Osteonecrosis is an idiopathic, debilitating, and progressive disease with various causes, including disruption of the blood supply or venous occlusion, that results in increased intraosseous pressure. Etiologies include alcoholism, blood disorders, trauma, radiation therapy, corticosteroid administration, dysbaria, and autoimmune diseases.
1
Bone marrow edema syndrome is an uncommon, self-limiting skeletal disease. Curtiss and Kincaid
2
first reported the syndrome in 1959. They referred to it as transient osteoporosis because of the osteopenic appearance of the affected bone on plain radiographs. Since then, this entity has been described using other terms, such as transient marrow edema syndrome.
3
The natural course of this entity shows spontaneous remission after 6 to 12 months.
4
The pathogenesis of osteonecrosis and bone marrow edema syndrome remains unknown, but several hypotheses have been proposed. For osteonecrosis, vascular disturbances with a mismatch between the arterial inflow and venous outflow, reduced vessel density, or thrombembolisms of the terminal vessels may play a role.
1,5
Traumatic injuries with an initial fracture of the subchondral bone that result in necrosis of the surrounding area may also cause osteonecrosis.
6
For bone marrow edema syndrome, thrombembolism, obstruction of arteriolar inflow or venous outflow, injury to the vessel wall secondary to vasculitis, altered lipid metabolism, and reduced fibrinolysis have been suggested as etiologic factors.
7
Some authors have postulated that bone marrow edema syndrome might be a reversible or a prestage form of osteonecrosis.
8,9
For both entities, several studies have indicated that coagulation abnormalities causing thrombophilia or hypofibrinolysis might contribute to the emergence of these entities. However, inhomogeneities in the patients’ collectives, the broad variety of determined parameters and differences in the diagnosis and therapy make a literature evaluation and comparison of data among these studies difficult. Hence, the purpose of the current work was to systemically review the literature reporting coagulation abnormalities at the site of osteonecrosis and bone marrow edema syndrome.
Results
Study Identification and Level of Evidence
A total of 45 studies with 2163 patients were identified (Table
1
).
10–54
Of these, 13 were case reports and 32 were original manuscripts (those with a large series of patients). Eleven studies were published before 2000 and 34 after. Fourteen studies were level IV, 22 were level III, and 9 were level II studies; no level I study was identified (Table
1
).
Etiology
The majority of studies focused on findings at the site of idiopathic or secondary osteonecrosis (41 studies; 1911 patients),
13–54
Three studies (60 patients) reported bone marrow edema syndrome.
10–12
One study reported a cohort of 145 patients with elevated lipoprotein(a) (Lp[a]) plasma values,
23
and another study described a cohort of 96 patients with
Morbus Perthes46
(Table
1
). Thirteen etiologies were identified as possible factors for the emergence of secondary osteonecrosis, and 1 study describes cases of osteonecrosis and
M Perthes
(Table
2
).
46
Localization of Osteonecrosis and Bone Marrow Edema Syndrome
The hip joint was reported as the most commonly affected joint in 40 studies, followed by the knee and shoulder joint in 8 and 4 studies, respectively (Table
1
).
Investigated Parameters
A total of 48 parameters were investigated (Table
3
). The most frequently investigated parameter was Lp(a) (n=14), followed by plasminogen activator inhibitor (PAI), protein S (n=7) and C (n=5), tissue plasminogen activator (tPA) (n=5), factor VIII (n=5), apolipoproteins ApoA1 (n=5) and ApoB (n=4), and various genes or gene mutations.
The Lp(a) serum values were frequently elevated in cases of osteonecrosis and bone marrow edema syndrome (Table
3
). However, some discrepancies are evident among the osteonecrosis studies. Glueck et al
24,27
reported that elevated Lp(a) values were evident in primary but not secondary osteonecrosis. Moreover, these findings only account for unifocal but not multifocal osteonecrosis.
27
Posan et al
46
reported elevated values in primary and secondary osteonecrosis. Jones et al
36
reported no differences in the serum Lp(a) values in patients with osteonecrosis compared with a control group.
The PAI has been investigated at the site of bone marrow edema syndrome and osteonecrosis. Berger et al
10
reported that decreased PAI levels were seen in 1 patient with bone marrow edema syndrome, whereas the levels were normal in 2 other cases. Jones et al
36
reported a significant difference in PAI activity between patients with osteonecrosis and healthy controls. Most of the studies reporting PAI involvement in the pathophysiology of osteonecrosis have been reported by Glueck et al
21,22,27
: high serum levels of PAI combined with increased protein function and antigen activity were found in unifocal idiopathic but not unifocal secondary osteonecrosis. However, in multifocal osteonecrosis, an elevated PAI activity was observed in cases of secondary osteonecrosis but not for idiopathic osteonecrosis.
30
Regardless of the etiology, a higher frequency of the 4G/4G polymorphism of the PAI-1 gene has also been identified as an osteonecrosis risk factor.
25,27,38
Chotanaphuti et al
16
reported a high prevalence of protein S deficiency in patients with idiopathic osteonecrosis. Elishkewich et al
20
made similar observations in a 36-year-old man. Familial protein S deficiency, causing a low level of free protein S, was also identified as a risk factor for idiopathic osteonecrosis by Glueck et al.
26,27,30
Studies by Pierre-Jacques et al
45
and Üreten et al
49
confirmed the contribution of low protein S levels to the development of idiopathic multifocal or unifocal osteonecrosis, respectively. Low protein C concentration levels also reportedly contributed to the emergence and maintenance of idiopathic osteonecrosis by Glueck et al,
24
Mehsen et al,
41
and Wermes et al.
51
Consistent data were found regarding the tPA in osteonecrosis: Glueck et al
22
reported a decrease of stimulated tPA function in 4 of 5 patients with idiopathic osteonecrosis. They also confirmed these findings for secondary osteonecrosis in larger patient cohorts.
24,25
Jones et al
36
also reported a significant decrease in tPA function among 45 patients with secondary osteonecrosis. Glueck et al
21
reported a higher stimulated tPA function in secondary osteonecrosis compared with idiopathic osteonecrosis.
Blood coagulation factor VIII was reported to be significantly elevated in patients with osteonecrosis.
47
Chotanaphuti et al
16
and Glueck et al
29,30
identified this parameter as a risk factor for idiopathic osteonecrosis. Coagulation factor VIII was also elevated in 1 patient with steroid-induced secondary bilateral femoral head osteonecrosis.
31
Controversial data have been reported for serum ApoA1 and ApoB levels. In patients with idiopathic osteonecrosis, serum ApoA1 levels were elevated in 5 patients and decreased in 1 patient, whereas elevated serum ApoB levels were observed in 2 patients.
22
Hirata et al
34
reported that the serum ApoA1 and ApoB levels were not associated with the development of osteonecrosis. Glueck et al
21
reported higher serum ApoB levels in idiopathic than in secondary osteonecrosis. Zalavras et al
52
reported significantly higher serum ApoB levels in patients with idiopathic osteonecrosis compared with a control group without osteonecrosis, but this effect was not evident for the serum ApoA1 levels. Miyanishi et al
42
reported that a high serum ApoB:ApoA1 ratio was a risk factor for the development of non-traumatic osteonecrosis compared with traumatic osteonecrosis; similar findings were reported by Hirata et al.
33
Ten genes or gene mutations were investigated in 18 studies (Table
3
). Several studies demonstrated a homozygosity or heterozygosity of methylenetetrahydrofolate reductase (MTHFR) polymorphisms,
15,20,26,40,53
Factor V Leiden,
13,19,29,30,46
and prothrombin 20210A mutation
13,19,50
in association with osteonecrosis; Kechli et al
37
reported no association of the aforementioned mutations with osteonecrosis during or after treatment for malignancy in a pediatric population. Dai et al
17
reported that various haplotypes of the tissue factor pathway inhibitor gene were associated with the emergence of idiopathic or alcohol-induced osteonecrosis. He and Li
32
reported significant differences for the P-glycoprotein gene ABCB1 in patients with steroid-induced osteonecrosis compared with a control group. Glueck et al
28,30
described the T786C polymorphism of eNOS in patients with idiopathic osteonecrosis. Hirata et al
33
reported a higher frequency of the T7623T and CT alleles of the ApoB gene in 34 patients with osteonecrosis than in control patients, resulting in a statistically significant elevated odds ratio. Zhang et al
54
reported an underexpression of the CHST2 and the GPCR26 gene in 3 patients with femoral head osteonecrosis.
In the majority of studies, coagulation parameters were determined in the peripheral blood. One study compared the serum laboratory findings with those locally determined in the affected bone and showed an increase of those in the bone marrow.
12
Another study solely investigated the gene expression in osteonecrotic femoral heads.
54
Treatment Procedures
Twenty-seven (60%) of 45 studies reported no data on the treatment procedures of osteonecrosis or bone marrow edema syndrome. In 9 (20%) studies, anticoagulation therapy was administered and various substances have been used (eg, warfarin, phenprocoumon, enoxaparine, fondaparinux, heparin, ticlopidin, and acetylsalicylic acid) (Table
4
). A core decompression of the affected region was reported in 4 (9%) studies. Total hip arthroplasty was performed in 4 (9%) studies (Table
4
).
Discussion
The current report systematically reviewed the literature for a possible involvement of thrombophilia and hypofibrinolysis in the etiology of osteonecrosis and bone marrow edema syndrome. Forty-eight thrombophilic and hypofibrinolytic parameters were identified in 45 studies with a total of 2163 patients. The most frequently reported laboratory findings included altered serum concentrations of Lp(a), ApoA1, and ApoB, decreased concentration and function of fibrinolytic agents (tPA, protein C, and protein S) and increased levels of thrombophilic markers (PAI and coagulation factor VIII). Furthermore, several single nucleotide polymorphisms (Factor V Leiden, methylene tetrahydrofolate reductase C677T, and prothrombin 20210A mutations) were identified in the molecular biological pathogenesis of osteonecrosis and bone marrow edema syndrome. Despite inhomogeneities in the reported results, patients’ collectives, and determined parameters, these data strongly suggest that coagulation abnormalities may play an important role in the emergence of both diseases.
Lp(a) was first reported in 1963 by Berg.
55
Lp(a) is a low-density lipoprotein-like particle in which ApoB-100 is bound with a disulfide bridge to ApoA.
56
This unique structural feature accounts for the potential atherogenic and thrombophilic activity of Lp(a).
10
In plasma, Lp(a) exists as peaks in the low-density lipoprotein range; a form of density intermediate between low- and high-density lipoproteins and another ApoE-rich fraction closer to the density of high-density lipoproteins.
56
Lp(a) is made by the low-density lipoprotein synthesis machinery in the endoplasmic reticulum and the ApoA moiety is added on the surface of hepatocytes.
56
Plasma concentrations of Lp(a) correlate inversely with the size of ApoA isoproteins.
57
Lp(a) levels are higher in women than in men; they do not appear to be affected by physical exercise.
56
Moderate alcohol consumption might lower Lp(a) concentration.
56
With regard to the pathogenesis of osteonecrosis and bone marrow edema syndrome, Lp(a) reduces fibrinolytic activity by competing with plasminogen at the fibrin surface for the common lysine binding domains, causing increased susceptibility to arterial and venous thrombotic events.
10
In accordance with this pathophysiological background, Berger et al
10–12
identified elevated levels of Lp(a) in patients with bone marrow edema syndrome. Although data reported in the literature are inconsistent regarding elevated Lp(a) levels in osteonecrosis, with some authors reporting a lack of significance of this parameter,
36,52
most studies reported increased Lp(a) concentrations in this patient cohort.
22,26,27,41,46
The majority of data for the increase of Lp(a) levels in the course of idiopathic
23,29
and secondary
21,24
osteonecrosis were reported by Glueck et al.
ApoB, the structural protein for the atherogenic lipoproteins (low and intermediate-density lipoprotein and large, buoyant, low-density lipoprotein and small, dense, low-density lipoprotein), is responsible for transporting lipids from the liver and gut to the peripheral tissues.
58
Each lipoprotein particle contains 1 ApoB molecule. Plasma ApoB levels increase with age
59,60
and are higher in men than in women.
59
In contrast, ApoA1 is the major structural protein for high-density lipoproteins and reflects the atheroprotective side of lipid metabolism.
58
ApoA1 is produced in the liver and intestine and is responsible for initiating reverse cholesterol transport, whereby excess cholesterol in peripheral tissues is carried back to the liver for excretion.
58
ApoA1 levels are reportedly higher in women than men.
61
An association of elevated serum low-density lipoprotein and ApoB, as well as decreased serum high-density lipoproteins and ApoA1, has been reported in coronary artery disease.
62
Against this background, ApoA1 may be regarded as protective for vascular diseases, whereas ApoB might have a deleterious effect. An elevated ApoB:ApoA1 ratio was reported to predispose individuals to the emergence of osteonecrosis and, therefore, should be ruled out before steroid administration.
33,42
Following the conversion of plasminogen into the active enzyme plasmin, fibrinolysis of blood clots is mediated by the degradation of matrix components and activation of procollagenases.
63
However, plasminogen activation may be hampered by PAI. In this respect, PAI-1 is the major fibrinolysis inhibitor.
64
Increased concentration of PAI-1, as well as enhanced PAI function, may cause arterial occlusion and ultimately lead to myocardial infarction.
65,66
In addtion, upregulation of this fast-acting inhibitor of fibrinolysis
38
is associated with an increased incidence of thrombophilia.
67
Furthermore, as early as in 1961, elevated PAI levels have been found to be involved in the pathogenesis of osteonecrosis,
68
possibly mediated by an increased intraosseous venous pressure that restricts blood flow to the subchondral bone regions and may culminate in osteonecrosis.
25,69
Over the past 3 decades, this finding has been confirmed by Glueck et al.
21,22,27,30
Single nucleotide insertion or deletion polymorphisms of the PAI-1 gene with prevalence of the 4G allele seems to be a risk factor for hypofibrinolysis and, consequently, osteonecrosis.
25,27,38
The tPA is another key player in the plasminogen activation system; reciprocally to the plasminogen activator inhibitor, tPA is considered the major stimulator of fibrinolysis.
24
This serine protease catalyzes the conversion of plasminogen to plasmin by cleavage of plasminogen at its arginine-valine peptide bond.
70
Clinically, recombinant tPA such as alteplase is approved by the US Food and Drug Administration for the treatment of myocardial infarction,
71,72
ischemic stroke,
73
or pulmonary embolism.
74
For the field of orthopedic research, decreased tPA function was found in idiopathic and secondary osteonecrosis.
21,22,24,25
Blood coagulation factor VIII is a glycoprotein released by the vascular, glomerular, and tubular endothelium and the sinusoidal cells of the liver.
75
Defects in its gene result in hemophilia A, a recessive X-linked coagulation disorder.
76
Patients with elevated levels of factor VIII are at increased risk for deep venous thrombosis and thromboembolism.
77
This thrombophilic potential has also been reported as a potential risk factor for the development of osteonecrosis in 5 studies included in the current article.
16,29–31,47
The anticoagulative effect of protein C was first reported by Mammen et al.
78
The activated form of this serine protease (activated protein C)
79
is capable of inactivating the coagulation factors Va and VIIIa, which are part of the prothrombinase complex, and, thus, are crucially involved in the generation of thrombin and blood clotting.
80
Protein S is an important cofactor of protein C in the inactivation of both coagulation factors.
81
Patients with protein C or S deficiency have a significantly higher risk of developing deep venous thrombosis or thromboembolism and disseminated intravascular coagulation,
82,83
and a high prevalence of protein S
16,26,27,30,45,49
and protein C
24,41,51
deficiency was detected in patients with osteonecrosis.
Besides a decreased concentration or function of protein C, the heritable resistance to activated protein C was reported by Dahlbäck et al
84
and is associated with familial thrombophilia. Most commonly, resistance to activated protein C is caused by a genetic mutation (replacement of arginine with glutamine at nucleotide position 506), resulting in a loss of the cleavage site of coagulation factor V and producing Factor V Leiden, a severe hypercoagulability disorder.
84,85
This disease is characterized by an elevated risk for venous and arterial thromboembolism.
86
However, several other genetic traits affect the anticoagulant response to activated protein C, but none cause the same severe resistance to activated protein C phenotype as Factor V Leiden, and their importance as risk factors for thrombosis is unclear.
79
A poor activated protein C response may also result from acquired conditions.
79
In the current review, resistance to activated protein C and the Factor V Leiden mutation were identified as potential risk factors for the development of osteonecrosis in numerous studies.
13,19,29,30,46
Another important gene mutation involved in the pathogenesis of osteonecrosis is the C677T polymorphism of the MTHFR. Replacement of cytosine with thymine at the nucleotide position 677 decreases the activity of this enzyme, interferes with the intracellular metabolism of homocysteine, and thereby mildly elevates the plasma homocysteine level.
15
Because hyperhomocysteinemia is an established risk factor for thrombotic events,
87
the increased incidence of osteonecrosis in patients with the C677T MTHFR mutation is coherent in this regard.
15,20,26,40,53
Mutation in the prothrombin gene (substitution of guanine for arginine at nucleotide position 20210) results in increased plasma prothrombin levels and is associated with venous thrombosis.
88
The frequency of the prothrombin 20210A gene mutation ranges from 6% to 12% in patients with deep venous thrombosis compared with a range of 1% to 4% in the general population.
89
The current authors believe that this specific prothrombin mutation is associated with the emergence of osteonecrosis.
13,19,50
The treatment strategy for osteonecrosis and bone marrow edema syndrome has been reported in 18 (40%) of 45 studies. With regard to painful bone marrow edema syndrome, Berger et al
10,11
reported the ineffectiveness of partial weight bearing for 6 to 8 weeks with regard to pain reduction and the necessity for surgical core decompression. To minimize bone loss during the acute episodes, calcitonin has been used.
90
Furthermore, the intravenous administration of the prostacyclin analogue iloprost yielded clinical success in patients with painful bone marrow edema syndrome of the knee.
4
Sowers et al
91
reported that the presence of subchondral bone marrow edema does not satisfactorily correlate with the presence or absence of knee pain. However, concomitant subchondral cortical bone defects in patients with bone marrow edema seem to have a stronger effect on the susceptibility to knee pain.
91
Therefore, the causes of pain in patients with bone marrow edema syndrome and a possible relation with laboratory or MRI findings will have to be elucidated in more detail.
For patients with osteonecrosis with coexistent thrombophilic or hypofibrinolytic disorders, no standardized treatment plan is available. Because the supposed pathogenesis with venous occlusion of the bone by fibrin clots, hypertension in the affected cancellous bone, and cell death by hypoxia
24
is similar to Legg-Calvé-Perthes disease,
92–94
several pharmacological substances have been tested to improve osseous perfusion and normalize laboratory disorders. Any conservative pharmacological treatment must be applied before irreversible collapse of the respective bone region (eg, Ficat stages I and II at the femoral head).
22
The most commonly applied treatments included oral anticoagulants, such as warfarin
20,21,31,39,45,48
and phenprocoumon,
51
with target international normalized ratio values ranging between 1.5
48
and 4.5.
51
Enoxaparine,
20,27,51
fondaparinux,
31
heparin,
45
ticlopidin,
43
and acetylsalicylic acid
43,48
were also given for anticoagulative therapy. Stanozolol, an anabolic androgenic steroid, potentially normalizes PAI- and tPA-function and Lp(a) levels
22
and has been reported to inhibit the progress of osteonecrosis at different anatomical sites by Glueck et al.
21,22,24
Surgical treatment options include core decompression,
20,45
bone resection and coverage of the osteonecrotic defect with platelet rich plasma,
50
alloarthroplasty,
26,31,35,37
and joint fusion.
37
When a systematic literature review is performed, some parameters have to be critically reconsidered. For example, besides its idiopathic form, bone marrow edema can arise in young, healthy, athletic patients after surgery or severe blunt injuries. However, this does not mean that all of these patients have coagulopathies that will cause the emergence of bone marrow edema. To clarify this topic, a multicenter, prospective, combined hematological-orthopedic and clinical-genetical study is strongly recommended.