Current State of Knowledge
Osteonecrosis (ON) was first recognized as a complication of HCT in 1987 (
56). Only a few papers have addressed the occurrence of ON in pediatric HCT recipients where the prevalence ranges from 1.3% to 14% (
9,
57–
59), with even higher occurrence (44%) being found in pediatric allogeneic HCT recipients who underwent routine screening for ON by MRI (
33). The true prevalence, however, is unknown as it can only be determined by prospective screening with MR, which is a much more sensitive method of detection of ON than plain radiographs (
12,
58,
60–
62).
In children, knees (31–40%) are the most frequent site of ON, followed by hips (19–24%), shoulders (9%), and other sites (
33,
63–
65). The majority of patients manifest ON in two or more joints (
12,
59,
63,
65). Typically, ON occurs within 3 years after HCT, the earliest time point being 1–6 months after the onset of steroid therapy, particularly if MRI is used for detection (
45,
59,
62,
63,
65–
68). A median interval for the development of ON is 11 months after HCT in children (
57).
Patients usually present with either vague, diffuse bone pain, presumably due to increased intraosseous pressure, or joint-related pain due to an effusion. Hip involvement is typically manifested by groin pain. Once subchondral collapse and articular deformity occur, arthritic-type joint pain predominates accompanied by functional limitation (limp, reduced range of motion) (
66,
69) However, during early stages of ON, patients may have mild transient bone pain during treatment or they may be completely asymptomatic and not necessarily progress to symptomatic disease (
33,
62). If left untreated, joint destruction usually occurs within 1–5 years after the onset of symptoms (
66,
70). Once disease progresses beyond a certain point, collapse of necrotic bone is inevitable. The reparative processes are usually ineffective, and actually counterproductive, leading to further separation of acellular necrotic bone tissue from viable tissue by a fibrous layer, preventing revascularization (
68,
69,
71). The risk of subchondral fracture of the necrotic bone leading to joint collapse is determined by the size (best assessed using the necrotic arc index) and location of the necrotic lesion (
72). For example, involvement of less than 10–15% of the femoral head and less than a third of the weight-bearing portion carries a good prognosis, while involvement of more than 25% of the femoral head or more than two thirds of the weight-bearing portion carries a poor prognosis (
68,
69,
73).
The pathogenesis of ON is multifactorial. Several mechanisms have been proposed, including increased intraosseous pressure or intraluminal obliteration that compromise intramedullary blood flow, leading to marrow ischemia, and ultimately necrosis (
65) (). The likely contributing mechanisms are defective bone repair due to damage to the bone marrow stroma, immunosuppression as well as radiation and drug induced injury to the vessel wall and vasculitis (
59,
66,
68,
69,
71,
74–
77).
There are multiple risk factors for the development of ON which occurs in pediatric HCT recipients at a median age of 14.4 years (
57). Higher incidence of ON has been reported in patients exposed to TBI-based conditioning regimens (
57,
59,
78), presumably due to radiation-induced microvascular damage (
79), in adult patients with acute leukemia and aplastic anemia compared to AML, CML, and other diagnoses (
59), and in recipients of allogeneic HCT (particularly unrelated) compared to autologous HCT (
66,
74,
78). The latter likely explains an increase in risk in patients transplanted after 1985 when unrelated donor HCTs became more common and newer immunosuppressive agents were introduced (
74). The data about the association between gender and the incidence of ON have been inconsistent. While GVHD, both acute and chronic, has been associated with ON, it is unclear whether it plays an independent pathogenic role since it is strongly correlated with the use of steroids (
57–
59,
66,
74,
80). An argument for an additional independent role of GVHD is that it increases the risk for microangiopathy (
81–
83). Corticosteroids are the strongest risk factor for ON, with both the cumulative dose and duration of treatment playing a role (
57,
58,
65,
66,
77,
78,
84–
86). Several mechanisms for this effect have been proposed, including altered lipid metabolism, adipocyte hypertrophy, stimulation of adipogenic differentiation of bone marrow stem cells at the expense of osteogenic differentiation, leading to the formation of fat emboli and fatty infiltration of the bone marrow, or a direct effect on osteocyte apoptosis (
67,
68). The risk of ON increases with the number of drugs used for immunosuppression, including prednisone, cyclosporine (CSA), tacrolimus (FK506), and mycophenolate mofetil (MMF) (
74,
78), due to their thrombogenic effects, as well as through vascular damage and dyslipidemia (
87–
90)
An association between low BMD and ON has not been characterized beyond the observation that both can coexist in pediatric HCT patients (
65). While the causative link is absent, it is conceivable that higher BMD would likely improve the biomechanical properties of the bone, and perhaps delay the collapse of the necrotic bone. There is a significant overlap between the risk factors for low BMD and ON. In addition, the two conditions may share pathogenic pathways, and therefore have additive effects. For example, impaired osteoblast activity contributes to reduced BMD in both pediatric HCT and adult HCT patients (
29,
43,
48,
91–
93). Reduced number of osteoblast precursors may in turn adversely affect the regenerative potential of the osteogenic compartment and the course of ON (
66,
77,
94,
95)
Gaps in Current Knowledge
The major limitation of current studies in pediatrics is the lack of an animal model, an insufficient number of patients to allow identification of patients at high risk for progression of ON, and inadequate assessment of asymptomatic disease. Furthermore, there is no consensus regarding optimal screening and treatment of early stage ON. Clearly MRI is the most sensitive screening modality, however, its cost effectiveness, especially in view of the lack of a reliably effective intervention in early stage disease, remains an obstacle.