, Yamazoe S, Sugano N, Fujioka M, Naruse S, Yoshimura N, et al. Initial MRI findings of non-traumatic osteonecrosis of the femoral head in renal allograft recipients. Magn Reson Imaging 1997;15:1017– 23. [44] Fujioka M, Kubo T, Nakamura F, Shibatani M, Ueshima K, Hamaguchi H, et al. Initial changes of non-traumatic osteonecrosis of femoral head in fat suppression images: bone marrow edema was not found before the appearance of band patterns. Magn Reson Imaging 2001;19:985–91. [45] Sakaia T, Sugano N, Tsuji T, Nishii T, Yoshikawa H, Ohzono K. Serial magnetic resonance imaging in a non-traumatic rabbit osteonecrosis model: an experimental longitudinal study. Magn Reson Imaging 2000; 18:897–905. [46] Sugano N, Nishii T, Shibuya T, Nakata K, Masuhara K, Takaoka K. Contralateral hip in patients with unilateral nontraumatic osteonecrosis of the femoral head. Clin Orthop Relat Res 1997;334:85–90. [47] Yamamoto T, DiCarlo EF, Bullough PG. The prevalence and clinicopathological appearance of extension of osteonecrosis in the femoral head. J Bone Joint Surg Br 1999;81:328–32. [48] Kim YM, Rhyu KH, Lee SH, Kim HJ. Can osteonecrosis of the femoral head be recurrent? Clin Orthop Relat Res 2003;406:123–8. [49] Hernigou P, Lambotte JC. Bilateral hip osteonecrosis: influence of hip size on outcome. Ann Rheum Dis 2000;59:817–21. [50] Shimizu K, Moriya H, Akita T, Sakamoto M, Suguro T. Prediction of collapse with magnetic resonance imaging of avascular necrosis of the femoral head. J Bone Joint Surg Am 1994;76:215–23. [51] Koo KH, Kim R, Ko GH, Song HR, Jeong ST, Cho SH. Preventing collapse in early osteonecrosis of the femoral head. A randomised clinical trial of core decompression. J Bone Joint Surg Br 1995;77:870–4. [52] Lafforgue P, Dahan E, Chagnaud C, Schiano A, Kasbarian M, Acquaviva PC. Early-stage avascular necrosis of the femoral head: MR imaging for prognosis in 31 cases with at least 2 years of follow-up. Radiology 1993;187:199–204.
P. Lafforgue / Joint Bone Spine 73 (2006) 500–507 [53] Nishii T, Sugano N, Ohzono K, Sakai T, Sato Y, Yoshikawa H. Significance of lesion size and location in the prediction of collapse of osteonecrosis of the femoral head: a new three-dimensional quantification using magnetic resonance imaging. J Orthop Res 2002;20:130–6. [54] Takatori Y, Kokubo T, Ninomiya S, Nakamura S, Morimoto S, Kusaba I. Avascular necrosis of the femoral head. Natural history and magnetic resonance imaging. J Bone Joint Surg Br 1993;75:217–21. [55] Mont MA, Hungerford DS. Non-traumatic avascular necrosis of the femoral head. J Bone Joint Surg Am 1995;77:459–74. [56] Hernigou P, Poignard A, Nogier A, Manicom O. Fate of very small asymptomatic stage-I osteonecrotic lesions of the hip. J Bone Joint Surg Am 2004;86-A:2589–93. [57] Iida S, Harada Y, Shimizu K, Sakamoto M, Ikenoue S, Akita T, et al. Correlation between bone marrow edema and collapse of the femoral
507
head in steroid-induced osteonecrosis. AJR Am J Roentgenol 2000;174: 735–43. [58] Churchill MA, Spencer JD. End-stage avascular necrosis of bone in renal transplant patients. The natural history. J Bone Joint Surg Br 1991;73: 618–20. [59] Nishii T, Sugano N, Ohzono K, Sakai T, Haraguchi K, Yoshikawa H. Progression and cessation of collapse in osteonecrosis of the femoral head. Clin Orthop Relat Res 2002;400:149–57. [60] Pritchett JW. Statin therapy decreases the risk of osteonecrosis in patients receiving steroids. Clin Orthop Relat Res 2001;386:173–8. [61] Glueck CJ, Freiberg RA, Sieve L, Wang P. Enoxaparin prevents progression of stages I and II osteonecrosis of the hip. Clin Orthop Relat Res 2005;435:164–70.