• avascular necrosis;
  • stem cell transplantation;
  • steroid treatment;
  • bone mineral density;
  • graft versus host disease;
  • fibroblast colony-forming unit


  1. Top of page
  2. Abstract
  6. Acknowledgements


The most debilitating skeletal complication of stem cell transplantation (SCT) is avascular necrosis (AVN).


Two hundred seven consecutive patients were evaluated prospectively for AVN. They survived disease free for more than 180 days after autologous or allogeneic SCT for hematologic malignancies. The diagnosis of AVN in suspicious cases was confirmed by magnetic resonance imaging. Possible correlations with treatments, bone mineral density (BMD), graft versus host disease (GVHD), and in vitro growth of fibroblast progenitors were investigated. Bone mineral density was evaluated by dual-energy X-ray absorptiometry in 100 transplanted patients, and the in vitro growth of fibroblast progenitors was monitored by a fibroblast colony-forming unit (CFU-F) assay in 30 patients after allogeneic SCT.


Twelve patients developed AVN 3–114 months (median, 26 months) following SCT: 10 (10%) after allogeneic SCT and 2 (1.9%) after autologous SCT (P = 0.04). Twenty-five joints were affected by AVN. All patients had femoral head involvement, which was managed with hip replacement in six of them. All but one patient who developed AVN after allogeneic SCT suffered from chronic GVHD (cGVHD). Avascular necrosis occurred 1–4 months after exacerbation or progression of cGVHD. Cumulative dose of steroids was similar in both SCT groups (including steroids given pretransplant for the basic disease), whereas treatment duration was significantly longer in the allogeneic SCT group. Avascular necrosis was related to the decreased number of bone marrow CFU-F colonies in vitro, but not to BMD values.


Avascular necrosis is a skeletal complication that occurs more often after allogeneic than after autologous SCT. Occurrence of AVN symptoms after clinical follow-up of cGVHD suggests that cGVHD requiring long-term steroid therapy is one of the main risk factors for AVN. Avascular necrosis may be facilitated by a severe deficit in the repopulating capacity of bone marrow stromal stem cells after SCT. Cancer 2003;97:2453–61. © 2003 American Cancer Society.

DOI 10.1002/cncr.11373

Avascular necrosis (AVN), a condition defined as nontraumatic ischemic bone necrosis, has been described in 3–41% of organ transplant patients. The most prevalent localization of AVN is the femoral head.1, 2 Acute avascular necrosis negatively influences the quality of life of organ transplanted patients.

In 1987, Atkinson et al.3 were the first to identify AVN after stem cell transplant (SCT) in a population of 50 patients surviving longer than 2 years. Since then, studies comprising a larger series of patients have identified a number of significant risk factors, such as the primary disease, age, total body irradiation (TBI), and corticosteroid administration.

The pathogenesis of AVN is still matter of controversy. It is considered to be the result of multiple triggering factors such as metabolic disorders, local vascular damage with temporary or permanent loss of blood supply to the bone, increased intraosseous pressure and mechanical stress leading to demineralization, death of trabecular bone, and collapse.4–6 The process is mostly progressive, resulting in joint destruction within 3–5 years if left untreated. Therefore, surgical treatment is a frequent requirement in these patients.

Although transplanted patients have multiple risk factors, prolonged administration of high-dose steroids may be the main cause of AVN. Conversely, AVN is rare in patients with Cushing syndrome, who have a high concentration of endogenous cortisol, as well as in patients receiving corticosteroid replacement for adrenal insufficiency.7

Allogeneic and autologous-transplanted patients often are pooled together in clinical studies on bone complications after SCT. However, there could be considerable differences between these two settings. An allograft compromises the host immune system more severely than an autograft, due to the intense immunosuppressive effect of the conditioning regimen to avoid graft rejection and to the prolonged administration of multiple immunosuppressive drugs to prevent graft versus host disease (GVHD). The frequent development of acute or chronic GVHD (aGVHD, cGVHD) after allogeneic SCT (allo-SCT) induces additional relevant alterations in the immune system.8 Finally, the patients treated by an allogeneic transplant are generally younger than patients who receive an autologous transplant.

One recent study found a strong relationship between prolonged bone loss and cGVHD in a population of long-term survivors after allo-SCT, whereas the duration and cumulative dose of corticosteroid treatment were marginally significant.9

This prospective study investigated the mechanisms or precipitating conditions that trigger AVN. Our goal was to predict which patients are at higher risk and would benefit from an early diagnosis. We evaluated the occurrence of AVN in a relatively large population of patients who had received autologous (auto-SCT) or allo-SCT. Clinical and radiographic features of AVN were compared with the type of transplant, the clinical course of cGVHD, treatment, bone mineral density (BMD), and the repopulating capacity of stromal stem cells after allo-SCT.


  1. Top of page
  2. Abstract
  6. Acknowledgements


Of 255 consecutive patients who received an SCT procedure between 1991 and 2002, 207 were alive and free of malignancy 180 days after transplant. Starting from the first diagnosis of AVN in 1994, all patients were observed prospectively for AVN during periodic follow-up at our institution. Additional data were obtained by reviewing long-term follow-up medical records starting from diagnosis up to 12 years posttransplant. One hundred patients received an allo-SCT (47 females and 53 males; age range, 21–55 years; median, 32 years) and 107 received an auto-SCT (49 females and 58 males; age range, 18–59 years; median, 39 years). The primary disease included acute or chronic myeloid leukemia (AML, CML), acute or chronic lymphocytic leukemia (ALL, CLL), Hodgkin disease (HD), and non-Hodgkin lymphoma (NHL). Allo-SCT recipients received unmanipulated bone marrow-derived stem cells from human leukocyte antigen-identical siblings. Auto-SCT patients received unmanipulated bone marrow cells or mobilized peripheral blood stem cells. Mobilization was achieved by chemotherapy and granulocyte–colony-stimulating factor at a dose of 16 μg/kg per day by subcutaneous injection. Clinical features of the patients and previous treatment history for the underlying disease are summarized in Table 1. Informed consent to investigate bone complications after SCT was obtained from all patients and the study was designed in accordance with the Helsinki II declaration on human experimentation.

Table 1. Characteristics of Patients Who Underwent Allogeneic or Autologous SCT and of Healthy Controlsa
CharacteristicsAllo-SCT (n = 100)Auto-SCT (n = 107)Controls (n = 200)
  • SCT: stem cell transplantation; BMI: body mass index; AML: acute myeloid leukemia; ALL: acute lymphocytic leukemia; CML: chronic myeloid leukemia; HD: Hodgkin disease; NHL: non-Hodgkin lymphoma; SAA: severe aplastic anemia; CLL: chronic lymphocytic leukemia; CsA: cyclosporin A; IDA: idarubicin; MITO: mitoxantrone; ARAC: cytarabine; ETO: etoposide; VCR: vincristine; DAUNO: daunorubicin; CYCLO: cyclophosphamide; MTX: methotrexate; ADRI: doxorubicin; ASP: asparaginase; HU: hydroxyurea; IFN: interferon-alpha; EPIR: epirubicine; BLEO: bleomycin, PDN: prednisone; CHL; chlorambucil; FLUDA: fludarabine.

  • a

    Values are expressed as mean ± standard deviation. Corticosteroid cumulative dose is expressed as prednisone equivalents. In autologous transplant patients the cumulative doses include steroids administered before transplant for treatment of the underlying disease.

  • b

    P < 0.005 vs. auto-SCT group.

Age at evaluation (yrs)31 ± 10.539.6 ± 11.834.5 ± 10.4
Time from transplant (mos)38.9 ± 12.237.8 ± 12.4
BMI (kg/m2)23.4 ± 1.725.9 ± 2.124.2 ± 1.9
Previous treatment history   
Corticosteroid dose (g/m2)7.2 ± 2.77.16 ± 1.4
Treatment duration (days)260.5 ± 32.3b167 ± 34.8
CsA dose (g/m2)67 ± 3.42
Treatment duration (days)308 ± 39

Conditioning Regimens, aGVHD Prophylaxis, and Treatment of aGVHD and cGVHD

All patients who underwent allogeneic or autologous transplantation for AML, CML, or ALL received a conditioning regimen of busulphan (16 mg/kg in 4 days) and cyclophosphamide (120 mg/kg in 2 days). Patients with HD, NHL, and CLL were conditioned with carmustine (300 mg/m2 on day 1), etoposide (200 mg/m2 in 4 days), cytarabine (400 mg/m2 in 4 days), and melphalan (140 mg/m2 on day 6). After allografting, aGVHD prophylaxis included cyclosporine A (CsA; 1 mg/kg intravenously from day −1 to + 21, followed by 10 mg/kg for 6 months orally) and short-course methotrexate (10 mg/kg for four doses). Diagnosis and grading of GVHD were defined according to clinical criteria.8 The clinical diagnosis was confirmed, if indicated, by histopathology of the skin, liver, or mucous membranes before starting immunosuppressive therapy. Acute GVHD was treated by pulse high-dose methylprednisolone therapy (2–10 mg/kg for 10 days), followed by slow-dose tapering in the following 3 months as the patients' clinical condition was monitored. Chronic GVHD was treated by prednisolone at doses of 1–1.5 mg/kg for at least 1 month in combination with CsA at doses ranging from 4 to 8 mg/kg per day, followed by 25% tapering every month during the following 6 months. Each organ involved by cGVHD was evaluated for response to the immunosuppressive therapy using the following criteria: major or minor responses were defined as complete resolution or improvement of cGVHD signs and symptoms, respectively, while on immunosuppressive therapy; no response and progression as no improvement or worsening in any disease signs or symptoms, respectively, despite immunosuppressive therapy; and exacerbation as reactivation of cGVHD symptoms after initial improvement while on immunosuppressive therapy. Fifty-six patients developed aGVHD of global grade 1–3 and 58 developed cGVHD (25 limited and 33 extensive form). All these patients were treated as stated above for a period ranging from 6 to 24 months (Table 1).

Diagnosis of AVN

Avascular necrosis was suspected in posttransplant patients after a clinical evaluation. To confirm the diagnosis, a magnetic resonance imaging (MRI) scan was performed, which involved anteroposterior and lateral scans of T1 and T2-weighted images of the involved site (Fig. 1). A staging system developed by the ARCO (Association of Research Circulation Osseous) was used.1099Tc-labeled methylene bisphoshonate triphasic bone/joint scans were obtained for six patients in Stage I–II.

thumbnail image

Figure 1. Magnetic resonance imaging (MRI) scan of bilateral femoral head osteonecrosis in a 39-year-old man who developed pain and functional limitation 2 months before the MRI scan was performed. There are diffuse focal hypointense lesions in epiphyseal and metaphyseal regions on both sides in the coronal T1-weighted image (A). (B) High intensity signaling in the T2-weighted image.

Download figure to PowerPoint

Densitometric Evaluation

Bone mineral density was evaluated in 100 patients (50 allo-SCT and 50 auto-SCT; 46 females and 54 males; age, 34 ± 10.8 years; Table 2). This group of patients was representative of the entire population of long-term survivors and included all patients with AVN. Densitometric values were compared with those of 200 healthy subjects matched for age, gender, and body mass index (BMI). The densities of the lumbar spine (L1–L4) and femoral neck were measured by dual-energy X-ray absorptiometry (DEXA), using a QDR 1000 densitometer (Hologic, Waltham, MA). Individual BMD values were expressed as grams per square centimeter, T-scores, and Z-scores.

Table 2. Bone Mineral Density Measured in a Subgroup of Patients after Allogeneic or Autologous SCTa
VariableAllo-SCT (n = 50)Auto-SCT (n = 50)P value
  • SCT: stem cell transplant; NS: not significant; BMI: body mass index; SD: standard deviation.

  • a

    Results are expressed as mean ± standard error of the mean.

Age (yrs)29.3 ± 1.638.9 ± 1.7NS
BMI (kg/m2)23.5 ± 0.525.9 ± 0.6NS
Femoral neck Z-score (SD)− 1.06 ± 0.14− 0.57 ± 0.170.041
Lumbar spine Z-score (SD)− 0.92 ± 0.17− 0.33 ± 0.20.017

Fibroblast Colony-Forming Unit (CFU-F) Assay

In vitro CFU-F growth assay was performed as previously described9 in 30 allotransplanted patients and 20 bone marrow donors. Briefly, isolated bone marrow mononuclear cells (BMMNC) were obtained through density gradient centrifugation using lymphocyte separation medium (Life Technologies, Gaithersburg, MD). The BMMNC were resuspended at a concentration of 2 ×106/mL in McCoy's 5A modified medium containing 10% fetal bovine serum and L-glutamine (Mesencult, StemCell Technologies, Vancouver, Canada) supplemented with 1 × 10−8 mol/L dexamethasone (Sigma, St. Louis, MO), which allows the recruitment of bone marrow stromal cells to the osteoblastic lineage, and plated in 25-cm2 tissue culture flasks. After incubation at 37 °C, 5% CO2 for 15 days in a humidified atmosphere, characteristic fibroblastoid cell aggregates of more than 50 cells were scored in situ as CFU-F under an inverted microscope. All cultures were performed in duplicate.

Statistical Analysis

Data are expressed as mean ± standard deviation (SD) or standard error of the mean. Risk factors for AVN including gender, age, underlying diagnosis, corticosteroid and CsA exposure, type of transplant, BMD, and GVHD occurrence were investigated. Analysis of risk factors was performed using the Pearson correlation coefficient for data expressed by parametric values and the Student t test for nonparametric variables. After stratifying the patients with and without AVN, the chi-square test was used to assess the association with specific clinical features. Statistical significance was considered for any P value less than 0.05.


  1. Top of page
  2. Abstract
  6. Acknowledgements

Twelve patients (seven males, five females) developed AVN after SCT: 10 in the allo-SCT group (6 males, 4 females) and 2 in the auto-SCT group (1 male, 1 female; 83% vs. 17%; P = 0.037). The first symptoms of AVN occurred 3–114 months after SCT (median, 26 months) and manifested as joint pain (at rest and/or night pain) in three patients and as joint pain plus functional limitation in nine patients. A total of 25 joints were affected (mean, two joints per person; range, 1–4 joints) and included the hip, ankle, shoulder, and knee. All patients had lower limb involvement and AVN was confined to the lower limbs in 10 patients.

Diagnosis by MRI Scan and Scintigraphy

An MRI scan was performed 1–10 months after onset of the first symptom. In three patients with unilateral inguinal pain, an MRI scan showed bilateral femoral head disease (contralateral silent involvement). An area with a low-intensity signal in the femoral head and/or the “double line sign” was observed in nine patients (Stage I–II) and collapse of the femoral head (Stage III–IV) occurred in three patients. Triphasic scintigraphy, which was performed in six patients, showed an abnormal uptake (increase uptake surrounding a cold area)11 in five of these patients.

AVN in Auto-SCT Patients

Two patients developed AVN 3 and 12 months after auto-SCT. In both patients, the femoral head was affected bilaterally. In addition, both patients had hypogonadism: one man had testicular dysgenesis and one woman had posttransplant ovarian failure. The cumulative corticosteroid doses were 7 and 8 g in the two patients with AVN and 6.76 ± 0.11 g in autotransplanted patients without AVN. The duration of steroid treatment was 240 and 175 days in the two patients with AVN and 167 ± 24.9 days in patients without AVN. Lumbar and femoral Z-scores were − 0.03 and − 2.39 SD and − 1.02 and − 0.76 SD, respectively, in patients with AVN. Patients without AVN had lumbar and femoral Z-scores of − 0.23 ± 0.04 SD and − 0.48 ± 0.038 SD, respectively.

AVN in Allo-SCT Patients

Ten patients developed AVN 10–114 (median 32) months after allo-SCT. Sites included both femoral heads in four patients, one femoral head in three patients, the bilateral femoral head and unilateral/bilateral humeral head in two patients, and the bilateral femoral head and right knee in one patient. All but one patient affected by AVN had cGVHD. Exacerbation or progression of cGVHD was documented in eight patients 1–4 months before AVN was diagnosed and all these patients had multiorgan involvement. The clinical characteristics of the allo-SCT patients with AVN and GVHD are summarized in Table 3. Liver function tests showed increased liver enzymes in all patients (alanine aminotransferase, 112–1200 U/L; aspartate aminotransferase, 67–912 U/L; gamma-glutamyl transpeptidase, 90–350 U/L) and cholestasis in two patients (alkaline phosphatase, 487–785 U/L; bilirubin, 5–9 mg/dL). Skin lesions included hypercheratosis, lichen planus, and focal epidermal atrophy. Ocular dryness (Sjögren-like syndrome) was the major ophthalmologic manifestation, whereas oral dryness and lichenoid lesions were the most frequent upper gastrointestinal tract symptoms.

Table 3. GVHD in Allotransplanted Patients with AVN
PatientGenderAge at SCT (yrs)Interval SCT-AVN (mos)aGVHD gradecGVHD formcGVHDcGVHD state at AVN diagnosis
Principal siteOther sites
  1. AVN: avascular necrosis; SCT: stem cell transplant; aGVHD: acute graft versus host disease; cGVHD: chronic graft versus host disease; M: male; F: female.

1M37540ExtensiveSkinEyes, oralStable
2M45530ExtensiveSkinLiver, oralExacerbation
4F19111–2ExtensiveLiverSkin, eyesProgression
5M34244ExtensiveGutSkin, liverProgression
6F211140ExtensiveLiverSkin, eyesExacerbation
8M19482ExtensiveLiverSkin, oralExacerbation
9M36280ExtensiveSkinLiver, eyesProgression

Patients with AVN received immunosuppressive treatments for longer periods compared with allotransplanted patients without AVN (CsA, 403 ± 89.3 vs. 272 ± 35 days; P = 0.079; corticosteroid, 448 ± 52 vs. 207 ± 40 days; P < 0.001). The cumulative steroid dose received before AVN was also significantly higher compared with patients without this complication (15.3 ± 2 vs. 5.9 ± 1.15 g/m2; P < 0.005). Two of four women with AVN did not receive hormone replacement therapy and they developed hypogonadism after SCT. Four and eight patients with AVN had a T-score less than − 1 SD at the lumbar spine and femoral neck, respectively. However, BMD values did not differ significantly between allo-SCT patients with or without AVN (lumbar, − 0.7 ± 0.53 vs. − 0.04 ± 0.3; P = 0.065; femoral neck, − 1.02 ± 0.5 vs. − 1.07 ± 0.3; P = 0.84).

Bone Marrow CFU-F in Allo-SCT Patients

The bone marrow compartment of stromal stem cells, measured by CFU-F assay, was decreased in transplant patients compared with normal donors (22.3 ± 3 per 105 mononuclear cells plated vs. 55 ± 4; P < 0.0001). We found the bone marrow CFU-F compartment markedly depleted particularly in allo-SCT patients affected by cGVHD (CFU-F: 43.2 ± 9.3 vs. 13.7 ± 12.8 in allo-SCT patients without and with cGVHD, respectively; P = 0.01). In addition, almost all transplanted patients with AVN showed less CFU-F compared with transplanted patients without AVN (CFU-F: 24.5 ± 3.5 vs. 12.4 ± 4.3 in allo-SCT patients without and with AVN, respectively; P = 0.042; Fig. 2).

thumbnail image

Figure 2. Clonogenic stromal precursors in allotransplanted patients with or without avascular necrosis and in normal donors. Bone marrow mononuclear cells were used for the fibroblast colony-forming unit assay. Each circle represents one patient who was studied.

Download figure to PowerPoint

Risk Factor Assessment

Groups with or without AVN were similar in terms of age, follow-up period, and BMI. Steroid cumulative dose did not differ in between allo-SCT and auto-SCT patients, whereas the duration of treatment was significantly longer among the allo-SCT patients (P = 0.001; Table 1).

The chi-square test revealed that the type of transplant and the dose and duration of steroid treatment were risk factors for AVN (Table 4). AVN was significantly more frequent after allo-SCT compared with auto-SCT (10% vs. 1.9%, χ2: 6.23; P = 0.044) and was related to steroid treatment dose and duration (P < 0.0001 and P = 0.0011, respectively). Chronic GVHD, particularly its extensive form, was related strongly to the presence of AVN (P = 0.018 and P < 0.0001, respectively). No significant influence was found regarding gender and CsA treatment. Exacerbation or progression of cGVHD was documented in 8 of 10 allo-SCT patients who developed AVN, whereas the other 2 patients had persistent mild liver cGVHD. The Z-scores for both lumbar and femoral BMD were significantly lower in allo-SCT than in auto-SCT patients (Table 2), suggesting greater bone loss in allo-SCT patients setting. Although 9 of 12 patients who developed AVN showed a Z-score value at the femoral neck less than − 1 SD, the BMD difference between the groups with and without AVN did not reach significance at the femoral site and was marginal at the lumbar spine (Table 4).

Table 4. Risk Factors for AVN as Assessed by the Chi-Square Test
VariableWith AVN (n = 12)Without AVN (n = 195)Chi-square testP value
  • AVN: avascular necrosis; CsA: cyclosporin A; GVHD: graft-versus host disease; BMD: bone mineral density.

  • a

    Data pertaining only to allotransplanted patients.

  • b

    Data measured in 50 allotransplanted and 50 autotransplanted patients.

Allogeneic transplant1090  
Autologous transplant21056.4660.039
 < 199 days347  
 ≥ 199 days7431.90.385
 ≤ 60 g347  
 > 60 g7431.9070.385
Steroid duration    
 ≤ 216 days2101  
 > 216 days109468.8< 0.0001
Steroid cumulative dose    
 < 6 g1103  
 ≥ 6 g11928.950.0011
Acute GVHDa650  
No acute GVHDa4400.2740.872
Any chronic GVHDa1049  
No chronic GVHDa0418.050.018
Extensive chronic GVHD924  
No/limited chronic GVHD16616.014< 0.0001
Lumbar Z-score    
 < −0.5 DSb644  
 > −0.5 DS6440.01.0
Femoral Z-score    
 < −1.075b941  
 > −1.0753473.410.182
Lumbar BMD (g/cm2)0.97 ± 0.150.991 ± 0.16t test0.667
Femoral neck BMD (g/cm2)0.78 ± 0.15b0.8 ± 0.12t test0.598


According to the ARCO classification,10 AVN was diagnosed at Stage I in five patients, at Stage II in four patients, at Stage III in two patients, and at Stage IV in one patient. Seven patients in Stage I-II were treated conservatively with bed rest, analgesics, and pulsing electromagnetic fields, but three patients experienced disease progression. Six patients with femoral head involvement required surgical treatment. Five hip replacements were performed in three patients (two bilateral) and core decompression was used in five joints in three subjects (one patient had unilateral replacement and contralateral decompression). All patients with AVN requiring hip replacement suffered from severe and prolonged cGVHD (duration range, 5–10 years). Treatment of other localization included rest, analgesics, and pulsing electromagnetic fields, with significant improvement of clinical symptoms.


  1. Top of page
  2. Abstract
  6. Acknowledgements

Symptomatic AVN occurred in 12 of 207 long-term survivors after SCT (5.8%) within 3–114 months (median, 26 months) after grafting, all but two of them having received an allo-SCT. An equal number of men and women were affected. A significant statistical association was found for the following AVN risk factors: an allogeneic transplant, presence and grade of cGVHD, and steroid treatment length and cumulative dose. These findings are consistent with previous reports on AVN in other series of patients after SCT, the frequency of AVN ranging from 3.7% to 24%.3, 12–18 All these studies point to multiple joint involvement, with prevalent localization at the femoral head. The high male-to-female ratio (8:1) described in “a total of AVN population”19, 20 appears to be lost in transplanted patients, regardless of the type of transplant.21–24 Surgical treatment was required in one-half of the patients. We found a relationship between AVN and decreased osteoblastic precursor regenerating capacity within the stromal stem cell compartment.

The inability to regenerate a normal osteogenic cell compartment may account, in part, for AVN occurrence after allo-SCT. A larger sample of patients and a comparison with CFU-F growth in the autotransplant setting is needed to confirm this issue. Furthermore, our study strongly suggests a close relationship between AVN and cGVHD. In fact, a recent exacerbation or progression of cGVHD was documented in 8 of 10 allo-SCT patients before AVN diagnosis. The high incidence of acute and/or cGVHD reported in various studies of patients with AVN suggests that steroid treatment of cGVHD was the most important factor responsible for AVN onset. However, it is difficult to separate the direct effects of GVHD from those of its treatments.

The relationship between AVN and potential risk factors emerging from previous studies is summarized in Table 5. Mori et al.25 reported that cGVHD exacerbation may be a possible cause of AVN, whereas Enright et al.12 hypothesized that an alloreactive immune response is a possible cause for osteonecrosis. In a cohort of 902 patients after allo-SCT or auto-SCT who were followed up between 1974 and 1988, Enright et al. reported an AVN incidence of 10.4% among allotransplant patients and 0% among autotransplant patients. In addition, they reported that 89% of the allotransplant patients with AVN had developed GVHD and 95% had been treated by steroids for GVHD prophylaxis or treatment. Older age and GVHD requiring steroid therapy were documented risk factors for AVN among allo-SCT recipients in the large multicentric study by Sociè et al.14 In addition, they showed that corticosteroids were also an independent risk factor by a multivariate analysis. The 1997 Sociè et al. study14 confirmed the 1994 Sociè et al.13 results in a cohort of 727 allo-SCT recipients followed-up in a single center. Both studies13, 14 showed that aGVHD was associated with the increased incidence of AVN. Similarly, in a study by Fink et al.,15 AVN occurred more frequently in allotransplanted patients than in autotransplanted patients (93% vs. 7%). Some studies identified TBI as a possible etiologic factor for AVN.12, 14, 15 However, a similar AVN frequency was observed in our cohort of patients, none of whom had been treated by TBI. The correlation between TBI and AVN found by simple regression analysis means that other factors must account for this association. Stern et al.26 reported an AVN frequency of 9.6% in a cohort of 104 allo-SCT recipients evaluated within 3 years. All patients in that study were treated with corticosteroids and CsA or tacrolimus.

Table 5. Risk Factors Reported by Major Studies on AVN
AuthorsNo. of patientsFollow-up period (yrs)SCT typeAVN prevalence (%)Assessment of corticosteroid roleAssociations found with AVN
  1. AVN: avascular necrosis; SCT: stem cell transplant; aGVHD, cGVHD: acute/chronic graft vs host disease; nr: data not reported; TBI: total body irradiation; BMD: bone mineral density; MPD: methylprednisolone.

Atkinson et al. (1987)3502Allotransplant5/50 (10)nraGVHD, cGVHD
Enright et al. (1990)1290214Allotransplant28/642 (4.4)Multivariate analysisAge, disease type, allo-SCT, TBI, cGVHD, alloreactivity
Wagener et al. (1991)30433Allotransplant8/33 (24)t testcGVHD, allo-SCT
Sociè et al. (1994)137271.5Allotransplant27/727 (3.7)Multivariate analysisAge, male gender, aGVHD
Sociè et al. (1997)14438819Allotransplant77/4388 (17.5)Multivariate analysisAge, initial diagnosis, TBI, aGVHD, cGVHD
Fink et al. (1998)15193917Allotransplant/autotransplant87/1939 (5)Logistic regressionAllo-SCT (93%) vs. auto-SCT (7%), TBI, aGVHD, cGVHD
Wiesmann et al. (1998)17272nrAllotransplant17/272 (6.3)nrAge, GVHD
Ebeling et al. (1999)18832.5Allotransplant4/52 (7.7)t testAllo-SCT, male gender, lower BMD at femoral neck
Stern et al. (2001)261043Allotransplant10/104 (9.6)nrCorticosteroid treatment
Torii et al. (2001)31100nrAllotransplant19/100 (19)t testYounger age, cumulative steroid dose, pulse i.v. MPD therapy, cGVHD

The exact mechanisms and risk factors for the majority of nontraumatic forms of AVN are still unknown. However, dysregulation of lipid metabolism, drug-induced injury to the vessel wall, and vasculitis have been identified as possible contributors.4–6, 13, 20 Lipid metabolism is altered by steroids and gonadal insufficiency and remains abnormal long after SCT. Vasculitis after SCT may have an autoimmune origin and may be related to cGVHD. Any organ or tissue can be affected by this complication, even in the absence of increased autoantibody production. Indeed, AVN develops frequently among patients with a dysregulated immune system, particularly during the posttransplant period,21–23 as well as among patients with systemic lupus erythematosus.27 However, it occurs rarely in other noninflammatory conditions in which long-term, high-dose corticosteroids have been used, such as in patients with neurologic degenerative disease or chronic pulmonary diseases. Steroid-induced AVN is difficult to replicate in experimental animal models, unless incompatible skin grafts are applied simultaneously,28 perhaps simulating the allograft reactivity of the posttransplant state.

Patients with cGVHD receive high steroid doses for a long periods of time. Therefore, on the basis of clinical data only, it is impossible to distinguish the separate effects of steroids and GVHD.

Cyclosporine A increased AVN incidence after kidney transplantation,29 whereas no significant relationship was found in patients after SCT. In the current study, significantly longer CsA treatment was found in allo-SCT patients who developed AVN.

Conversely, no relationship was found with gender, age, and gonadal function. Although significantly lower BMD values were found in allotransplanted than in autotransplanted patients at both skeletal sites evaluated, a direct relationship was not documented between BMD values and AVN. Similar results were reported by Stern et al.,26 who found osteopenia in some patients with AVN. In the current study, four patients (two auto-SCT and two allo-SCT) with AVN had been hypogonadic for a long period. There were multiple negative effects on bone status, but a relationship was not found between AVN and gonadal status, which supports previous evidence that AVN is not a complication of postmenopausal osteoporosis.19

Of the 12 patients who developed AVN, 9 were diagnosed in Stage I–II, shortly after symptom appearance. These patients received early treatment intervention, whereas the remaining three patients had severe and prolonged cGVHD and required hip replacement. Early diagnosis by MRI (shortly after the onset of symptoms) may help to detect AVN at an early stage, and early intervention and monitoring may prevent the need for surgery.

The results of the current study confirm those of other studies and focus on several aspects that were not investigated thoroughly before. The pathogenesis of this condition is likely multifactorial with various precipitating conditions. Corticosteroid treatment may be a possible contributing factor to the development of AVN. However, the strong relationship among AVN, allo-SCT, and cGVHD recurrence documented in this study suggests that immune-mediated mechanisms may also operate in AVN development after SCT.


  1. Top of page
  2. Abstract
  6. Acknowledgements

The authors thank the nursing staff of the Division of Hematology (Federico II University, Naples, Italy) who provided care for all transplanted patients


  1. Top of page
  2. Abstract
  6. Acknowledgements
  • 1
    Bewick M, Stewart PH, Rudge C, Farrand C, McColl I. Avascular necrosis of bone in patients undergoing renal allotransplantation. Clin Nephrol. 1976; 5: 6672.
  • 2
    Hawking KM. Avascular necrosis of bone after renal transplant [letter]. N Engl J Med. 1976; 294: 397.
  • 3
    Atkinson K, Cohen M, Biggs J. Avascular necrosis of the femoral head secondary to corticosteroid therapy for graft-versus-host disease after marrow transplantation: effective therapy with hip arthroplasty. Bone Marrow Transplant. 1987; 2: 421426.
  • 4
    Mont MA, Hungerford DS. Non-traumatic avascular necrosis of the femoral head. J Bone Joint Surg [Am]. 1995; 77: 459474.
  • 5
    Chang CC, Greenspan A, Gershwin ME. Osteonecrosis: current perspectives on pathogenesis and treatment. Semin Arthritis Rheum. 1993; 23: 4769.
  • 6
    Mankin HJ. Nontraumatic necrosis of bone (osteonecrosis) N Engl J Med. 1992; 326: 14731479.
  • 7
    Vreden SG, Hermus AR, van Liessum PA, Pieters GF, Smals AG, Kloppenborg PW. Aseptic bone necrosis in patients on glucocorticoid replacement therapy. Neth J Med. 1991; 39: 153157.
  • 8
    Sullivan KM. Graft versus host disease. In: FormanSJ, BlumeKG, ThomasED. Bone marrow transplantation. Boston: Blackwell Scientific, 1994: 339362.
  • 9
    Tauchmanovà L, Serio B, del Puente A,et al. Long-lasting bone damage detected by dual energy x-ray absorptiometry, phalangeal osteosonogrammetry and in vitro growth of marrow stromal cells after allogeneic stem cell transplantation. J Clin Endocrinol Metab. 2002; 87: 50585065.
  • 10
    Stulberg BN. Editorial comment on fifth international symposium on bone circulation. Clin Orthop. 1997; 334: 23.
  • 11
    Dumont M, Dunais S, Taillefer R. “Doughnut” sign in avascular necrosis of the bone. Clin Nucl Med. 1983; 9: 4447.
  • 12
    Enright H, Haake R, Weisdor D. Avascular necrosis of bone: a common serious complication of allogeneic bone marrow transplantation. Am J Med. 1990; 89: 733738.
  • 13
    Sociè G, Selimi G, Sedel L, et al. Avascular necrosis of bone after allogeneic bone marrow transplantation: clinical findings, incidence and risk factors. Br J Haematol. 1994; 86: 624628.
  • 14
    Sociè G, Cahn JY, Carmelo J, et al. Avascular necrosis of bone after allogeneic bone marrow transplantation: analysis of risk factors of 4388 patients by the Société Francaise de Greffe de Moëlle (SFGM). Br J Haematol. 1997; 97: 865870.
  • 15
    Fink JC, Leisenring WM, Sullivan KM, Sherrard DJ, Weiss NS. Avascular necrosis following bone marrow transplantation: a case-control study. Bone. 1998; 22: 6771.
  • 16
    Mascarin M, Giavitto M, Zanazzo GA, et al. Avascular necrosis of bone in children undergoing allogeneic bone marrow transplantation. Cancer. 1991; 68: 655659.
  • 17
    Wiesmann A, Pereira P, Bohm P, Faul C, Kanz L, Einsele H. Avascular necrosis of bone following allogeneic stem cell transplantation: MR screening and therapeutic options. Bone Marrow Transplant. 1998; 22: 565569.
  • 18
    Ebeling PR, Thomas DM, Erbans B, Hopper JL, Szer J, Grigg A. Mechanisms of bone loss following allogeneic and autologous hemopoietic stem cell transplantation. J Bone Miner Res. 1999; 14: 342350.
  • 19
    Mazières B. Osteonecrosis. In: KlippelJH, DieppePA. Rheumatology. London: Mosby, 1994: 47.147.10.
  • 20
    Calandriello B, Grassi G. Idiopathic osteonecrosis of the femoral head. Epidemiological and aetiological factors. Ital J Orthop Trauma. 1982; 8: 918.
  • 21
    Shane R, Rivas MDC, Silverberg SJ, et al. Osteoporosis after cardiac transplant. Am J Med. 1993; 94: 257264.
  • 22
    Lee AH, Mull RL, Keenan GF, et al. Osteoporosis and bone morbidity in cardiac transplant recipients. Am J Med. 1994; 86: 3541.
  • 23
    Braun WE, Richmond BJ, Protiva DA, Gifford RW, Straffon RA. The incidence and management of osteoporosis, gout, and avascular necrosis in recipients of renal allografts functioning more than 20 years (level 5A) treated with prednisone and azathioprine. Transpl Proc. 1999; 31: 13661369.
  • 24
    Grotz W, Breitffennnldt K, Cybulla M. Immunosuppression and skeletal disorders. Transplant Proc. 2001; 33: 992993.
  • 25
    Mori A, Hashino S, Kobayashi S, et al. Avascular necrosis in the femoral head secondary to bone marrow infarction in patient with graft-versus-host disease after unrelated bone marrow transplantation. Ann Hematol. 2001; 80: 238242.
  • 26
    Stern JM, Sullivan KM, Ott SM, et al. Bone density loss after allogeneic hematopoietic stem cell transplantation: a prospective study. Biol Blood Marrow Transplant. 2001; 7: 257264.
  • 27
    Zizic TM, Marcoux C, Hungerford DS, Dansereau JV, Stevens MB. Corticosteroid therapy associated with ischemic necrosis of bone in systemic lupus erythematosus. Am J Med. 1985; 79: 596604.
  • 28
    Gosling DC, Sampson WFD, MadLeod M, Hanby AM, Slapek M. Susceptibiliy of the rabbit and the rat to steroid osteonecrosis: an experimental study. Transplantation. 1987; 43: 751753.
  • 29
    Sato K, Satomi S, Oguma S, et al. Relationship between aseptic necrosis of femoral head bone and immunosuppression therapy, especially CsA administration. Nippon Geka Gakkai Zasshi. 1993; 94: 832839.
  • 30
    Wagener P, Schulte D, Link H, Kirchner H, Stoll M, Polwoda H. Musculoskeletal manifestations in patients after bone marrow transplantation. Initial clinical rheumatologic observation. Z Rheumatol. 1991; 50: 199203.
  • 31
    Torii Y, Hasegawa Y, Kubo T, et al. Osteonecrosis of the femoral head after allogeneic bone marrow transplantation. Clin Orthop. 2001; 382: 124132.